{"_id":"doc-en-kubernetes-88466bc637a805045217da94feabbdbd3df0ba52cd0d9904e95d3984c4cf4efa","title":"","text":"( noticed this) HardPodAffinitySymmetricWeight should be in the , not in KubeSchedulerConfiguration If/when fixing this, we need to be careful to do it in a backward-compatilble way.\nDo you mean ? It seems scheduler policy config's args can not passed to FitPredicates; it need to update related interface. For the : mark to be deprecated and set scheduler policy config to HardPodAffinitySymmetricWeight if policy config is empty.\nAre you working on this? if no, I can take it.\n, nop; please go ahead.\n/ / / , When reviewing the PR, there's one question on : the default value of is 1, which means the weight (1) will always add to priorities if term matched (), does it our expectation? And the user has to set it to 0 to disable it. IMO, the default value of should be 0.\nThe value 1 (instead of 0) was chosen so that if you say A must run with B, there is an automatic preference that B should run with A. This seems like reasonable behavior.\nThanks very much; reasonable case to me. But it seems we can not disable it: if set to 0, the default funcs will set it to default (1); similar result to un-set. Anyway, no user complain this now :)."}
{"_id":"doc-en-kubernetes-599cbd9914ac52a34ee08ac617d1f4bd774f52f2d5bdcb07fa1a0e76f7194545","title":"","text":"On a large cluster, the routecontroller basically makes O(n) requests all at once, and most of them error out with \"Rate Limit Exceeded\". We then wait for operations to complete on the ones that did stick, and damn the rest, they'll be caught next reconcile (which is a while). In fact, in some cases it appears we're spamming hard enough to get rate limited by the upper-level API DOS protections and seeing something akin to , e.g.: We can do better than this!\nc.f. , which is the other leg of this problem\nI took a look into code and it seems that we are doing full reconciliation every 10s. If we need to create 2000 routes at the beginning it seems to frequently. I think that we need to: first compute how many routes we have to create based on that try to spread them over some time to reduce possibility of causing 'rate limit exceeded' I will try to put some small PR together as a starting point.\nI'm reopening this one, since this doesn't seem to be solved. So basically something strange is happening here. With PR we have 10/s limit on API calls send to GCE. And we are still rate-limitted. However, according to documentation we are allowed for 20/s: So something either doesn't work or is weird.\nWhat is super interesting from the logs of recent run: It starts to creating routes: After 2 minutes from that, we are hitting \"Rate limit Exceeded\" for the first time: And only after 9 minutes we create the first route ever. The second one is created after 15 minutes, and from that point we create a number of them So it seems that nothing useful is happening for the first 2 minutes, and then between 3m30 and 9m30\nOK - so I think that what is happening is that all controllers are sharing GCE interface (and thuse they have common throttling). So if one controller is generating a lot of API calls, then other controllers may be throttled. has a hypothesis that it may be caused by nodecontroller.\nOne thing my PR didn't address (because it was late): Now we have some obnoxious differences in logging when you go to make a request, because the versus could actually be absorbing a fair amount of ratelimiter."}
{"_id":"doc-en-kubernetes-bf24c8733a67a958496376416be2eb636c6e4bfb94999815173220461c93fd0c","title":"","text":"It could before because of the operation rate limiter and operation polling anyways, but now could be worse.\n- yes I'm aware of that. And probably this is exactly the case here.\nI got fooled by that - I guess we should fix it somehow.\nOne approach to fixing it is a similar approach to the operation polling code, but in the : warn if we ratelimited for more than N seconds (and dump the request or something? not clear how to clarify what we were sending).\nI found something else - Kubelet is also building GCE interface: We have 2000 kubelets, so if all of them send multiple requests to GCE, we may hit limits.\nAny idea in what circumstances Kubelet will contact cloud provider?\nHmm - it seems it contacts GCE exactly once at the beginning. So it doesn't explain much.\nI think it's only here: I've rarely seen GCE ratelimit readonly calls, in practice, and kubelets come up pretty well staggered.\nOK - so with logs that I , it is pretty clear that we are very heavily throttled on GCE api calls. In particular, when I was running 1000-node cluster, I see lines where we were throttle for 1m30. I'm pretty sure it's significantly higher in 2000-node cluster. I'm going to investigate a bit more where all those reuqests come from.\nActually - it seems that we have a pretty big problem here. Basically, Kubelet seems to send a lot of requests to GCE. So to clarify: every time Kubelet is updating NodeStatus (so every 10s by default) it is sending an API call to GCE to get it's not addresses: that means, that if we have 2000 nodes, kubelets generate 200 QPS to GCE (which is significantly more than we can afford) We should think how to solve this issue, but I think the options are: increasing QPS quota or calling GCE only once per X statusUpdates.\nThe second issue is that we are throttled at the controller level too, and this one I'm tryin to understand right now."}
{"_id":"doc-en-kubernetes-b1dad1b374e4c58d925d9fef62d9846d5e0e5eda04aecdd78a124036a36cf31c","title":"","text":"- that said, any rate-limitting on our side, will not solve the kubelet-related problem\nHmm - actually looking into implementation it seems that NodeAddresses are only contacting metadata server, so I'm not longer that convinced...\nAlso, I think that important thing is that a single call to GCE API can translate to multiple api calls to GCE. In particular - CreateRoute translates to: get instance insert route a bunch of getOperation (until this is finished)\nOK - so I think that what is happening here is that since CreateRoute translates to: get instance insert route a bunch of getOperations That means that if we call say 2000 CreateRoute() at the same time, all of them will try to issue: getInstance at the beginning. So we will quickly accumulate 2000 GET requests in the queue. Once they are processed, we generate the \"POST route\" then for any processed. So It's kind of expected what is happening here. So getting back to the Zach`s PR - I think that throttling POST and GET and OPERATION calls separate is kind of good idea in general.\nBut I still don't really understand why do we get \"RateLimitExceeded\". Is that because of Kubelets?\nI think that one problem we have at the RouteController level is that if CreateRoute fails, we simply don't react on it. And we will wait at least 5 minutes before even retrying it. We should add some retries at the route controller level too. I will send out some short PR for it.\nBut I still don't understand why do we get those \"RateLimitExceeded\" error\nOK - so as an update. In GCE, we have a separate rate-limit for number of in-flight CreateRoute call per project. is supposed to address this issue.\nOK - so this one is actually fixed. I mean that it is still long, but currently it's purely blocked on GCE.\nAs a comment - we have an internal bug for it already."}
{"_id":"doc-en-kubernetes-631bc5da562ebc33bd49790f3078648639a2126878c5fc7773cd8143bd3bdf27","title":"","text":"TestRoundTripSocks5AndNewConnection Disabled by No response No response No response /sig\nThere are no sig labels on this issue. Please add an appropriate label by using one of the following commands: - Please see the for a listing of the SIGs, working groups, and committees available. <detailsInstructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\nThis issue is currently awaiting triage. If a SIG or subproject determines this is a relevant issue, they will accept it by applying the label and provide further guidance. The label can be by org members by writing in a comment. <detailsInstructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\nthere is a race shutting down the socks5 server"}
{"_id":"doc-en-kubernetes-523d59ec5104a53f44aa3efaa79fa37e525dbabb537793d9e95c31ff14a3e1c1","title":"","text":"The prestop and poststart exec tests volume mount /tmp from the host and attempt to create a file there from inside the pod. This is blocked by selinux. Failure:\nwe need to have a mechanism to say that a test should not be run if selinux is enabled (if legitimately required) or fix each test (where appropriate)"}
{"_id":"doc-en-kubernetes-c5b2d72b2e785256d7a2577807fdbd43ef000a4da594f69c949ac55e4186d114","title":"","text":"[x] two weeks soak end date : 18 March 2021 According to this APIsnoop query, there are still some remaining RESOURCENAME endpoints which are untested. with this query you can filter untested endpoints by their category and eligiblity for conformance. e.g below shows a query to find all conformance eligible untested,stable,core endpoints Note: Community feedback for the e2e test has lead to a number of improvements including extending the test coverage to all outstanding endpoints listed above. As each endpoint is tested a &rsquo;watch&rsquo; confirms that the result is valid before testing the next endpoint. readCoreV1NamespacedServiceStatus (get /status) patchCoreV1NamespacedServiceStatus (patch /status) replaceCoreV1NamespacedServiceStatus (update /status) patchCoreV1NamespacedService Create a Service with a static label Patch the Service with a new label and updated data Get the Service to ensure it&rsquo;s patched Upate the Service with a new label and updated data Get the Service to ensure it&rsquo;s updated Delete Namespaced Service via a Collection with a LabelSelector This query by apisnoop shows that all outstanding endpoints as list at the start of this document have been hit by the e2e test. Note that the results do include other service endpoints that have been addressed in other conformance tests. If a test with these calls gets merged, test coverage will go up by 4 points This test is also created with the goal of conformance promotion. /sig testing /sig architecture /area conformance\nCan you please review the proposed test and advise if we could go ahead with a PR\n/assign\nIt looks like it's patching metadata and spec but not status?\nIssues go stale after 90d of inactivity. Mark the issue as fresh with . Stale issues rot after an additional 30d of inactivity and eventually close. If this issue is safe to close now please do so with . Send feedback to sig-testing, kubernetes/test-infra and/or . /lifecycle stale\n/remove-lifecycle stale\nI've been retesting this mock test and apisnoop does log as been hit. From my testing below it looks like the patchStatus is working as expected or have I missed something?"}
{"_id":"doc-en-kubernetes-6c98fb90bee2cd0d2fc404edc784deab934fddaa81022ed1883862c51f5dae6d","title":"","text":"Using the following code before the patch request Gives this response Then after running the patch ServiceStatus Then running The response shows the patched changes.\nDuring the Conformance Meeting on 12th of January 2021, reviewed approach with Clayton, and we will start on the test PR this week!\n/reopen\nReopened this issue. <detailsIn response to : /reopen </details\n/triage accepted"}
{"_id":"doc-en-kubernetes-cf86a1652325e783afd2e6de68ee19a134c3d5bc67e73b574ad8b67c1852b0fb","title":"","text":"/kind bug What happened: I deployed a fresh custom HA kubernetes v1.10.2 cluster with 3 nodes (all controller and worker currently, more workers planned). I then created a namespace called \"stummi\" and a rolebinding with \"user=stummi clusterrole=admin\" in this namespace. User stummi (successfully authenticated by TLS client certificate) was not able to list pods or anything else in his namespace: While trying to figure things out, I downgraded to v1.8.11 (a version without cluster role aggregation, that worked) and also tried v1.11 alpha2 (which did not work). I never purged my etcd and I sadly do not have logs from before the different versions. Shortly after posting on slack and before writing this ticket, I dumped the cluster roles in question: Note the rules. What you expected to happen: I expected user stummi to be able to do most things in his namespace. How to reproduce it (as minimally and precisely as possible): Anything else we need to know?: I posted this problem on slack and helped me and wanted this issue opened. /sig auth The cluster is deployed with ansible and I can give the playbooks when required. Environment: Kubernetes version (use ):Cloud provider or hardware configuration: 3 virtual machines (2 cores, 8GB memory) OS (e.g. from /etc/os-release): Debian Stretch Kernel (e.g. ): Install tools: custom ansible playbooks Others: Please let me know if you need more information. I'm happy to provide everything you need to fix this :)\n/assign"}
{"_id":"doc-en-kubernetes-6cbf7dcb5e6486cb532532a59cda0ac4f93a00ec9c96bc20136b3a20d89cc61b","title":"","text":"<!-- Please only use this template for submitting reports about continuously failing tests or jobs in Kubernetes CI --Which jobs are failing: ci-kubernetes-e2e-gci-gce-ipvs Which test(s) are failing: [sig-network] Services should only allow access from service loadbalancer source ranges [Slow] Testgrid link: https://k8s- Reason for failure: Anything else we need to know: The test is failing consistently, however, there is also test that is failing a lot but not always, may it be related?\n/sig-network /area ipvs /cc /help wanted\n/sig network\n/triage unresolved Comment when the issue is assessed and confirmed. I am a bot run by vllry. ‍\n/assign\n/remove-triage unresolved\nHi, , turnout this is because mark drop rule in be deleted, I raied a pr to fix this issue, PTAL. : )\n/close as duplicate of\nClosing this issue. <detailsIn response to : /close as duplicate of Instructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\n/reopen this was affecting ipvs only before , let's keep it open until we verify is not failing anymore\nReopened this issue. <detailsIn response to : /reopen this was affecting ipvs only before , let's keep it open until we verify is not failing anymore Instructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\nThis issue is currently awaiting triage. If a SIG or subproject determines this is a relevant issue, they will accept it by applying the label and provide further guidance. The label can be by org members by writing in a comment. <detailsInstructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\n/assign Ok I tested this out with kind in ipvs-mode with metallb. In iptables mode I had zero failures but ipvs does not pass at all during the timeout window. This is likely due to"}
{"_id":"doc-en-kubernetes-1611cacb700fc7bde4bed341ff7dc392aeb6c35e6c0f63ce6a10b84981e8d04a","title":"","text":"<!-- Please use this template while reporting a bug and provide as much info as possible. Not doing so may result in your bug not being addressed in a timely manner. Thanks! If the matter is security related, please disclose it privately via --What happened: Docker has a hardcoded 16K limit on log line length, which splits lines longer than that into multiple messages in the log file. in , however, doesn't account for this when is set to and prints the timestamp in front of every entry in the log file instead of every logical line. This leads to random timestamps appearing in the middle of log lines, potentially breaking the applications that parse these logs. What you expected to happen: inserts timestamps once per logical line, so original log data is preserved. Alternatively, allow API to request the logs in a sane format like JSON, which would allow merging the lines back and combining them with the timestamps more intelligently on the application side. How to reproduce it (as minimally and precisely as possible): Create : apiVersion: v1 kind: Pod metadata: name: log-lines spec: containers: name: main image: busybox command: ['sh', '-c', 'while sleep 1; do echo -n ; for i in $(seq 1 ); do echo -n A; done; echo ; done'] Start the pod with Fetch the logs with , or using the API () Observe random timestamps inserted in the middle of the log lines. Environment: Kubernetes version (use ):\nReiterating the mentions to trigger a notification: <detailsIn response to : Instructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\nI will work on it. /sig cli /remove-sig node\nThe logs are being corrupted by the API endpoint in kubelet, so shouldn't it be in scope of SIG node?\nThe log is not corrupted by k8s. The same log could be got using \"docker container logs -t CONTAINERID\" I think you should post your requirement to Docker."}
{"_id":"doc-en-kubernetes-495f248c0af4a1a8db07569505db46eac9f11be4907de183bc6923cd7882b142","title":"","text":"Please refer following code Line 50 ~ 52 if config.Timestamps { logLine = append([]byte(msg.Timestamp.Format(jsonmessage.RFC3339NanoFixed)+\" \"), logLine...) }\nkubelet doesn't rely on Docker when reading logs, it reads the files itself (see the code linked in the issue). I don't think Docker would change the forced line splitting (see for example), so this should be fixed in Kubernetes.\nYour are right and kubelet read log file from host such as /var/log/pods/pods/NAMESPACEPOD-NAMExxx/....(JSON format) I will post a PR for it. /remove-sig cli /sig node\nIssues go stale after 90d of inactivity. Mark the issue as fresh with . Stale issues rot after an additional 30d of inactivity and eventually close. If this issue is safe to close now please do so with . Send feedback to sig-testing, kubernetes/test-infra and/or . /lifecycle stale\n/remove-lifecycle stale\nIssues go stale after 90d of inactivity. Mark the issue as fresh with . Stale issues rot after an additional 30d of inactivity and eventually close. If this issue is safe to close now please do so with . Send feedback to sig-testing, kubernetes/test-infra and/or . /lifecycle stale\n/remove-lifecycle stale\nIssues go stale after 90d of inactivity. Mark the issue as fresh with . Stale issues rot after an additional 30d of inactivity and eventually close. If this issue is safe to close now please do so with . Send feedback to sig-testing, kubernetes/test-infra and/or . /lifecycle stale\n/remove-lifecycle stale\nIssues go stale after 90d of inactivity. Mark the issue as fresh with . Stale issues rot after an additional 30d of inactivity and eventually close. If this issue is safe to close now please do so with . Send feedback to sig-testing, kubernetes/test-infra and/or . /lifecycle stale\n/remove-lifecycle stale\nIssues go stale after 90d of inactivity. Mark the issue as fresh with . Stale issues rot after an additional 30d of inactivity and eventually close. If this issue is safe to close now please do so with ."}
{"_id":"doc-en-kubernetes-2f614c9ad06da31fe9313a5b2eaea6d6901d55cba8508cf5ebe57105d301c4f6","title":"","text":"Send feedback to sig-testing, kubernetes/test-infra and/or . /lifecycle stale\n/remove-lifecycle stale\nIssues go stale after 90d of inactivity. Mark the issue as fresh with . Stale issues rot after an additional 30d of inactivity and eventually close. If this issue is safe to close now please do so with . Send feedback to sig-testing, kubernetes/test-infra and/or . /lifecycle stale\n/remove-lifecycle stale\nIssues go stale after 90d of inactivity. Mark the issue as fresh with . Stale issues rot after an additional 30d of inactivity and eventually close. If this issue is safe to close now please do so with . Send feedback to sig-contributor-experience at . /lifecycle stale\n/remove-lifecycle stale\nIssues go stale after 90d of inactivity. Mark the issue as fresh with . Stale issues rot after an additional 30d of inactivity and eventually close. If this issue is safe to close now please do so with . Send feedback to sig-contributor-experience at . /lifecycle stale\nStale issues rot after 30d of inactivity. Mark the issue as fresh with . Rotten issues close after an additional 30d of inactivity. If this issue is safe to close now please do so with . Send feedback to sig-contributor-experience at . /lifecycle rotten\nI saw the attempt fixing this but it was scary (echo this comment: ). I suggested a way it may be improved. Also the proposed fix didn't introduce any e2e tests to validate that runtime behavior is not changed and runtimes keep the no-newline-on-splitted-log logic. are you or anybody on this thread are still interested to work on this? Please re-open if so. Meanwhile closing this ahead of bot doing it because the inactivity. /close\nClosing this issue. <detailsIn response to : I saw the attempt fixing this but it was scary (echo this comment: ). I suggested a way it may be improved. Also the proposed fix didn't introduce any e2e tests to validate that runtime behavior is not changed and runtimes keep the no-newline-on-splitted-log logic. Instructions for interacting with me using PR comments are available ."}
{"_id":"doc-en-kubernetes-2e8714433462894a36e35ec9c24fb7c5a56d0558d2953a09bb86920b184c167e","title":"","text":"If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\n/reopen\nYou can't reopen an issue/PR unless you authored it or you are a collaborator. <detailsIn response to : /reopen Instructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\n/reopen\nReopened this issue. <detailsIn response to : /reopen Instructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\nThe Kubernetes project currently lacks enough active contributors to adequately respond to all issues and PRs. This bot triages issues and PRs according to the following rules: After 90d of inactivity, is applied After 30d of inactivity since was applied, is applied After 30d of inactivity since was applied, the issue is closed You can: Reopen this issue or PR with Mark this issue or PR as fresh with Offer to help out with [Issue Triage][1] Please send feedback to sig-contributor-experience at . /close [1]:\nClosing this issue. <detailsIn response to : The Kubernetes project currently lacks enough active contributors to adequately respond to all issues and PRs. After 90d of inactivity, is applied After 30d of inactivity since was applied, is applied After 30d of inactivity since was applied, the issue is closedReopen this issue or PR with Mark this issue or PR as fresh with Offer to help out with [Issue Triage][1] Instructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\n/remove-lifecycle rotten /reopen will fix it.\nReopened this issue. <detailsIn response to : /remove-lifecycle rotten /reopen Instructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\n/triage accepted\nThe Kubernetes project currently lacks enough contributors to adequately respond to all issues and PRs."}
{"_id":"doc-en-kubernetes-422d52914e9725e7355cb5e937e13c24e15f273f63951d92fe52d86a167487cb","title":"","text":"This bot triages issues and PRs according to the following rules: After 90d of inactivity, is applied After 30d of inactivity since was applied, is applied After 30d of inactivity since was applied, the issue is closed You can: Mark this issue or PR as fresh with Mark this issue or PR as rotten with Close this issue or PR with Offer to help out with [Issue Triage][1] Please send feedback to sig-contributor-experience at . /lifecycle stale [1]:\nThe Kubernetes project currently lacks enough active contributors to adequately respond to all issues and PRs. This bot triages issues and PRs according to the following rules: After 90d of inactivity, is applied After 30d of inactivity since was applied, is applied After 30d of inactivity since was applied, the issue is closed You can: Mark this issue or PR as fresh with Close this issue or PR with Offer to help out with [Issue Triage][1] Please send feedback to sig-contributor-experience at . /lifecycle rotten [1]:\nThe Kubernetes project currently lacks enough active contributors to adequately respond to all issues and PRs. This bot triages issues and PRs according to the following rules: After 90d of inactivity, is applied After 30d of inactivity since was applied, is applied After 30d of inactivity since was applied, the issue is closed You can: Reopen this issue or PR with Mark this issue or PR as fresh with Offer to help out with [Issue Triage][1] Please send feedback to sig-contributor-experience at . /close [1]:\nClosing this issue. <detailsIn response to : The Kubernetes project currently lacks enough active contributors to adequately respond to all issues and PRs. After 90d of inactivity, is applied After 30d of inactivity since was applied, is applied After 30d of inactivity since was applied, the issue is closedReopen this issue or PR with Mark this issue or PR as fresh with Offer to help out with [Issue Triage][1] Instructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\n/reopen\nReopened this issue. <detailsIn response to : /reopen Instructions for interacting with me using PR comments are available ."}
{"_id":"doc-en-kubernetes-0e28d298340708e57a780633dba41fa2a744d19c9ab3d061a118f6b1e6c3d902","title":"","text":"If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\nI received a bug report about this issue. The root cause is that the buffer reads up to a new line. If the line is greater than 4096 bytes and '--timestamps' is turrned on the kubelet will write the timestamp and the partial log line. We will need to refactor the ReadLogs function to allow for a partial line read."}
{"_id":"doc-en-kubernetes-b0087f3d33811f5a8bf21a1757c4fff7aaef7add29731940c9c3bd61ec9b29c4","title":"","text":"Is this a BUG REPORT or FEATURE REQUEST?: What happened: Now the and has been consistent in by PR . There is no need to convert v1.ObjectReference to clientv1.ObjectReference in . What you expected to happen: Remove this function and cleanup the codes which call this function. How to reproduce it (as minimally and precisely as possible): Anything else we need to know?:\nThere are no sig labels on this issue. Please by:<br(1) mentioning a sig: <bre.g., for API Machinery<br(2) specifying the label manually: <bre.g., for sig/scalability<br<brNote: method (1) will trigger a notification to the team. You can find the team list and label list\nWill submit the cleanup after PR merged.\n/sig api-machinery\n/sub"}
{"_id":"doc-en-kubernetes-083944d00199b655f25fb35f51e7ea3786d4181f487cac5cdbda9f2140ce4497","title":"","text":"<!-- Please only use this template for submitting enhancement requests --The built-in and cluster roles here are deprecated in 1.13 when this issue was resolved: /sig storage /milestone v1.17 /priorit important-longterm\nYou must be a member of the GitHub team to set the milestone. If you believe you should be able to issue the /milestone command, please contact your and have them propose you as an additional delegate for this responsibility. <detailsIn response to : <!-- Please only use this template for submitting enhancement requests -- /milestone v1.17 /priorit important-longterm Instructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\n/cc /priority important-longterm\nIssues go stale after 90d of inactivity. Mark the issue as fresh with . Stale issues rot after an additional 30d of inactivity and eventually close. If this issue is safe to close now please do so with . Send feedback to sig-testing, kubernetes/test-infra and/or . /lifecycle stale\n/remove-lifecycle stale /assign\nBug triage for 1.17 here with a gentle reminder that code freeze for this release is on November 14. Is this issue still intended for 1.17?"}
{"_id":"doc-en-kubernetes-32f73fe3230f64533d19b392edefbaf0b0e1a8feaecfa7ce5826559c53a1a616","title":"","text":"Which jobs are flaking: ci-kubernetes-node-kubelet-serial-hugepages Which test(s) are flaking: all Testgrid link: https://k8s- Reason for failure:\n/sig node\n/assign /cc\nGitHub didn't allow me to assign the following users: cynepco3hahue. Note that only , repo collaborators and people who have commented on this issue/PR can be assigned. Additionally, issues/PRs can only have 10 assignees at the same time. For more information please see <detailsIn response to : /assign /cc Instructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\ntestgrid alert:\n/cc\nI started to look into logs and the issue that one of the kubelet restarts fails on: I will try to reproduce the issue locally.\nany news? Have you managed to reproduce the issue locally?\nI already have a fix that is waiting for merge -"}
{"_id":"doc-en-kubernetes-563adcd071a3c6fe544aec5f1bf507590a5c621259a5e37f4edab1ba22b77da9","title":"","text":"Park of ginkgo-ization\nPer IRC, is working on this\nPushed an update yesterday to address comments.\nwhat's the status of this? There's an issue open to de-flake this test (), but it's tough to do that if the code is in the middle of being moved to ginkgo.\nThis has been merged."}
{"_id":"doc-en-kubernetes-3c17ea4b4f4a9bb23915ee9fbb41703ef0532ac00becc05b8dda4d23a9ea9da2","title":"","text":"As the RBAC role need to be related to resources. I think we could use a command to get the supported resources. ref to\nThis would be cool.\nI thought we a command that listed resources in the discover doc a long time ago.\nis just the versions, but principle is the same. We want to list all the types, their resources, and tie that back to \"explain\".\nWhat's the best way to determine this information currently? I'm having difficulty figuring out the exact set of permissions (and how they would be written in an RBAC role) for any given kubectl command. The documentation for this right now is split across multiple places and certain things are not explained at all. Specifically, it's difficult to determine which \"API verb\" is used for which operations. Take as an example. It's unclear if the way to allow this is to use the \"proxy\" subresource of a pod or service with \"get\"/\"create\"/etc. verbs, or to use the \"proxy\" verb with the resource (pod, service) directly. is a similar case. The almost works for this purpose, but it's still difficult. Operations for a given resource are grouped into categories like write, read, status, and proxy, but these things don't obviously correlate with API verbs. Again, proxy is especially confusing as there is no mention of a proxy \"verb\". The API operations are only shown using their HTTP verbs. One thing that could help on this document would be a section on the right column under \"kubectl command\" that shows exactly the rule needed in an RBAC role to allow this command via kubectl. And even then, that assumes an exact 1-to-1 mapping between API endpoints and kubectl commands, which I doubt is always the case. (For example, port fowarding to a pod seems to require both a GET on the pod resource and a CREATE on a port forward subresource.) There also needs to be better documentation about how this information could have been determined in the first place. I've resorted to manually looking through API requests made with but even then it's difficult to figure out. Amusingly, I was the one who wrote which is referenced in the original issue description (as well as the table explaining verbs in )."}
{"_id":"doc-en-kubernetes-20fd632aaf82ba3efee7609ebeda29ca6df3199abf73ef8bcfe417660974a0ce","title":"","text":"tl;dr: It's too difficult right now to go from \"I want a user to be able to run \" to \"here is the RBAC rule to use for that\" for any case where there isn't a specific example shown in the docs.\nIssues go stale after 90d of inactivity. Mark the issue as fresh with . Stale issues rot after an additional 30d of inactivity and eventually close. Prevent issues from auto-closing with an comment. If this issue is safe to close now please do so with . Send feedback to sig-testing, kubernetes/test-infra and/or . /lifecycle stale\nIt seem that relative PR stop updated. If this issue still valuable, i will help fix it\nStale issues rot after 30d of inactivity. Mark the issue as fresh with . Rotten issues close after an additional 30d of inactivity. If this issue is safe to close now please do so with . Send feedback to sig-testing, kubernetes/test-infra and/or . /lifecycle rotten /remove-lifecycle stale\nRotten issues close after 30d of inactivity. Reopen the issue with . Mark the issue as fresh with . Send feedback to sig-testing, kubernetes/test-infra and/or . /close"}
{"_id":"doc-en-kubernetes-9fca6378f8b8df6d20b226b62d39deddc0dd6ba6166f0f70fb99239f64c63f5d","title":"","text":"All of our external load balancer tests failed for the better part of a day due to the GCE project that our tests run in hitting its quota for number of target pools. While the target pools have been deleted, the fact that we hit our quota indicates that there must be a leak somewhere, and we need to plug it.\nFor reference, the previous issue was\nCould the leakage be related to ? The quota exhaustion occurred about 22 hrs after that.\nI'd guess that this is due to GetTCPLoadBalancer returning that the load balancer doesn't exist if the forwarding rule is gone, ignoring the target pool, while the service controller considers its deleting done if the cloud provider says that the load balancer doesn't exist. I that logic to the cloud provider because the other implementations of DeleteTCPLoadBalancer weren't/won't be idempotent in the sense of returning ok if the LB doesn't exist. To fix it, we'll either need to make sure all Delete implementations are idempotent and remove the Get call in the controller, or add a method along the lines of IsTCPLoadBalancerDeleted to ask the cloud provider whether all pieces of it have been completely cleaned up. We can't make GetTCPLoadBalancer check that because it's relied on to tell us whether or not the balancer is fully created.\nCopying over my commentary from on why the GKE e2e tests are leaking so many forwarding rules and target pools vs the GCE tests:\nIs there a way for the services test to wait for the service controller to clean up the external resources before finishing (without just doing a sleep)?\nNot without polling GCE directly for whether the resources still exist, although that may be worth it as a workaround. What would be perfect is that clean-up-all-cloud-provider-resources API method that we talked about a couple months ago.\nCan we not perhaps use GCE namespaces or projects as a cheap way of getting this here?\nI don't believe so. We can't use projects due to the fact that projects can't be programatically created (they're intended to be long-lived). I'm not aware of anything along the lines of a GCE namespace other than networks, which won't work because target pools and forwarding rules unfortunately don't seem to be scoped to networks."}
{"_id":"doc-en-kubernetes-d79b213781c20847fbb47c058c0afaba60ed2f1d9f5b51e77a0a4cda944c52e0","title":"","text":"The target pool leak in the code might be fixed, but the forwarding rule + target pool leak from deleting GKE clusters too quickly is certainly still happening. Reopening for the sake of that half of the issue.\nCan we just add a short sleep to jenkins after the last test has run before it deletes anything?\nYes, you could add a sleep to Ginkgo's AfterEachSuite(): But I'd be concerned about what the upper bound on the sleep period is. Can we get the tests and code in sync with a 1 minute latency here? (e.g. if the controller garbage collects every 60 seconds, then the test only needs to sleep for 60 sec?) And add a BEEG comment to the test's sleep config noting that increasing the sleep time will not improve things, otherwise I'm pretty sure that someone noting flakes in this test in future will think \"maybe I should just increase the sleep period a little bit.\"\nActually, thinking about this a little more, it would be better to put the \"sleep\" logic in the k8s master teardown procedure (i.e. not in the e2e tests or in Jenkins), because otherwise users are going to run into the same resource leakage problems as we are. So master deletion blocks on cleanup of backend cloud resources.\nThere's no way to block master deletion from the master itself. The test framework (or user) just pulls the rug out from underneath the master. See\nRobby's right -- is the real solution to this, but until we have that the best we can do is to have the wait happen elsewhere. The options seem to be: GKE itself Ginkgo's post-testing logic Jenkins' teardown logic I don't like putting it in GKE, because arbitrarily slowing down our API makes for a poor experience for the majority of people that don't still have resources being deleted. I'm not eager to put it in Gingko either, since won't that sleep also take effect whenever anyone runs an e2e test locally against a dev cluster? Putting it in the Jenkins logic makes the most sense to me. I'll send a PR and cc you.\nWe may also want to change the kube-down logic to bring down the master and then nodes in parallel to both speed it up and more closely resemble GKE's teardown logic.\nI've deleted the old forwarding rules and target pools from the project."}
{"_id":"doc-en-kubernetes-6b81898435dc239c0c98669b4836ce5b1423be788127f142aa9eb15c198137cf","title":"","text":"I'll periodically check to see whether anything is leaked after this point. I apologize if the deletions affected any running tests - everything looked to only be from GKE and parallel e2e runs."}
{"_id":"doc-en-kubernetes-ba159de838e276a42915cbe831ad277f6b9150cbf7cfdd40b9245b21c2eb7560","title":"","text":"<!-- Please use this template while reporting a bug and provide as much info as possible. Not doing so may result in your bug not being addressed in a timely manner. Thanks! If the matter is security related, please disclose it privately via --What happened: kubelet terminated unexpectedly during hack/local-up-cluster kubelet log as a result apiserver cannot find the node to schedule pods What you expected to happen: kubelet runs successfully and hack/local-up-cluster creates a single node cluster How to reproduce it (as minimally and precisely as possible): run hack/local-up-cluster from HEAD Anything else we need to know?: additional cgroups check seemed to be introduced in ; hack/local-up-cluster works when I check out the parent commit have CI for hack/local-up-cluster and , so I'm not sure if it's an error on my side. It would be greatly appreciated if someone could explain it to me Environment: Kubernetes version (use ): Cloud provider or hardware configuration: OS (e.g: ): Debian GNU/Linux Kernel (e.g. ): Install tools: Network plugin and version (if this is a network-related bug): Others: /sig node cc\nThanks for raising such a detailed issue! cc any thoughts here? Confusingly, I think the kubelet has been assigned a different cgroup for memory and for cpu... however the fact that its using makes me think the user is relying on system defaults. out of curiosity, have you made any mods to how your system assigns cgroups?\nIt sounds like we might want to handle the error more gracefully. If we can't get the kubelet's cgroup, we should fall back to collecting metrics from all cgroups.\ncollecting metrics from all the cgroups on the system? Any concerns about how many cgroups there could be? Also, or any interest in implementing this fix (assuming we go with the fallback behavior)? If not, I can try.\nthanks for looking into this! Sorry I'm not familiar with cgroups. I'm don't know any modification to my system assigns cgroups in particular\nI can put out a PR this week\nxref\nThanks!"}
{"_id":"doc-en-kubernetes-cecee0a132f84dbf6d5783d328e94b49c8666ebba8eef4ded1687e53b0451e97","title":"","text":"/kind failing-test <!-- If this is a flake, please add: /kind flake --/sig autoscaling Looks like it just started happening today (@ 2020-09-15 03:54:19 +0000 UTC) !\ncc the namespace from the provided object \"kube-system\" does not match the namespace \"horizontal-pod-autoscaling-830\". You must pass '--namespace=kube-system' to perform this operation. the namespace from the provided object \"custom-metrics\" does not match the namespace \"horizontal-pod-autoscaling-830\". You must pass '--namespace=custom-metrics' to perform this operation. the namespace from the provided object \"custom-metrics\" does not match the namespace \"horizontal-pod-autoscaling-830\". You must pass '--namespace=custom-metrics' to perform this operation.\nrelated fix:\nNo I don't think that's the fix for this (that is just a little cleanup). Actually I think we have to delete namespace parameter in line : Because before , the will have no actually effect. But now with the change of , it will have namespace : at the end of the command. But the namespace in yaml file is which is different from . We should not have parameter when applying from file because this test didn't have before, right? Related PR:\n/assign"}
{"_id":"doc-en-kubernetes-a166ddd65322b2b4c2d8cbe35891249f4d033568413f3b77fc1f46ee1d1ee8b8","title":"","text":"/kind cleanup /sig storage /sig node Description: Feature gate VolumeSubpathEnvExpansion was GA in 1.17 and is marked for removal in 1.19 Related Issues: GA Release:\n/assign\n/cc"}
{"_id":"doc-en-kubernetes-43f40e78041c5b4f7b71fe448bfb25306103aa8e8278b5a81de2f92cd8b688c4","title":"","text":"In the new scheduler performance testing framework, (bool) is provided to benchmark performance of pods created in an \"operator\": (L11 below) However, the initialized Pods created in L5~L7 seem to be also benchmarked, although it doesn't specify . (please see the reproduce steps below). I think it's b/c once scheduler starts, the associated Prometheus artifacts are also up, and it's a single in-memory datastore. So no matter we start to collect the metric in which step, the prometheus data is already up-to-date. In the above workload config, the final data have benchmarked 1500 pods instead of 1000 pods. I guess it won't work either if running in the pod initialization step or not. Unless we manage a breakpoint at the end of each step, and calculate delta histogram for the step that specifies . Pods created in step without are not benchmarked. Apply the following diff: And then run: You will get an output sth. like: /sig scheduling\nThis issue is currently awaiting triage. If a SIG or subproject determines this is a relevant issue, they will accept it by applying the label and provide further guidance. The label can be by org members by writing in a comment. <detailsInstructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\nCould you please help confirm if it's an issue?\ncc\nUnfortunately, does not apply for metrics gathered from . As you mentioned. The default gatherer collects all metrics produced by the scheduler. It's the scheduler itself who knows pod specs and would have to decide which pods will get ignored when updating metrics. The question is if computing metrics for a subset of pods processed by the scheduler is a valid use case.\nThis is probably what we want. Is something like that possible ?\nI think being able to absolutely ignore certain pods for metrics is an overkill.\nThe use case is to ignore the pods created in steps that don't specify , so as to collect the Pod metrics create at particular step(s) (). Otherwise, the semantics of option is quite misleading. Nope, that's not the intention. We don't need to target certain pods, like using label selector or whatever. Just as I mentioned, measure pods created at particular steps."}
{"_id":"doc-en-kubernetes-a8b513dba6e655a29138693e1746ea6f16e628420cb4b4aca138e84f0ed453fc","title":"","text":"I would do something like the following: if a step has collectMetrics:false, we wait for them to finish scheduling, and then we take a snapshot of the histogram. Then for the next step that has collectMetrics:true, we take the diff with the latest snapshot to report the results. This might not be perfect, as we don't control if the metrics have finished syncing when we take the snapshot. But I think that's fine. Would doing this be possible? Now, if we have steps that run in parallel, one with collectMetrics:false and another with collectMetrics:true, then this doesn't work. Note that I don't know the details of how the gatherer works.\nThat adheres to what I have in my mind - we can only manage this on client side. I think it's fine. As we are already fetching the \"snapshot\" when pods get scheduled when equals true. True. An alternative is to design the case to be like: the step with to employ Pods that are all unschedulable, and then specify a custom metric \"schedulere2eschedulingdurationseconds[result=\"unschedulable\"]\" to target a particular histogram instead of all histograms (default option) inside a histogramVec. Note this can only be possible when and some subsequent changes get in place.\nWe might just call Histogram.Clear() method which will clear what has been collected so far at the end of phase: Assuming individual phases have no pods in common and each phase can clear metrics without affecting other phases. Though, as said, it will work only for serial case.\nWDYT\nI'm afraid doesn't work. That was my initial thought as well, but no luck. I guess it's b/c, theoretically, we cannot reset a metric on the server side - Promoetheus' model doesn't support this. But let me know if I miss anything.\nThe performance tests assume all the metrics all gathered and accessed in the same binary. As long as that's the case, works."}
{"_id":"doc-en-kubernetes-c977cd9286b4a197ec98b02fe65342de07359c1835b69599916b95554e8a5e53","title":"","text":"We already call in the code:\nThat was my understanding as well, but no luck :( Here is a diff to call in case when \"concreteOp.CollectMetrics\" is false: <details<summaryDiff:</summary<p</p</detailsAnd below is the output - it's still 1500 (500 init pods plus 1000 measured pods) instead of 1000 being benchmarked: <details<summaryOutput:</summary<p</p</details\nAlso, if you run the tests and together, the result proves doesn't work as expected:\nBut I do find a way to fix this: ."}
{"_id":"doc-en-kubernetes-0cad02d163e61aff3cc2c1257b68701a42fb1145e0b51f6816fd9a85adca78e6","title":"","text":"I create a ResourceQuota with two pods limit in my namespace. Then, I create a statefuleSet with 3 replicas. As expected, two pods are created successfully, and the third pod failed to create. But, the ReadyReplicas is not 2 ,but 1. The ReadyReplicas is correct 1Create a ResourceQuota with two pods limit in a new namespace. 2Create a statefulSet with 3 replicas 3Check if the ReadyReplicas of statefulSet matchs the number of stateful running pods. 1According to the logic of the statefulSet controller, new pod created in current sync is not be immediately counted into the ReadyReplicas. 2Failure to create pod will cause to return error, then skip updating status. <details</details<details</details<detailskind v1.23.3 </details<details</details<details</details<details</details\nThis issue is currently awaiting triage. If a SIG or subproject determines this is a relevant issue, they will accept it by applying the label and provide further guidance. The label can be by org members by writing in a comment. <detailsInstructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\n/sig apps\n/assign"}
{"_id":"doc-en-kubernetes-e8eec0a3f7a81297ed919db6952e73e9801661fbfe386a524b19b8189beefe88","title":"","text":"When ResourceClaimParameters or ResourceClassParameters fail validation in resource driver during PodSchedulingContext sync loop, the error is visible only in the events of the PodSchedulingContext. The error should be visible on the object that fails validation as well, ResourceClaim or ResourceClass respectively. Creating ResourceClaim with unsupported parameters' values or unsupported No response <details</details<details</details<details</details<details</details<details</details<details</details\n/assign\n/sig node\n/triage accepted\nKEP:"}
{"_id":"doc-en-kubernetes-3bda2fba5ce2ad90b89601bf2aa62ec37353e4edef7a62972ac2a85cc1aff832","title":"","text":"The following PriorityClass endpoints was Promoted it 1.14 without Conformance tests, and remain untested: replace-, read-, patch-, list-, delete-collection- SchedulingV1PriorityClass According to this APIsnoop query, there are still some remaining PriorityClass endpoints which are untested. with this query you can filter untested endpoints by their category and eligiblity for conformance. e.g below shows a query to find all conformance eligible untested,stable,core endpoints ! createSchedulingV1PriorityClass deleteSchedulingV1PriorityClass Promoted to Conformance 1.19 with /sig testing /sig architecture /area conformance\nIs the ultimiate goal to use conformance tests to explicitly issue all CRUD requests against PriorityClass? And, is equivalent to \"update PriorityClass\"?\nall APIs must have Conformance tests unless they are not eligible for Conformance due to the There is other APIGroups that have both Replace and Update as endpoints. That is the prerogative of the SIG to decide.\ncan you please point us to similar conformance tests used for core objects such as pods and nodes?\nThis is a life cycle test for Create, List, Update, Patch, DeleteCollection More can be found in the test/e2e folder. Also look for examples on the Conformance at previously Promoted Conformance test Reach out if we can help with more information on our approach to Conformance testing.\nI remembered about this PR. Last week a few lines of code was to cover the last of the \"Lease\" endpoint group for conformance. The one file contain the Conformance tests for the complete life cycle of Lease endpoints.\n/assign I will work on it.\n/reopen Will monitor in test grid to move to Conformance after 2 weeks soak.\nReopened this issue. <detailsIn response to : /reopen Will monitor in test grid to move to Conformance after 2 weeks soak. </details\n/triage accepted"}
{"_id":"doc-en-kubernetes-23d2f241356f8f089aacd04cb02b22714ea2023d38ddcb3dae2268a8e3824c6f","title":"","text":"Is this a BUG REPORT or FEATURE REQUEST?: /kind bug This is a followup from filed separately now that we're not tracking this as part of 1.9 release. The tests have been disabled with [DisabledForLargeClusters] tag for now, and this fact will be mentioned in the scalability validation report for v1.9. This issue tracks fixing the test so the tag can be removed and the test ran again in the large cluster. /assign /kind e2e-test-failure\nIssues go stale after 90d of inactivity. Mark the issue as fresh with . Stale issues rot after an additional 30d of inactivity and eventually close. If this issue is safe to close now please do so with . Send feedback to sig-testing, kubernetes/test-infra and/or . /lifecycle stale\n/remove-lifecycle stale\nIssues go stale after 90d of inactivity. Mark the issue as fresh with . Stale issues rot after an additional 30d of inactivity and eventually close. If this issue is safe to close now please do so with . Send feedback to sig-testing, kubernetes/test-infra and/or . /lifecycle stale\n/remove-lifecycle stale\nIs this e2e test still broken?\nSorry for the confusion here. Currently this e2e test is disabled on extra large cluster due to some scalability issues with the load balancer related components. At the moment the issues haven't been completely resolved yet.\nok, adding priority so that I know it's not a 1.13 blocker. /priority important-longterm\ncc - this is why the relevant tests for are not part of the e2e suite.\nThis issue is still valid --- some action items on top of my head: Figure out a plan to make service controller properly handle operations in parallel. Check on the status for cloud providers (especially GCP, which the test ran against) on handling large number of nodes as LB backends. /remove-triage unresolved\nIssues go stale after 90d of inactivity. Mark the issue as fresh with . Stale issues rot after an additional 30d of inactivity and eventually close. If this issue is safe to close now please do so with . Send feedback to sig-testing, kubernetes/test-infra and/or ."}
{"_id":"doc-en-kubernetes-82e0e423f6c651003322d5e3ca8baa34b91f0522dd0c015fbde1e81f949704e6","title":"","text":"/lifecycle stale\nStale issues rot after 30d of inactivity. Mark the issue as fresh with . Rotten issues close after an additional 30d of inactivity. If this issue is safe to close now please do so with . Send feedback to sig-testing, kubernetes/test-infra and/or . /lifecycle rotten\n/remove-lifecycle rotten /lifecycle frozen\n/reopen Reopening as is being reverted in\nReopened this issue. <detailsIn response to : /reopen Reopening as is being reverted in Instructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\nthese tests are flaking in xref\ncc since these were enabled by\nYeah - though these failures doesn't seem to be scale related, they seem to be test related to me. This basically means nodes without public ips: It's pretty much the same thing as (just for a different test). should be addressing these kinds of errors. [Not a release blocker for sure]\nBTW - has been opened directly for this one.\n... so since is closed, Can we close this and re-enable for large clusters ? (i.e. remove the Tag in that disables this test) ? I'm not up to speed on the history but 2 related notes I came across digging into related test today... 's a couple of TODOs in the code referencing this specific issue, but your above direct issue is now closed, so we can close this as well right?... And if so, we can remove the TODO on the test... -enabling the test ... not sure wether the timing is good right now (given that is open right now , which sparked my stumbling upon this issue :))\nWe tried to enable all of them and that caused us a bunch of issues and those had to be reverted: A couple tests were enabled later in: and those seem fine. The remaining (mostly session-affinity ones) had some issue, which so far we didn't have time to debug and I haven't heard a good explanation why those were failing. It's possible that they were improved with , but I doubt those would really cause the failures we've seen in the past. So until this is debugged (or at least understood) I'm reluctant to enabling them."}
{"_id":"doc-en-kubernetes-d99484417892d904a3f5edbb73fa06d2de654c460b279ad73f511e028afa6a6f","title":"","text":"There are CVEs happening in different images - we already migrated Kubmeark from busybox in , but we should go one step further and migrate the base image to distroless. There are some things that would need to be changed mentioned here: /sig scalability\nCan I work on this? /assign\n/remove-help\n- sure that would be great.\nHi based on the discussion in , the kubemark image has dependency on and maybe more other utilities. So it seems impossible to migrate the base image to distroless? Or do we need something like debian-hyperkube-base?\nThe mount one we may get rid of - basically we should be able to switch to FakeMounter:"}
{"_id":"doc-en-kubernetes-fb3d610f127a7a4f9ff1a7fc03ca6b60768656de779ed421c207f21093cd272e","title":"","text":"<!-- This form is for bug reports and feature requests ONLY! If you're looking for help check and the . --Is this a BUG REPORT or FEATURE REQUEST?: What happened: Currently when run kubectl create with --raw, it will raise \"--raw cannot read from a url \" if the filename has a prefix \"http\", which is not proper as we may have filenames like \"httptest\" or similar. What you expected to happen: Restrict the url check with \"https://\" or \"http://\" prefix. How to reproduce it (as minimally and precisely as possible): Run 'kubectl create --raw URI -f' with a filename which has a \"http\" prefix.\n/sig cli\n--raw should start with\nthe message could be improved, but I think the check is reasonable (I'd actually change it to error if the raw path doesn't start with )\nThis is the filename url check, not the raw path. I think you mean this"}
{"_id":"doc-en-kubernetes-25b4b43b77be18dfeeb1e5d60f56671fd9593ee9c291ed7dcecd0e9048479b76","title":"","text":"What happened: We will now always delete the pods from unschedulable queue no matter the Add operations succeed or not in function. What you expected to happen: Delete pods from unschedulable queue only when add operations succeeds\n/sig scheduling\n/assign\nWill the pods be to the queues again somewhere if we delete them by mistake here ? It seems that can not make sure to add them again.\nI am not sure if we will add them back again somewhere. If not, not only but also need to be updated. I have already send a PR here:"}
{"_id":"doc-en-kubernetes-b97e647702b8859b86e9b0127fc694e7b8feffe09fe2ccd8837b28995c7af595","title":"","text":"Is this a BUG REPORT or FEATURE REQUEST?: /kind feature What happened: I need to know to which Pod(s) a PVC is being mounted to, without having to go through each Pod one by one checking the Volume mounts and the PersistentVolumeClaim. The kubectl describe pvc and get pvc, only tell you to which Persistent Volume the PVC is bound to, but not if is being used by a Pod(s) What you expected to happen: Add a new field in the describe or get that lists the pods that are using the PVC as a volume mount How to reproduce it (as minimally and precisely as possible): Have a PV/PVC bound and create a pod using a volume mount Anything else we need to know?: Not sure if this is only for Azure Disk, which is the Storage Class I am using. Environment: Kubernetes version (use ): 7 Cloud provider or hardware configuration: Azure OS (e.g. from /etc/os-release): Ubuntu 16.04 Kernel (e.g. ): 4 Install tools: Azure Kubernetes Services Others: -n/a\n/sig storage\n/sig cli\nI think the kubectl describe can just construct this on the fly. Another angle is should we add this information to PVC object itself\nThat will be great, as right now the only way to correlate a PVC to a Pod is via a \"describe\" on the Pod.\nI can work on a PR for this!\nGitHub didn't allow me to assign the following users: clandry94. Note that only and repo collaborators can be assigned. <detailsIn response to : I can work on this! Instructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\n+1 for the feature. cc\nHow to find \"what pods are using particular PVC through command line\"\n+1 for the feature.\nthis feature will be in Kubernetes 1.12\nok kubernetes 1.12 released, how to? which command?\nCan someone point to the documentation or paste command to get the pods which have the PVC in use\nsee the associated PR to see an example usage of this The line in should show it.\nWorkaround for the meantime:\nThanks , this workaround is great for those people who still use old kubernetes version."}
{"_id":"doc-en-kubernetes-d7becb5323d04fe79bf1f030fc587d9b0aefb9105f1724f943828245df0352e1","title":"","text":"However it will meet an error if the pod have no volumes, so I improves it by adding a to check whether the volumes exists or not.\npython script to list pods with pvc in namespace(s): Is a part of a cli to set pods with velero/restic annotations. Hope it helps someone :)\nFor posterity... There is a Used By: field in the output of describe.\nclient-go didn't give this api... ,PodPVCIndex can solve the problem\nHi thanks worked for me ;) but I want to make it reverse my workarounds with adding after didn't worked do you have any ideas\n$ kubectl describe pvc -A Used\" Name: ocs-deviceset-lso-volumeset-0-data-1b45dn Namespace: openshift-storage StorageClass: lso-volumeset Used By: rook-ceph-osd-1--8fwpd Also useful ^\nthe reason it didn't work is because each return does, in point of fact, have a key defined for claimName. That key is just set to null value. So, what you'd actually want to do is test for nulls. As nice as answer is, it's a PITA to parse unstructured human-readable command output, as opposed to command output in a markup language that well known shell parsers like jq and yq can work with. Towards that end, because I had use of this, I came up with the following: That simple not null test at the end plucks out all your nulls. Obviously invert the last conditional if you want all the nulls, but you'll get all sorts of temp things that way that you probably won't care about."}
{"_id":"doc-en-kubernetes-a574ac2a261fa7b4cd59d0907178e287efcaa3989e734104cb2eecf5d89ff460","title":"","text":"should have the JSON attribute name \"items\", not \"minions\"\nDoh! Fix now and break clients?\nDefine Items and Minions and use Convertors to resolve it?\nSure. I'll make a fix."}
{"_id":"doc-en-kubernetes-8bc8d69d49aedfb4a9bf3991de80f7c339e96bfbefe5b1f250dae70fa0393729","title":"","text":"See for more context. Currently, kubelet compares the spec of a mirror pod (created via apiserver) with that of a static pod (config stored as a file) to decide whether the two are considered identical. If they differ, kubelet would delete the mirror pod and re-create it with the correct spec. Due to the fact that apiserver may set default values for some fields in the spec of the mirror pod, the specs could deviate undesirably. To avoid problems like this, we can store the config of a static pod as an annotation in the mirror pod, and compare the static pod with the annotation.\nWould a hash of the config be sufficient if we're only going to compare? The whole spec would generally be more useful, but I wonder if that is too much data to put in an annotation.\nThis seems totally reasonable to me, but I'm still desperately curious why you can't just bypass defaulting when you generate mirror pod from static pod? (i.e. just use the static pod's configuration verbatim)\nIt's not kubelet who applies the default values here. The admission plugin was only run at the apiserver, and to bypass that, you have to ask apiserver to treat mirror pods differently and potentially at multiple places.\nOn May 22, 2015 10:38 AM, \"Yu-Ju Hong\" wrote: Understood, but is it really multiple places? If so then I agree it's a bad idea. Otherwise it might be worth considering as an alternative.\nIt's untenable for Kubelet to assume to know/circumvent all fields that will be set by other entities: admission control, defaulting, late initialization, auto-something, whatever. It needs to obey the same rules as other clients.\nI don't understand why you would ever want to apply any of the things you mentioned to static pods. Static pods should be copied verbatim into the mirror pods, right? That's what I'm proposing.\nWhat if the master is updated before the kubelet, and the master adds a new API field with a default value?\nThat case should be discussed in\nI agree that mirror pod should reflect the true status of the static pod running on the node, and no default values should be set in this case."}
{"_id":"doc-en-kubernetes-0491e123884f0f20f50d0a788bca7a6a29e2fa08227703a00221ecbd3bcf483c","title":"","text":"However, I think it'd be harder to implement this way since, as stated above, that there are many things to circumvent in the apiserver itself. It could also be harder to keep this special case from being broken by a yet another new feature/plugin. Keeping the hash of the config in the annotation seems like a good compromise even though it may not reflect the true status 100% accurately. I think this is acceptable. What do you think?\nSure, if you think the annotation approach is easier, I would go with that. I guess in the case Brian mentoined (master adds a new API field with a default value), you would not update the mirror pod, because the static pod definition has not changed?\nYes, a hash would be sufficient.\nIt shouldn't be necessary to update the mirror pod in the case that a new default field was updated, because that field should be populated automatically by the apiserver on decode from etcd.\nWe should fix this so we don't have this problem again in 1.2."}
{"_id":"doc-en-kubernetes-fe4a53f9c12ffb58184de5c048a5217a0a9a95660f59883741a14147feb2210e","title":"","text":"Execute make test WHAT=./pkg/api/testing/ And it failed with information: /kind failing-test\n/sig api-machinery\nwhat's the urgency of fixing this? Is it a release blocker?\ni don't see it as a release blocker. but the problem is in client-go, so it's important to get fixed."}
{"_id":"doc-en-kubernetes-c21754b57b54e9212f2cf7773a0f449c984b1bc5da8671633611701b2ce62a1e","title":"","text":"Given a request with a Content-Type header set to , the server will respond with an error because it does not exactly match the existing strings. Some browsers (Firefox, see: ) will automatically add to a Content-Type header when making an XHR request."}
{"_id":"doc-en-kubernetes-fc6177451e9ae3dd109f95c34ca28c9d9a0c4aae10f2acbc42a784aa55bc42f8","title":"","text":"seems to belong to sig node PTAL and fix ASAP\n/assign as it seems like having something to do with"}
{"_id":"doc-en-kubernetes-4b4094f56c55cfa490f4fdbec1ba6d40dc83e208d3fe25d4f173459fce063c48","title":"","text":"Is this a BUG REPORT or FEATURE REQUEST?: What happened: Currently the vsphere cloud provider blocks the controller-manager from running if its running on a non-vmware host, because it reads and checks if its content starts with . What you expected to happen: The controller manager to start How to reproduce it (as minimally and precisely as possible): Start a on a non-vmware environment with the flags set Anything else we need to know?: Environment: Kubernetes version (use ): Cloud provider or hardware configuration: OS (e.g. from /etc/os-release): Kernel (e.g. ): Install tools: Others:\n/sig vmware\ncan you please share some more context- Why do you want vSphere cloud provider to work with non-vmware host (I believe you mean ESXi)? Where is this requirement coming from.\nWe are building a Kubernetes cluster management solution which does Kubernetes in Kubernetes and works by creating the control planes in one cluster, then creating worker instances on one of the supported platforms, among them vsphere. Currently the cluster hosting the control planes (We call it seed cluster) has to run on ESXi, otherwise the controller-manager crashes. Having the check in place for the Kubelets is completely fine and definitely makes sense, but for the controller-manager it should IMHO not be required."}
{"_id":"doc-en-kubernetes-5fef938b0452d0f6b58e576cd11f17813fb6d6b970f83a7868d905b7784a5e39","title":"","text":"Presently, the elasticsearch example Kubernetes configurations target elasticsearch v1.7.1, which is quite old. As of this writing, the latest elasticsearch release is v5.6.2, with many deployment changes since then. Could the elasticsearch examples be updated to target the v5 series of elasticsearch? See for a more up to date reference configuration.\ncc\nhey it's been a long while since I last contributed to the docs. i can do it soon but i need pointers on how to properly do it.\nhey, so we need to update the stuff under examples/elasticsearch/ *&repos=kubernetes So basically update the images, yaml files etc and test that everything works (and submit a PR)\n/sig testing\nAh, so there's no documentation updates anymore. OK I'll take care of it, assign it to me.\nIssues go stale after 90d of inactivity. Mark the issue as fresh with . Stale issues rot after an additional 30d of inactivity and eventually close. If this issue is safe to close now please do so with . Send feedback to sig-testing, kubernetes/test-infra and/or . /lifecycle stale\nGitHub didn't allow me to assign the following users: me. Note that only and repo collaborators can be assigned. <detailsIn response to : /assign me Instructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\n/assign pires\n/remove-lifecycle stale\nit seems that the example moved to I am now reading that repo guidelines and will update this there. I'll also open a PR that cleans this folder in the current repo and closes this issue."}
{"_id":"doc-en-kubernetes-13eba344b12142e2782a67745f08ac2e814f1785c70b127779bdac58248d8b7b","title":"","text":"<!-- This form is for bug reports and feature requests ONLY! If you're looking for help check and the . --Is this a BUG REPORT or FEATURE REQUEST?: /kind feature What happened: Currently, we flush ipvs services by deleting them one by one. That's because there was no flush API in libnetwork. Now, libnetwork supports flush API, I think it's time to rsync. What you expected to happen: How to reproduce it (as minimally and precisely as possible): Anything else we need to know?: Environment: Kubernetes version (use ): Cloud provider or hardware configuration**: OS (e.g. from /etc/os-release): Kernel (e.g. ): Install tools: Others:\nThere are no sig labels on this issue. Please by: a sig: e.g., to notify the contributor experience sig, OR the label manually: e.g., to apply the label Note: Method 1 will trigger an email to the group. You can find the group list and label list . The in the method 1 has to be replaced with one of these: bugs, feature-requests, pr-reviews, test-failures, proposals\n/sig network\n/area kube-proxy\n/assign"}
{"_id":"doc-en-kubernetes-5d8539be161cacf223528437869590f13423a31b1176c381211bcba6b79e0cca","title":"","text":"When trying to edit multiple resources, for example: will panic. Ref as the possible cause cc\nif was the cause, any fix needs to happen in 1.1 as well, was already cherry picked there\nAlso shows that we need a test for this case.\nis the cause. When we edit multiple resources (i.e., list-type objects), the list's Accessor.annotations (type ) is nil and will panic when UpdateApplyAnnotation tries to SetAnnotations ()."}
{"_id":"doc-en-kubernetes-8c9ebf4e41021efd0692132f5b2f5e7ec08451127a302fc405c42cb972b9ed2e","title":"","text":"<!-- Please use this template while reporting a bug and provide as much info as possible. Not doing so may result in your bug not being addressed in a timely manner. Thanks! If the matter is security related, please disclose it privately via --What happened: I created several Workloads which require up to ten ports for both UDP and TCP protocol. Now my logs get spammed with . What you expected to happen: should consider -combination instead of only . How to reproduce it (as minimally and precisely as possible): Deploy any service with a node-port with both protocols, e.g.: Anything else we need to know?: This was already acknowledged by here: However, as this thread addresses a different bug and seems kind of stuck, i'd like to suggest that this might be addressed in a new context. Environment: Kubernetes version (use ): Cloud provider or hardware configuration: OS (e.g: ): Kernel (e.g. ): Install tools: Deployed via Rancher v2.4.3\n/sig api-machinery\n/remove-sig api-machinery /sig network\n/triage unresolved Comment when the issue is assessed and confirmed. I am a bot run by vllry. ‍\nHow did you get into this situation? The allocator should not have granted the same node port twice?\nOh right, that issue. The only real way out is to de-allocate the NodePorts and then re-allocate them, I think\nAs this happens for all of my ports in the said Production Application (the one where my log gets spammed) as well as complete fresh installation (where i verified the bug), i think there's no \"way out\" except fixing the bug. As said, for me it's necessary to assign both protocols to the same port. This might be a niche-use-case, but checking the it's actually very likely for ports to allow tcp AND udp. I'm not sure about the history of this, but it might be that issue was the origin, which was adressed in PR (\"Allow services which use same port, different protocol to use the same nodePort for both\"). But this PR doesn't contain the adjustment of the said , so still considers the behaviour as invalid, while it's actually allowed and supported."}
{"_id":"doc-en-kubernetes-33f53ed726ab1a89c68c77aae3b932484545b038842bda14047a639f615b725d","title":"","text":"/assign\nAhh, I thought you meant that you had accidentally been allocated the same nodePorts and I was very confused when you said \"all of them\". I have reproduced this issue, and I can see it by inspection. The obvious fix would be to collect ports split by protocol, but I think that's not right since we don't store it split. An even simpler fix would be for to build a . Then for each protocol, for each port, build that contains all the port numbers, but remove any duplicates ONLY IF they are duplicated across protocols. Duplicates in the same protocol should be retained (to report an error later). I don't have time to tackle this tonight (and write a test), but it seems like a good bug.\nperhaps I can fix this problem /assign\nthings may be a little more complicated. Right now the is not aware of the protocol of the port so I guess simply rewriting function to return is not enough, the port allocator will still return on same node port even with different protocol. perhaps we need to refactor the to make it protocol-aware.\nI don't think we need to do make the allocator be protocol aware. IIRC we allow you to use the same nodeport on different protocols ONLY within a single service. This means (I think) that we can do hyper-local deduplication as we collect the nodeports for each service. If we find the same nodeport number with different protocols, just remove the duplicates because it is allowed. I.e. a service that looks like: return - deduped. if ANY OTHER service tried to use it would be an error at the \"rebuilt\" allocator Likewise if a single Service claims to use the same nodeport number more than once on the same protocol, it should be an error: return - NOT deduped.\nobviously, we want a great unit test here :P\nah got it! thanks for your detailed explanation!\nHi! I have filed a PR to resolve this issue, could you take a look?"}
{"_id":"doc-en-kubernetes-0e6152e65eb2ac92729fbf36045ced1b3b82ff210ce22134b0c0a6ef8bf6047e","title":"","text":"kubectl works mostly fine without /version. But a number of commands (apply, create, convert, ....) are wired to this: A 404 on /version let's the kubectl to fail hard. For environments where might be a security problem (cf. ) this is a nogo. Relates to minimal apiserver routes:\n/cc\n/version is desirable (we think all servers should have it) but it should not be required for the proper functioning of a client to be able to retrieve it."}
{"_id":"doc-en-kubernetes-16c95be4925a0d298a67709c93efbb137b71325e5e3041edcf306d6572e4ae65","title":"","text":"Which jobs are failing: pull-kubernetes-integration Which test(s) are failing: TestWebhookConverterWithDefaulting /assign /priority important-soon /area custom-resources /sig api-machinery /kind flake /milestone v1.16\n/cc\nseveral of the webhook converter tests have similar flakes. it looks like the wait for storage version to become effective times out occasionally Rare flakes seen in all of these tests: TestWebhookConverterWithWatchCache/nontrivial-converter TestWebhookConverterWithWatchCache/metadata-mutating-converter TestWebhookConverterWithDefaulting/metadata-mutating-converter TestWebhookConverterWithWatchCache/noop-converter TestWebhookConverterWithDefaulting/empty-response TestWebhookConverterWithoutWatchCache/nontrivial-converter/check-2 TestWebhookConverterWithoutWatchCache/noop-converter/check-0/v1beta2 TestWebhookConverterWithoutWatchCache/nontrivial-converter/check-0/v1beta2\ncc\nYikes, I’ll sort it out.\niiuc sending an empty patch gets short-circuited and doesn't bump CR generation, even if the storage version gets changed, which means we do not re-write to etcd and go through the encoding path this would be racing and doing noop's if the first pass of didn't get the expected storage version I will send a fix to make the patch do actual mutation\nIt doesn't bump generation, but the bytes to persist in etcd would still be different if the storage version had changed, so they should persist. You can verify that by changing the storage version on a CRD yourself, do an empty patch, and see if the resourceVersion changes.\nyou're right. Generation is for non-metadata change only, and the RV changes. I queried etcd locally and the apiVersion did change for an empty patch (sidetracking: another observation is after changing the storage version (say v1 -v2; noop convertor), empty-patching v2 endpoint would bump generation, while empty-patching v1 endpoint wouldn't)\nanother theory: our cached storage strategy could be stale if the CRD informer merges multiple events together (e.g. the informer does a re-list)."}
{"_id":"doc-en-kubernetes-4d6bef5a272bfef04fa7527d92c495ee5b0e5d06d06a593b8089b98a3b84ddb9","title":"","text":"Two scenarios: -create-update events merged into an update event: strategy with UID:a was created on demand for CRD Foo Foo was deleted and recreated strategy with UID:b was created on demand for the new Foo new Foo got updated (e.g. switching storage version) the strategy with UID:b is supposed to be deleted and , to reflect the last update. But if the CRD informer merges the delete-create-update events into , the strategy with UID:a will be deleted, but the strategy with UID:b won't. -update events merged into a create event: strategy with UID:a was created on demand for CRD Foo Foo was updated (e.g. switching storage version) the strategy with UID:a is supposed to be deleted and , to reflect the last update. But if the CRD informer merges the create-update events into , the strategy with UID:a won't be deleted (as we don't react on create events). another race found in , which could happen in a single update.\nI think I :"}
{"_id":"doc-en-kubernetes-d7e6fda1614d5720cc8c8b15824f08c5e880f7950d0c5cc5cd811c2fcceb82ab","title":"","text":"Because of this scheme mutation, this leads to concurrent use of the same scheme to cause the program to panic. It would be ideal to do this in some concurrent safe manner or add a warning or similar\n/sig api-machinery\n/assign\n/assign /triage accepted"}
{"_id":"doc-en-kubernetes-aa59e6a5adfc1b233e34e0c1fe9e3a0ec7f2fe6a4636f877a41f4c9aef45f71d","title":"","text":"/kind feature The equivalence cache feature is planned to move to beta in 1.11, and it would be useful to have some metrics attached to it. The equivalence cache should improve scheduling throughput, so we want to be able to correlate e.g. throughput with eCache hit rate. /sig scheduling /sig instrumentation\n/assign misterikkit\nIssues go stale after 90d of inactivity. Mark the issue as fresh with . Stale issues rot after an additional 30d of inactivity and eventually close. If this issue is safe to close now please do so with . Send feedback to sig-testing, kubernetes/test-infra and/or . /lifecycle stale\n/remove-lifecycle stale"}
{"_id":"doc-en-kubernetes-a8f4a2c4efafea09945dcba88cce946e365cd281546f6c4fd8280bc12f2c4ef5","title":"","text":"<!-- Please use this template while reporting a bug and provide as much info as possible. Not doing so may result in your bug not being addressed in a timely manner. Thanks! If the matter is security related, please disclose it privately via --What happened: TestLoopbackHostPortIPv6 fails when run in the machine where there is no IPv6 loopback device configured, with following error: Especially running in the containers where IPv6 is not really configured for the runtime What you expected to happen: Unit tests shouldn't really depend on the environment it is running, How to reproduce it (as minimally and precisely as possible): Anything else we need to know?: Environment: Kubernetes version (use ): master\n/sig testing /cc\n/assign\n/remove-sig testing /sig api-machinery EDIT: This is a specific SIG's unit test that is environment dependent, not a systemic test issue or the framework etc.\n/cc"}
{"_id":"doc-en-kubernetes-70cc8a837504b60a5c6c5f5140cd4a6de426afd5bd1cc3a810b5342e232bd03d","title":"","text":"integration-master ci-kubernetes-integration-master.OverallChanges 11/19/2021 23:54 CET see this as example to get information about the failure No response No response_\ngreat catch /sig api-machinery /priority critical-urgent it seems all tests fail with how can this fail consistently only in periodics and not in presubmits? is an external dependency?\nI would say this is the origin Server Side Strict Field Validation \npresubmits only run short tests, periodics run all long ones... looking if there's a long test that needs fixup\nfix in"}
{"_id":"doc-en-kubernetes-1b1e05089c93c6eb59b848626d58302483d33d2f2e5531ac5b621da204277493","title":"","text":"Is this a BUG REPORT or FEATURE REQUEST?: /kind feature This is a followup on , but for Go 1.9 We should do the same kind of testing now to possibly catch bugs in either Go or Kubernetes when doing this kind of dependency version bump. Ideally we'd have automated tests against Go tip (), but we don't have that at the moment, so we have to do it kind of manually. We should also decide which k8s version we should target. k8s v1.8 or v1.9? What Go versions have been used so far? Kubernetes v1.7.0 - go1.8.3 Kubernetes v1.6.0 - go1.7.5 Kubernetes v1.5.0 - go1.7.4 Kubernetes v1.4.0 - go1.6.3 Kubernetes v1.3.0 - go1.6.2 Kubernetes v1.2.0 - go1.4.2 Seems like the trend has been to stay on the same Go version for two releases and then bump. That'd mean we'd switch to Go1.9 in k8s v1.9 Everyone happy with that? Or is there any reason for us to go start using go1.9 in k8s v1.8? In any case, we should indeed start testing go1.9 out now... cc\nLet me know if you find any bugs\nFWIW pretty tight to code-freeze I would prefer punting out a cycle fwiw.\nI don't think we're that close to code freeze given that we have 5 (of 8) weeks left in this coding cycle But I'm happy to defer this to v1.9 if others think that. I guess the main reason for a bump would be scalability goals, but I'm not sure we have the need to expand above 5000 (and that Go would be the bottleneck in that case) for v1.8\nmade the Go benchmarks 9% faster with 15% larger binaries. I'm interested to see how that affects our numbers.\nmid-stack inlining was not included in Go 1.9. It made the compiler too slow. It's being targeted to Go 1.10 now instead, after some compiler optimization work.\nOh, thanks for the correction!"}
{"_id":"doc-en-kubernetes-018b01245292fe1977b7eec8bea53846947ed737ea149d849054911d63924f58","title":"","text":"Regardless, we need to run the scalability tests against the RC so we can report issues earlier-- the last upgrade was particularly painful because of an HTTP regression that we caught way too late.\nyeah. I filed for that now\nyeah, my recollection is that the Go releases usually occur too late in our release cadence and end up getting bumped to the next Kubernetes release. (Also we inevitably find something wrong, though testing RCs would help with that. Though I'm not sure anyone has cycles to test RCs and fix all of the issues that do come up.)\nBased on our current time frame, it looks like go1.9 support will have to slip to k8s 1.9. is open to dig into and start things testing, but there's a number of things that have to move.. I'll try to document those this time around to help with future go version bumps. So far: bazel build rules needed to be updated to support go1.9 () bazel version possibly needs to be updated? some of our custom bazel rules broke\nWhen done, we should update this:\n/priority important-soon\nUpdating the dependency is blocked pending bazel 0.6.0 - .\nCan we upgrade to 0.6.0 now? Does bazel 0.6.0 have multiarch support btw?\nWe can't update to depend on bazel 0.6.0+ until we can figure out how to support different versions of the bazelbuild and kubekins-e2e images in PR jobs in prow based on what branch is being tested. cc\nNoting that we're working on branching jobs, is testing one option here:\n[MILESTONENOTIFIER] Milestone Issue Current <details open<summaryIssue Labels</summary- : Issue will be escalated to these SIGs if needed. : Escalate to the issue owners and SIG owner; move out of milestone after several unsuccessful escalation attempts. : New functionality."}
{"_id":"doc-en-kubernetes-69273b286facac9ab246a3deb71fa6e24d55618a503d411a1618742e8a027c28","title":"","text":"</details<details<summaryHelp</summary<ul<li<a href=\"\"Additional instructions</a</li<li<a href=\"\"Commands for setting labels</a</li</ul</details\nI approved this for v1.9, this is something we definitely want in order to keep up with Go (and for improved performance)\nwe're starting to use the forked images now /cc this should be more feasible very soon Edit: test-infra jobs have also been using bazel 0.6 (now 0.6.1) / a recent rules_go to soak things a bit\nWe've identified a failing test as a result of the WIP PR to upgrade the codebase to go1.9: The admission webhook test fails with certificate errors under go1.9.1, but runs successfully under go1.8.3. cc: who wanted to be alerted to potential bugs\n- please see for a possible fix"}
{"_id":"doc-en-kubernetes-6eb9bac2f3074e091af4e33eda553e3139bfcfe5ecf153737d95dc4605072155","title":"","text":"This avoids drift between Kubernetes components and allows for cluster management tools that pass uniform feature gates to all components. xref:\n/assign\nHey I f you stuck anywhere. Please free to involve me in this."}
{"_id":"doc-en-kubernetes-b8a3e3a5c22dc8491069aa612d226f113bfee5a5c0b17ff9cb35a90b4b6a10ba","title":"","text":"<!-- Please only use this template for submitting reports about failing tests in Kubernetes CI jobs --Which jobs are failing: , Which test(s) are failing: Since when has it been failing: since this PR introduced them Testgrid link: Note that this is the flakiest test in both dashboards, and failing quite frequently. Reason for failure: unknown Anything else we need to know: cc\n/sig node\nsee: , both getting empty PID and multiple PIDs.\nopened revert at\nfor discoverability for the release team /priority critical-urgent /milestone v1.17"}
{"_id":"doc-en-kubernetes-ceb20554e653f6657baad1bbb36c40e66d8a76887956e3c674b77febc82d4455","title":"","text":"There's some practical limit on the number of concurrent access tokens that can be used in a Kubernetes cluster before it begins to spend all of its time in the authn stack verifying them. Absent fixes to it was < 500; with the fixes linked there it's still not infinitely scalable-- I still hit (what I believe to be) with ~ 5k unique tokens @ ~700QPS. This request is to this dimension an official name & definition, and list it with others, e.g. here: it it have a useful value (I think it should be 10k) /kind bug /sig scalability\n/cc\noh and 4. add tests which verify it is as high as believed.\ncc since projected tokens will increase the number of unique tokens\n/sig auth\nLRUs are not good at scaling and sharding has some major pitfalls as pointed out. I would suggest that we try to drop the LRU cache for successful authentication results in the token cache. The main benefit of using an LRU here is bound cache size. Entries in the token cache have a fast expiration independent of the LRU policy. The records probably take up O(2kb) (guesstimate) of memory. Given that these are successful attempts with presumably valid tokens, I don't think we are getting any benefit from the LRU. If we don't have a constraint on fixed cache size, we can back the token cache with a sync.Map and run a single background goroutine to periodically walk the map and remove expired entries. This would fix the problematic locking and eviction due to nonuniform distribution of tokens over shards. The failure cache is less important and we don't cache failures by default. If we consider this the \"slowpath\", we can probably leave it as. If that's not good enough, we have a couple options (with various tradeoffs) to improve it.\nIf off by default is not enough, more reasoning why the failed cache is not super useful: it might help with an accidental DoS but it doesn't prevent anything actively malicious DoS. I think we can focus our improvements on the success cache.\n/assign He is already working on adding an explicit test for it.\nhas taken it over and is actively working on this. Unfortunately, I cannot assign this issue to him as he's not a k8s member yet\nIt's not closed - all testes must be submitted and appropriate documentation must be\nThanks everyone!!"}
{"_id":"doc-en-kubernetes-3e7e95e6f142ca39e6e3ca0b78c3932bd6cbdfdc898eed9d3ffda27d28f904c7","title":"","text":"We should make the change in PR API reference links to should use HTTPS, not HTTP Visit h and look at the IANA hyperlinks Once this fix is made, a PR to is needed to update the docs (by autogenerating them). This typically would be deferred until the next Kubernetes release. v1.23 not applicable not applicable not applicable not applicable not applicable\n/sig docs\nThis issue is currently awaiting triage. If a SIG or subproject determines this is a relevant issue, they will accept it by applying the label and provide further guidance. The label can be by org members by writing in a comment. <detailsInstructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\nI opened this issue so that someone else can open a new PR that is equivalent to .\n/retitle Update links to IANA for Service Name and Transport Protocol Port Number Registry\nI would like to work on this issue. May I assign it to myself?\nAnyone - including you - is welcome to work on this issue! So long as you've you can open a pull request to propose a fix.\nThanks\n/assign\nI have made the PR. I hope this works. If not then will be happy to do the necessary changes.\nHad to close the previous PR because of issues, here is the new one I made."}
{"_id":"doc-en-kubernetes-3f173d1022a4408031efac17c8299f04d835794ab8a77d69bb7667b0107a08bf","title":"","text":"Some kubelet metrics use non-standard prometheus collectors unnecessarily. Specifically, we have which can be simple gauges. Since we cannot migrate non-standard prometheus collectors using the internal metrics stability framework, we will need to convert these metrics into standard prometheus variants. /sig instrumentation /assign\nGitHub didn't allow me to assign the following users: solodov. Note that only and repo collaborators can be assigned and that issues/PRs can only have 10 assignees at the same time. For more information please see <detailsIn response to : Some kubelet metrics use non-standard prometheus collectors unnecessarily. Specifically, we have which can be simple gauges. Since we cannot migrate non-standard prometheus collectors using the internal metrics stability framework, we will need to convert these metrics into standard prometheus variants. /assign Instructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\nrecommends that we generate those metrics in the GenericPLEG during the relist. We can generate the data directly from the pods call there ().\nI would be interested to work on this issue if no one else is working on it currently\nGo for it!\n/assign"}
{"_id":"doc-en-kubernetes-572d0164a11730b1ec949ed970afee14cdb12823b4b7a698fae66a86d82f748e","title":"","text":"1) start with a single stack cluster 2) create a PreferDualStack service 3) configure API server with a 2nd service CIDR 4) Confirm the above service is not dual-stack 5) Add a label to the above service 6) Observe that the service becomes dual-stack I think the correct answer is to have a \"repair loop\" that upgrades (and downgrades!) PreferDualStack Services to match the config. I think this is a GA blocker.\n/triage accepted The current thought is: of type with are special new breed of api objects that I don't think we have any similar precedent in the apis, i could be wrong. Their behavior is tied to how the cluster is configured. Thus their behavior must change as the cluster behavior change. a way the user is stating . If it is cluster then i am good with two ClusterIPs. If it is single stack cluster then I am also good with a single ClusterIP. One can assume that if the user was ok was with getting a additional IP when possible then they must be ok with taking it away when it is no longer possible. Based on the above I believe best course of action is create a new repair loop or piggyback on any of the existing (or better: fix it in rest stack during list and get calls). The new logic takes care of matching to the .\nreproducer and regression test if somebody takes this, feel free to add it\nack - Scheduled to tackle 3rd week of May. Will definitely include that patch."}
{"_id":"doc-en-kubernetes-b17ee1dc189a7b4f3a1873cc63d0b98225c1c3184abd41b75f3a03dd8e681b66","title":"","text":"<!-- This form is for bug reports and feature requests ONLY! If you're looking for help check and the . --Is this a BUG REPORT or FEATURE REQUEST?: What happened: Failures - https://k8s- Part of pause is not used properly for tests for other arches than x86. As the current pause image is not a manifest image, so we need to use a different image for different archs like for , for etc.. With this, I'm using logs-generator from repo What you expected to happen: How to reproduce it (as minimally and precisely as possible): Anything else we need to know?: Environment: Kubernetes version (use ): Cloud provider or hardware configuration: OS (e.g. from /etc/os-release): Kernel (e.g. ): Install tools: Others:\n/sig testing\n/assign"}
{"_id":"doc-en-kubernetes-45049cecee14166a5520f4dc4cbecae5a5a4a44f20c23557ee81cd45d6979f23","title":"","text":"This test is failing in both gce serial suite: http://k8s- /sig network /priority failing-test /priority critical-urgent /kind bug /status approved-for-milestone cc /assign PTAL\nxref\n/milestone v1.10\nthis also fails in http://k8s-\nFailing to connect to Services at all. I can't even ping pod IPs:\nSo I can ping pods same-node and cross-node right up until it kills the apiserver. Rebooting fixes a given node. The apiserver demonstrably comes back - I can query it for info...\n- I could use help - have to shift attention soon.\nFWIW, I am testing on a Calico-enabled node\nI'll take a look at this as well.\nWeird, now even after a reboot, this condition is sticky. I have a cluster in this state, if you want to poke.\nugh, now it is resolved.\nI just repro'ed without the e2e - have a cluster, proved ping works, killed apiserver ping fails.\nInteresting - docker shows the pods not running, but apiserver says they are alive.\nUpdate: docker sees no running pods. API server thinks kubelet is running them. kubelet is getting \"\" from apiserver, even after kubelet restart.\nUpdate and recap: Bring up a cluster Run some pods Prove that pods are running and services work pkill -9 apiserver Wait. Watch kubelets \"lose\" pods that were assigned to them Observe that pods are not running and services do not work Observe that apiserver says the pods are running pkill -9 kubelet Observe that it is still broken kubectl delete one pod Observe that kubelet is fixed for that one pod, but not others Q: Why is kubelet not getting a full listing when it restarts?\nWhat exactly does it mean for it to be fixed for a deleted pod?\nKubelet started running the pod that was deleted - it got the update for that one.\nThe cluster I was testing was a week or 2 old. I built a cluster from HEAD and I can not repro - said the same. Is this test running against head of master?\nI also ran the test manually against my new cluster and it passed\nFWIW I was also unable to reproduce this yesterday on a cluster from HEAD."}
{"_id":"doc-en-kubernetes-a335c4245d7a15c6e680ffd06bca4d469bc98515fa6a43286dbf7f7a3bfccc88","title":"","text":"Did some investigations and had some findings for this specific failure: Traced down a bit on the history, the first occurrence was https://k8s- From the , seems suspicious because it modified . Also found reported a relevant symptom: The fix () for that was merged a day ago and may explained why we can't repro from HEAD anymore.\nnice, is this issue reproducible without the fix ()?\nSorry should have had confirmed this, bringing up a cluster now.\nQuick update: I was able to repro the same failure () from HEAD with reverted. Though this test is now flaky on some other errors: https://k8s- https://k8s- They all seem to failed while communicating with apiserver.\nThanks ! So fixed the issue and I am curious about what introduced the issue. Do you happen to know when did this test start to fail? Just trying to find out what went wrong in between. We have some other possibly related issues so finding out the root cause of this one would be super helpful.\n/assign\nThe first occurrence (for that exact failure) I noticed was https://k8s-, which happened on 2018-02-15 09:49 PST.\nYou all rock! I see a couple greens on it now, but still mostly red?\nThe test is now flaky mostly on creating an exec pod after restarting apiserver. Some sample failures: To elaborate a bit more, a snippet of the test procedure is: ssh to master and (sending a SIGTERM) Keep polling on /healthz on apiserver until 200 is returned. Create an exec pod for later steps (and failed on various erros --- EOF, connection refused, apiserver terminating). According to above test failures, my suspicion is that the test is assuming the apiserver is restarted too soon. It is possible that apiserver captured SIGTERM and did a graceful termination, so that polling on /healthz may succeed even in the termination phase. I'm not surprised trying to create a pod via a terminating apiserver will hit those errors. Could anyone confirme the SIGTERM theory?\nSpoke with folks, sounds like after SIGTERM apiserver does wait a bit for remaining requests to finish. I will take a stab to refine the test. cc\nSent .\nTest is now passing on gce but failing on gke."}
{"_id":"doc-en-kubernetes-937105a0fb62d0b30f0dcf82865ffcd7dde374c2d2e0895384450ffe72527abd","title":"","text":"Apparently we don't have a way to trigger apiserver restart on gke at this moment. Sending a PR to skip this test on gke.\n[MILESTONENOTIFIER] Milestone Issue Needs Attention Action required: During code freeze, issues in the milestone should be in progress. If this issue is not being actively worked on, please remove it from the milestone. If it is being worked on, please add the label so it can be tracked with other in-flight issues. Note: This issue is marked as , and must be updated every 1 day during code freeze. Example update: <details<summaryIssue Labels</summary- : Issue will be escalated to these SIGs if needed. : Never automatically move issue out of a release milestone; continually escalate to contributor and SIG through all available channels. : Fixes a bug discovered during the current release. </details<details<summaryHelp</summary<ul<li<a href=\"\"Additional instructions</a</li<li<a href=\"\"Commands for setting labels</a</li</ul</details\nCommented too quick. On gke we do a same-version master upgrade. Will send another fix :/"}
{"_id":"doc-en-kubernetes-532b1798e6d44b2cb365a3a392f829271af9f2d30a6550736395566aab22d4dd","title":"","text":"<!-- Please only use this template for submitting enhancement requests --What would you like to be : kube-controller-manager currently does not expose work queue metrics Why is this needed: queuing delay and queue length are important information to debug control plane performance in large scale cluster /assign zhan849 /sig apps"}
{"_id":"doc-en-kubernetes-b6e1a63eb087c1fdbaf4f20789cf4dc422d7eb3a250658bf836af9dda6b010da","title":"","text":"<!-- Please only use this template for submitting enhancement requests --What would you like to be modified: image to be based on anything other than Docker Hub sourced Why is this needed: With Docker Hub rate-limit policy in-place, we probably should consider removing this dependency to stay in sync with other base images in Kubernetes project. Folks building their own binaries (i.e. with ) might run into this as well. Notes: Prior discussion happened on An idea from is to use the should probably be renamed to avoid confusion with k/release (h/t Currently, is set to , so migrating to may not be a drop-in move However, , so maybe we don't have to start from scratch /sig testing /area conformance /sig release (cc'ing because builds conformance image, though my quick search seem to show there isn't any build pipelines referencing )\n/assign\ndo we still want to do this? I read somewhere that the images built in Google Cloud are being mirrored from Docker Hub, right? This would mean that the API rate limiting will not hit us any more.\nWomp yeah I think DockerHub rate limit is not really the big motivator at this point. However, I think I still want to consider using distroless / go-runner to build and produce conformance image to reduce the noise from CVE scanning done in downstream places. What do you think?\nYeah I think this seems a valuable addition. Are you planning to work on this issue? :)\nI will, though probably won't start until next week. If you have ideas and would like to start early feel free to go for it! Otherwise, I'd be happy to work on and deliver this.\nThank you Wilson! Feel free to ping again if you need any assistance. /assign\n/triage accepted\none thing I need a little help with is to find a reference for using go-runner image that's used everywhere -- the one that I've found is inside conformance directory which seems to be different from what everyone else is using\nWe maintain a go-runner in k/release:\noops, I forgot to get back on this last week -- I meant to say that I need a reference for image which I can use as , so something like on Dockerfile. Is that a possible thing to do?"}
{"_id":"doc-en-kubernetes-5d193977829c25729bf0cf7e65192f8c845e68df9f65fd9e6c218e9e5725e8bf","title":"","text":"after searching around, I found this -- going to try and use and see how it works :smile:\nStarted a PR -- plenty of things I'd like to be discussed but we should probably have the discussions there\nIssues go stale after 90d of inactivity. Mark the issue as fresh with . Stale issues rot after an additional 30d of inactivity and eventually close. If this issue is safe to close now please do so with . Send feedback to sig-contributor-experience at . /lifecycle stale\n/remove-lifecycle stale there is work in progress"}
{"_id":"doc-en-kubernetes-cc18fa8624c9153eb5caf2824e8d5840bb7b2767a3286fa9c73b8b488eed7889","title":"","text":"/kind bug What happened: Every time a health check is called on the APIServer via the endpoint, an etcd healthcheck is performed. However, as this just starts at the first server in the list passed to the apiserver binary all the healthchecks generate read load on a single etcd member. What you expected to happen: The first server that is checked should either increment (and wrap around) each time or be randomised to spread load around the cluster\n/assign\nI think a random check is ok.\n/sig api-machinery"}
{"_id":"doc-en-kubernetes-8be37f8ebf3b929ba13e3d88ac8f62b2b24f9ed008212f39c0aaa7e2ec850c18","title":"","text":"Is this a BUG REPORT or FEATURE REQUEST?: /kind bug What happened: I'm testing the new option in kubernetes 1.12 and it seems that kubelet always requests the signing of a new CSR and tries to rotate the certificates when it is started. This is happening even though there's a valid (brand new) certificate on disk issued by kube-controller-manager. From what I can tell, this is happening because kubelet implements and is set after a successful certificate rotation, but is never persisted to disk. If kubelet is restarted, then the last requested CSR is nil. What you expected to happen: Ideally, kubelet should only request the signing of a new CSR if it doesn't already have a serving certificate or it's about to expire. How to reproduce it (as minimally and precisely as possible): Restart a kubelet instance that had it's serving certificate issued by kube-controller-manager Check and see that it requested the signing of a new CSR Anything else we need to know?: kubelet is being run with . , , and are all omitted to use the default values. The TLS bootstrapping for client certificates is currently not being used--just the bootstrapping for serving certificates. Environment: Kubernetes version: v1.12.0 - built from source at the tag () Cloud provider or hardware configuration: Bare metal OS: CentOS 7.5.1804 Kernel: Install tools: Custom RPMs built from this git repo with no additional patches Others:\n/sig node\n/sig auth\nNot sure if this is the ideal fix, but I've patched it so that it compares the current CSR template to the last requested template only if it doesn't already match the current cert.\ncc\nthanks for the proposed fix. I think it looks good. Could you send a PR? We should also remove entirely and only look at current cert and template. does that sound reasonable?\nSure thing, I'd be happy to do that. EDIT: I will submit a PR after approval from my employer.\nI think that's probably ok, as long as it checks everything we include in the template and we consider a certificate missing anything we request to be cause for re-requesting another cert (an issuer couldn't choose to issue a cert that drops some SANs and us not request another cert) left some comments in\nThis regression is a frequent flyer."}
{"_id":"doc-en-kubernetes-1b55ad000b296f12b046608a2d187d8503e900f5b67dbcaa87fc7ef8e9366d19","title":"","text":"I'd love to see an integration test that would catch this (but that shouldn't block a fix).\nThanks for the comments! Now that I have the CLA signed, I'll address those and submit a PR.\nJust submitted the PR:"}
{"_id":"doc-en-kubernetes-98dd9e78520525bf65ff2afc2de58a6375036f9264791def3d02921656a321fe","title":"","text":"this job failed with It seems it is imssing some dependency The job history confirms it is totally broken Originally posted by in\nThis issue is currently awaiting triage. If a SIG or subproject determines this is a relevant issue, they will accept it by applying the label and provide further guidance. The label can be by org members by writing in a comment. <detailsInstructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\n/sig testing"}
{"_id":"doc-en-kubernetes-226c0928db9d5aa6f1446838b8788eb7c60cd41b1b8246f4bd1a4e2614bb7834","title":"","text":"/kind bug Using , if a block device is symlinked to the discovery directory e.g. , then MapVolume will fail with since the containerized kubelet cannot see the path on host. is mounted by containerized kubelet, so if before we try to map the device we set devicePath to the evaluation of symlink and end up with a devicePath of , containerized kubelet will see the device and successfully map the device. Are there any issues with this solution? This is on Kubernetes 1.10\n/sig storage\nah yes containerized kubelet has many issues when dealing with symlinks. Check out Mounter.GetMode(), it calls a nsenter EvalSymlinks which should be able to do what you need. But actually, one thing to consider, is do you need to do the losetup on the host, or does inside the container work?\nYes it appears to work from inside the container, on the host I run which disappears after I delete the pod using the device. My understanding of running losetup in MapVolume is not that great but I suppose that as long as that /dev/loop0 is created we are fine. Thanks for the quick response, I'll create a PR using that function and tag you\nDoes it create it on the host or in the containerized kubelet? What happens when containerized kubelet is restarted? The purpose of losetup is to do \"refcounting\" on the block device to prevent possible naming collisions when devices are attached/detached under unhealthy conditions. Ie, if a device showed up as /dev/sdb and you gave that to a pod, and then for some reason it become unhealthy and disappeared (while the pod is still running), another device (maybe even another volume type) could get attached to the node and reuse /dev/sdb. Now the pod has access to a device it's not supposed to. With filesystem volumes, the refcounting is done with mount points, but for block device, loopback is the equivalent.\nIt is created on the host (and also visible in the container), and stays around through restart. I understand now, thank you.\nActually after I restart it becomes /vdb instead of /dev/vdb in the losetup output, idk what that is about."}
{"_id":"doc-en-kubernetes-e6271b0ba1dda755d5bb536a727016a621c7c29c599386bc1e9c22a230980cbf","title":"","text":"There is no such thing as /vdb and /dev/vdb is still intact of course. Seems inconsequential since it is disappearing after all pods using the device are deleted, i.e. it's still serving as a refcounter"}
{"_id":"doc-en-kubernetes-375711dd290c57e850707ef6745091e3080fe654060082beb69e30fe80db0e88","title":"","text":"ci-kubernetes-e2e-gce-scale-correctness Kubernetes e2e suite: [sig-node] SSH should SSH to all nodes and run commands kubetest.Test Jul 10 14:01:51 UTC+0200 Kubernetes e2e suite: [sig-node] SSH should SSH to all nodes and run commands: more . Test /sig node\n/cc\n/triage accepted\nUhhh, this may help here\n/milestone v1.22\n/sig scalability FYI\nI got annoyed that I couldn't click through to see the other logs, so now you can, ref: (spyglass doesn't render build- because it's 500MB, and the limit for rendering is 100MB)\n/assign SIG Scalability chairs/TLs, please find someone to address or call this a non-release-blocking failure\nAssigning to current scalability oncall (): /assign\nThis is sig-node issue - should be addres by SIG-node folks. /sig node\nis being discussed as a fix for this and I'm not sure it's correct\nis it sig scalability's intent that 5k node clusters should support SSH?\nThe SSH test in question passes in the \"default\" job: So the fact that this failure is unique to a scalability job means to me that scalability needs to own at least some part of this\nI did a little bit of digging and ci-kubernetes-e2e-gce-scale-correctness uses option in kubetest, which according to seems to be setting KUBESSHBASTION env. So I believe this is main difference between our scalability job and \"default\".\nDebugging the test in and providing a fix there. The crictl host PR () is doing something different."}
{"_id":"doc-en-kubernetes-c52a82aeae83a0ba9ff68393d2821e656f88f6d5fbc5e134627056e17acb1808","title":"","text":"What happened: When the duration is exactly 2 years or 3 years, human readable duration is returned as or . But when the duration is exactly 4 years, 5 years, etc, the human readable duration is returned as , , etc. What you expected to happen: The format of an exact year should be consistent. Remove when duration is an exact year: should be should be Check other years... 4, 5, 6, 7... not sure if they also have the problem. How to reproduce it (as minimally and precisely as possible): Because this is time-dependent, it can be difficult to reproduce via the CLI, but you can see the problem illustrated in which shows the current behavior: The and test cases should be changed to want and , and the code should be changed to make that pass. Anything else we need to know?: Note, this is not a problem when the duration is exactly 1 year because the human readable duration is still shown in days. So nothing needs to change to handle the case of exactly 1 year. Environment: Kubernetes version (use ): 1.18 Cloud provider or hardware configuration: n/a OS (e.g: ): n/a Kernel (e.g. ): n/a Install tools: n/a Network plugin and version (if this is a network-related bug): n/a Others: n/a\n/sig api-machinery\n/good-first-issue\nThis request has been marked as suitable for new contributors. Please ensure the request meets the requirements listed . If this request no longer meets these requirements, the label can be removed by commenting with the command. <detailsIn response to : /good-first-issue Instructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\nCould I pick this up as my first issue?\nAbsolutely. Just use to assign it to yourself.\nAlso I just updated the issue description. Check the other exact years 4, 5, 6, 7 because they might also have the problem, I can't remember. I know 8 and beyond don't have the problem because it just stops showing days after 8 years.\n/assign bitscuit\nmight recall a related PR?"}
{"_id":"doc-en-kubernetes-f6c1495fe924460d9eb2ac0fa6ac2f987b00b8ce0493b8b416038d6dd8a8b830","title":"","text":"(also, I have to leave the pedantic note: a year isn't always 365 days, so it's surprising to me to see the abbreviation exists at all)\nI can't find it. Maybe I was remembering\nYeah it was probably that one. At first I had the tests and the fix. But it sat for a while. I changed it to just the tests of existing behavior so it could be approved more easily. Then I made this issue for the fix but decided it would probably be a good first time issue for someone instead of me doing it.\nso should leap years be considered by the code?\nThere's no way to know how long a year is without knowing the date, so it's not a good unit--in the abstract, there's no way to know if it's a leap year or not. The go language deliberately omits it IIRC. Henry Li <:\nWhat's the intention of HumanDuration? From what I've seen, it is just an approximation of a timestamp solely for humans to understand, so assuming 365 days per year sounds about right to me. This issue is about consistency though, so it could be solved the other way. Instead of abbreviating when it's an exact year (2y, 3y, etc.), exact years could just all have 0d (2y0d, 3y0d, etc.)\nI think it is just used for a ballpark duration. It loses accuracy as the duration increases anyway because it changes the units. I guess my intention for opening the issue was to make it consistent with itself. For example it doesn't say 2d0h or 2h0m, so it should not say 2y0d.\n(FTR I agree that Xy0d looks worse than Xy.) Brian Pursley <:\nHi I have made the required changes. Shall I raise the PR if you haven't started working in it ?\nalready assigned it, so I think it would be good to let him submit the PR, unless he unassigns or does not submit a PR in a reasonable amount of time.\nThanks I was planning on submitting the PR over the weekend\nHello, I am a first timer in open source contribution community. I think I made the required changes for that issue. However, I don't know how assignment rules work."}
{"_id":"doc-en-kubernetes-e51351c05e6844b765acf34eaa6a0653d5a4e3452baed4722f32f9c5caa3ece5","title":"","text":"/kind bug What happened: Wrote a that returned a faulty json patch which specified an our of range index (e.G. when there is only one container). This resulted in the apiserver crashing. Stacktrace: What you expected to happen: No crash /sig-api-machinery\n/sig api-machinery\nThis maybe a bug in . I will follow this question.\nThis bug had been fixed in PR: you can checkout it.\nThanks for your comment but I don't think the linked PR fixed the issue, as it got merged almost a year ago and I am running Kube 1.10.1 which does contain that patch but still panics. Just from quickly looking at the code I the linked PR I believe there are two things wrong: The check if the index is valid only happens when a negative index was given should be\nOnce the PR of is merged I will update kubernetes.\n/kind bug"}
{"_id":"doc-en-kubernetes-52b8d78c627cb5787b11bea8197d2293333b64736fbbbb287954cc79911301a5","title":"","text":"release-master-informing: capz-windows-containerd-masterKubernetes e2e suite: [It] [sig-node] Container Lifecycle Hook when create a pod with lifecycle hook should execute prestop https hook properly [MinimumKubeletVersion:1.23] [NodeConformance] Kubernetes e2e suite: [It] [sig-node] Container Lifecycle Hook when create a pod with lifecycle hook should execute poststart https hook properly [MinimumKubeletVersion:1.23] [NodeConformance] Both the tests have been failing during most of the runs since 20th October Kubernetes e2e suite: [It] [sig-node] Container Lifecycle Hook when create a pod with lifecycle hook should execute prestop https hook properly [MinimumKubeletVersion:1.23] [NodeConformance] | 1m6s -- | -- {Oct 31 18:42:02.425: Timed out after 30.001s. Expected <errors.errorString | 0xc0015283b0: { s: \"failed to match regexp \"GET /echo?msg=prestop\" in output \"I1031 18:41:06. 8988 ] Started HTTP server on port 9090nI1031 18:41:06. 8988 ] Started UDP server on port 9091n\"\", } to be nil failed (0xc000f1d000) +0x469 () +0x335} Kubernetes e2e suite: [It] [sig-node] Container Lifecycle Hook when create a pod with lifecycle hook should execute poststart https hook properly [MinimumKubeletVersion:1.23] [NodeConformance] | 5m11s -- | -- {Oct 31 19:05:50.205: timed out while waiting for pod container-lifecycle-hook-6892/pod-with-poststart-https-hook to be running and ready failed (PodClient).CreateSync(0xc000eb5920, 0x22?) +0x94 (0xc0045e8c00) +0x73 () +0x325} No response_ /sig windows cc\n/triage accepted /cc\nThese failures happen also with Kubernetes master + containerd master with Flannel CNI. Testgrid link:\nI figured out why these new failures appear."}
{"_id":"doc-en-kubernetes-468407671b0c3bb4dc0cf6c503c02539c6d31cfb69250c40c5c81f059af779f1","title":"","text":"I've submitted the fix via PR\nThis issue is also affecting 1.25 tests: If possible, we should cherry-pick to the release-1.25 branch once it's merged on master."}
{"_id":"doc-en-kubernetes-3a5f1791fd28a186fcf23c773fa97525e98fb83cc70e7098ac61d449614dcefb","title":"","text":"opencontainers/runc has fixed three CVEs: Latest runc is pulled by k/k Check runc version in file No response n/a n/a <details</details<details</details<details</details<details</details\n/assign /sig security architecture release\n/triage accepted"}
{"_id":"doc-en-kubernetes-aa0efe62d493b3611250d6c4d1b34c71d8fe0e732e78e4445af443d5c2b2f03c","title":"","text":"<!-- Thanks for filing an issue! Before hitting the button, please answer these questions.--What keywords did you search in Kubernetes issues before filing this one? controller /etc/ssl/certs kubeadm /etc/ssl/certs Is this a BUG REPORT or FEATURE REQUEST? (choose one): BUG REPORT <!-- If this is a BUG REPORT, please: Fill in as much of the template below as you can. If you leave out information, we can't help you as well. If this is a FEATURE REQUEST, please: Describe in detail the feature/behavior/change you'd like to see. In both cases, be ready for followup questions, and please respond in a timely manner. If we can't reproduce a bug or think a feature already exists, we might close your issue. If we're wrong, PLEASE feel free to reopen it and explain why. --Kubernetes version (use ): 5 custom built master I also use a custom built hyperkube image instead of the single app images for apiserver, controller, ... Environment: Cloud provider or hardware configuration: azure OS (e.g. from /etc/os-release): CentOS 7.2 Kernel (e.g. ): Linux ma-kub8ms0 3.10.0-327.36.3.el7.x8664 SMP Mon Oct 24 16:09:20 UTC 2016 x8664 x8664 x8664 GNU/Linux Install tools: kubeadm Others: deployed with Azure RM template and ansible What happened: controller is not able to do https, including calls to the Azure RM API What you expected to happen: https should work as usual How to reproduce it (as minimally and precisely as possible): Use kubeadm to install kubernetes on CentOS exec into the controller and try \"curl \" Anything else do we need to know: It looks like caused this issue. On CentOS, I see these files on the host OS: I suspect the symlinks to /etc/pki/... being the problem here, as they are not available inside the hyperkube Debian image. I also tried to verify this with It seems to not be based on Debian and thus gives a different error: I see 2 possible solutions: of host mounting /etc/ssl/certs into the controller pod, the ca certificates could be installed into the image."}
{"_id":"doc-en-kubernetes-41c6565c36e0283a9e2054713b809a870281386c810e77c1faf5476befe20dbf","title":"","text":"This is already done for hyperkube and would work out of the box when the host mount would be omitted. could try to figure out if /etc/pki should also be host mounted and add this to the controller manifest if required. I'd be happy to provide a fix by myself, I just need to know which solution you would prefer.\nSame problem seems to apply to the apiserver. Not sure however if it actually does https calls.\nimplements solution 2.\nThis issue was moved to kubernetes/kubeadm"}
{"_id":"doc-en-kubernetes-7c5448078f172c3dba4d1f6f6ed5d66f21d2fc0ea9dd38dbb1ed9b4cc49eea15","title":"","text":"Right now, we have a mixture of user documentation and developer documentation in the top-level directory. We should make it easier to find a just user documentation. Step 1 would be to move developer documentation into a subdir like docs/development or something like that. User docs would stay in the top-level docs dir.\nmove to"}
{"_id":"doc-en-kubernetes-5e93ba1a338583c6193506912a68a67fcc908451a37c483694071bc120df2bcb","title":"","text":"When a --resolv-conf (DNS resolver config) is passed to kubelet on a Windows node, the config content is used as part of sandbox initialization (through getHostDNSConfig()). Now with the recent changes (), it seems now to fail with an error like: E0315 10:22:16. 4764 ] \"Error syncing pod, skipping\" err=\"failed to \"CreatePodSandbox\" for \"img-pull-(-b40a-459d-9033-)\" with CreatePodSandboxError: \"Failed to generate sandbox config for pod \"img-pull-(-b40a-459d-9033-)\": Unexpected resolver config value: \"C:etckubernetescni\". Expected \"\" or \"Host\".\"\" pod=\"img-puller-3842/img-pull-\" podUID=-b40a-459d-9033- This is showing on all the runs of GCE testgrid (latest): Reading/using the passed resolv config on Windows Run a Windows pool with DNS resolv config being set as part of kubelet params. No response_ 27 alpha.3 1.27 beta.0 master <detailsGCE </details<details</details<details</details<details</details<details</details\n/triage accepted /milestone v1.27 /sig node /sig network\n/sig windows /cc\n/priority criticial-urgent as this is likely a release blocking\nThe label(s) cannot be applied, because the repository doesn't have them. <detailsIn response to : /priority criticial-urgent Instructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\n/priority critical-urgent\nHmm, so you're setting the [1]. This PR introduced the capability to read the DNS config options from the Windows host registries while removing the capability to read from a file, introducing a regression. I've sent a PR addressing this issue: [1]\nIs the file you are pointing to a UNIX-style or something else?\nWas the file changing the behavior of the node?"}
{"_id":"doc-en-kubernetes-a55794e1d2c420cf7484c297920657922704eb45d045ae8237364febdd60d634","title":"","text":"Yeah, what used on GCE/GKE, seems to be a Linux-like conf, eg: nameserver 10.128.0.1 search c.<project_name.internal google.internal ` We haven't tried to remove from kubelet to see the behavior.\nThis Issue/PR has been marked as release blocker. Please have the fix merged by eod today otherwise the feature has to be reverted. For the Issue/PR owner: please have a revert/feature disable PR open today in case the fix failed to get in. Thank you! See the -- Kubernetes 1.27 release manager\nI opened up a revert PR in case this PR does not merge by code freeze.\nThanks for the resolution! both 2019 and 2022 grids seems to be back to green now!\nThanks for confirming! I opened to ensure gets updated too"}
{"_id":"doc-en-kubernetes-6248148605ea242e531a4bb3c184ba4481934859726617f8259287d94988a4e9","title":"","text":"When setting for other container runtime, won't start kubelet correctly because cri is disabled : Is there any reason of disabling cri at ? cc\nno, that wasn't supposed to go in. I was testing something and it slipped in accidentally. I'll open a PR to fix.\nAh I see your PR for it. Thanks!"}
{"_id":"doc-en-kubernetes-1e6e7a922459aa1ae762d90ee2c0c561e09f7928de6f5ad4ad5cc72934b7dcd1","title":"","text":"New tests for subpath consistently fail in ci-kubernetes-e2e-gce-multizone: https://k8s- /kind bug /sig storage\nIs this a blocker for 1.10?\nI would expect so, yes. /milestone v1.10\nI'm on it /assign\nThe problem is likely due to inline GCE PD volumes not having enough zone information. Considering changing volumes to PVs instead.\nfix was in progress in to change to a PV\n/unassign\n[MILESTONENOTIFIER] Milestone Issue: Up-to-date for process Note: This issue is marked as , and must be updated every 1 day during code freeze. Example update: <details open<summaryIssue Labels</summary- : Issue will be escalated to these SIGs if needed. : Never automatically move issue out of a release milestone; continually escalate to contributor and SIG through all available channels. : Fixes a bug discovered during the current release. </details<details<summaryHelp</summary<ul<li<a href=\"\"Additional instructions</a</li<li<a href=\"\"Commands for setting labels</a</li</ul</details"}
{"_id":"doc-en-kubernetes-b8d2c0da68304e83135ba886da3cc34043a1f75faa72e9753c7e99fa612dae72","title":"","text":"From etcd3: May need to retry poststarthook a few times to avoid bootstrapping failures like this."}
{"_id":"doc-en-kubernetes-a3c7d28dcea4cbdaac9e2be0d7d2d60e791f1250fe870aa2d2d4769d84db6cae","title":"","text":"What would you like to be : Currently Mutex lock is acquired in for accessing different maps. Some functions only read map entry: We can utilize RWMutex where the above function acquires read lock. Why is this needed: For concurrent calls to IsFitPredicateRegistered / IsPriorityFunctionRegistered, read lock would lower contention.\n/sig scheduling\nfunctions getFitPredicateFunctions, getPriorityFunctionConfigs, ListRegisteredFitPredicates, ListRegisteredPriorityFunctions and GetAlgorithmProvider can also use read lock.\nI happened to raise earlier today. Anyone can finish resting ones.\nGood to know,\nThanks for raising this ! My PR covered and . Feel free to raise another PR to cover others.\nOnce is merged, I will send my PR thanks\nit's merged now."}
{"_id":"doc-en-kubernetes-600b05087b3e5a02baef82de4ad7a72820fecd2de58582bba171390ab13fae8d","title":"","text":"The namespace seems to be hard coded :\ncc:\nThanks. We can improve it after v1 release."}
{"_id":"doc-en-kubernetes-0d085a3ee589d9a38ade8c650d9534110265a809ec41b82f41ce81cebea870e6","title":"","text":"is a library yet is is instead of release tag. Please consider vendoring by tag whenever possible. Thanks.\nWe vendored it that way because the commit that fixed a bug (which we needed to be fixed and vendored back in) was This was merged on June 13. The release 1.1.4 () was released on July 1. If we want to update it to 1.1.4, I believe it should be fine. Between the commit we have vendored right now and the latest release, there are only three commits: adds tests merges (?) the case-sensitive fix - I'm not really sure what this commit is doing. fixes renaming when naming strategy is set (looks fine). Also, there are also no licensing changes. opinions?\n/sig api-machinery\n+1 to what said, it was a fire drill and 1.11.0 release was in trouble. Though in hind sight, we should have opened a issue to track that. something better to do next time. cc\nI'm indifferent to updating this. I'm okay with moving up if it doesn't break anything, or I'm fine with leaving things as is. We don't require deps to match semver. We can version out any commit in a deps tree, as we rely on our vendored copy.\nIt is always a good idea to match semver and it should be a default policy. Even if it doesn't matter to you, it should, because it is a best practice. Moreover downstream it matters much more as dependencies may be un-vendored in favour of tested re-usable packages.\n/cc\nIssues go stale after 90d of inactivity. Mark the issue as fresh with . Stale issues rot after an additional 30d of inactivity and eventually close. If this issue is safe to close now please do so with . Send feedback to sig-testing, kubernetes/test-infra and/or . /lifecycle stale\nSomeone please take care of this or bot will eventually close this valid issue without action...\nStale issues rot after 30d of inactivity. Mark the issue as fresh with . Rotten issues close after an additional 30d of inactivity. If this issue is safe to close now please do so with . Send feedback to sig-testing, kubernetes/test-infra and/or ."}
{"_id":"doc-en-kubernetes-3d634ec524d1253e3e5f3f0b99b6d74925e6b15b79d6d989e49b255bf837eee0","title":"","text":"/lifecycle rotten\n/lifecycle rotten\n/remove-lifecycle rotten PR open (thanks,\noops! sorry, I meant to remove it. thanks!\nThanks.\nplease do feel free to open issues/prs of this sort early in the release cycle, so we can work through things needed well before the milestones/freeze etc. we are getting some heat about updating dependencies and we kind of settled on some guidelines here: if you think that needs updating, we could definitely debate that as well :) (on a PR against that doc)\nit have nothing to do with release cycle or justification for upgrade if/when dependency has been vendored incorrectly. It is basically a vendoring bug. No debate is necessary although that kind of issues usually meet some resistance and even hostility here... Bugs are filed when they are discovered, regardless of release cycle. It would be wrong to sit on the bug without reporting it in order to synchronise with release cycle."}
{"_id":"doc-en-kubernetes-bb7c21eb048735f6a80e177a210d26530b5959539babed9d56a66ccfff61adf4","title":"","text":"<!-- Please only use this template for submitting reports about continuously failing tests or jobs in Kubernetes CI --Which jobs are failing: Which test(s) are failing: Since when has it been failing: Testgrid link: https://k8s- Reason for failure: Anything else we need to know: /cc /milestone v1.19 /priority important-soon /sig network\nRef:\n/triage unresolved Comment when the issue is assessed and confirmed. I am a bot run by vllry. ‍\nmy fu points me to this commit [] Fix multiple conversion tests that belongs to this PR , that seems related can you confirm if that PR is causing this test to fail? (should we gate on this test?)\nYes, that PR is related to the failure. The test utility is trying to convert extensions/v1beta1 data read from various files directly to a go type. This was never a valid thing to do, and happened to silently succeed because the two structs had an identical structure. This is exactly the type of bug was designed to flush out\nthe correct fix is to update the test/e2e/testing- files to\n/assign\nIt was my PR that broke this - I can also fix that.\nThanks, o it was the commit that was good but the test that was wrong :upsidedownface: , At least I got right the commit :sweat_smile:\nis out for review (talked with offline, he was aware that I'll send that out)"}
{"_id":"doc-en-kubernetes-05fa43eac8c17b852831e6979a2d93db8ab276262ace733bb7d146872dff7701","title":"","text":"pointed out that for scheduling and kubelet admission, we should consider resources of terminated-but-not-yet-deleted pods to still be in use, because it may take an arbitrary amount of time for the system to deallocate resources. We didn't check to see if this is currently how it is implemented or not. cc/ /sig scheduling\ncc/\nFor terminated-but-not-yet-deleted & phase is case, the Pod is still in scheduler's cache, so it's handled well in scheduler; the Pod will be removed from cache when it's deleted. But in kubelet, it filtered terminated-but-not-yet-deleted out () when using scheduler.predicates. The option in my mind is to ~pass all Pods (including terminated-but-not-yet-deleted) to each admin, and let each admin decide whether filter terminated-but-not-yet-deleted out~ only filter out finished (Failed or Succeed) pods for Admits.\nIn -, it already consider by . So both scheduler & kubelet handles terminated-but-not-yet-deleted pods well.\nGreat, thanks for looking into it!\nshould this issue be closed then?\nI think so.\nJust to clarify: Kubelet correctly accounts the resources used by terminated-but-not-yet-deleted pods, i.e, kubelet does not count the resource usage of those pods towards total usage on the node. Scheduler, however, does not do the same. Scheduler waits for a pod to be deleted, and then it deducts its resource request from the associated node total usage. Upon deletion of a pod, scheduler is notified and acts: So, scheduler and kubelet behaviors are not exactly the same. I am not sure if scheduler can receive an event when a pod's container status changes. If it could, we could make the logic the same as that of Kubelet.\nI guess the current behavior is OK because it is conservative (scheduler considers the resources to be in use even though kubelet considers them free). And actually it seems that scheduler is more \"correct\" than kubelet, since the resources really aren't usable until the pod is deleted, right? (IIUC that was Vish's original point)\nI agree. The current behavior is correct, but a bit conservative. Basically, the scheduler side is more conservative and waits until a pod is deleted from the cache before considering its resources released."}
{"_id":"doc-en-kubernetes-0f8f1ec2a7e3c6ce351bcfa7358868ba1886341c66515327ef662e361aa12de0","title":"","text":"That said, I believe resources of terminated-but-not-deleted pods are usable on nodes (should be verified). So, the current behavior is a bit less than ideal in terms of efficiency, but AFAIK the duration between termination and deletion is only a few seconds. So, it may not be worth optimizing.\nI thought was saying the opposite (see my first entry in this issue). Maybe he can clarify."}
{"_id":"doc-en-kubernetes-4ac48b480971aca26f0a1896d0fa1bdd3b9ed6e78abeeb1cea4ee05f140cc101","title":"","text":"<!-- Please use this template while reporting a bug and provide as much info as possible. Not doing so may result in your bug not being addressed in a timely manner. Thanks! If the matter is security related, please disclose it privately via --What happened: controller-manager is stuck in deleting an incorrectly configured LB type service. it cannot be recovered until restarted. What you expected to happen: controller-manager should recover. At least it should allow us to create the service without restarting it. How to reproduce it (as minimally and precisely as possible): put a random IP in above yaml and make sure the resource group doen't exist. Anything else we need to know?: Environment: Kubernetes version (use ): 1.12.4 Cloud provider or hardware configuration: AKS OS (e.g: ): Kernel (e.g. ): Install tools: Others:\n/sig azure\nI also want to add that the above repro is stuck in 403. It's possible for it to be stuck in 404 as well when the SP has the right permission.\nCan you share the log for 404?\nThanks for reporting the issue. /assign Do you mean other services couldn't be configured when this happens?\ncorrect.\nFYI, we had the same issue when the Idle Timeout annotation was set to \"3\" which isn't a supported case. Will this PR also cover this edge case ?\nUnfortunately , this is still not fixed. The idle timeout is not 403 or 404. We need another fix for this (and possibly more for other annotations)"}
{"_id":"doc-en-kubernetes-73e94f1f9679cf5fca554c149f1c357d9b16156cc7199a4d351eb3b22980b589","title":"","text":"We had rather strange case today when after rolling update one of the kubelet's started wrong container. For this specific replicationcontroller we are using What happened was that after rolling-update when i started to debug the issue i discovered that one of the kubelet's did not pull the image: When looking into what image it actually started i saw: When checking from the actual node i saw this It seems that instead of starting myimage with tag 2208 kubelet started image tagged as 1393 since the docker image id starts with 2208."}
{"_id":"doc-en-kubernetes-dc8aac09a0e4bb26f9dc5f50aef7f845ea9d7c4b925d34b9f6eda044925f80f5","title":"","text":"TestRunOneVolumeDetachFailNodeWithReadWriteOnce 4 times since 03-19 23:00 PDT, intermittently ! https://k8s- Still investigating whether this is a tolerable flake or not with more history than testgrid view, or otherwise would appreciate input from anyone with expertise in this test, especially as regards to whether this should be a blocker for tomorrows beta cut. maybe: /sig storage /sig testing\nThis issue is currently awaiting triage. If a SIG or subproject determines this is a relevant issue, they will accept it by applying the label and provide further guidance. The label can be by org members by writing in a comment. <detailsInstructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\nis already open for this, looks like it is attempting to fix it\nthank you!!! closing this in favor of that one, which I also to our CI signal project board for visibility. For completeness: since it's a longer standing issue, well identified as a flake and not a failure and also actively in progress, NOT considering it a blocker for the go/nogo signal for the beta tomorrow. /close\nClosing this issue. <detailsIn response to : thank you!!! closing this in favor of that one, which I also to our CI signal project board for visibility. For completeness: since it's a longer standing issue, well identified as a flake and not a failure and also actively in progress, NOT considering it a blocker for the go/nogo signal for the beta tomorrow. Instructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details"}
{"_id":"doc-en-kubernetes-aef7ec00dc93616f97177ca3d23f28217445848b34c429e5746962ea28329f43","title":"","text":"pull-kubernetes-unit TestRunOneVolumeDetachFailNodeWithReadWriteOnce Jan 7th, 2022 Check volume <fake-plugin/volume-nameis reported as attached to node <fail-detach-node, got false, expected true No response_ /sig storage\n/cc\nthis is the top unit test flake according to\ncc\nThe Kubernetes project currently lacks enough contributors to adequately respond to all issues and PRs. This bot triages issues and PRs according to the following rules: After 90d of inactivity, is applied After 30d of inactivity since was applied, is applied After 30d of inactivity since was applied, the issue is closed You can: Mark this issue or PR as fresh with Mark this issue or PR as rotten with Close this issue or PR with Offer to help out with [Issue Triage][1] Please send feedback to sig-contributor-experience at . /lifecycle stale [1]:\nI think this test is no longer flaky by checking the test grid, close it now.\nUnfortunately this does still seem to be flaky: Here was the output from : /reopen\nReopened this issue. <detailsIn response to : Unfortunately this does still seem to be flaky: Instructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\n/remove-lifecycle stale\nstill the case\nHow about adding some logging to help troubleshoot this? I looked at it a little, but it is hard to tell which part of the test is failing. PR\n/\nThis is still happening\n/triage accepted\nFrom the logs, in the OneVolumeDetachFailNodeWithReadWriteOnce test, succeeded, but the next verification failed. In verifyVolumeAttachedToNode, the attachState is from calling and within asw, it calls . In verifyVolumeReportedAsAttachedToNode, the volumes are obtained from calling , it seems that the node is not in asw.nodesToUpdateStatusFor at this step. Cannot reproduce locally. I tried to run the test 100 times locally with debug.Stack()... 3 * 100 times all passed.\nTried to run stress test on my linux machine as well but still cannot reproduce. Command: Result: Also tried running n process in parallel. Command: Result:\nIt seems like it happens more frequently."}
{"_id":"doc-en-kubernetes-62e47fd83b7fc1cc39259be1688a9f133b9cd2dc48247eb5d858fae7fd13955b","title":"","text":"But I may have no time to dig into the cause. Let's see.\ncc as the test author.\nThe problem is that this test has a lot of dependency on the sleeps that are hardcoded with a very narrow scope, but our CI seems to be able to stress it much more. As soon as you slightly change one of the sleep duration it starts to flake locally The test can not make an assumption that timing is perfect, even more in the sub second range can someone in sig-storage review this test and make it less dependant on those tight sleeps?\nAccording to most failures, it failed at Line 660. Let me run some stress locally. with 250ms sleep with 100ms sleep\nI agree this test case is too brittle due to its specific timing dependencies. To fix this, I suggest reworking the test case to: 1) Add retry with timeout to wait for final conditions 2) Remove the intermediate timing-based state transition checks. Instead, we can keep track of state transitions in the Fakes and validate those stats. wdyt?\nFriendly ping from 1.27 CI Signal that this test is still a noisy flake on the master-blocking dashboard as CI Signal has flagged it in and"}
{"_id":"doc-en-kubernetes-12565e8dbeb532aef2fc81c48820204edabd27cddb58e2c25eb51a8447b79e73","title":"","text":"introduced somethng that really should be e2e tested.\nWould be nice to get this in in the 2 weeks we have after 1.2 branch cut, if no more flakes crop up (but I'm not too optimistic). Bumping to next-release.\n/sig network I'm assuming sig-network owns kube-proxy\nI thought I noted a change related to this recently, but I may be mistaken.\nIssues go stale after 90d of inactivity. Mark the issue as fresh with . Stale issues rot after an additional 30d of inactivity and eventually close. Prevent issues from auto-closing with an comment. If this issue is safe to close now please do so with . Send feedback to sig-testing, kubernetes/test-infra and/or . /lifecycle stale\nStale issues rot after 30d of inactivity. Mark the issue as fresh with . Rotten issues close after an additional 30d of inactivity. If this issue is safe to close now please do so with . Send feedback to sig-testing, kubernetes/test-infra and/or . /lifecycle rotten /remove-lifecycle stale\n/lifecycle frozen /remove-lifecycle stale\n/remove-lifecycle rotten\nGoing to try to take this on.\nthanks!\nI've had some trouble constructing a test for this. should I determine the timeout? There doesn't appear to be an existing mechanism for waiting for service creation. seem to time out rather than be refused - I'm confused by this, as it seems to be the opposite of the intended behavior.\ncan look at the endpoints + wait for longer than kube-proxy sync time. refused means packets got to their destination but there was nothing listening there. If traffic is blocked, then you will just get a timeout.\nI created a service with no matching pods to test manually outside of the e2e suite. # curl test-api curl: (7) Failed to connect to athena-api port 80: Connection timed out # curl --connect-timeout 8 test-api curl: (28) Connection timed out after 8000 milliseconds Looking back at original example in I see that I'm getting the \"before\" behaviour. An iptables REJECT should send a denial, not a timeout. Am I looking at the wrong thing here? To me this looks like a regression, but I don't think that's likely."}
{"_id":"doc-en-kubernetes-075a02cd819ce7b09a20727c244a2aff5570644c9d21afadee17b14b07c37e5c","title":"","text":"I have a cluster from HEAD as of 2 weeks ago: So indeed it seems to not work. If I hit that same service IP from the VM itself I immediately get curl: (7) Failed to connect to 10.0.199.54 port 80: Connection refused. If I insert an iptables jump to KUBE-SERVICES at the front of FORWARD, it correctly rejects. This seems in accordance with the iptables flowchart. I'll do some spelunking to see if there's a reason NOT to reject in this case. Vallery Lancey <:\nCan you check IPVS mode?\nThis is why we need tests. Sigh.\nProposed fix is - still needs your e2e :)\nThanks Tim, if I have time I’ll get a PR up tonight.\nSure will check IPVS mode.\nBase on my test, IPVS has no such issue."}
{"_id":"doc-en-kubernetes-27479256c4cd4a81489a76c239bc26b6dc52b8a0b6a444f52adbe9ce38513013","title":"","text":"<!-- Please use this template while reporting a bug and provide as much info as possible. Not doing so may result in your bug not being addressed in a timely manner. Thanks! If the matter is security related, please disclose it privately via --/kind bug /sig cloud-provider /area provider/azure Azure in-tree CCM fails to delete a load balancer: The corresponding logs: The clouprovider config is: The Nodes are actually Azure VMs that belong to a VMO (aka vmss with flexible orchestration mode) - regular VMs and managed independently of the vmss. So those VMs are managed on Azure as . LoadBalancer deletion works with the very same setup with v1.21.1 of the in-tree Azure CCM. After recent cherry-picks, with v1.21.4 we observe the failure above. a cluster using VMs that belong to a VMO a Service of type LoadBalancer the Service of type LoadBalancer sure that the deletion fails with: Kubernetes version (use ): v1.21.4 Cloud provider or hardware configuration: OS (e.g: ): Kernel (e.g. ): Install tools: Gardener Network plugin and version (if this is a network-related bug): Others:\n/assign /triage accepted"}
{"_id":"doc-en-kubernetes-c9313de66f3508f8886cd07268b253d5e2cfa321359afb89838b6984a0afe883","title":"","text":"The recently merged PR a version cache in dockertools, which makes the test  flaky. The reason is that injected error into fakedockerclient and then expected to see the error from right after that (See ). However, the error was consumed by the goroutine which periodically updates the version cache (See ). The docker daemon version is rarely updated, and we only need version once for each container creating, during which we also do container create, start, inspect #etc. 1) Do we care about the cost of the call so much that we have to introduce the complexity? 2) Kubelet periodically calls , can we rely on that to update the cache? Anyhow, at least, we should disable the cache or have a fake one in the unit test. :) /cc\nI once considered move this cache to runtime level, while thus may need to change runtime interface. What if only cache apiversion?\nWe can only get via , so there will still be the data race. :) One option is to update the version cache in , now kubelet periodically call to check the runtime status. Another better option is to cache the version in layer, now is under our control, we can refactor it with an additional argument indicating whether to use cached value or get directly from the daemon. If we cache the version only for performance, I prefer the second option which is cleaner and more efficient. :)\nOK, will look into this tonight\nThe unit test itself should not start a goroutine to update the cache (unless you're actually testing the goroutine). We should disable it anyway, as we often do in our unit tests. It also sounds like we don't need a goroutine dedicated for this. If we can tolerate the additional overhead from time to time, we can set a expiration time (e.g., 1 minute), and only call the runtime if the entry expires.\nTotally agree. :) Maybe a new function in wrapping the update logic, or wrapping the logic in the .\nexpiration time sounds good in this case\nThis is happening quite frequently - increasing priority to P0."}
{"_id":"doc-en-kubernetes-eb797bc1cabb351f81b78965d26c745c1a384907fc5b719bed4a9e14b0656c34","title":"","text":"The compute project and zone in the prerequisites check are incorrectly calculated: !!! Error in hack/make-rules/test-e2e- Error in hack/make-rules/test-e2e- 'tee -i \"${artifacts}/build-\"' exited with status 1 0 Call stack: 1: hack/make-rules/test-e2e- main(...) Exiting with status 1 blocked by the following errors in each case: 1- Could not find gcloud compute/zone when running: .... 2- Could not find gcloud project when running: ..... 1- gcloud config unset compute/zone 2- gcloud config unset project 3- make test-e2e-node REMOTE=true LIST_IMAGES=true gcloud --version Google Cloud SDK 453.0.0 alpha 2023.10.27 beta 2023.10.27 bq 2.0.98 core 2023.10.27 gcloud-crc32c 1.0.0 <details</details<details</details<details</details<details</details<details</details<details</details\nThis issue is currently awaiting triage. If a SIG or subproject determines this is a relevant issue, they will accept it by applying the label and provide further guidance. The label can be by org members by writing in a comment. <detailsInstructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\n/sig node /assign"}
{"_id":"doc-en-kubernetes-8781beaf9cecb5a2a5891763e358d97bbb8efac5ebba9fc986604965efa0dda5","title":"","text":"Currently, it could not delete multiple pods at once, need to delete them one by one, it's very inconvenient if I have tons of pods to delete. Unlike docker, docker can multiple stop/rm containers/images at once.\nThis is possible using label selectors as of , which is better if you have tons of pods.\nI agree that using label selector is better to delete all of pods under the label. But to delete multile pods at once is still usable: I just want to delete 1/2 pods under the label, label selector will not work selector could not span over other selectors or pods Another reason is docker rm/stop x y z works, I will wonder why kubectl cannot. This is not urgent, but I hope it can be implemented.\nI agree. Thanks for the suggestion.\nWould this be a suitable task for someone new to k8s development? If so, I'd be willing to take a look at this, if someone could provide a couple of hints where to start.\nFurther to my last message, I have a functional work in progress I would like to share, if anyone is interested?\nPRs are welcome. See\nYes, this is a reasonable place to start. Since its in the client, you won't get much orientation to the core server code, but it will be a good place to orient to the client side API. There us also prior art that you can model off of. The kubectl create ... command already supports multiple resources. A good command to adapter would be the 'label' command. You can find it in Many thanks in advance! Brendan On Feb 19, 2015 8:21 AM, \"Eric Tune\" wrote:"}
{"_id":"doc-en-kubernetes-5b9acc75297a25b96aff671a444795dd08937bfc26a86f521e5dacf125ddfa55","title":"","text":"<!-- Please only use this template for submitting reports about failing tests in Kubernetes CI jobs --Which jobs are failing: gce-new-master-upgrade-cluster-new-parallel Which test(s) are failing: Varies, but all subpath failures, including: [sig-storage] CSI Volumes [Driver: csi-hostpath-v0] [Testpattern: Dynamic PV (default fs)] subPath should fail if subpath with backstepping is outside the volume [Slow] [sig-storage] CSI Volumes [Driver: csi-hostpath] [Testpattern: Dynamic PV (default fs)] subPath should fail if subpath directory is outside the volume [Slow] [sig-storage] CSI Volumes [Driver: csi-hostpath-v0] [Testpattern: Dynamic PV (default fs)] subPath should fail if subpath file is outside the volume [Slow] [sig-storage] CSI Volumes [Driver: csi-hostpath] [Testpattern: Dynamic PV (default fs)] subPath should fail if subpath file is outside the volume [Slow] ... and pretty much every other subpath test, but never all of them at once. There's also a few other storage tests failing, such as: [sig-storage] Volume expand [Slow] Verify if editing PVC allows resize [sig-storage] Detaching volumes should not work when mount is in progress Since when has it been failing: 11/22 Testgrid link: https://k8s- Reason for failure: These flakes started around the time that merged, but doesn't match up with the exact merge stamp, so it's probably coincidental. The subpath test failures seem to be mostly timeouts: ... so possibly this is just GCE fail. Anything else we need to know: This test job has always been flaky, with around a 40% failure rate. <!-- DO NOT EDIT BELOW THIS LINE --/kind flake /sig storage /priority important-soon cc\ncan you please help investigate this. It looks to me like the tests are flaking for and . I wonder if the v0 tests are racing with the non-v0 tests for something in this test?\nThe other tests including for the other CSI driver and non-CSI in-tree plugins are solid green which also makes me think this is a issue with the hostpath tests or driver."}
{"_id":"doc-en-kubernetes-b96111816628bb5b6b036ae1cc6a1fd4cca373f6f47aaa901c194d255325d8fd","title":"","text":"(just adding additional similar flakes here for tracking) Similar flakes in :\nadding to the milestone for triage since the flakes look new\nThese all 4 subpath tests call and it fails in finding a specific failed event by using . As uses to check if the specific error event happens, and , I guess that tests may flake if other failed event happens before the expected one. Also, already wait for the event, we might be able to delete  in . I will create a PR to fix above.\n/priority critical-urgent /remove-priority important-soon /milestone v1.13\nwhile we should update the test with your PR, I'm not sure if it's going to fix the failure. Looking at one failing test log, these are the events I see on the failing pod: So it seems to be failing before even reaching the subpath logic.\nTimeline of events: Test pod started getting processed by kubelet: It took a minute after the test pod was created for the csi driver to finish registering and mount to succeed, and then subpath event to trigger: So should address the issue. I was initially confused because the test logs didn't show the subpath event for some reason\nI opened up to investigate if 1 minute for csi plugin registration is expected\nI've a tracker issue for the complete list of storage flakes: This is so that we can track whether we're addressing most of the flakes with the subpath fixes, or if additional fixes are required.\nkeeping open until flakes are verified cleaned up\n/reopen Reopening this issue until we can verify the fix in CI\nAs of this morning we are seeing CSI \"subpath should fail\" flakes in the following job Multiple \"subpath should fail\" flakes in this slow job alone Multiple \"subpath should fail\" flakes in the 1.13 slow job, similar to master. The following 2 tests are flaky only in this upgrade job can you help us determine if these are cases are flaky tests that you are fixing or is there an underlying CSI bug that needs addressing?\nroot cause of this is a delay in an external sidecar containers and the tests not waiting long enough to accommodate the delay. There are no changes required in (outside of test) to handle this. These should not be 1.13 blocking. We'll continue to look in to getting them non-flaky today."}
{"_id":"doc-en-kubernetes-bf7ca95c6d3fe47cf47149272c15b0cd4379bd169d397f0222144854aadb75e6","title":"","text":"In the meantime, we can mark these tests as flaky to reduce the noise, if you'd like. looks like the same issue () is still being hit. See : And logs from (): We need to make the tests handle this condition more gracefully. The suggestion here is one possible solution.\nWould be good to capture this as a known issue in release notes.\nto followup on capturing this in Release notes\nper , can you create a targeted fix increasing the timeout? Looks like marking these tests as flaky will be difficult for 1.13, but if that doesn't fix it we will need to do that.\nShort term fix to bump the test timeout:\nAs per discussion in release burndown meeting, this is not a 1.13 blocker. please move this issue to 1.14 if there are no pending issues you are concerned about\nOk looks like the test still flaked. Looking at , the mount succeeded after a minute and proceeded to subpath processing, which (correctly) failed. However, for some resaon, the test is not seeing the subpath event and still failed after 4 minutes. On kubelet, subpath processing failed after a minute and generated the event: But the test doesn't see it for some reason, and still times out after 4 more minutes:\nI saw the same on CSI volume mount succeeded at: Kubelet correctly failed the pod because of subpath But test still failed: Because the test failed to find the pod event. This looks like a test issue. Not a 1.13 blocker.\ncan you PTAL at this\nIs there a way to check if events are overloaded and not able to handle new events? The MountVolume retries while the driver is not registered is very frequent and could have overloaded the event server.\nChecking for the existence of an event seems like the wrong way to test this. Would checking pod status alone be insufficient? Ultimately that is what we really care about.\nChecking for the specific event guarantees that we are failing with the correct error. But perhaps events are not reliable enough to depend on. Checking if the Pod never came up running would probably reduce the flakes but could potentially mask false negatives. cc for any thoughts."}
{"_id":"doc-en-kubernetes-134801dda9ebcf7b8cdf97d148e6b55c98d4c0b9989e7755b1ae0bf2d394c600","title":"","text":"Tests are still flaking and actually it seems like in general storage flakes in slow are flaking more since 11-21 or so: https://k8s- My theory is that the driver registration delay is causing a flood of events, which causes tests that check for events to fail. I think this is revealing multiple issues in both the tests and product, and we should work on fixing them all, although any one of the fixes should mitigate the test flakes: Driver registrar can sometimes take a minute to come up: Volume manager operation executor doesn't seem to be throttling retries or backing off: Tests should not rely on events: Tests should wait for CSI driver to be registered on all nodes before creating test pods:\nand have been merged and will address the biggest issues. has been backported to 1.13 already, backport is still pending.\nIssues go stale after 90d of inactivity. Mark the issue as fresh with . Stale issues rot after an additional 30d of inactivity and eventually close. If this issue is safe to close now please do so with . Send feedback to sig-testing, kubernetes/test-infra and/or . /lifecycle stale\nStale issues rot after 30d of inactivity. Mark the issue as fresh with . Rotten issues close after an additional 30d of inactivity. If this issue is safe to close now please do so with . Send feedback to sig-testing, kubernetes/test-infra and/or . /lifecycle rotten\nRotten issues close after 30d of inactivity. Reopen the issue with . Mark the issue as fresh with . Send feedback to sig-testing, kubernetes/test-infra and/or . /close\nClosing this issue. <detailsIn response to : Rotten issues close after 30d of inactivity. Reopen the issue with . Mark the issue as fresh with . /close Instructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details"}
{"_id":"doc-en-kubernetes-22d12c96031ea08416343124eb9503ad24bb849f7be271f6e9f581299968b0bd","title":"","text":"Example: kubectl cluster-info shows a redundant line: The cluster is already up, so the phrase \"Starting cluster\" is confusing in context of the command. (My kubernetes test cluster is running on AWS i.e. KUBERNETES_PROVIDER=aws. I had used kube- to create a test kubernetes cluster on aws). Log of my shell session:\nThanks for the bug report I've filed a PR to fix it."}
{"_id":"doc-en-kubernetes-818250efc38f3e1e4f69f5ba9f53826c59d17d9aa3970f389bfaed5f09e5cc10","title":"","text":"Is this a BUG REPORT or FEATURE REQUEST?: What happened: use kubectl search log with follow option get api latency (apiserverrequestlatenciessummary{quantile=\"0.99\",resource=\"pods\",verb=\"GET\"} )high to give a alert in my cluster. the apiserver log is below and we can see it costs 1m25s. What you expected to happen: do not make this api latency account in getting pod request api latency How to reproduce it (as minimally and precisely as possible): use kubectl to search a pod log with follow option long enough and you will get this metric apiserverrequestlatenciessummary{quantile=\"0.99\",resource=\"pods\",verb=\"GET\"} higher than the ordinary case. Anything else we need to know?: I am happy to fix it. Environment: Kubernetes version (use ): 1.6 Cloud provider or hardware configuration**: bare-metal OS (e.g. from /etc/os-release): ubuntu 16.04 Kernel (e.g. ): 4.4 Install tools: kubeadm Others:\nThere are no sig labels on this issue. Please by: a sig: e.g., to notify the contributor experience sig, OR the label manually: e.g., to apply the label Note: Method 1 will trigger an email to the group. You can find the group list and label list . The in the method 1 has to be replaced with one of these: bugs, feature-requests, pr-reviews, test-failures, proposals\n/sig API Machinery\n/area apiserver\n/sig api-machinery\n/assign Thanks... looking forward to your PR!\nfixed with PTAL"}
{"_id":"doc-en-kubernetes-a834f77ddd825b3d04a660aa7c11d0734893e4dcde6e0d4a7399e3b3daf56428","title":"","text":"<!-- This form is for bug reports and feature requests ONLY! If you're looking for help check and the . --Is this a BUG REPORT or FEATURE REQUEST?: What happened: the pause-amd64:3.0 image on has an empty schema2 manifest, which prevents is from being pulled by proxy/caches which only support V2 registries. What you expected to happen: Being able to pull ` using a V2 manifest compatible client. Anything else we need to know?: The proxy/cahce is requesting this\nReiterating the mentions to trigger a notification: <detailsIn response to : Instructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\nAfter talking with , he mentionned the following\nAFAIK, does not support well. After inspecting the image, you can find is still using manifest v1.\n/cc FYI\nIm not sure which exact version x schéma X is needed. What I know is that we can download through the proxy/cache all other Kubernetes kube-system images, only the pause image is giving us trouble. Therefore, I would say that having the image available using the same format as the other kube-system images would be the request here.\nYeah, we haven't updated the pause image in a while, since it doesn't really ever change. I'm not sure whether rebuilding + pushing over the existing tag is safe, or if we should bump to or something. thoughts?\nIf we do rebuilding/updating the image, that would help and a bunch of similar issues.\nit's better/save to go with new tag\nI didn't realize there were changes to the pause binary that were never actually deployed. It seems like we should probably just do that (and bump the tag).\nthat sounds great. Any idea when the existing 3.0 tag could be pushed again?\nWe shouldn't push over the 3.0 tag; we should push a 3.1 or 4.0 tag instead. I'm not sure who owns the pause container who should drive this. Maybe sig-node and/or\nIt seems like it should be sig-node. I'll look into it.\nThe new container has been built, can you test with\nIts working! thanks! I was able to pull it through our artifactory proxy , Lee Verberne <:\nexcellent, thanks for checking. /assign"}
{"_id":"doc-en-kubernetes-7628cd26f0dafa992a4ea43a6c68b9bcdbf056fca43ee32b40ab60cef4154c17","title":"","text":"Is this a BUG REPORT or FEATURE REQUEST?: /kind bug What happened: pause does not reap zombie processes using the default pause container provided by What you expected to happen: should include the up to date pause code to reap zombie processes when using docker = 1.13.1 found here How to reproduce it (as minimally and precisely as possible): a.out will leave a zombie process that is not cleaned up Anything else we need to know?: Building the pause container from master solves this issue Related: Environment: Kubernetes version (use ): 1.7.0+ Cloud provider or hardware configuration**: N/A OS (e.g. from /etc/os-release): CentOS 7 Kernel (e.g. ): 4.10.13+ Install tools: Others: docker = 1.13.1\nReiterating the mentions to trigger a notification: <detailsIn response to : Instructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\n/assign\ni get a simliar problem when i run crond service in pod container, every day it makes some [anacron] zombie progresses UID PID PPID C STIME TTY TIME CMD root 1 0 0 Nov14 ? 00:00:00 /pause root 7 0 0 Nov14 ? 00:00:00 /usr/local/bin/dumb-init /usr/lo root 13 7 0 Nov14 ? 00:00:00 nginx: master process /usr/local nobody 21 13 0 Nov14 ? 00:00:01 nginx: worker process nobody 22 13 0 Nov14 ? 00:00:07 nginx: worker process nobody 23 13 0 Nov14 ? 00:00:07 nginx: worker process nobody 24 13 0 Nov14 ? 00:00:05 nginx: worker process root 77 1 0 Nov14 ? 00:00:00 [anacron] root 1 0 00:01 ? 00:00:00 [anacron] <defunctroot 1 0 01:01 ? 00:00:00 [anacron] <defunctroot 1 0 02:01 ?"}
{"_id":"doc-en-kubernetes-6179feaaee46a15cd909bb088c82657b9b59fba8efdee1296995076588512032","title":"","text":"00:00:00 [anacron] <defunctroot 1 0 03:01 ? 00:00:00 [anacron] <defunctroot 0 0 11:49 ? 00:00:00 /bin/bash root 0 1 19:15 ? 00:00:00 /bin/bash root 0 19:15 ? 00:00:00 ps -ef !\nDo we know why this was never resolved / was never deployed?\nI just fought through this issue with a rabbitmq helm chart. I had to custom build a pause container from master to fix it. The rabbitmq helm chart uses for its liveness checks. It's a thin wrapper around a call to the rabbitmq-server daemon, which is not PID 1 (/pause is PID 1). The rabbitmq-server process forks some children to do the work and then they exit. Since the pause-amd64:3.0 is not reaping, those processes stick around as zombies. The liveness check runs every 5 seconds. After about 4 hours, there are about 27K zombies and the node begins failing to launch new containers because it's starved for resources. The pause container built from current master fixes that behavior, no zombies at all in the container.\nThe plan is to build a new version of pause-amd64 and then update the image referenced by Kubernetes. This means the default image will change in the next Kubernetes release. We can make the new image available once the freeze lifts and we can merge .\nLooks like this made it into 1.10"}
{"_id":"doc-en-kubernetes-6ad186288ef256533240c5bfccb6ba92929a47585cc346f791c0fbaeeb93c979","title":"","text":"Though I have successfully installed kubernetes via :smiley: , still want to point out this doc missing the step to install . Thus, if user just follow the guide in the doc without having much knowledge about kubernetes, he/she would get error \"It looks as if you don't have a compiled kubectl binary\" during the step of . Manually install kubectl on at least master node should be necessary (suggest to let user also install on minion node for convenience), any way, the installation step is .\nI forget to mention that.I will update the doc this tuned\nI update the scripts in pr , you can try it now. I copied the kubectl binary from the kubernetes release and put it in , so there is no extra step for user to manually download . Actually, in the first version of ubuntu k8s deployment, I copied the to every node, but I think it is an overhead, so I removed it.I think this is just enough.\nDoc update : Move the kubectl into the PATH, it may need sudo privilege so I decide not do for the user.\nHave the same issue, please update the doc ASAP.\nGreat job :+1:\nAlso ran into this bug after recklessly running ( instead of ) :)"}
{"_id":"doc-en-kubernetes-2bba8e4fb458ddf612a1ae84873c0951878ae27f68765b5b6efc434715382796","title":"","text":"If one target in multipath volume gets down and a pod that uses the volume is deleted and quickly recreated (e.g. by a Deployment), the new pod never starts. How to reproduce it (as minimally and precisely as possible): Deployment with single replica with one multipath iSCSI volume one target (e.g. switch the machine off). Multipath will mark appropriate path as failed: the deployment pod. will quickly start a new pod for the Deployment. new pod will be (or ) forever, What happened: Kubelet tries to WaitForAttach / SetUp volume for the new pod. The volume is still attached. Kubelet goes through all targets and tries to find out if it's logged in or not. Eventually it tries to read of the failed path and that returns error: What you expected to happen: Kubelet assumes that the path is failed / not logged in and tries to recover it. If it fails it continues with multipath with 2 available devices and 1 failed one. IMO, simple instead of will do its job just fine. /assign /sig storage cc for iSCSI expertise\nFix:"}
{"_id":"doc-en-kubernetes-a36678f387087dd8252401a5249426873ced2ebb26644cae4efea96ce80ca1f6","title":"","text":"reported that conformance testing with the latest version of sonobuoy against a version 1.29.0-alpha.3 cluster fails on the support the 1.17 Sample API Server using the current Aggregator test: , related to I found the reason why the test is working fine in Kubernetes CI jobs but fails with sonobuoy is because the service account used by sonobuoy \"sonobuoy-serviceaccount\" doesn't have read permission of the non-resource url \"/\", which is used by the test to confirm the removal of group path: The permission of \"sonobuoy-serviceaccount\" is as below: There may be two ways to fix it: the RBAC rule of sonobuoy to include \"get\" of \"/\". However, could other test runners encounter this issue? Is there a doc listing the required permissions for running e2e tests? the e2e to skip checking the removal of the group path if the service account doesn't have permission to access the root path. Which way do we usually prefer? Running e2e with sonobuoy should succeed. a v1.29.0-alpha.3 cluster. No response <details</details<details</details<details</details<details</details<details</details<details</details\nimplemented this solution.\n/sig api-machinery\nthank you for opening this issue and working on a solution. I'm probably not the right person to decide on which solution is preferred. From my perspective, either solution is fine.\nNote: Sonobuoy is not owned by any Kubernetes SIG. At most it is between SIG Architecture and the CNCF to choose whether to recommend it for running tests. SIG Testing owns the e2e image that sonobuoy can be used to schedule but not sonobuoy. IMHO this is a bug in running the tests via a pod, which is not how we develop the tests (they're run outside the cluster, by a cluster admin account typically). I would say 1) is the correct answer. xref: /sig architecture /sig testing /triage accepted cc\nthanks for your comment. I guess 1) means non of the existing sonobuoy versions can pass K8s conformance test against a cluster 1.29.0-alpha3, as the RBAC rule is hardcoded."}
{"_id":"doc-en-kubernetes-4545476cbce0605bfe46d5a01c41b8f4cf7b5ac07772bfb1748611bc506e90ef","title":"","text":"I will file an issue to sonobuoy first.\nquestion here is what are the set of client permissions required to run the conformance suite? perhaps a level of access or a more granular recommendation should be documented. what is your take on the root path requirement as approver for: EDIT: slack thread:\nI agree, we already require e.g. privileged pods for a long time in conformance tests and we should make it clear that we expect a highly privileged client.\n+1 for what is saying, that we should expect that the user running the tests has required permissions. These tests are not built for being run unprivileged.\ni wanted to PR the admin note in our conformance docs under sig-arch, but then i saw this: so broke this guarantee. sounds like SIG Arch needs to comment.\nThis statement surprises me. Root discovery is one of the least-privileged permissions you can have on a cluster. For years this was even open for unauthorized users on the internet (if your api endpoint has been exposed). In other words, just add the permission. While thinking about it, it is surprising that the user in question does not have this access already. Don't we allow that for every authorized user?\nIn contrast, I hope that lots of privileged operations are tested by conformance today. If we don't test things like admission webhooks for example, we have much bigger problems. And admission webhooks are basically equivalent to cluster-admin.\nthat is outdated, we use hostNetwork pods one some of the tests IIRC\nlooks like we don't. before this ticket, i was under the impression that the conformance suite already requires admin and that all testers (like Sonobuoy) use admin level credentials. it is non-privileged (e.g., does not require root on nodes, access to raw network interfaces, or cluster admin permissions) looks like we may have to change that sentence completely.\nWould have expected it here: The PR in question explicitly fixes the result returned by . If that is part of conformance (IMO it should), every conformance test needs access it obviously. Whether we require admin, or we add it manually, either would work."}
{"_id":"doc-en-kubernetes-06abc1218304c92779dae2744e33107a4d26263c1c54e4453f3f92a1949c2aa2","title":"","text":"adding to milestone and making release blocking ... we should not release 1.29 in a state that the most commonly used conformance tool will fail\nIMO we should clarify the state and go fix the tool, but we should keep the test and we should NOT add skip. Conformance docs are not a release artifact, and sonobuoy release lifecycle is not a conformance artifact. If we can't get sonobuoy working in a reasonable time, we already know the tests can be run by scheduling the image (which is ours) directly with kubectl instead and we can document that in the upstream conformance docs. We need to fix the for this image: ,%20master%20(dev) And it should probably be to release-blocking in the future.\nit looks like the e2e change made in was not consciously made part of conformance, it was just trying to functionally test the bugfix being made. For 1.29, I think we should revert the change to and make an integration test that validates the fix instead. We can revisit whether content of is part of conformance or not for 1.30\nI opened to revert the e2e change and move the functional test to an integration test\nFiled to try to get the image job passing again.\nWe probably need another issue to track 1.30 follow-up?\nIf no one has done it, I could create another issue and follow up with the items that have been discussed here: outdated privilege requirement in commonly used conformance tools like sonobuoy aware of the required priviledge to avoid similar situations in the future. permission of GET \"/\" to authorized users via the role for root discovery."}
{"_id":"doc-en-kubernetes-dfff84e3b11879758919fecc393710655267c5015ee34e69e71565465ba5db90","title":"","text":"The latest release of CNI plugins is , but we still have . I'll attempt to create a PR for this based on the .\nThis issue is currently awaiting triage. If a SIG or subproject determines this is a relevant issue, they will accept it by applying the label and provide further guidance. The label can be by org members by writing in a comment. <detailsInstructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\n/sig network\n- assigning to you :)"}
{"_id":"doc-en-kubernetes-5056a2515a29befdf589a74f13db031ed4142bbabfa8d2c37cafac2de4fccaea","title":"","text":"I started the federation-apiserver and went to and got the following output: Expected: I should have got an empty v1.ServiceList Any idea on what we are doing wrong? Have you guys seen this error?\nI think your deep copiers or conversion functions are not getting registered properly. If you look at my change in , some places are inappropriately registering with api.Scheme instead of the passed in scheme. Maybe you have that problem or a similar one, since I think you guys copied some code. , Nikhil Jindal wrote:\nDeep copy generation was slightly broken as well - you may want to double check that they are registered. On May 23, 2016, at 5:50 PM, Daniel Smith wrote: I think your deep copiers or conversion functions are not getting registered properly. If you look at my change in , some places are inappropriately registering with api.Scheme instead of the passed in scheme. Maybe you have that problem or a similar one, since I think you guys copied some code. , Nikhil Jindal wrote: — You are receiving this because you are on a team that was mentioned. Reply to this email directly or view it on GitHub\nI got the same error with latest branch, as I was swamping on service controller, so I use a previous version and worked around it.\ndeep copies were indeed missing, but adding them hasnt fixed the issue. Will debug more. is my PR to add the deep copies cc Please feel free to file issues for any such problem you face. We need to fix all these issues. Its good if we know about them early. Thanks!\nfixes this"}
{"_id":"doc-en-kubernetes-9df8e5ba2e9f919d9bec92d550979c4e420a21d16e91da208714cd0aae742f47","title":"","text":"<!-- This form is for bug reports and feature requests ONLY! If you're looking for help check and the . --Is this a BUG REPORT or FEATURE REQUEST?: /assign What happened: The tests are failing again because the scheduler messages have changed yet again. I'm going to remove the check for the exact messages because this is at least the 3rd time this has happened. Instead depend on unit/integration tests to check for specific failures. Also there's no need for these tests to run across all types of local volumes because it's just testing the scheduler part."}
{"_id":"doc-en-kubernetes-19a61d82ec14336213ef377c4e39e11807c2ce9673880ce3d82a192d5a783624","title":"","text":"Typically, the dns and dashboard is in crash loop. Ref cc\nif you check the file, does it contain lines like ?\nPrevious comment is ref: from I'll copy that to container-linux as well."}
{"_id":"doc-en-kubernetes-b90c168e06a7b711f67a139e334802368690ddadfe42c413624f21a9b2a8ae1a","title":"","text":"What happened: When debugging timeouts related to the stopping of cgroups, the current logs from -tryKillingCgroupProcesses do not provide any correlation between the PIDs which were killed and the cgroup they belonged to. What you expected to happen: Log the cgroup name along with the PID How to reproduce it (as minimally and precisely as possible): N/A Anything else we need to know?: N/A Environment: Kubernetes version (use ): master\n/sig node"}
{"_id":"doc-en-kubernetes-8720ed8a574bdce7c104bd3a62cc93935fad36eef53bf2f2ff98186bbe4abc40","title":"","text":"The metric used to report acutal pod in the label , but now I see the podname prefix only. The full pod name in the label, so I can trace the metric to the failing pod. Check raw api and verify the actual pod name with uids. <details</detailsAzure AKS <details</detailsNo response No response No response_\n/sig node\n/triage accepted /priority important-soon /cc As discussed this is a change in functionaloty that wasn't highlighted in release notes. Need to decide on what do do with it. I understand it's a change of functionality. what is the side effect of this for you? Does the label without last part not informational enough?\nI've been using the metric in my custom grafana dashboard that shows CPU, RAM, IO usage of pods with graph of increases in probes. If the apps are stable, the probe graph looks steadily. I can see if some irregularities in system metrics are influencing probes of respective pod. Also, if readiness probe is failing, the pod is removed from service, but is not restarted so I must have knowledge of pods with failing readiness probes. I attach the screenshot of my graph. ! So I'd welcome the possibility to pair the probes with exact pod and thank you for looking into it.\nare we thinking that this resulted in an undesirable change? Spologies for being a bit late to reply, I'm returning from a brief paternity leave staycation.\nThis issue is labeled with but has not been updated in over 90 days, and should be re-triaged. Important-soon issues must be staffed and worked on either currently, or very soon, ideally in time for the next release. You can: Confirm that this issue is still relevant with (org members only) Deprioritize it with or Close this issue with For more details on the triage process, see /remove-triage accepted\nWell this is awkward... I still believe the actual behavior does not respect the intended idea...\nThis issue makes it hard to link probe metrics with right pods. The value of pod label in and are also inconsistent with how it is used in other kubelet metrics, see: vs Definitely a bug, regression since 1.25.\nI can't get anyone to look at the fix/revert: you may have better luck\n/milestone v1.28\ndiscussion in\nYou've pushed it through, congrats!"}
{"_id":"doc-en-kubernetes-be027d483b5644a511b34dd42acd191f0a8a8265cd239e919307434a4dbdd401","title":"","text":"<!-- Please use this template while reporting a bug and provide as much info as possible. Not doing so may result in your bug not being addressed in a timely manner. Thanks! If the matter is security related, please disclose it privately via --What happened: kube-proxy doesn't validate the configurable IPVS tcp, tcpfin and udp timeouts. For example, when kube-proxy starts with , it will always log a warning: . What you expected to happen: kube-proxy should validate --ipvs-tcp-timeout, --ipvs-tcpfin-timeout, --ipvs-udp-timeout flag values at start. How to reproduce it (as minimally and precisely as possible): add flags to kube-proxy Anything else we need to know?: Environment: Kubernetes version (use ): Cloud provider or hardware configuration: OS (e.g: ): Kernel (e.g. ): Install tools: Network plugin and version (if this is a network-related bug): Others:\n/sig network /area ipvs\n/triage unresolved Comment when the issue is assessed and confirmed. I am a bot run by vllry. ‍"}
{"_id":"doc-en-kubernetes-c62a800befd6b705c74ea2036c66907711f830c03e53ca86df0e2014d6541d16","title":"","text":"<!-- Please use this template while reporting a bug and provide as much info as possible. Not doing so may result in your bug not being addressed in a timely manner. Thanks! If the matter is security related, please disclose it privately via --What happened: Node external IP of azure VMSS cannot be obtained when cloud-controller-manager is enabled. Corresponding logs are shown as follows:\n/sig azure\n/milestone v1.16"}
{"_id":"doc-en-kubernetes-f651131826f05881213cb9e37bd90a3c65eb2d3012bfdb53d9a39fc66f409d28","title":"","text":"In 1.1, if you specify \"current-context: i-do-not-exist\" in a kubeconfig, and there's no matching context by that name in the contexts, kubelet silently uses what I assume is either an empty or a default configuration. This can result in hilarity like wondering why your kubelet isn't using the TLS settings you provided, and spending several hours debugging the whole stack only to discover that you typo'd current-context :). It would be better if kubelet and other users of errored out on such a misconfiguration, to make the failure obvious and trivial.\ncc\nI think kubelet silently uses an empty configuration. Guess we should give a waring here.\nDid you start to work on this?\nBelow is related code: can you confirm this? Besides if we want to log a waring in kubelet, we may need to get in kubelet.\ncc\ncc\nI believe kubelet already uses . Adding a check in that current context exists if specified sounds like the right approach.\nAre there any backwards compatibility implications here? would this break users who have existing configs (that fallback to default namespace)? , Jeff Lowdermilk wrote:\nI think kubelet directly use not . Also already checks whether the user-specified current context exists, the problem is that if not exist, we just skip. We need to do something if the current context not exist, like log a warning.\nErroring on a specified and bad seems reasonable. If it is not specified and there's only one option, I don't think we should fail.\nyeap, the case is that user specify the but not exist, in such case we should just fail.\nIf no one is working on this I can take it.\nI think this is the root cause. Basically we should error on this not exist context. If you have extra bandwidth, please go ahead :)\nThis is related to - the defaulting is what allows us to continue. When that merges, only kubectl will have the \"use localhost:8080\" as defaulting."}
{"_id":"doc-en-kubernetes-e1455c8231c3d7b8e80cabcafbaabb5d8dbfa173f57128f88fd42bbd8e6a09fe","title":"","text":"<!-- Please use this template while reporting a bug and provide as much info as possible. Not doing so may result in your bug not being addressed in a timely manner. Thanks! If the matter is security related, please disclose it privately via --What happened: Following the example in , I created a pod with 3 sysctls specified in the pod's security context. Setting the sysctl resulted in the pod failing to start. What you expected to happen: I expected the pod to start with no issues. If there is something missing in the docs, I appreciate updating There is also a discussion at that suggests that kernel does not namespace certain net.* sysctls. If that's the case, can we document which sysctls are supported? I assumed is a supported sysctl because it's in the official k8s docs. How to reproduce it (as minimally and precisely as possible): Following the pod creation, the pod doesn't start: Once I delete the following two lines from the pod securityContext, the pod starts fine: Anything else we need to know?: You can see that I've enabled the unsafe sysctls with kubelet: Environment: Kubernetes version (use ):Cloud provider or hardware configuration: VMware Fusion OS (e.g: ):Kernel (e.g. ):Install tools: A single-node k8s cluster created with kubeadm Network plugin and version (if this is a network-related bug): Calico Others:\n/sig node\nThank you for the incredibly detailed issue report :) I did some poking around, and if is accurate and still current, then I believe the issue is that is not namespaced (i.e. cannot be set on a per pod/container level). If I'm understanding correctly, we'd need to make changes to the Linux kernel in order for it to be namespaced. As a disclaimer, I'm not positive if this parameter became namespaced in new kernels... it could be interesting to investigate if this still happens on a machine with the latest kernel. W.r.t. fixes... I would agree with you that short term, we need to update this documentation so that it doesn't include . do you have any interest in doing so? If not, I'm happy to."}
{"_id":"doc-en-kubernetes-29c4d3d8cbaaaaa0e62ad55e361fb5835540d2eb3a9414143acd94f95c7db66b","title":"","text":"Long term, I think it would be great to have better documentation re which sysctl parameters are namespaced (and thus supported) and which ones are not, and also a more descriptive error message if you try and utilize a non-supported parameter.\nThis historic issue also seems relevant: Perhaps we were premature to close it.\nMatt, thanks for confirming the problem! I found the following comments very helpful: Looking at , there doesn't seem to be any difference between Linux 4.15 and 5.1. I'm happy to remove the references to in the docs, but before doing so, should we confirm sysctl worked at some point in the past? As you said, we should enhance the error message and the docs with the supported sysctls. I understand documenting the namespaced sysctls can be a moving target, but not everyone likes to check the Linux source code to see what's namespaced or not. :)\nIt's probably better to change the docs to set a namespaced sysctl like or . A good way to figure out what network sysctls are namespaced is to pass to kubelet and see all the sysctls listed under inside the pod.\n/triage unresolved Comment when the issue is assessed and confirmed. I am a bot run by vllry. ‍\nAgree with all your statements here :) I'm also surprised that ever worked, but perhaps we could sync with the original author of the docs? +1 to in the meantime changing the docs to a namespaced sysctl.\nI don't think ever worked as it's hardly to believe a sysctl got unnamespacified in the kernel (it's always the other way around). is a good example of namespaced networking sysctl. However, the above suggested is not in some cases - it's unnamespaced in kernel 4.14 and became namespaced in 4.15. In general, this is hard to document because there are hundreds of parameters and some of them is kernel version dependent and you always need to lookup the kernel code for answers. I think what we should do now is to with in the example in the doc. in the doc to something like . I think this should be very close ( is an exception) - I couldn't verify all of them as there are so many sysctls and they can also be by loading kernel modules.\nSGTM, can you drive that?"}
{"_id":"doc-en-kubernetes-2ba48dfbc5ff8b560d4d7fc12720c1689ab1c2d3bbdba3005ad7428905707669","title":"","text":"Is this a BUG REPORT or FEATURE REQUEST? (choose one): BUG REPORT Kubernetes version (use ): HEAD Since v1.7.0-alpha.3, control plane images are only pushed for amd64. Compare: to This is a release-blocker and should be fixed ASAP. I guess the actual regression code-wise happened in test infra. cc\n/assign\nalso cc\n/remove-sig testing\nStill looking, but \"Only tag release containers if the name is different.\" makes me suspicious.\noh, this is a bug introduced by . fix should be easy.\nI can open a new issue as well to track what I sig-testing for but I want us to set up the required testing to make sure this never happens again\nthe removal was under the guise of \"sig testing doesn't control the cutting of bits\", if there are tests that need to be put in place to ensure all the right things are cut/released, I would expect sig-release to track that eg: if there's code that needs to be changed in kubernetes/release, it's a sig-release issue; if there's code that needs to be changed in kubernetes/test-infra, it might be a sig-testing issue (or we can triage to the right group)\nSounds good\nDetails about the exact bug and the fix are in . The reason we didn't notice this sooner is that anago only if the required image doesn't exist - it does so because we don't support all arches for all release versions, and we didn't want to hard-code a list of expected versions into the release repo. (Also, anago automatically , so we were still pushing the legacy versions - we were only dropping the non-amd64 images with this bug.)\nCan we add a after the warns? Otherwise even is the user of anago sees the warning, there's nothing he can do about it.\nAnago is moving toward being less interactive. If any checks need to be done, they should be done at build time. 1) Determine what should be built 2) Stop right there if the build fails for whatever reason. Adding redundant checks comes with the cost of maintaining hardcoded lists in multiple places. The push step (and release infra in general) should be agnostic and just push what was built.\nBut in this issue, the build was successful, but some images were removed due to the bug."}
{"_id":"doc-en-kubernetes-ecee26c3d0c9764e1c3372f00c48d9f570739b7afcc76ce0d3ff7c1362459190","title":"","text":"I'm worried that when building beta.2 this Wednesday, if a similar problem happens, i won't be able to stop anago from pushing the incomplete image sets.\nI'm reopening this unless we are really sure that it actually works...\nHow can we create the testing required to stop this from happening again?\nI'm doing a dry run today.\nanago could have some sort of sanity checking, though I'm not sure how we'd do that without maintaining hardcoded lists, as notes. I guess the \"real\" fix is having some automated testing somewhere to consume the non-x86 docker images.\nAfter anago dry run, i saw these images on my machine, so it looks fixed.\nThanks and :tada:!"}
{"_id":"doc-en-kubernetes-e1df151812a738ca8d7e9f97519d671c643638f3ca9d9400cc430f5f7a7d204c","title":"","text":"When user creates a static PV using folder name in path: on vSAN backed datastore then detach disk does not work because check fails - The code expects folder name to be UUID but it does not and hence it determines disk is not attached to the node and hence no detach is ever performed. /sig storage cc"}
{"_id":"doc-en-kubernetes-0282c95c8ddc11607ba516e065f4121126017f786957d03a031978171c244a62","title":"","text":"The replicas is 3 but just create 1 pod 'web-0' and one pvc, then kube-controller-manager panic happend. Should create 3 pod and 3 pvc add '--feature-gates=StatefulSetAutoDeletePVC=true' arg for kube-controller-manager, then create statefulset with 2 replicas and volumeClaimTemplates. No response <details</details<details</details<details</details<details</details<details</details<details</details\n/sig apps\nI can reproduce the issue and want to fix it. It seems like variables cause . /assign\n/triage accepted\nGreat find! Does this only happen on an upgraded cluster? I'll spin up a cluster and see if the e2e test catches this problem.\nThe e2e tests are still passing on a 1.24 cluster, so I suspect this is an upgrade problem. I wonder if there are other holes here.\nI installed the Kubernetes 1.25 cluster directly using and still reproduced this issue, so it was probably not caused by upgrading the cluster.\nInteresting. I thought that would set the policy to be non-nil. Let's keep this bug open after you submit your PR as I think something else is going on here---I don't think we should be seeing the null policy in the first place.\nI suspect that somewhere accidentally set it to nil, and I'll keep looking at this code.\nI tried the following (and also with parallel pod management), and it didn't seem to cause the panic. This was a 1.24 cluster, maybe this is only happening on head versions?\nI just tried this on a kind cluster running on k8s head, and the problem didn't seem to reproduce. But this was setting the feature flag default in I wonder----does the api server need to have the feature gate as well? That would explain why defaulting wasn't happening.\nAs far as I'm concerned,I just set the in kube-controller-manager manifests to reproduce this issue. I repeat the yaml used as follows: I tested that in statefulset yaml, setting or not would cause the component to panic. The panic logs are the same as above, so ignore them here."}
{"_id":"doc-en-kubernetes-bacc559e7a05833801486ab81cebdf65b8969406a4ba99bead3e0d4a0613defb","title":"","text":"statefulset with : The environment\nI believe you---this doesn't seem to be well-documented, but it's my understanding that you need to set the feature gate on all of the system components. Obviously failing to do so shouldn't cause a panic, so thank you again for finding & fixing this bug! Let me ask around to see what the expected behavior is. I know there may be skew between kubelets and master versions, but I didn't expect to have to deal with skew between master components.\nI read the code, and now I understand what you mean. This feature gate must be enabled in both the and components. We need to describe it in the to prevent someone from ignoring the component, which also needs to enable the feature gate.\nI don't think we need all the system components, like the component, right? I read the original KEP documentation. This feature gate only involves components and :\nI had a bit of discussion about this with others. Because upgrades don't occur atomically, any feature gate has to work regardless of how the other system components are configured (things shouldn't panic!). I suppose if the cluster administrator wants to track what needs to be enabled on each feature they can try to sequence which component has which feature gate. If it were me I'd try to make all the components consistent though.\nThis is certainly a method, but it is a bit more cumbersome for the administrator. It is best to have a place for the document to be queried by the administrator and avoid confusion. I don't think everyone is very familiar with the k8s mechanism. I suggest adding a partial description to this section of the : Before: After: I can submit this PR on .\nThis is going to be true of any api change. When you tried to use it with only adding the controller manager feature gate, I think you would be able to set the retain policy, but there wouldn't be any defaulting or validation (?).\nI've seen other feature gates that say what components are needed to enable it to use the new feature. For example: If it's not necessary, then I close .\nonly adding the controller manager feature gate, I think you would be able to set the retain policy, but there wouldn't be any defaulting or validation (?). I remember just enabling it in kube-controller-manager."}
{"_id":"doc-en-kubernetes-dc6539ce5e1daf74890724a45ab3f8f9d1791d28b0765a687405f604711fc05a","title":"","text":"it using the retain policy, but the statefulset does not have the policy from output of (It should be the apiserver authentication forbids it)\nYeah I think that makes sense. Thanks for finding this issue! After your PR with the panic fix is submitted, I think we can close this?\nYes. I agree. /close\nClosing this issue. <detailsIn response to : Yes. I agree. Instructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details"}
{"_id":"doc-en-kubernetes-4d10931e14d2065e9a15e6880eb555f03f2fd6b09352947daeddff0435b576bc","title":"","text":"Follow up from: See details there. When is , there is no need to discover the NUMA topology. /sig node /priority backlog /kind feature /help-wanted\n/cc\n/triage accepted\nI would like to take this up also can I get more brief information about it? code review comments from asking that inside the NewManager() implementation, we need to skip NewPolicyOptions and NewNUMAInfo but when the NumaTopology comes kubelet issue also comes. I need some more information to start with this!\n/cc\nWhen there is no topology manger policy specified, NewManager should be a no-op. Just skip all checks.\nExactly. Just to add: we may need few iterations over the change to get a clean flow in the call.\n/assign"}
{"_id":"doc-en-kubernetes-3e2b7d9568ee0bf267de4bcdac04137844677bca3ba3f5eceb5026951de8c86d","title":"","text":"Event for a pod: The cni plugins do exist at /opt/cni/bin. They cannot be executed and bash is complaining:\ncc\nI'm guessing this is a shared lib problem, since /usr/grte/v4/lib64/ld-linux-x86- doesn't exist on my gce vm. We should update docs to make it clear how someone would install manually (if they run into such problems again) and statically link (if the problem is indeed the shared lib) with something like Unsure how we'd support different GOOS (vend different tarballs vs checking in and building the plugin inline with our make scripts)\nuggh. These binaries need to be ultra-static or else we need to containerize them.\nEssentially doing would've prevented this bug :)\nAnd see for the CNI_ARGS problems\nalso , re ... however, I'm not too sure if your comment was mean to be on this ticket.\ncc\nI have a working version. I uploaded my cni binary with bug fix on GCS. change source to change sourcehash to Welcome to give it a try.\nAlso see problem:"}
{"_id":"doc-en-kubernetes-06e17cf980b6b4fe370857f0b2e0687536ee3b3b60ed1f78548a694def2f21f6","title":"","text":"Failed here /retest Originally posted by in This fails because the pod tries to bind port 80, that is a very demanded port, but there are two tests doing that, that increase the chance to collide\n/sig network\nClose when merges\n/triage accepted"}
{"_id":"doc-en-kubernetes-0882877f8aa87dd9f2fab7aa6c2163d630c073e98d0481f7c1762728faaae3b6","title":"","text":"Failed due to an apparently unrelated PR: https://travis-\nProblem here is that I left a small chance of a double-call in my parallelized cache. Didn't realize this was being tested. Will fix."}
{"_id":"doc-en-kubernetes-c0ae7c5a9b926980f7fbe5b2efcf245b830bbe6bb1b2519b4aac2b3a38b48ce6","title":"","text":"Golang 1.13 uses TLS 1.13 by default (opt-out), which introduces RSA-PSS. RSA-PSS does not work with keys that are too small. The issue was uncovered during Fedora mass rebuild which uses Golang 1.13: See /sig api-machinery\nWhat is the failure mode if a smaller key is used?\n/sig auth /sig scalability\n/sub"}
{"_id":"doc-en-kubernetes-4d7af6d1d5d1ba0c7ab59da9ea13925b54c282d46d48402ebb3f5409fd2b1204","title":"","text":"<!-- Please only use this template for submitting reports about failing tests in Kubernetes CI jobs --Which jobs are failing: kubernetes/node-problem-detector/node-problem-detector-ci: ci-npd-e2e-node See the Which test(s) are failing: [] NodeProblemDetector [NodeFeature:NodeProblemDetector] [] SystemLogMonitor should generate node condition and events for corresponding errors Since when has it been failing: Dec 03. was at 19:50 PST, was at 21:51 PST. Testgrid link: https://k8s- Reason for failure: Apparently is the culprit. Anything else we need to know: This is likely caused by some library upgrade. Nothing big.\n/sig node\n/assign\nSeems like the failure is caused by . /cc /cc /cc /cc\n/priority critical-urgent\nThe particular bug I care about has been fixed in . I'm gonna still work on submitting , as the same error should also be affecting a few other tests."}
{"_id":"doc-en-kubernetes-3936c52aacccf581fb5174635de31acc9a85b981ebc339b83016fc7b6087f92b","title":"","text":"PR introduced an annotation for fluentd migration from static pod to the daemonset to avoid two instances of fluentd on a single node during the transition stage. I didn't read the pr carefully, but based on the code at , every node object has one label be default: \"\". I think this is wrong. 1) Not everyone use fluentd as the logging agent. 2) It is hard-coded in Kubelet code as part of kubeletnodestatus. No matter running fluentd as a static pod or a daemonset on the node, Kubelet shouldn't treat it different from other pods. The right approach is introducing an external controller to apply this cloud-provider / kubernetes vendor (i.e. GKE) specific label to the node. cc/\nThe string \"fluentd\" shouldn't appear anywhere in our code. Please remove it ASAP.\nAbsolutely agree with you, thanks for bringing this up! Is it possible to apply labels to the node as part of the node startup script/salt?\nTo clarify, I'm thinking about possible ways to provide a fix for that ASAP. Writing external controller is a best solution, but here we have a release coming and this label is intended to be only a temporary measure to perform upgrade from using fluentd as static pods to using daemon set.\nThere are many ways to achieve this outside the core system. But I believe GKE already has an existing mechanism to add this kind of labels to the node, and the consistency is important for GKE management. cc/\nOK, thanks! However, I believe we should be sure not to break everyone's upgrade process, i.e. after master upgrade to 1.6 on any platform, there should be only one fluentd agent on each node. I propose changing the configure scripts and salt to set up needed labels in kubelet's flag (). WDYT?\nIt seems to me like upgrades could survive without the explicit dependency on fluentd: I'm not sure if upgrades call initialNode or not, my own ignorance ? If they do, and this is removed, can't an external controller just do the work of re--adding that label if its dropped?\nthanks for raising this. Let's sync f2f\ncc\nWDYT about exposing kubelet version as a label and have a selector in fluentd ds config targeting version 1.6+?"}
{"_id":"doc-en-kubernetes-4cb16c5ce2a173a3021e859767a51a1df9544dd94429e93c357c1a133ef72828","title":"","text":"and I talked about this issue a bit for migrating kube-proxy. One idea we came up with was that of running a controller that will identify kubelet versions and apply taints to nodes running static pods. The daemonset version would not tolerate that taint and that way we keep daemon set pods off of nodes running static pods. Have you considered that approach?\nIt doesn't differ from a controller applying labels. I'd like to avoid necessity of introducing such controller, since it's not clear what is it's responsibility (generic controller for applying labels on nodes based on some rules? a custom controller just for fluentd purposes). In GKE we can do it on GKE control plane side, but we need a story for open source.\nMove applying labels to startup scripts/salt seems to be the least intrusive and easiest to implement way.\nActually what I proposed in is not an option since nodeSelector in PodSpec is old style selector.\nI'm not a big fan of approach with salt, but it seems there is no better option.\nRe: Instead of exposing Kubelet version as a label, why not using field selector. Kubelet version is part of NodeStatus already. I just simply don't want to have a duplicate info for the same object.\nis it possible to schedule based on fieldselector? I'm only aware of using field selector while listing objects.\nAlso node-labels option is in alpha\nBut it's present since 1.2 What is the status of this flag? Can we use it for fluentd purposes or better avoid this?\nMaybe it does make sense to implement a temporary ad-hoc controller? It's the easiest to deploy solution, just another in fluentd addon.\nI know. It totally doesn't make sense to introduce anything custom. I'll use just the flag. It's rather hacky but unfortunately I can't see anything better.\nCool thing about controller is that you can deploy it into any cluster and it will work, with CA also. You don't have to set up your node labels manually then or restart the cluster.\nThere is a number of open question with having a new controller: it run on the master or a node? we guarantee that it will be always running? it have any resource requirements? I'd like to avoid discussing it right before the release.\nHaving a controller will help with kube-proxy transition too in the future."}
{"_id":"doc-en-kubernetes-3203f30994cf42c5336d5a6d0dc1585171cc98cff1719959478ccaf13036f1c4","title":"","text":", Piotr Szczesniak <:\nis this a 1.6 release blocker?\nI think yes. The PR is already sent and approved. I need to fix e2e tests because it doesn't work well with the test infra (e2e tests passes locally). EDIT: I'll do it hopefully today.\nThe PR is ready to be merged and lgtmed. Waiting for approvals."}
{"_id":"doc-en-kubernetes-518f837785f17a5fd9706a6c0a0614adac703d65f6bb6045d79a025795dcb97b","title":"","text":"Files storing latest versions for minors are not updated since infix was to the version string. Example Similar issues can be found in 1.25 and 1.24 builds (last 1.23 build infix so it has not happened there yet): I expect the files to be overwritten with newer version of the build. Manually run build job for one of the minors. No response Not related. Not related. No response No response No response No response\nI believe this may be also responsible for breaking kubemark tests: since they fail due to version mismatch\n/sig release\nThe label(s) cannot be applied, because the repository doesn't have them. <detailsIn response to : /priority high Instructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\nThanks for raising the issue! /cc who is working on before\nI assume this will be fixed automatically with the next patches, because then we would compare with . Anyway, I think we should fix that intermediately by manually updating the version markers to match their correct version. I did that now for the bucket\nWe had to update the versions markers again, because those were pointing to two commits earlier and didn't include the Go updates. The latest version markers are: v1.24.12-35+ v1.25.8-37+ v1.26.3-41+\nlatest-1.27 has been affected by this as well. I updated it manually to v1.27.0-9+. See this thread for more details:\n/triage accepted\nThis should be fixed now, closing it. Feel free to reopen if we missed anything.\nThis issue is still present, see this Slack thread for more details: /reopen\nReopened this issue. <detailsIn response to : This issue is still present, see this Slack thread for more details: /reopen Instructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\nWe have a similar situation with latest-1.28 which is currently pointing to: . This is one commit before the release commit: ! It's mostly like that the release step didn't update version markers, we should look into this."}
{"_id":"doc-en-kubernetes-5a00e106d863b6b2d6ffe122805a64f76a2709c3a46b9a75024b175401623671","title":"","text":"As of now, I'm going to manually update this version marker to unblock 1.28.1 patch release. I'll also take a look into this. /assign\nThe PR that fixed this issue got reverted so this is still relevant: /reopen\nReopened this issue. <detailsIn response to : The PR that fixed this issue got reverted so this is still relevant: /reopen Instructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\nWe're running into this issue again with stable-1.29 marker. It's currently pointing to which is It should instead point to I'm going to manually change it following the same procedure as the last time. Version marker updated to\nThis issue is labeled with but has not been updated in over 90 days, and should be re-triaged. Important-soon issues must be staffed and worked on either currently, or very soon, ideally in time for the next release. You can: Confirm that this issue is still relevant with (org members only) Deprioritize it with or Close this issue with For more details on the triage process, see /remove-triage accepted\n/triage accepted\nSame issue for 1.30, where I had to update the version marker manually again."}
{"_id":"doc-en-kubernetes-33260ec64a7d900f1342ac27e4dcabb354b23351ec3e8ae604c63a04e69f42d4","title":"","text":"is a bash script linter which could help catch common problems in our (numerous) bash scripts. I know we'd like to reduce our use of bash scripts, but until then I think being stricter about script quality would be helpful. /cc\ncc you were talking about this earlier. It has to only check modified lines.\nFalse positives are a problem. For instance, not being able to follow a call through an environment variable causes an error. I would be much more comfortable with a non-blocking way to point out these issues, and leave to it human judgement.\nYeah, I've been interested in doing this for a while, but can be pretty persnickety and I haven't yet taken the time to configure it to be less noisy (or figure out if it can be). It's good, though.\nYeah, it would definitely need to be configured to ignore a lot of types of errors, but I still think it could be used to check for some types of errors. We could start with a pretty minimal set.\nI would settle for something that just produced a quick-fix format artifact in the GCS bucket, and then the PR munger reading from it and commenting on the PR. We could re-use that for other linters too.\nhas been pretty accurate historically.\nshellcheck, not spellcheck\nreading too many issues in a on\n/sig testing /priority awaiting-more-evidence Might be more noise than it's worth, but we do pylint our .py scripts now\nIssues go stale after 90d of inactivity. Mark the issue as fresh with . Stale issues rot after an additional 30d of inactivity and eventually close. Prevent issues from auto-closing with an comment. If this issue is safe to close now please do so with . Send feedback to sig-testing, kubernetes/test-infra and/or . /lifecycle stale\nStale issues rot after 30d of inactivity. Mark the issue as fresh with . Rotten issues close after an additional 30d of inactivity. If this issue is safe to close now please do so with . Send feedback to sig-testing, kubernetes/test-infra and/or . /lifecycle rotten /remove-lifecycle stale\nRotten issues close after 30d of inactivity. Reopen the issue with . Mark the issue as fresh with ."}
{"_id":"doc-en-kubernetes-7c2396e8c525e9f1f7bfa4cd4f910bc5e5cdd53da779129188a0590e81ff81c5","title":"","text":"Send feedback to sig-testing, kubernetes/test-infra and/or . /close\nI'd like to reopen this, and eventually implement (first we need to deal with the failures ...) I've easily found a lot of bugs in test-infra and kubernetes/kubernetes bash with shellcheck. See also /assign\n/reopen /remove-lifecycle rotten /priority important-longterm\nReopening this issue. <detailsIn response to : /reopen /remove-lifecycle rotten /priority important-longterm Instructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\nWe also have non-blocking and optionally-run presubmits, even just having an easy way for reviewers to request a shellcheck result for changed files seems like something we should do. I also think we should fix the existing failures. Glancing through some k/k bash with shellcheck in my editor, these generally seem to be things that are actually wrong.\nIssues go stale after 90d of inactivity. Mark the issue as fresh with . Stale issues rot after an additional 30d of inactivity and eventually close. If this issue is safe to close now please do so with . Send feedback to sig-testing, kubernetes/test-infra and/or . /lifecycle stale\nStale issues rot after 30d of inactivity. Mark the issue as fresh with . Rotten issues close after an additional 30d of inactivity. If this issue is safe to close now please do so with . Send feedback to sig-testing, kubernetes/test-infra and/or . /lifecycle rotten\n/remove-lifecycle rotten coming back to this hopefully\nThis is done now. Only took two years :^)"}
{"_id":"doc-en-kubernetes-73496a383d34eb1ab5962a20be4dce66952c6d20058c4ff89ce94afc1cb5683d","title":"","text":"We are getting the following error when the apiserver connects the webhook on localhost (configured via URL). We should only use the SSL tunnel for the connections to nodes when the webhooks are running as services. /sig api-machinery /kind bug /assign\n/close"}
{"_id":"doc-en-kubernetes-c9c1c743d3f3434b5b176b4daffd6034698e7589ba7079b19c7793ebc64267f8","title":"","text":"Kubernetes e2e suite. Kubernetes e2e suite.[It] [sig-autoscaling] [Feature:HPA] Horizontal pod autoscaling (scale resource: CPU) [Serial] [Slow] ReplicationController Should scale from 1 pod to 3 pods and then from 3 pods to 5 pods and verify decision stability [sig-autoscaling, Feature:HPA, Serial, Slow] Kubernetes e2e suite.[It] [sig-autoscaling] [Feature:HPA] Horizontal pod autoscaling (scale resource: CPU) [Serial] [Slow] ReplicationController Should scale from 5 pods to 3 pods and then from 3 pods to 1 pod and verify decision stability [sig-autoscaling, Feature:HPA, Serial, Slow] Kubernetes e2e suite.[It] [sig-autoscaling] [Feature:HPA] [Serial] [Slow] Horizontal pod autoscaling (non-default behavior) with autoscaling disabled shouldn't scale down [sig-autoscaling, Feature:HPA, Serial, Slow] Kubernetes e2e suite.[It] [sig-autoscaling] [Feature:HPA] [Serial] [Slow] Horizontal pod autoscaling (non-default behavior) with autoscaling disabled shouldn't scale up [sig-autoscaling, Feature:HPA, Serial, Slow] Kubernetes e2e suite.[It] [sig-autoscaling] [Feature:HPA] [Serial] [Slow] Horizontal pod autoscaling (non-default behavior) with both scale up and down controls configured should keep recommendation within the range with stabilization window and pod limit rate [sig-autoscaling, Feature:HPA, Serial, Slow] Kubernetes e2e suite."}
{"_id":"doc-en-kubernetes-50a764361af76177e3b0fd7754c28175d0d0117b957f7afce14b1a8f0955f9e1","title":"","text":"[It] [sig-autoscaling] [Feature:HPA] [Serial] [Slow] Horizontal pod autoscaling (non-default behavior) with both scale up and down controls configured should keep recommendation within the range over two stabilization windows [sig-autoscaling, Feature:HPA, Serial, Slow] 11/3/2023 <*fmt.wrapError | 0xc000812080: client rate limiter Wait returned an error: context deadline exceeded { msg: \"client rate limiter Wait returned an error: context deadline exceeded\", err: <context.deadlineExceededError{}, } looks similiar to /sig autoscaling /sig windows\nThis issue is currently awaiting triage. If a SIG or subproject determines this is a relevant issue, they will accept it by applying the label and provide further guidance. The label can be by org members by writing in a comment. <detailsInstructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details"}
{"_id":"doc-en-kubernetes-1c37f3b13ef66319b53623e24b4fece0550cb85e7e62ca9b694924c6c433e0c8","title":"","text":"/kind bug What happened: does not show the description for the new field in What you expected to happen: should describe the new field. How to reproduce it (as minimally and precisely as possible): Environment: Kubernetes version (use ): master\n/sig auth\nGitHub didn't allow me to assign the following users: fanzhangio. Note that only and repo collaborators can be assigned. <detailsIn response to : /assign Instructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\nIf nobody has started yet, I wanna to pick it up.\nwas working on it AFAIR.\nIndeed. if you don't mind, I'd take over."}
{"_id":"doc-en-kubernetes-65b87d6ab9dc4eafe4d20be92d7fc50e93b7f8fc56cfe96572cf225758153ef7","title":"","text":"Is this a BUG REPORT or FEATURE REQUEST? (choose one): BUG <!-- If this is a BUG REPORT, please: Fill in as much of the template below as you can. If you leave out information, we can't help you as well. If this is a FEATURE REQUEST, please: Describe in detail the feature/behavior/change you'd like to see. In both cases, be ready for followup questions, and please respond in a timely manner. If we can't reproduce a bug or think a feature already exists, we might close your issue. If we're wrong, PLEASE feel free to reopen it and explain why. --Kubernetes version (use ): Client Version: version.Info{Major:\"1\", Minor:\"4\", GitVersion:\"v1.4.5\", GitCommit:\"\", GitTreeState:\"clean\", BuildDate:\"2016-10-29T01:38:40Z\", GoVersion:\"go1.6.3\", Compiler:\"gc\", Platform:\"linux/amd64\"} Server Version: version.Info{Major:\"1\", Minor:\"4\", GitVersion:\"v1.4.5\", GitCommit:\"\", GitTreeState:\"clean\", BuildDate:\"2016-10-29T01:32:42Z\", GoVersion:\"go1.6.3\", Compiler:\"gc\", Platform:\"linux/amd64\"} Environment: Cloud provider or hardware configuration: vSphere OS (e.g. from /etc/os-release):Kernel (e.g. ):Install tools: None Others: None What happened: When I create a deployment with vSphere cloud provider dynamic provisioning, it creates a volume spec with volume path - [datastore] kubevols/volume which contains a white space in it. In this case the cloud provider queries the getDeviceNameFromMount() to return the path of the volume mounted. Since getDeviceNameFromMount() queries the filesystem to get the mount references, it returns a volume path \"[datastore]040kubevols/volume\". Later the kubelet searches for this volume path in both the actual and desired states. Th actual and desired states contains volume with path \"[datastore] kubevols/volume\". So, it couldn't find such volume path and therefore kubernetes stalls unable to make any progress further similar to one described in ."}
{"_id":"doc-en-kubernetes-4472e5c92e91fa48fbd9ae4b2faf4fe2f973ca70781a7072044a8b4998a8c4c5","title":"","text":"Also similar issue described in What you expected to happen: All the pods in the recreated deployment comes up. How to reproduce it (as minimally and precisely as possible): See above. Anything else do we need to know: None\ncc"}
{"_id":"doc-en-kubernetes-856660a825f3ffe3a03d6e2754ae87c832572ac27681226648a4feba9d3ff6d0","title":"","text":"I tried to create a kubernetes deployment with vSphere volume with volume path \"[datastore] kubevols/redis-master\". In this case the cloud provider queries the getDeviceNameFromMount() to return the path of the volume mounted. Since getDeviceNameFromMount() queries the filesystem to get the mount references, it returns a volume path \"[datastore]040kubevols/redis-master\". Later the kubelet searches for this volume path in both the actual and desired states. Th actual and desired states contains volume with path \"[datastore] kubevols/redis-master\". So, it couldn't find such volume path and therefore kubernetes stalls unable to make any progress further similar to one described in . However, when I create a kubernetes deployment with volume path - \"[datastore]kubevols/redis-master\" which doesn't contain any spaces, it works perfectly fine. So, the basic problem is to format and return the correct volume path from getDeviceNameFromMount() function. Alternatively, we can format it properly in the cloud provider code but this has to be done individually in all the cloud provider code base. Inorder to prevent it, it would be better to fix the getDeviceNameFromMount() function to return the correct formatted volume path.\nCC Kubernetes should handle volumes with spaces (looks like vsphere is the first such volume) gracefully. Non a blocker issue for 1.5.\nyou should make sure that the 1.5 release notes capture that \"vsphere volumes w/spaces have problems\". Known issue accumulator in\ngetDeviceNameFromMount is function which queries /proc/mount to get mount points. 040 - is an escape sequence. 040 is the ascii code, in octal(base 8) number system, of the space. The mount command interprets 040 as literal space in the mount point. : Can't we provide a fix for this in 1.5 release itself? There a 2 options: can provide a fix in getDeviceNameFromMount. can provide a fix to vSphere cloud provider to fix it after the call to getDeviceNameFromMount. For option 1, we have to make sure it works for all cloud providers where as for option 2 its not needed. Let me know which option you think is feasible. I am ok with anything."}
{"_id":"doc-en-kubernetes-f0c75eb090ebda051cbb8d4dcd548ddf167d040dce30f4ff6f2c05a5971aa5a4","title":"","text":"1 is the right answer. But we're very close to 1.5.0 release (<1 week). I'd also like whoever addresses this to ensure the entire system works end-to-end for volumes with spaces, not just patch one little corner. I'm afraid it would be a large change and we're not going to have enough time to validate it in 1.5. You can certainly start working on it. If you get a PR in time, we can consider it for 1.5.0. Otherwise it'll have to go in a patch release."}
{"_id":"doc-en-kubernetes-9deab690131da58e4566d26be974a26500a50c4525ef4b542cd9399a25bc940c","title":"","text":"[x] Test PR : [x] Two weeks soak end date : 6 April 2022 [X] Test promotion PR : According to this APIsnoop query, there are still some remaining Job status endpoints which are untested. There are conformance tests for CronJob status endpoints already the existing as a template, a new for job test has been created. The e2e logs for this test are listed below. This query shows the endpoints hit within a short period of running the e2e test If a test with these calls gets merged, test coverage will go up by 3 points This test is also created with the goal of conformance promotion. /sig testing /sig architecture /area conformance\n/triage accepted\n/reopen\nReopened this issue. <detailsIn response to : /reopen Instructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details"}
{"_id":"doc-en-kubernetes-c6679bcd8f59ea95eb0b6a4e2bf394545d09b32db20c61705567c0727bbb7817","title":"","text":"It seems to be valid to add here. Would you like to open a pull request for it? this is like a cleanup, as the set insert will skip and the validation will do twice. is used to track bugs. I suggest you open a PR for that if you like.\n/sig storage\nI think adding ` == true and we didn't skip, it will insert set twice I believe? wasn't notice that at first glance. However, I can file a PR now\n/triage accepted\nThe label cannot be applied. Only GitHub organization members can add the label. <detailsIn response to : /triage accepted Instructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\n/triage accepted\nPlease feel free to submit a PR."}
{"_id":"doc-en-kubernetes-638d244f4d44257351ae4786f705d88f47a0e12c74ce71a459164cb07c9c1943","title":"","text":"<!-- This form is for bug reports and feature requests ONLY! If you're looking for help check and the . If the matter is security related, please disclose it privately via --Is this a BUG REPORT or FEATURE REQUEST?: /kind bug What happened: kube-cm kube-scheduler does not exit when kube-apiserver suspended What you expected to happen: kube-cm kube-scheduler should exit after LeaderElector stops leading How to reproduce it (as minimally and precisely as possible): Anything else we need to know?: Environment: Kubernetes version (use ): Cloud provider or hardware configuration: OS (e.g. from /etc/os-release): Kernel (e.g. ): Install tools: Others:\n/sig api-machinery /kind bug\n/assign\n/cc"}
{"_id":"doc-en-kubernetes-2c9ef8984ebaf899659836b925069334113c910fc47ebe0516a7d659674f044a","title":"","text":"xref In 1.16: change the metadata-proxy addon to use change the kube-proxy addon to use change the masq-agent addon to use stop setting , , and in kube-up\nHey I'm trying to start to contribute to kubernetes and this seem to be an easy task. Can I help with this one?"}
{"_id":"doc-en-kubernetes-3fbd6fe4c52deabe3c900459c29c36a04807fdd1453263d11d2244d21f4a4f4a","title":"","text":"Public IPs are expensive resources. Given that we're running scalability tests on 5k-node clusters, we are unnecessarily consuming 5k IPs which is a lot. We should switch our tests so that nodes have only private IP addresses. We are going to do that by configuring NAT so that nodes can talk to outside world (e.g. to download image from dockerhub). But that also requires changing some correctness tests which assume public IPs.\nTo summarize recent work: setting up private gce clusters with nat, feature is gated via the env test and dashboard with correctness tests on gce private cluster - https://k8s- 27 tests fail on fixing the tests - should fix majority of them\nIssues go stale after 90d of inactivity. Mark the issue as fresh with . Stale issues rot after an additional 30d of inactivity and eventually close. If this issue is safe to close now please do so with . Send feedback to sig-testing, kubernetes/test-infra and/or . /lifecycle stale\n/remove-lifecycle stale /lifecycle frozen\n- is there anything missing to be done here?\nNo, we can close this one. It was done long time ago. Thanks for catching /close\nClosing this issue. <detailsIn response to : No, we can close this one. It was done long time ago. Thanks for catching Instructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details"}
{"_id":"doc-en-kubernetes-2d0ebe90b3c755e0edf33c6811bf431b4c51fd00f4f7b6a1d49649dd59cb487b","title":"","text":"<!-- Please use this template while reporting a bug and provide as much info as possible. Not doing so may result in your bug not being addressed in a timely manner. Thanks!--What happened: Used the endpoint /BumpMetric to create fake metrics for Prometheus. When I tried to curl /metrics I got the following response: Bad request. Not a POST request What you expected to happen: I expected to see something like this: # HELP info message. # TYPE gauge 0. How to reproduce it (as minimally and precisely as possible): Following the docs, after having it running inside kubernetes: Anything else we need to know?: Looks like introduced this by mistake. Environment: Kubernetes version (use ): Cloud provider or hardware configuration: OS (e.g. from /etc/os-release): Kernel (e.g. ): Install tools: Others: <!-- DO NOT EDIT BELOW THIS LINE --/kind bug\n/sig testing"}
{"_id":"doc-en-kubernetes-56027fa26048aea64718f2bd7d7c5d6f58ddca1472c1fc738c03141b6dea3ad6","title":"","text":"Currently it assumes it can read ContainerManifest - it should try both\nWhy not hold on this until we have Kubelet-read-Pod in?\nAlso the HTTP source has the same problem.\nIt should read Pod. Changed the title.\nIdeally, I'd like all Pods created directly via Kubelet to be to etcd and surfaced via the API. If updates are made via the API, I'd like those to propagate back to the node. We can discuss whether the config file should be updated or ignored. At some point, I'd like Kubelet to persistently cache pods received from the apiserver, perhaps by just writing them all out as config files.\nis the issue to sync local pods back to apiserver.\nthis is somewhat related to other issues we talked about.\nDetail: We want locally specified pods to be able to use YAML config files, but we also want to rip YAML out of the API server . The config-file behavior should be as similar to the kubectl experience. Kubelet could use parts of the kubectl library to translate YAML to JSON. Vision: Eventually, I'd like Kubelet to use the apiserver library and support exactly the same Pod API as the master, at which time I'd like kubectl be able to create pods directly on Kubelets, as well as reading from config files. I also expect to have a per-node controller . I'd like to be able to use the per-node controller to update pods started via config files (or direct kubectl-kubelet pods), overriding those original files. For availability and self-hosting, we'll eventually need local checkpointing , ideally just using serialized JSON for each object. If Kubelet used apiserver, this could be done using a custom registry implementation.\nI think that is a prerequisite for it.\nIf noone picks it up, I'm going to start working on it tomorrow.\nthanks, wojtek-t\nfixed it for http channel"}
{"_id":"doc-en-kubernetes-55af5a538910186290479e8a37adc800d26766e6d09583162f76a512f0433ab7","title":"","text":"<!-- Please only use this template for submitting enhancement requests --What would you like to be : Actually, the parameter depends to all namespaces in the cluster. Filtering is only possible with labels. When running the should only depend on the objects defined in . Why is this needed: A typical use-case is having one repository and namespace per project. When working with declarative objects it should be possible to remove objects and get them pruned like having them described in the respository. This would also help when renaming objects within the repository - which actually needs a manual on the host for the old objects. This even could be a better solution for the problem described in\nReiterating the mentions to trigger a notification: <detailsIn response to : Instructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\n/sig api-machinery\ncc\n/assign\nGitHub didn't allow me to assign the following users: Liujingfang1. Note that only and repo collaborators can be assigned and that issues/PRs can only have 10 assignees at the same time. For more information please see <detailsIn response to : /assign Instructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\nWe have been aware of the issues in prune and we want to improve it. Recently we opened a for improving pruning in . With the proposed solution, the problem from current will be fixed.\nIssues go stale after 90d of inactivity. Mark the issue as fresh with . Stale issues rot after an additional 30d of inactivity and eventually close. If this issue is safe to close now please do so with . Send feedback to sig-testing, kubernetes/test-infra and/or . /lifecycle stale\n/remove-lifecycle stale\nIssues go stale after 90d of inactivity. Mark the issue as fresh with . Stale issues rot after an additional 30d of inactivity and eventually close. If this issue is safe to close now please do so with . Send feedback to sig-testing, kubernetes/test-infra and/or ."}
{"_id":"doc-en-kubernetes-cb54288fa32d0be96e99fcb10abda15ff9759ea2461e12485115b7e51b53055d","title":"","text":"/lifecycle stale\n/remove-lifecycle stale\nI just deleted a bunch of things in another team's namespace (:blush:). I'd would love to have this feature. I had been using dry-run in another context, but then forgot to check it again after switching contexts.\nIs there any workaround for this? I want to prune resources within a single namespace only.\nWe're still working on improving prune, the current mechanism is broken enough that we don't recommend people us it.\nAlso look like it's been fixed recently.\nCan someone confirm if this is implemented? I tried using this feature with kubectl 1.18.2, which has , yet as with previous versions kubectl it returned an error about pruning a nonNamespaced object of kind Namespace.\nI assume the problem is that \"Namespace\" is a non-namespaced resource, and therefore is not filtered by namespace, so gets deleted by \"prune\". As a workaround it's possible to manually pass a (whitelist of resources allowed to be deleted), and not include \"Namespace\" in it: The list of resources is taken from .\nWould wrapping this loop with achieve the same result?"}
{"_id":"doc-en-kubernetes-3bd2c8928ffbfbb644dae4df3d9fa0aa73d567a5bc560c59938e22bd3dd510ca","title":"","text":"Which jobs are failing: pull-kubernetes-e2e-gce Which test(s) are failing: [sig-network] Services should have session affinity work for NodePort service [LinuxOnly] [sig-network] Services should be able to switch session affinity for service with type clusterIP [LinuxOnly] [sig-network] Services should be able to switch session affinity for NodePort service [LinuxOnly] [sig-network] Services should have session affinity work for service with type clusterIP [LinuxOnly] Since when has it been failing: At least since 9/30 (). But that's almost at the end of what's currently visible on the testgrid, so maybe longer Testgrid link: Not sure what I'm supposed to put here. ? Reason for failure: The test case connects to a service every 2 seconds for 1 minute and checks if it hit the same endpoint at least 15 times in a row. But the individual checks are taking unexpectedly long and so sometimes the test doesn't even manage to run 15 times total in 1 minute."}
{"_id":"doc-en-kubernetes-be859aec3019c8f94fa574b44096bac92d128f839b0ed157072d5b09f5ea53f4","title":"","text":"eg: Oct 8 13:32:01.402: INFO: Running ' --server=https://34.82.143.14 -- exec --namespace=services-1060 execpod-affinityklpfn -- /bin/sh -x -c curl -q -s --connect-timeout 2 http://10.40.0.5:/' Oct 8 13:32:07.627: INFO: stderr: \"+ curl -q -s --connect-timeout 2 http://10.40.0.5:/n\" Oct 8 13:32:07.627: INFO: stdout: \"affinity-nodeport-2z9hv\" Oct 8 13:32:07.627: INFO: Received response from host: affinity-nodeport-2z9hv Oct 8 13:32:09.402: INFO: Running ' --server=https://34.82.143.14 -- exec --namespace=services-1060 execpod-affinityklpfn -- /bin/sh -x -c curl -q -s --connect-timeout 2 http://10.40.0.5:/' Oct 8 13:32:16.027: INFO: stderr: \"+ curl -q -s --connect-timeout 2 http://10.40.0.5:/n\" Oct 8 13:32:16.027: INFO: stdout: \"affinity-nodeport-2z9hv\" Oct 8 13:32:16.027: INFO: Received response from host: affinity-nodeport-2z9hv Note the gap between the \"Running\" line and the \"stderr\" line. It's not clear if the slowness is in connecting to the execpod or in connecting from the execpod to the affinity service. It is also not particularly consistent between attempts. (Here it was 6s, 1s, 2s, 3s, 5s, 7s, 7s, 3s, 4s, 6s, 7s, 8s.) Anything else we need to know: It looks the checks are also slow in successful runs, it's just that they're not quite as slow, so they pass."}
{"_id":"doc-en-kubernetes-0f765c22d0d19db3b6df2fd444c9da8d7f7f8af548830f46d73b17c37e323eed","title":"","text":"I'm not sure if the problem is \"something is making the cluster slow and so this totally valid test is taking too long and failing\" or \"e2e-gce has grown over time and now cluster load has reached the point where this formerly-passing test now has unreasonable performance expectations\". /sig network"}
{"_id":"doc-en-kubernetes-2b4d325d4814fe81f2a99615b844cff823940de918e3b360e7f108b94b4fe81b","title":"","text":"/kind bug What happened: walteraa [6:55 AM] Hi, The controller manager of the federation using Google cloud DNS is crashing, the log showed is that: the hyperkube command is that: I have created this in the Google DNS Someone knows why the DNS zone query is getting the 404 error?? :disappointed: What you expected to happen: It should have created the missing domain. How to reproduce it (as minimally and precisely as possible): As per above. Anything else we need to know?: Environment: Kubernetes version (use ): 1.7.4 Cloud provider or hardware configuration**: OpenStack (irrelevant) OS (e.g. from /etc/os-release): irrelevant Kernel (e.g. ): irrelevant Install tools: irrelevant Others: irrelevant\nThe bug is related to this: The above code assumes that the provider code returns no error but an empty list if the domain does not exist, but instead the provider returns a 404. Need to fix the provider and/or the above code.\n/sig federation\nI am unlikely to have time to work on this soon, so if anyone else wants to, go right ahead.\n/assign\nAn additional information: I have created a which is based on the credential JSON file created through instruction. After that, I have configured the Federation controller manager to mount this secret as a volume in the container and set the environment variable to the JSON file inside this volume. Before doing that, I was getting an authentication error so the google DNS provider couldn't be instantiated, and after doing the instructions that I mentioned before, this authentication error was solved but now I'm getting 404 error when Google DNS provider tries access the specified managed zone. The file that I have generated follow this structure:"}
{"_id":"doc-en-kubernetes-ddb9bacde18c9f97d2ebd4306efde2d358d8075e932023fee86edf729006549a","title":"","text":"Which jobs are failing: Which test(s) are failing: Since when has it been failing: Testgrid link: Reason for failure: Anything else we need to know: /sig api-machinery /assign\n/cc\nwant to take a look at the audit event issue first? I can dig in after that if issues remain.\nyes, I'm looking at the audit event issue now\nactually the failure message suggests that the missing audit log is caused by unexpected admission invocation: I think it's unrelated to auditing Note that there are two tests named \"no reinvocation of webhook B when in-tree or prior webhook mutations\" :( and from the audit log (), it also suggests that the test is hitting a wrong endpoint\nThanks I'll dig in."}
{"_id":"doc-en-kubernetes-5266c57682848ccf4bbfd7d438a66f7fa505006f21647746bff2e69ab7956c41","title":"","text":"In cluster/saltbase/salt/kube-proxy/kube-proxy.manifest, we have this: But kube-proxy also has a flag. Is there a reason to use the approach instead of using the flag? cc as you wrote the code, in\n/cc who's working on moving kube-proxy to a DaemonSet\nIn general, oomscoreadj is not expected to be directly updated by pods and kubelet is expected to manage it. If kube-proxy runs in the Guaranteed QoS class, then it's OOM score adj will be set to a low number. Until that happens, kube-proxy process may want to be shielded from OOM kills to an extent (this matters only when kube-proxy's memory usage memory request) Meanwhile, it does not make sense to have two means to configure the same thing. I'll let decide which option he prefers and sunset the other option.\nThanks for clarifying. At this point, I prefer to use the built-in flag instead of relying on composing bash commands. Sent for this."}
{"_id":"doc-en-kubernetes-486ca0159276eedd0d2aa336961ee5fd6fb83d0ac97af7f3f9c8c863d43fcef2","title":"","text":"What would you like to be : Periodically remove cleanupInfos for old containers from dockershim's map. One way of doing that would be to just forget about s older than say an hour, as they should always be short-lived. Why is this needed: introduced a new in-memory map for the dockershim, that could potentially (in pathological cases) cause memory leaks - for containers that use GMSA cred specs, get created successfully, but then never get started nor removed. This might be okay for the alpha release of the GMSA feature, but should be addressed in the beta. See\n/assign wk8\n/sig windows\n/milestone v1.15\ngot moved out. let's move this issue with it as well /milestone v1.16\nHello! I'm bug triage lead for the 1.16 release cycle and considering this issue is tagged for 1.16, but not updated for a long time, I'd like to check its status. The code freeze is starting on August 29th (about 1.5 weeks from now), which means that there should be a PR ready (and merged) until then. Do we still target this issue to be fixed for 1.16?\n: yup we'd love to have this in 1.16, still waiting for kind reviewers on :) ..."}
{"_id":"doc-en-kubernetes-d047193f0eb4348dad25266c8f06cc7d0100df3011e6281006c74e70d78d442e","title":"","text":"<!-- Please only use this template for submitting reports about flaky tests or jobs (pass or fail with no underlying change in code) in Kubernetes CI --pull-kubernetes-unit : (PR with no go code changes, should not have failed any unit tests) links to appreciated links to specific failures in spyglass appreciated <!-- Please see the deflaking doc () for more guidance! --/sig storage\nThis issue is currently awaiting triage. If a SIG or subproject determines this is a relevant issue, they will accept it by applying the label and provide further guidance. The label can be by org members by writing in a comment. <detailsInstructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\n/assigne\n/assign\n/assign tried to fix it in"}
{"_id":"doc-en-kubernetes-2e8b5198754cc6270240a247dfa503fa4a3eeae2d5481e425503db2c2c68474d","title":"","text":"Here is a list of TODOs that could be fixed for Ginkgo migration from v1 to v2. [x] Lost the validation on the SIGDescribe: [x] Ginkgo V2 currently cannot get the number of specs that will run in , this might be fixed in the Ginkgo's minor release of V2 in the future. [x] <delGinkgo dry-run mode has dropped the test spec in V2 (), consider to switch back to use to generate test spec when the feature is ready in . </del--- This has been fixed in Ginkgo 2.1.4 [x] <delMigrate all the custom reporter in , should be lightweight and not sure where those custom reporters are used. </delFixed here: [x] Replace all the camel case flags to kebab case flags, e.g. to . [x] <delRemove the legacy e2e test binary's parameter and in 1.28, those parameters were used to manage report, this is no longer needed for Ginkgo v2. </delno longer needed. [x] upgrade vendor when is migrated to ginkgo v2, this will help to remove the dependency of - [x] <delrevisit to explore the possibility to cover the test in Ginkgo V2. </delfixed by [x] remove the workaround introduced in Ginkgo V1 ( is not called after hitting ^c): partially addressed by: , some cleanup needed for each sig before removing the hook entirely. I think has the pr to fix this entirely, but I cannot get the link for that, mark it as resolved. : [x] New work item (?): capture only a single output text output stream () in JUnit (this issue was resolved along with another one in ) [ ] Due to the junit format change, some tool which use to grab \"value\" directly cannot get the failure reason as what it does in v1. [x] Test failures are missing information since GinkgoV2 migration [ ] is clustering stack traces, not failure text () Address those leftovers if there is any update from Ginkgo in the future. see the relevant link on the each of the issues."}
{"_id":"doc-en-kubernetes-c4753cc2146447a07d108ff02ffe6d67b2be2a6ba7adffaf5026d0a94c1c787a","title":"","text":"No response_ <details</details<detailsall </details<details</details<details</details<details</details<details</details\nThis issue is currently awaiting triage. If a SIG or subproject determines this is a relevant issue, they will accept it by applying the label and provide further guidance. The label can be by org members by writing in a comment. <detailsInstructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\n/sig testing\n/cc\n/cc\nhello! I would like to work on this issues\nI think we cannot do anything now except the last two, and those two should be based on top of the:\nI would like to work on task 5 of this issue\nno problem, pls wait for an while until that PR got merged, then you can create a pr to fix those deprecated parameters.\nRegarding : searching for points out the places where the workaround is used and needs to be replaced once is solved.\nI just noticed this comment in\nseems like is not used in k/k, but in other sigs.\nhmm... I just grepped the test code, will push a pr to fix this.\n/open\n/re-open\nokay, this is not resolved actually, but I have no way to re-open it.\n/reopen\nReopened this issue. <detailsIn response to : /reopen Instructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\n~ it seems that some more jobs needs to be adapted to not use parallel~ [repo][0]=kubernetes/test-infra~ ~ACTION REQUIRED: Ginkgo V2 doesn't accept go test's flags to parallelize Ginkgo specs, please switch to use or instead.~ Sorry for the noise, the flag is passed to kubetest, and it correctly translate it to , it is not the issue here\nnp, thanks for the update!\nNew work item (?): capture only a single output text output stream () in JUnit.\nupdated the issue, haven't got a chance to take a close look at the item though.\nThis should be resolved by \"open stdout\" now should only appear if a test intentionally records SpecReports.\nmark it as resolved.\nis this complete now?"}
{"_id":"doc-en-kubernetes-138d08f1aeca8e81cbf8aa639b51baa29b1653208e0d897ccfc4810c9421dc3e","title":"","text":", Only two of them left, I need to follow this one, here is the issue for this one opened in the Ginkgo: I think this should be fixed in the third party tools instead of Kubernetes, if we agree with that, we can mark it as completed as well. For the first one, I will continue to follow until there is a update from Ginkgo's side, for the later one, there is no consensus yet IIUC, we can open a separate issue for that, I am gonna to close this one since there is no much thing we can do now. /close\nWe agreed to fix the tool, but had no issue open for it. I just created one:\nAnother issue for"}
{"_id":"doc-en-kubernetes-2bd3e1d9e94a5358a6c2d5b4c6b5864c879bb8cabae55e6531ce8f2da7e1ce93","title":"","text":"What happened: Occasionally we get a Daemonset stuck with less Current replicas than Desired, causing a Daemonset sync tends to fix it, e.g (add a new node). This happens to all our Daemonsets. I'm using fluentd as an example. There are no Daemonsets pods in pending nor are there any errors when describing the Daemonset What you expected to happen: The Daemonset Pod to be scheduled How to reproduce it (as minimally and precisely as possible): Reproducing this state is very tricky, I've had the most success by terminating nodes in our ASG causing a new one to come up. However I think I've identified why this is happening. The issue appears to be related to daemonset Pod deletion, a Pod gets deleted causing to trigger, this lowers the expectations by as expected. The delete pod event then triggers a sync of the daemonset, in which it finds which involves fetching pods from the informer cache This is somehow returning the deleted pod. You can see in the logs here thinks that it needs to delete the pod that was deleted. It is also setting the expectation back to after the delete pod event previously lowered it by 1. At this point you're stuck as the function can't get past this line As the expectation can not be satisfied as syncNodes set it back to and there's nothing it to get it back to . So I believe this is what's causing us to get less replicas than expected, the function is stuck due to the expectation so it can't perform a full sync to schedule the new pods. There's also not much to trigger a new sync once the expectation has expired until a new node comes up or one goes down. Environment: Kubernetes version (use ):Cloud provider or hardware configuration: AWS OS (e.g: ):Kernel (e.g. ):\n/sig apps\nI investigated this as much as I could without getting some external input, it feels like it's some kind of race condition in the informer cache. If anyone has ideas of if places to dig deeper that would be handy.\nReiterating the mentions to trigger a notification: <detailsIn response to : I investigated this as much as I could without getting some external input, it feels like it's some kind of race condition in the informer cache. If anyone has ideas of if places to dig deeper that would be handy."}
{"_id":"doc-en-kubernetes-cf3a9a922f150fcd06b37026fcb1a4fe5fd4fd3de4493928a0abbcee85e2e3a6","title":"","text":"Instructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\nThis problem is definitely related to node termination, it's trying to delete the already deleted pod from the node that's been terminated. The problem seems to go away if is set, I think although this kind of race condition may have always been possible, it's being frequently caused via Which is finding the non-existent pod in the pod cache and seeing that it is assigned to a non-running node, resulting in a request to delete the non-existent pod. cc\nReiterating the mentions to trigger a notification: <detailsIn response to : This problem is definitely related to node termination, it's trying to delete the already deleted pod from the node that's been terminated. Instructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\n/assign could you please take a look at this? It seems that it is related to the code you in .\nThanks for reporting the bug. Bug was introduced in which deleting pods without nodes. There is race condition between removing pod from cache and listing pods for DaemonSet. Bug occurs during following scenario: is deleted. for node is deleted by PodGCController. handler is executed by informer () set is to queue by delete handler () is processed and pods are taken from cache before deleted pod is removed. for deleted pod is not found so we add deleted pod to podsToDelete list (). are set to del:1() succeeds as pod is not found(). are never met so we need to wait till expectations expire. As fix I will send a change to so only pods that can't be deleted by PodGCController are deleted by DaemonSet controller."}
{"_id":"doc-en-kubernetes-45f48f64f34b41c2ff28ff0a971f436119bbad623219cbe5456b39a5b105f849","title":"","text":"<!-- Please only use this template for submitting enhancement requests --What would you like to be : Node E2E tests for Kubelet lock contention flags: --exit-on-lock-contention --lock-file Why is this needed: These flags are being graduated to KubeletConfiguration on , we should add E2E tests for these flags before proceeding.\nThis issue is currently awaiting triage. If a SIG or subproject determines this is a relevant issue, they will accept it by applying the label and provide further guidance. The label can be by org members by writing in a comment. <detailsInstructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\n/sig node /wg component-standard\nIssues go stale after 90d of inactivity. Mark the issue as fresh with . Stale issues rot after an additional 30d of inactivity and eventually close. If this issue is safe to close now please do so with . Send feedback to sig-contributor-experience at . /lifecycle stale\nStale issues rot after 30d of inactivity. Mark the issue as fresh with . Rotten issues close after an additional 30d of inactivity. If this issue is safe to close now please do so with . Send feedback to sig-contributor-experience at . /lifecycle rotten\nRotten issues close after 30d of inactivity. Reopen the issue with . Mark the issue as fresh with . Send feedback to sig-contributor-experience at . /close\nClosing this issue. <detailsIn response to : Rotten issues close after 30d of inactivity. Reopen the issue with . Mark the issue as fresh with . /close Instructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details"}
{"_id":"doc-en-kubernetes-7ee140198c665b218e0dfb7e3cd2f2624912887cfcfab5f30828d12c6130aa35","title":"","text":"What happened: Set up a HA cluster with 1.9 and 1.10 API servers (occurs on any apiserver skew boundary where a cohabitating resource's storage version changed groups) Observed errors when modifying workload objects: when modified via the 1.10 server, saw this error in the 1.9 server: when modified via the 1.9 server, saw this error in the 1.10 server: When this error was encountered, all watchers for the resource in question would be dropped and have to re-establish, and the internal watchers feeding the watch cache would drop and re-establish What you expected to happen: Watch events from all supported storage versions are processed correctly How to reproduce it (as minimally and precisely as possible): Start a server from master (for inspectability, store in json). With Create a deployment (e.g. kube-dns in kube-system) Establish a watch via extensions (kubectl get deployments.extensions --all-namespaces -w) Establish a watch via apps (kubectl get --all-namespaces -w) Tail the kube-apiserver log Change the apps/v1 deployment in etcd to extensions/v1beta1: Observer client watches drop and server log errors: Anything else we need to know?: Environment: Kubernetes version (use ): Cloud provider or hardware configuration: OS (e.g. from /etc/os-release): Kernel (e.g. ): Install tools: Others: /sig api-machinery /cc /priority critical-urgent\nerror is triggered by this returning false: constructed here: the info about expected cohabitating groups is not included in the list of acceptable inputs"}
{"_id":"doc-en-kubernetes-4d5dc3f2fd1aa2418695125e98dd263b9bbd4b25db7ea8b25723d335bfe8c79b","title":"","text":"<!-- This form is for bug reports and feature requests ONLY! If you're looking for help check and the . --Is this a BUG REPORT or FEATURE REQUEST?: /kind bug What happened: Statefulsets are sometimes continually rolling pods and I don't know why statefulset created: storageclass referenced: kubectl describe statefulset statefulset-test: As you can see from the above describe, the pod in the statefulset has been restarted a lot of times. Observing the created pod, 1) The pod comes up ready 2) immediately transitions to terminating 3) Is terminated 4) A new pod is created 5) Go back to (1) (repeats forever) Despite no updates to the statefulset spec ocurring. Making a change to the statefulset template (for example a trivial change adding a nonsense annotation) stops the rolling and fixes the issue When this statefulset was created, there was a mutating webhook enabled. After the observed pod rolling. I disabled this by making its configuration invalid and making it non required, and the rolling persisted. The status of the statefulset: : You can see that there is a discrepency between currentRevision and updateRevision. currentRevision: statefulset-test- updateRevision: statefulset-test- A short while later the status is changed to contain currentRevision: statefulset-test- updateRevision: statefulset-test- This change happened at Sun Apr 1 19:48:48 2018 Presumably after the rollout of statefulset-test- is complete."}
{"_id":"doc-en-kubernetes-25284c3702a45280c2a40fae68a3ba4d1c026c9e5d96ab43c2dee09fe3985aa0","title":"","text":"Here are all the controllerrevisions from above: kubectl get controllerrevision statefulset-test- -o yaml kubectl get controllerrevision statefulset-test- -o yaml: kubectl get controllerrevision statefulset-test- -o yaml: The differences between the above controllerrevisions are insignificant (the hash, the creation timestamp, the name, the resource version, the uid, the self link) Interestingly all these controllerrevisions were created very close together in time shortly after I created the statefulset The current controllerrevisions are After watching the state change for a while, it actually seems like the statefulset is just cycling through the updateRevision being one of the above 10 controllerrevisions Leaders during this time From kube-controller-manager logs during this time, here is one excerpt: The error sync in there is suspicous, and as you can see the controllerrevisions are changing in the logs I have the full logs from a 5 minute window on either side of the above if necessary. The reason I'm inclined to think this is a kubernetes bug is that any update to the statefulset template fixes the problem. How to reproduce it (as minimally and precisely as possible): Not really sure. Worth just trying to create the above statefulset (requires creating the I tried deleting and recreating the above statefulset and it didn't happen, so something flaky is going on Anything else we need to know?: kubernetes 1.9.3 (client and server) cloud provider: aws\n/sig apps\nI realize this will probably need some further debugging, happy to get some direction\nConsistent repro: Create a statefulset that creates a controllerrevision whose hash is between max int32 and max uint32 (this shouldn't be that hard to do, possibly a 50/50 chance of this). Manually create a controllerrevision identical to the first (same revision), incrementing both the name and hash by one. See that the statefulset controller will continually switch which controller revision should be the target of the statefulset.\nDue to the issue fixed here: , under some circumstances controllerrevisions whose revision is the same were being considered equal."}
{"_id":"doc-en-kubernetes-59ace9008f72ca5d82f22a23445c41e6f5a4a98690d88922022af0dc9f3faebe","title":"","text":"This ends up causing this bug to manifest as follows: ListRevisions finds many controllerrevisions with the same revision (unclear how this state happens, but it can happen) The sort here randomly sorts them Due to the linked bug , FindEqualRevisions incorrectly finds that all these controller revisions are equal to the proposed udpate. The updateRevision selects the last revision found here Therefore, updateRevision, is randomly selected and causes the bug. This fixes the immediate issue, but the sort should probably be made stable (i.e not just look at the revision number, perhaps just consider the hash for tie breakers)\nAs an update, fixing the equality check will not be sufficient because it is possible for multiple controller revisions to be created that are actually equal, including the same hash, that have the same revision number. I believe the fix will need to be equal controllerrevisions with the same revision number from being created sort the controllerrevisions so that the controllerrevision that is the target of the statefulset is stable\nIssues go stale after 90d of inactivity. Mark the issue as fresh with . Stale issues rot after an additional 30d of inactivity and eventually close. If this issue is safe to close now please do so with . Send feedback to sig-testing, kubernetes/test-infra and/or . /lifecycle stale\n/remove-lifecycle stale\nping I use the K8S which version is 1.11.4 and encounter the same problem. I want to reproduce the problem and find your way: Manually create a controllerrevision identical to the first (same revision), incrementing both the name and hash by one. Could you tell me how to mannually create a same controllerrevision identical to the first? Thanks in anvance."}
{"_id":"doc-en-kubernetes-82432c9a9613bf4aed980cc8db469ba43857eb8e48c1e0dd0f39492115206ab6","title":"","text":"generation for local KEK based on counter; also rotate once a week even if the counter does not hit 2 million? using AES-GCM encryption of local KEK ref: The implementation would be to the directory.\n/sig auth /triage accepted\n/assign\nGitHub didn't allow me to assign the following users: ibihim. Note that only , repo collaborators and people who have commented on this issue/PR can be assigned. Additionally, issues/PRs can only have 10 assignees at the same time. For more information please see <detailsIn response to : /assign Instructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\n/assign\nthank you !\n/milestone v1.26\nHi bug-triage shadow team member for 1.26 here. Looks like this issue is in the 1.26 milestone, but has been moved to 1.27. Is there something else or should this issue go to 1.27 as well? Thank you!\nI just forgot to move this issue /milestone v1.27\n/reopen is the partial implementation.\nReopened this issue. <detailsIn response to : /reopen Instructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\nMy suggestion on the reference implementation, after reading 's comments is like so... The GRPC server, which implements the expects an implementation that satisfies . The local key logic can be a middleware. So it expects a it can fallback to, if it doesn't know a key or needs to encrypt its own local KEK. So the local implementation would be need to: have a storage of s have a reference to the current in use (if we put it into the cache, it might vanish) have a reference to another , which should be the remote KMS. So the sub-stories could look like: write Middleware write Transformers that can get exhausted (by time, by usage) write 's storage write some dummy for remote KMS? I have some sketches of what I am talking about ."}
{"_id":"doc-en-kubernetes-b75b4d558a76c05a2f299aa9e4208bfe14cbcf4125785eeef841ee037778035f","title":"","text":"I would implement that in terms of the interface so that it can be used both in testing (via a local key) and in real code (by delegating to a HSM).\nHi bug-triage team member for 1.27 here. This issue is still open but has had no recent activity, and no active PRs. Is everything on track for the Code Freeze in March 14th? Thank you!\n/milestone clear\nCould be closed, right? :)\nWe still need metrics and tracing support for the reference implementation (though it is all a lot less important now).\n/retitle [KMSv2] Add metrics and tracing in for plugins\nThis is duplicate of /close\nClosing this issue. <detailsIn response to : This is duplicate of Instructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details"}
{"_id":"doc-en-kubernetes-56dc15a3fe2161c6cf8927a51d2d0dce03fea80f01ef1e0345dcc3ca50d1379f","title":"","text":"Is this a BUG REPORT or FEATURE REQUEST?: /kind bug What happened: a cluster, And node1 with resource 1 DaemonSet, request resource 4 run , There is no pod created. resource to the node: run , There is still no pod created. What you expected to happen: In step 5, There is one pod created. How to reproduce it (as minimally and precisely as possible): see above description Anything else we need to know?: Environment: Kubernetes version (use ):1.10.5 Cloud provider or hardware configuration: OS (e.g. from /etc/os-release):CentOS7 Kernel (e.g. ): Install tools: Others: /sig app\nxref\nThanks! is not work here! Because: DaemonSetsController didn't watch the changes.(we use to updateNode.) we restart kubelet, The DaemonSetsController watch the changes, but it seems that the node info is not updated, and still failed to create pod:\nI found it is a bug here, ignore the changes. I made a fix , PTAL! Thanks!\nAny workarounds for old cluster versions? Up: looks like in my case helped but after some delay.\n, which version of k8s are you using? If it's sunset, you have to merge the pr yourself :). I think restart kubelet trigger delete/add to make sure it sync .\nSorry, I didn't say\nI think we still support 1.12 for now :) let's cherry pick it if not and wait for next 1.12 release."}
{"_id":"doc-en-kubernetes-63751a7023e8b04f1af20ab3ae3207ec2b4f40a0fbf73a42db07ba523eb8b990","title":"","text":"changed the default behavior of scheduler framework's - from stopping at the first failed filter to always running all filters. This change breaks Cluster Autoscaler at scale. Cluster Autoscaler imports the scheduler framework to simulate filters on artificial nodes. In a (pretty common) one-pod-per-node scenario, the number of calls is ~quadratic wrt. the number of nodes needed for unschedulable pods - it can be a lot of calls. Before , in such scenario the vast majority of calls would stop at failing Fit or NodeAffinity filters, and not run the rest of them (e.g. PV-related filters). After the change, every call to runs all filters, which completely breaks Cluster Autoscaler at scale (an example scenario of 300 1-pod-per-node pods and 200 node groups takes 10min in 1.24, and ~50s in <1.22). Can we get the old behavior back? Or re-introduce the config option so that Cluster Autoscaler can set it accordingly? Ideally, we'd also like to get this backported as far as possible (the change was introduced in 1.24).\nThis issue is currently awaiting triage. If a SIG or subproject determines this is a relevant issue, they will accept it by applying the label and provide further guidance. The label can be by org members by writing in a comment. <detailsInstructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\n/sig scheduling\nThis seems to be a regression. I would be interested in adding a toggle flag that allows one to toggle on that behavior if this was done on purpose.\nThat PR wasn't supposed to change the behavior actually. Scheduler is always defaulted to stop upon the first Filter failure, and it's just we offered an option to continue go through other Filter plugins upon failures, and that option was removed recently. Let me check the code to see if there is a regression.\nWhen removed the following changes: It should have kept . I will submit a fix along with test, and backport to 1.24, 1.25, 1.26. Thanks for reporting!"}
{"_id":"doc-en-kubernetes-64891c2c5f731cd49a5bc282c860dc0eba8f556ac4ce7c46f89d192c47705edc","title":"","text":"cc /assign /kind bug /kind regression /priority critical-urgent /triage/accepted\nThanks Kuba, this is indeed a regression on the scheduler end.\n/open for patches\n/re-open /reopen\nReopened this issue. <detailsIn response to : /re-open /reopen Instructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\nThanks for the reporting, and thanks for the fix.\nReally sorry about this regression, and hope not to happen again. I think there's no e2e tests between kubernetes scheduler and cluster-autoscaler, so can we also have testcases of pods unschedulable in scheduling perf-test, maybe this can also help to prevent this happening again. UT benchmark test also makes sense, but it's not graphical.\nThanks for the quick fix! Unfortunately I'm afraid this kind of issue would only be caught in a large-scale test scenario, which is not cheap to run continuously.\nFor this particular issue, UT ( in the PR) is able to catch it. And hence doesn't need to be that complicated to test with cluster-autoscaler. However, standing from a broader angle, why we didn't catch the regression in perf test is b/c the triggering actor is unschedulable Pods, instead of schedulable Pods (which are the happy path most perf tests covered). So going forward, I'd like to see more perf tests to mixin unschedulable Pods, and hence more likely to surface this kind of issue. (we have preemptio preemption test suite but doesn't seem adequate)\nThis is not for the particular issue, but keeping an eye on whether there's any performance degradation between two kubernetes components, like we do for scheduling performance. but it's actually out of the scope of scheduler. I'll add it as one of the goals of\nActually, the schedulerperf has a test case . And, the schedulerperf collects the metric (), which has the label in it (). But, the problem is currently we don't distinguish the metric values by label on each data; we just collect/aggregate it regardless the label."}
{"_id":"doc-en-kubernetes-e7a581bff1a17589b65e3454390d917eb3d7571d932e457d6db360afa738d97e","title":"","text":"(We can distinguish the metric values by any labels in the schedulerperf as we do in the metric ()) I'd say just distinguishing the happy path and unhappy path with label in metric should be enough to observe the particular performance degradation in the unhappy path.\ncould you help share the perf-dash link where this metric () is exposed?\nHm, cannot see any scheduler metrics result from the current perf-dash I see some configuration that seems to be to parse the schedulerperf result file though. https://perf- TBH, I haven't visited the perf-dash so much, so not sure whether it's not working for a long time, or got broken recently. Also, another thing, I noticed the perf-dash doesn't show all test cases. All test cases after haven't been executed.\nSo... probably, something wrong happens in  job (probably in the test case) and the schedulerperf is failing. The schedulerperf doesn't make the result file when fails, so I'm guessing that's why we cannot see any scheduler metrics results from the perf-dash. Okay, I found it. Let me open another issue for this.\ncc is this actually a regression, and is it resolved yet?\nIt's resolved by , and also cherry-picked to 1.24.x ~ 1.26.x. /close\nClosing this issue. <detailsIn response to : It's resolved by , and also cherry-picked to 1.24.x ~ 1.26.x. Instructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details"}
{"_id":"doc-en-kubernetes-12f7047d8d873a7d6bac7023fc38517c350449071b3fe2b6d53f3ea14de5351f","title":"","text":"<!-- Please use this template while reporting a bug and provide as much info as possible. Not doing so may result in your bug not being addressed in a timely manner. Thanks! If the matter is security related, please disclose it privately via --What happened: EndpointSlice for masters was not created after enabling EndpointSlice on a pre-existing cluster. Only those EndpointSlices managed by kube-controller-manager were created. What you expected to happen: EndpointSlice for masters should be created after enabling EndpointSlice on a pre-existing cluster, the same as the EndpointSlices managed by kube-controller-manager. How to reproduce it (as minimally and precisely as possible): kubeadm to create a cluster: the EndpointSlice feature by editing the kube-apiserver, kube-controller-manager, kube-proxy manifest kube-apiserver and kube-controller-manager is restarted, the EndpointSlice in namespace will be missing. Anything else we need to know?: Environment: Kubernetes version (use ):Cloud provider or hardware configuration: OS (e.g: ): Kernel (e.g. ): Install tools:Network plugin and version (if this is a network-related bug): Others:\n/triage unresolved Comment when the issue is assessed and confirmed. I am a bot run by vllry. ‍\n/remove-triage unresolved /assign"}
{"_id":"doc-en-kubernetes-f91ac3317a925ac3a02c50dc39021d458d6020da9f69361ec2732a68459b1b7c","title":"","text":"<!-- Please use this template while reporting a bug and provide as much info as possible. Not doing so may result in your bug not being addressed in a timely manner. Thanks! If the matter is security related, please disclose it privately via --What happened: is the recommended way to configure the outbound networks together with Azure standard loadbalancer (SLB). However, for load balancer rules are disabled by default. That means all frontend IPs would be used as the SNAT IPs for virtual machines, according to . This would introduce troubles for various security scenarios, as there's no way to get a stable list of egress IPs. Azure cloud provider should support this scenario, and disable the outbound SNAT when outbound rules are configured by provisioning tools. What you expected to happen: How to reproduce it (as minimally and precisely as possible): Anything else we need to know?: Environment: Kubernetes version (use ): Cloud provider or hardware configuration: OS (e.g: ): Kernel (e.g. ): Install tools: Others: /kind bug /sig azure /priority critical-urgent"}
{"_id":"doc-en-kubernetes-42eca355948acd2923f06d9ab50c5116c5afebe4d73334e949a0d732f7352d29","title":"","text":"The server-side validation does not check the type of values with . Create a CRD like: Then create a CR: It is not rejected, but should because must not be a boolean."}
{"_id":"doc-en-kubernetes-99c81b4f57584ea95d5f96a4bd9eeeb1b9aeab22e948ca7c4386b5070fb15e2e","title":"","text":"Kubernetes e2e suite [It] [sig-network] Firewall rule should have correct firewall rules for e2e cluster See also a similar cluster with the same test failing: /kind failing-test /kind flake /sig network\nThis issue is currently awaiting triage. If a SIG or subproject determines this is a relevant issue, they will accept it by applying the label and provide further guidance. The label can be by org members by writing in a comment. <detailsInstructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details"}
{"_id":"doc-en-kubernetes-30073b5bdd6536f3426706d62259848d2143e49f7e7d427f9e7ac1b4ee8d409d","title":"","text":"https://k8s- Failed: [] Kubectl client [] Kubectl should be able to talk to api server kubectl running in a pod could talk to api server [Conformance] {Kubernetes e2e suite}\nhttps://k8s- Failed: [] Kubectl client [] Kubectl should be able to talk to api server kubectl running in a pod could talk to api server [Conformance] {Kubernetes e2e suite}\nhttps://k8s- Failed: [] Kubectl client [] Kubectl should be able to talk to api server kubectl running in a pod could talk to api server [Conformance] {Kubernetes e2e suite}\nhttps://k8s- Failed: [] Kubectl client [] Kubectl should be able to talk to api server kubectl running in a pod could talk to api server [Conformance] {Kubernetes e2e suite}\nhttps://k8s- Failed: [] Kubectl client [] Kubectl should be able to talk to api server kubectl running in a pod could talk to api server [Conformance] {Kubernetes e2e suite}\nhttps://k8s- Failed: [] Kubectl client [] Kubectl should be able to talk to api server kubectl running in a pod could talk to api server [Conformance] {Kubernetes e2e suite}\nhttps://k8s- Failed: [] Kubectl client [] Kubectl should be able to talk to api server kubectl running in a pod could talk to api server [Conformance] {Kubernetes e2e suite}\nhttps://k8s- Failed: [] Kubectl client [] Kubectl should be able to talk to api server kubectl running in a pod could talk to api server [Conformance] {Kubernetes e2e suite}\nhttps://k8s- Failed: [] Kubectl client [] Kubectl should be able to talk to api server kubectl running in a pod could talk to api server [Conformance] {Kubernetes e2e suite}\nhttps://k8s- Failed: [] Kubectl client [] Kubectl should be able to talk to api server kubectl running in a pod could talk to api server [Conformance] {Kubernetes e2e suite}\nhttps://k8s- Failed: [] Kubectl client [] Kubectl should be able to talk to api server kubectl running in a pod could talk to api server [Conformance] {Kubernetes"}
{"_id":"doc-en-kubernetes-190b617a6f6f5bf58cdcd7e3b6ef0448e95fbe4e5f55ef7f7041e83808b7f11a","title":"","text":"e2e suite}\nhttps://k8s- Failed: [] Kubectl client [] Kubectl should be able to talk to api server kubectl running in a pod could talk to api server [Conformance] {Kubernetes e2e suite}\nhttps://k8s- Failed: [] Kubectl client [] Kubectl should be able to talk to api server kubectl running in a pod could talk to api server [Conformance] {Kubernetes e2e suite}\nhttps://k8s- Failed: [] Kubectl client [] Kubectl should be able to talk to api server kubectl running in a pod could talk to api server [Conformance] {Kubernetes e2e suite}\nhttps://k8s- Failed: [] Kubectl client [] Kubectl should be able to talk to api server kubectl running in a pod could talk to api server [Conformance] {Kubernetes e2e suite}\nhttps://k8s- Failed: [] Kubectl client [] Kubectl should be able to talk to api server kubectl running in a pod could talk to api server [Conformance] {Kubernetes e2e suite}\nAre you the correct assignee for this? If so, can you please give an update. If not, can you please reassign it to the correct person. CC\nhttps://k8s- Failed: [] Kubectl client [] Kubectl should be able to talk to api server kubectl running in a pod could talk to api server [Conformance] {Kubernetes e2e suite}\nhttps://k8s- Failed: [] Kubectl client [] Kubectl should be able to talk to api server kubectl running in a pod could talk to api server [Conformance] {Kubernetes e2e suite}\nhttps://k8s- Failed: [] Kubectl client [] Kubectl should be able to talk to api server kubectl running in a pod could talk to api server [Conformance] {Kubernetes e2e suite}\nYes, I guess so. I'm looking into it."}
{"_id":"doc-en-kubernetes-dd35fe5a7b4eb5003d2805a282d33ab299b7be8f45ced73e66ae92981d37585e","title":"","text":"https://k8s- Failed: [] Kubectl client [] Kubectl should be able to talk to api server kubectl running in a pod could talk to api server [Conformance] {Kubernetes e2e suite}\nhttps://k8s- Failed: [] Kubectl client [] Kubectl should be able to talk to api server kubectl running in a pod could talk to api server [Conformance] {Kubernetes e2e suite}\nhttps://k8s- Failed: [] Kubectl client [] Kubectl should be able to talk to api server kubectl running in a pod could talk to api server [Conformance] {Kubernetes e2e suite}\nhttps://k8s- Failed: [] Kubectl client [] Kubectl should be able to talk to api server kubectl running in a pod could talk to api server [Conformance] {Kubernetes e2e suite}\nhttps://k8s- Failed: [] Kubectl client [] Kubectl should be able to talk to api server kubectl running in a pod could talk to api server [Conformance] {Kubernetes e2e suite}\nhttps://k8s- Failed: [] Kubectl client [] Kubectl should be able to talk to api server kubectl running in a pod could talk to api server [Conformance] {Kubernetes e2e suite}\nWe need to get this fixed soon and picked up in release-1.5 since it is consistently red in\nack i have a PR currently running\nhttps://k8s- Failed: [] Kubectl client [] Kubectl should be able to talk to api server kubectl running in a pod could talk to api server [Conformance] {Kubernetes e2e suite}\nIf you found the root cause, we will close my pending PR that printing out some debug info. What is the PR # of the commit you mentioned above?"}
{"_id":"doc-en-kubernetes-34e45dcd30e91b287f55b34ca0f7c874b506aed582f8cf03d9c09b8eddbaff00","title":"","text":"https://k8s- Failed: [] Kubectl client [] Kubectl should be able to talk to api server kubectl running in a pod could talk to api server [Conformance] {Kubernetes e2e suite}\n: here it is\nhttps://k8s- Failed: [] Kubectl client [] Kubectl should be able to talk to api server kubectl running in a pod could talk to api server [Conformance] {Kubernetes e2e suite}\nhttps://k8s- Failed: [] Kubectl client [] Kubectl should be able to talk to api server kubectl running in a pod could talk to api server [Conformance] {Kubernetes e2e suite}\nWe timeout waiting for pod sometimes. Is it reasonable to use a longer timeout? Use for in that test.\nhttps://k8s- Failed: [] Kubectl client [] Kubectl should be able to talk to api server kubectl running in a pod could talk to api server [Conformance] {Kubernetes e2e suite}\nhttps://k8s- Failed: [] Kubectl client [] Kubectl should be able to talk to api server kubectl running in a pod could talk to api server [Conformance] {Kubernetes e2e suite}\n: sure thing, do you want to do that in a fresh PR?\nI updated my pending PR .\nhttps://k8s- Failed: [] Kubectl client [] Kubectl should be able to talk to api server kubectl running in a pod could talk to api server [Conformance] {Kubernetes e2e suite}"}
{"_id":"doc-en-kubernetes-16e9d6242dd147113f7225a4b24c8c9581a1b180a6b05f0edc47ba4e0c49bde0","title":"","text":"master-blocking: ci-kubernetes-unit It starts flaking today 30-05-2024 The test fail is This failure may be caused by the error message's order sensitivity. The error message's order of Resource Classes versus is different. N/A /sig node /sig scheduling\nThis issue is currently awaiting triage. If a SIG or subproject determines this is a relevant issue, they will accept it by applying the label and provide further guidance. The label can be by org members by writing in a comment. <detailsInstructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\n/remove-sig testing /sig scheduling (sig testing doesn't own all of the unit tests, just test frameworks, tooling, CI infra, etc) This appears to be a flake in scheduler unit tests which would be sig scheduling ( )\n.. and it appears to be DRA specific subset of scheduling so /sig node as well\nFYI - looks like this is happening every 3-6 hours give or take. Yesterday it was every six hours and today it seems to be every 3-4 hours. Looking into it a bit more, it looks like it may just be an issue with a connection via the test while it's trying to hit the server. I don't think the test itself is failing, I think the connection to the server is failing. ! CC\nHey All - hope you're well. Michael here from the Release Signal team. Just wanted to touch base and make sure we're good to go for the cut on the 4th. CC /sig-node"}
{"_id":"doc-en-kubernetes-b4459e221a04f6723f7b97c6a9d43d61c0d65417b08d143223c7057699636055","title":"","text":"is this a Node bug? cc as next oncall to follow up on this.\nThis is a node bug due to init containers\nI may need help debugging - I see:\nSo I think the problem is in the error case - we're not properly recording that the start failed, and the next time we come in we somehow think the first container succeeded. I'll add some debugging to see if we can't recreate it.\nLet's discuss on instead."}
{"_id":"doc-en-kubernetes-dcc88d1d89328d1da3366de828afcd8ab0373228cc4437ea655e2cda79030f86","title":"","text":"Kubelet issues a docker inspect call after creating a container, but for very short lived containers there is a window in which they could exit before being inspected in the docker manager. If the container PID is 0, we should check the ExitCode and other relevant fields and handle them appropriately. [1] This is the suspected cause of several flakes: /cc\nlooking at the logs 2 weird things: race described second container ran, which suggests inconsistent reporting of initialized status of init1 and choosing init2 as the next container ( vs ). I think 1 got it right and 2 got it wrong. From logs: I'd expect the second one to not have run in this case? Node logs:\nInit containers have an implicit assumption that container statuses include results of previous SyncPod invocations. If that's not the case we may need to fix that (or at least mitigate it). On May 18, 2016, at 8:28 PM, Prashanth B wrote: looking at the logs 2 weird things: race described second container ran, which suggests inconsistent reporting of initialized status of init1 and choosing init2 as the next container ( vs ). I think 1 got it right and 2 got it wrong. From logs: May 18 15:58:33.380: INFO: pod-init--1d4b-11e6-80a4- started at 2016-05-18T15:58:04-07:00 (2+1 container statuses recorded) May 18 15:58:33.380: INFO: Init container init1 ready: false, restart count 0 May 18 15:58:33.380: INFO: Init container init2 ready: true, restart count 0 I'd expect the second one to not have run in this case? Node logs: — You are receiving this because you were mentioned. Reply to this email directly or view it on GitHub\nThis should be true. Is there any evidence that suggests the opposite? only shows conflicting view of whether init1 has run.\nIf the container exits already and it's not an infra container, kubelet should not error out. Although even if it does, I failed to see how this could affect the readiness of the init container without checking the code further."}
{"_id":"doc-en-kubernetes-ade5deee53831749acf4a20dbe39f9ef1836656ae43e40b0bb8594efa4e81e92","title":"","text":"Each syncPod invocation, we inspect the list of init container in order looking for lastTerminationState. There could be a bug in that logic, or if we didn't get the full status, we might make a bad decision. When determining which container to run next, we always look for the last init container in the chain that has success, then run the next container (unless the pod is RestartNever). , Yu-Ju Hong wrote:\nSaw this flake again. How do we verify that container status is not lost between syn pods?\n... between calls to syncPods\nThe second sync clearly saw the init1 container, and decided to launch init2. See \"InitContainersToKeep\" in the message below.\ncontainers to keep can include failed containers. , Yu-Ju Hong wrote:\nDiscussed on slack. In summary, the problem has to do with how we convert the kubecontainer.PodStatus to the api.PodStatus. If a container is terminated, the function makes the decisions: the pod restart policy is \"Always\", and there is a \"failure reason\" recorded for the given container, this means the pod worker tried to (re)start the container but failed. Set the state to \"Waiting\" and populate the reason. , the container state is simply \"terminated\". Init containers don't obey the restart policy like regular containers, and the logic will need to be amended to work with them. is working on a PR. In addition to that, kubelet should not error out if a user container exited before it could apply the oom score (which is best effort in this case). fixes that.\nAhh, thanks for digging into the issue. This is the exact the concern I had when I reviewed the initial implementation of init container, which blocks pr for a while. I even asked to help me verify this case, unfortunately we didn't reproduce the issue back then. :-(\nI wish I'd seen the flakes in the runs prior to merge.\nWe didn't notice this until we started running the tests against docker v1.11.1, which is much faster and hit the race condition differently.\nI suppose that's a good thing...\nIt is hard to reproduce this."}
{"_id":"doc-en-kubernetes-0814a4e75b48cf2d27d20141c180e04fbc9b55d5398cce31b037f460c93acb1b","title":"","text":"The extra check of last terminated container in your initial code makes me nervous, but we couldn't reproduce it due to that small race windows until we have a faster docker running on the node. :-)\nI got \"Pods should invoke init containers on a RestartAlways pod\" flake (referenced from ): Logs:\nAll the fixes went in. Please reopen this issue if you see any more flakes."}
{"_id":"doc-en-kubernetes-932c798c1f1bc917ff850ea55bf430820901f4817711595aba278a9098d6e82b","title":"","text":"<!-- This form is for bug reports and feature requests ONLY! If you're looking for help check and the . If the matter is security related, please disclose it privately via --Is this a BUG REPORT or FEATURE REQUEST?: What happened: In order to migrate to a new version of kubernetes on GKE. I cordoned(Kubectl cordon <node) all nodes in the old node pool(1.8.9) in order to drain the node pool(pods on the node) with an intention to cause the pods to run on a new version of the node pool(1.10.2). As soon as I cordoned all the nodes in the old node pool, it caused the downtime of ingress(nginx). The nginx controller pods where running fine on the cordoned node pool. But the cordon caused the nodes to be removed from the target pool of ingress load balancer. Below are the documents that suggest the above same method that I followed as best practice for zero downtime upgrade. 1). 2). What you expected to happen: I expected \"kubectl cordon <node\" not to have the node/s removed from the target pool of the loadbalancer because the nginx controller pods where running absolutely fine on these nodes. How to reproduce it (as minimally and precisely as possible): 1).Cordon all the nodes in an old node pool that runs nginx controller. Anything else we need to know?: Environment: Kubernetes version (use ): 1.10.2 Cloud provider or hardware configuration: GKE OS (e.g. from /etc/os-release): Container OS Kernel (e.g. ): 4.14.22+ Install tools: Others: Cordon causes to remove the nodes from the target pool of the ingress load balancer. The health checks fail on the nodes of the target pool, which should not be the case. Because the nginx controller pods run fine on these nodes.\nTechnically, this isn't a Kubernetes bug as this is working as intended; however, it does beg the question why we remove unschedulable nodes if they're marked as ready since we perform the to determine LB backend nodes. is it scheduable? is it not a master? is it not marked for exclusion? is it marked as ready?"}
{"_id":"doc-en-kubernetes-3337d8d773bd260e0255505bc31356b49bc0519e69eafd73453232de84478830","title":"","text":"If any of those are a no, then it'll be removed from the pool on the next sync. Perhaps someone from can shed some light. /sig network\ncc This problem extends beyond GCP/ingress controller and to anyone using pools to upgrade nodes. A typical workflow I see is you'd cordon a node pool so as you drain each node the pods don't go to another node in the pool that'll shortly be drained as well. The current logic does not support that and complicates cluster administration because cordoning does not respect PDB's.\n/remove-sig apps /remove-sig gcp (since the service controller falls under sig networking)\nThose labels are not set on the issue: <detailsIn response to : /remove-sig apps /remove-sig gcp Instructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\nRef I thought a general practice for draining the old pool is to move over one node at a time in a rolling fashion. Shouldn't there always be certain backend pods being able to receive & serve traffic? cc\nWRT draining, it is recommended to do it one node at a time ; however, in real world scenarios doing it one node at a time can cause a lot of avoidable container churn if you don't cordon off the nodes you at least plan to drain so that pods aren't being scheduled on a node you'll be draining after the next one. According to the comments in the service controller it seems the unschedulable part exists because it was how we differentiated masters in the past; however, we have labels for that now. We also have the service exclusion label (can't find when/if that'll be promoted though as i don't think there's a ticket on it).\nI'm having problems with this issue as well. The fact that nodes are removed from target pools when cordoned means that I sometimes get brief interruptions of the traffic to a service when I cordon a node. I'm guessing the node isn't gracefully removed from the load balancer which results in some traffic being cut off trying to route through the recently cordoned node. The same problem occasionally happens when deleting nodes / node pools or when the cluster autoscaler deletes a node. We are running GKE ver. 1.10.2-gke.1."}
{"_id":"doc-en-kubernetes-434e849b429a310bf0eaaee7ac5356c43a18352f5a8ac7f41c382993cffab77f","title":"","text":"Our best defence against these issue have so far been to limit node changes as much as possible and not use any auto scaling.\nI've posted a question regarding this issue in SO some weeks before it was created. The suggested action was to use node taints () instead of cordon, that way nodes will still be marked as .\nThank you for the link, it Certainly helped. Perhaps, if thats the case(as mentioned in the accepted answer of the above link), then to cordon all the nodes of old node pools should not be a recommended action, at least in the GKE upgrade docs.\nonly says: And I can't find what more it does in the documentation. So I would expect that it has exactly the same behaviour as a taint. Why is removing running pods from the balancer working as intended? If it isn't a bug then the documentation is certainly unclear in this regard.\nThis just caused a major outage for us too. I agree with the previous posts, while this may be working as the code intended, it does not follow the patterns and documentation k8s provides. When I cordon a node it should not impact the traffic going to the node or any pods on it. I should be able to gradually remove traffic by draining nodes/pods and respecting PDBs.\nthis issue touches on the fact we are removing masters as well and inadvertently removing cordoned nodes. The logic in the service controller mentions we only filter out unschedulable nodes since that's how masters were setup.\nWe wanted to replace all of the kubernetes cluster nodes to different node instance types, so we started off by cordoning all of the nodes, but it surprisingly caused an outage for us. As pointed out, the only documentation around cordoning is , which I took to mean that new pods can not be scheduled on those nodes. It was surprising behavior to find out that, not only does it not allow new pods to be scheduled on those nodes, but that it also causes the service controller to remove unschedulable nodes from load balancing pools. I understand that it's working as intended, but if that was also documented as part of the cordon operation, I would have been able to avoid an outage.\nI finally caught up to this bug. Fun one. I agree that this is working-as-intended and that it stinks. We can do better."}
{"_id":"doc-en-kubernetes-33075f048c6031b18ea2c4d42a5821aded2e74dadef5a72c2d24382b09d40a2a","title":"","text":"As a short-term workaround, you can taint the nodes before cordoning one-by-one. That will stop new work from arriving, but will leave old work there. If we agree on an appropriately named taint, we can have nginx tolerate it by default. Am I missing any reason why this would not work? If that flies, we should fix docs ( those are GCP docs). I am happy to help with that. Longer term maybe we can do something smarter. I am not sure what that would be, just now. Ideas?\nIs there any reason we can't just leave the nodes in the load balancer pool when the node is cordoned? And when the pods are terminated the nodes can be removed? I may be missing some context here?\nThe presumption is that a node that is unschedulable is going away, or needs some other repair. Using it as a target for load-balancers means there could be sessions routed THROUGH that node. Even as the pods are drained. Right up until the node suddenly dies, causing an outage of a different flavor. What should really happen is that a node which becomes unschedulable starts draining connections. Once all the pods are moved off, the node can be removed from the pool entirely. I don't think any cloud providers, even GCP, handle this just right today. David Radcliffe <:\nWait, if you brought up a new nodepool, all those new nodes should be in the TargetPool. Cordoning the old nodepool should only terminate the existing connections, new connections should be fine. Something doesn't add up...\nWas this using\nYes\nRight, well, that explains it. So even draining won't do what you want - you need new connections to be allowed until you have capacity in the new pool. The good news is that you can do this entirely yourself, with taints. We just need nginx to tolerate the taint, which you can also specify yourself. The only real alternative I see would be, as you said, to leave nodes in the LB pool until the node dies. Given that we never specified this, it is probably different on every provider, which would make it hard to change (especially if the provider already provides clean drains while unschedulable)."}
{"_id":"doc-en-kubernetes-971dd5a381fee05df7a732c55096ac94f03e1a58bc9cf8aa911af5dca195b6a4","title":"","text":"IIUC, we don't want new pods to land on the tainted/cordoned nodes. In that case, we can place \"NoSchedule\" taint on the nodes. We don't need any tolerations on nginx. This will prevent new instances of nginx from landing on the tainted nodes, but the existing ones will keep running there.\nIssues go stale after 90d of inactivity. Mark the issue as fresh with . Stale issues rot after an additional 30d of inactivity and eventually close. If this issue is safe to close now please do so with . Send feedback to sig-testing, kubernetes/test-infra and/or . /lifecycle stale\n/remove-lifecycle stale\nThis just surprised us also. I understand the technical reasons behind it but the command name and associated documentation is misleading. To me it isn't intuitive that the command does not remove compute (pods) but does remove some networking. I would expect cordon to keep new pods from getting scheduled to it and keep it from getting to new services/ELBs/etc. The command would then cause pods to be removed and also take the node out of networking paths. Almost need a new state for a node of that is clearly defined.\nMy main issue is with the docs, and also GKE docs. They should explain this pitfall. After I got aware of the issue I understand why it works as it does. A simple workaround is to taint the nodes you are going to rotate first.\nJust chiming in; We as a GKE customer have today been bitten by this issue in production. The documentation of (in the context of GKE) is not extensive enough so we could be aware of this issue. This should be addressed. Luckily the outage was not severe because we were using the node-by-node approach (and accepted the container churn this would cause). We explicitly took this approach because our previous nodepool upgrade was a total outage also caused by in a different way. (for those interested in our total outage, not related to this issue): Due to the fact that also results in the re-calculating the amount of schedulable nodes and thus scaled the amount of pods in our cluster (in our case: from 110 back to 2) resulting for us in major DNS outage internally."}
{"_id":"doc-en-kubernetes-e282766302629c93406295497fcf5631aa302acfa1c106d008828355db43428f","title":"","text":"Be warned that for a nodepool upgrade can have a lot of unwanted & unexpected side effects.\nIt looks like this happens only to AWS and GCP but not on Azure? Can anyone confirm this? I just tested in both AWS and Azure, and while I see logs like in AWS Kubernetes, I do not see any logs like that in Azure. Is it intended? If so isn't behavior different for each cloud provider? i.e.) cordoning a node in AWS Kubernetes prevents loadbalancer from routing traffic to the affected node but cordoning a node in Azure Kubernetes will still let the loadbalancer route traffic to the affected node. Is it intentional?\nBesides running a test like the above, how would I determine what the behavior is on AWS? I haven't found anything in the node administration docs or the Service that would imply that would shift traffic away from a Pod. In fact, the doc for explicitly states, emphasis mine: If I hadn't come across this Issue due to reports from GKE, I suppose it would only have been a matter of time until we had a high-impact production outage. There's a disconnect between Kubernetes API and the LoadBalancer as to whether Pods running on a 'd node are Ready. From all the docs I've seen, I would expect that either a) would evict Pods over to a Node before setting the instance as , or b) LoadBalancer would not equate with\nHi All, What approaches people using too workaround this? From reading the comments I was hoping to follow the custom taint approach. However, I've noticed executes a cordon making it difficult to drain the node safely with this approach.\nI was lucky in that I experienced this issue while cordoning our testing environment, so we didn't experience a production outage. I think what I'm going to try is the following... it's going to be time-consuming but I think it's the safest bet. Double the number of nodes in the server group to bring online the new nodes first Place a custom taint on each \"legacy\" node As you said, executes a first, so that's a no-go. This is the time-consuming part... for each \"legacy\" node I am going to each pod, and let my PDBs take care of maintaining minimum number of pods. I'm hopeful to try this approach this afternoon, I'll comment back here if I do!"}
{"_id":"doc-en-kubernetes-e2e1ecb80d8f290360c9ff319629a62c8ce0008d9f4455b71f27e28e2654d975","title":"","text":"Same as myself! Pity neither of us have a story for Since my comment this morning I've been trying a few approaches, I've ended up with the following: Create a new node pool of equal size in GKE with the upgraded kube version Set the current node pool to not be able to grow beyond its current size - if your using regional nodes this can be annoying, let it bring up new nodes anyways and just have them taint as NoSchedule straight away Taint all old node pool nodes with NoSchedule Go through each old node pool node executing a taint of NoExecute which will cause the node to drain and go onto available nodes i.e. your newly provisioned node pool Ensure your nginx ingress is deployed with tolerances:\nHi All, I'm at a bit of a loss on this one. I've started using custom taints to drain the nodes without getting them removed from the load balancer. My application can hold a HTTP connection open for up to 600 seconds. I'm using nginx as a daemonset, have it configured with a keepalive timeout of 620s, and have . I drain the application first using a taint on each node. I have nginx on each node tolerate this taint and remove it separately at a later point using a different taint. I'm waiting 660s between drains of each node for each taint to ensure nothing is dropped - this all works fine as expected. However, once I kill the node pool or do a kubectl drain after the node is empty of all my stuff plus nginx ingress I start seeing connection issues. I'm a bit lost where to go at debugging them, best guess is something to do with kube-proxy. Can anyone shed some light on this? is there anything on the Google Layer 4 load balancer I'm missing here? Is there some way to correctly drain kube-proxy (or does that even make sense?)\nThis doesn't address the network issues you're seeing, but for those trying to \"roll their own cordon\" using taints: Using a NoExecute taint to \"drain\" the node won't be orderly and won't respect any PDBs. You want to replicate the logic from which tries to do evictions of every pod in a loop. (I realize this is a burdensome idea, but it is also the safest idea.)"}
{"_id":"doc-en-kubernetes-11147f85e8d1488f3d03a6c51d3d3b27940ae8960669e7fe9226d15b33379ba5","title":"","text":"To add on to ,: One approach that does not use node taints to force migrate, is to change for the requirements of your app. A change in node selector (e.g. based on instance type), along with a PDB (e.g. 66% / out of 3 replicas availability) should migrate you over fine. This worked out better for me than to drain or add taints to nodes to force push apps out. After the migration completes, you should be able to the nodes, and delete the node pool. FWIW, I too saw connection issues during load testing of a migration, and at least on my setup it was due to using an AWS NLB for the NGINX Ingress Controller. Switching to a classic ELB instead worked, as the NGINX LB Service seems to hit some sort of by my accounts when using NLBs. self-plug: I happened to recently cover a similar topic in a that you may find relevant. Hope it helps\nFor better or worse, for in my case pods that are behind an ingress resource are using node affinity to preferably not schedule them onto the same node. Originally this was done to ensure traffic is balanced across the different nginx ingresses. It has had a positive affect in this case too. Very interesting! Thank you for sharing, I'm not sure that swapping out the Layer4 GCE load balancer is an option here, but you have me thinking. I'm wondering if the TCP keepalive between the LB and nginx is potentially keeping connections open. , and suggest that there may be a 600 second TCP keepalive going on. Will retest doing a sleep of 620s after nginx removal and before node drain.\nHas anyone come up with a strategy to safely move workloads between two pools? Reading this bug, it seems there's no way to reliably do this without causing an outage if you follow the recommendation of + . In my case I'm not using an nginx ingress but services with type . From what I can see, adding to a node causes the Service Controller to consider the node not ready and thus removes it from a target group. This would mean that between and there is no way to add a taint to the node that will leave it in the LB until it's drained."}
{"_id":"doc-en-kubernetes-badabe4a617de62187355502ba8431d2071284df2978e7fc5a1318bb9d8c5a02","title":"","text":"Using doesn't work in this case since the thing I'm draining might just come back and I'm not sure if the service controller will still consider the node ready in that case, but I believe it does. The one strategy I can think of is using labels on the node and an anti-affinity on the pod. You'd then add the label to a node that would push the pod away and then delete the pod. As long as the anti affinity uses this should be safe to do in preparation of a migration. Once the pods have been deleted you could then the nodes.\nIssues go stale after 90d of inactivity. Mark the issue as fresh with . Stale issues rot after an additional 30d of inactivity and eventually close. If this issue is safe to close now please do so with . Send feedback to sig-testing, kubernetes/test-infra and/or . /lifecycle stale\n/lifecycle frozen /remove-lifecycle stale /remove-lifecycle rotten\n- As I mentioned in I set up a script that first taints all nodes in the old node pool and then cordon+taint one by one node. If your ingress services are redundant this should work well.\nIs there a plan for cordon behavior to improve on GCP? We have relied on this behavior for years on our AWS clusters and it was very surprising to find out.\nThis is a bug with service load balancers - not something we should document. We defined a label in that cluster admins can use - Unschedulable has nothing to do with SLB inclusion. which is beta and on by default in 1.19.\nDisagree. Unschedulable means \"no new workloads\". It has nothing to do with existing workloads, that is why we have drain (but we should have something in kube that communicates nodes are draining). I have opened to remove this check - we have an alternate approach now and the old logic is wrong (the comment in the code even says \"this is for masters!\" and that's wrong now too as part of the kep).\nA node status condition?\nPossibly, or a taint, or a label. Something concrete.\n/milestone v1.19"}
{"_id":"doc-en-kubernetes-31eecfa3591815d34ddfcff16c3269edcba1de74efc4a8809ac7fe48d40e0f43","title":"","text":"When using a generic ephemeral volume, the volume binder in kube-scheduler typically processes a new pod before the generic volume controller in kube-controller-manager has a chance to create the PVC. This leads to events for pods using generic ephemeral volumes: It would be nice to suppress these warnings for a short while (duration to be determined) after the pod creation. Later they should be emitted because then they indicate that the generic volume controller isn't working properly. This makes nicer to use.\nThis issue is currently awaiting triage. If a SIG or subproject determines this is a relevant issue, they will accept it by applying the label and provide further guidance. The label can be by org members by writing in a comment. <detailsInstructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\n/sig storage"}
{"_id":"doc-en-kubernetes-4279a6f285666ec67ce8924ca9393c19176af7de5707dd42a5f4c239ba12ead2","title":"","text":"/kind bug We're having trouble retrieving images from a private registry on DockerCloud. For nearly a year the following has worked flawlessly. Last Friday I pulled down the 1.9.0 CLI (via brew) and since then have not been able to access any image in a private registry on DockerCloud. I've tried this on 10+ brand new clusters created with kops v1.8.0. The Kubernetes cluster version have been 1.8.0, 1.8.4, 1.8.5 and 1.9.0 (which is not supported by kops yet) and all exhibit the same problem. We create the secret as such: And verify it using: Which yields: And create a test service with with these contents: The pod fails immediatly with: I've reviewed the 1.9.0 release notes and nothing jumps out as the root cause. For reference I've tried putting quotes around the value when creating the secret to no avail. I've also tried prefixing the value with , so the full value would read: also to no avail. Environment: Kubernetes version: Cloud provider and installation info: AWS cluster built with Kops 1.8.0, using (to isolate this issue) nearly the default settings for everything.\nReiterating the mentions to trigger a notification: <detailsIn response to : Instructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\nI have same problem, and I \"solved\" it creating a docker-registry secret in an old kubernetes version and pasting it on new kubernetes\nI'm recently (I know it worked perfectly about last month) facing issues with as well. For me it ended in a similar manner saying: I was able to confirm, that this is a kubernetes CLI related issue and not end user. Based on and this I was able to create properly formatted auth information looking like this: . When encoding this one-liner using (MacOS workstation) and apply like this works fine and images can be pulled from private repository again."}
{"_id":"doc-en-kubernetes-784638a373a6d25c569537512411337ba05226e31cffa99d23b394dffb534564","title":"","text":"However this breaks all automation and can't be considered as solution Confirmed working with kubectl 1.8.5 and server 1.8.4 (guess 1.7.9 as well)\nlooks like there's an extra \"auths\" element when adding the secret using 1.9 kubectl\nAdded in /assign\nMany thanks for the quick fix on this. Any idea when this might get rolled out? Also, it might be a good idea if this was to the known issue list for v1.9.0. It would have saved us a bunch of time trying to track down the root cause. Just trying to make sure no one else needs to do the same.\nlinking to the \"v1.9.0 known issues\" list\nack, asking Mehdy since he is the patch manager for v1.9 branch. Mehdy, any idea when we can ship 1.9.1 with this fix?\nworkarounds for folks until 1.9.1 is released: if you have a .dockerconfigjson for your private registry already, you can manually specify the type and data key (e.g. ) if you don't have a .dockerconfigjson file already, you can fix up the secret produced by manually: the type from to the data key from to the modified secret with\n1 will be released tomorrow\nI can confirm solution works, thanks so much!\nI am still having issues in 1.9.1. This worked in the 1.8.x branches: results in the following error:\nyou have to remove the gitlab-registry secret before you can apply an update... type is immutable, and the fix in 1.9.1 required changing the type of the secret"}
{"_id":"doc-en-kubernetes-82d56ce8e3638d906303613bbb222b7918f374f3f5d17197cc57429b2467d07d","title":"","text":"<!-- Please use this template while reporting a bug and provide as much info as possible. Not doing so may result in your bug not being addressed in a timely manner. Thanks!--What happened: I'm trying to test Eviction functionality in my program, but running into some inconsistencies around how Evictions are handled in the fake SimpleClient in the client-go testing functionality. Most notably, creates a different from , and both are missing information. issues a with set to (seems to be the only Fake that has this mismatch) on the subresource of . It is missing all the extra information in the eviction, i.e. the . issues a on the subresource of the resource. This is wrong, because aren't created, they're only posted and handled immediately. The Action is missing a Namespace. What you expected to happen: I want to be able to add my own reactor to the SimpleClient to handle this special case. I propose we change both cases to a with set to the . This will ensure it's not handled by the current SimpleClient ObjectReaction, and makes it so users can implement whatever they want themselves. I'd be happy to make a PR with these changes. How to reproduce it (as minimally and precisely as possible): Output: Environment: git revision of : <!-- DO NOT EDIT BELOW THIS LINE --/kind bug\nPOST requests are create requests. It doesn't matter if the object is persisted or not.\n/sig api-machinery\ncc"}
{"_id":"doc-en-kubernetes-bbec53075a30148d3cbdbc5512f8af7c37c0a3695f84aff12aac6489eaa4282e","title":"","text":"Not sure if this concided with a Jenkins restart.\nNot sure what we can do about this. Retry a few times?\nGitHub seemed to hiccup last night and fail a bunch of our builds. I'm not sure a retry would've helped, unless we kept retrying for many minutes.\nThis is the sort of thing that just happens occasionally. We've seen this a lot in OpenShift CI runs. I'm not sure there's much you can do other than try to avoid external network access (e.g. bake tools into the VM image).\nIt seems to have occurred twice in the last 12h so retry w/ timeout can't hurt?\nIt can't hurt... but it may or may not help depending on the extent of the external issue\nIf the problem is really bad, we could toss it somewhere on GCS?\nLooks like it failed on : We should probably put this in a retry loop or else do something more fancy"}
{"_id":"doc-en-kubernetes-10c13016eb5239afc7a330cfdab91dbf78232a49f2993f71669df735dbb939e1","title":"","text":"<!-- This form is for bug reports and feature requests ONLY! If you're looking for help check and the . --Is this a BUG REPORT or FEATURE REQUEST?: /kind feature The is not user friendly. There has been to make it better, but we still print in the pending message. We should change these errors to a more human readable message. For example, should be changed to something like . /sig scheduling"}
{"_id":"doc-en-kubernetes-e1c6f7407b346df6fb6212f23a2f930e04198e198bb03b9cff5041967802082d","title":"","text":"The PR above moved everything related to into because that was where most of the code already was, but didn't renamed the files to avoid a larger diff and merge conflicts. Once is merged, we can do the renaming. Code clarity. /sig node\n/triage accepted /priority important-soon\n/assign\na PR can be created for this now.\nwill do"}
{"_id":"doc-en-kubernetes-5018642be16ab87784521fef6d356d625aa306ba1fff6efc2d4048e42ef1a94c","title":"","text":"<!-- Please use this template while reporting a bug and provide as much info as possible. Not doing so may result in your bug not being addressed in a timely manner. Thanks! If the matter is security related, please disclose it privately via --What happened: When I tried to create a pod with env variables referencing to annotations: What you expected to happen: Should Proceed. As in It stated is a valid option How to reproduce it (as minimally and precisely as possible): Anything else we need to know?: Environment: Kubernetes version (use ): v1.18.2 Cloud provider or hardware configuration: OS (e.g: ): Kernel (e.g. ): Install tools: Network plugin and version (if this is a network-related bug): Others:\n/sig docs\nThe API docs need to be updated to indicate the correct supported values. See for correct values\nYou are a lifesaver. Thank you!\n/assign I can try to update the api documentation"}
{"_id":"doc-en-kubernetes-65858ef6270e846dbad645a3cdb03be72308acbcbd2d437380bbe522248f5825","title":"","text":"FYI. Currently the field of TPR is marked as which is wrong, actually if you try to create a TPR with empty versions, it will cause api server panic. Versions should be necessary field of TPR, we should validate this somehow."}
{"_id":"doc-en-kubernetes-970debf69dc4e7e663c051298279ebf93b58654f6c393d8683796c73cb4f1c74","title":"","text":"Kubernetes version (use ): Environment: Cloud provider or hardware configuration: Vagrant OS (e.g. from /etc/os-release): CoreOS Kernel (e.g. ):Install tools: Vagrant & Others: What happened: What you expected to happen: Success How to reproduce it (as minimally and precisely as possible): Followed instructions at Tried to run Received: Anything else do we need to know: From googling, it seems several of the fields in the config are for when using OpenShift origin. But creating the replication controller is a prerequisite to launching OpenShift Origin. Chicken & egg problem."}
{"_id":"doc-en-kubernetes-460ceb7b52a1ab0cadc1b29d7eee5cc1013b8d23be9bca0d145119ffcf7b27c6","title":"","text":"Look at one of the directories under (Please don't pick more than one directory for one PR as it will cause issues for reviewers) Find all the files named (example find -name ) See the following link it has a this tells tools how this package needs to be imported. So we should ensure that each has an import Add a description appropriate to the package (similar to the one below) Look through sub directories and see if there are any other important directories that are likely to be used by outside users and add a Why are we doing this? We perfer to use vanity urls instead of directly using to refer to code, so this helps ensure that tools honor our wish. See It would be good to have some package level documentation since these directories under staging/ are meant to be for use by folks outside of kubernetes github org. /good-first-issue\n/kind documentation /sig architecture\ncurrently working my through Not going to use / assign since there could be a number of PRs for this issue.\n/lifecycle active /good-first-issue edit: hmm..\nThis request has been marked as suitable for new contributors. Please ensure the request meets the requirements listed . If this request no longer meets these requirements, the label can be removed by commenting with the command. <detailsIn response to : /lifecycle active /good-first-issue Instructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\nMay I try on\nIIUC this means that there needs to be a even at the root of the repo? Then how does work? :thinking:\nHi , can I make a PR for this directory ?\nIssues go stale after 90d of inactivity. Mark the issue as fresh with . Stale issues rot after an additional 30d of inactivity and eventually close. If this issue is safe to close now please do so with . Send feedback to sig-testing, kubernetes/test-infra and/or . /lifecycle stale\nStale issues rot after 30d of inactivity. Mark the issue as fresh with . Rotten issues close after an additional 30d of inactivity. If this issue is safe to close now please do so with . Send feedback to sig-testing, kubernetes/test-infra and/or ."}
{"_id":"doc-en-kubernetes-5f31f1c27d7198dc83b313154ff681043a8b41e17eb5d3ddf32cd84bc023f931","title":"","text":"/lifecycle rotten\nI guess this is still not done, was supposed to partially fix this issue. /reopen\nReopened this issue. <detailsIn response to : I guess this is still not done, was supposed to partially fix this issue. Instructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\n/remove-lifecycle rotten /area code-organization"}
{"_id":"doc-en-kubernetes-fe3f123845ebc52583ef0b18b019ca5c9cf0487720bfc7eeadb38f7ce5aa77c7","title":"","text":"<!-- This form is for bug reports and feature requests ONLY! If you're looking for help check and the . --Is this a BUG REPORT or FEATURE REQUEST?: What happened: is consistently failing because of PodPreset feature tests, which require settings/v1alpha1, which is no longer enabled by default What you expected to happen: . How to reproduce it (as minimally and precisely as possible): . Anything else we need to know?: This was caused by and kubernetes/test-infra Environment: . Kubernetes version (use ): . Cloud provider or hardware configuration**: . OS (e.g. from /etc/os-release): . Kernel (e.g. ): . Install tools: . Others: .\nThere are no sig labels on this issue. Please by: a sig: e.g., to notify the contributor experience sig, OR the label manually: e.g., to apply the label Note: Method 1 will trigger an email to the group. You can find the group list and label list . The in the method 1 has to be replaced with one of these: bugs, feature-requests, pr-reviews, test-failures, proposals\n/sig service-catalog\ndoes the alpha CI job enable alpha API versions?\nfile has \"AllAlpha=true\" Maybe would be better?"}
{"_id":"doc-en-kubernetes-0b2213d42768c43adcb1b4f15cf67b8722c82d28c3ec9792f5837d41705eb847","title":"","text":"<!-- Please use this template while reporting a bug and provide as much info as possible. Not doing so may result in your bug not being addressed in a timely manner. Thanks! If the matter is security related, please disclose it privately via --What happened: kubectl version --client=yes command is giving the following error message. Error message is repeated at the start and the end and also containing programming jargon Error: invalid argument \"yes\" for \"--client\" flag: strconv.ParseBool: parsing \"yes\": invalid syntax Examples: Print the client and server versions for the current context kubectl version Options: --client=false: Client version only (no server required). -o, --output='': One of 'yaml' or 'json'. --short=false: Print just the version number. Usage: kubectl version [flags] [options] Use \"kubectl options\" for a list of global command-line options (applies to all commands). invalid argument \"yes\" for \"--client\" flag: strconv.ParseBool: parsing \"yes\": invalid syntax What you expected to happen: The error message can be improved with one of the two variants invalid argument \"yes\" for \"--client\" flag Examples: Print the client and server versions for the current context kubectl version Options: --client=false: Client version only (no server required). -o, --output='': One of 'yaml' or 'json'. --short=false: Print just the version number. Usage: kubectl version [flags] [options] Use \"kubectl options\" for a list of global command-line options (applies to all commands). (or) Invalid argument \"yes\" for \"--client\" flag How to reproduce it (as minimally and precisely as possible): Run the command kubectl version --client=yes Anything else we need to know?: Environment: Kubernetes version (use ): 1.15 Cloud provider or hardware configuration: AWS OS (e.g: ): Amazon Linux Kernel (e.g. ): Install tools: Network plugin and version (if this is a network-related bug): Others:\n/sig cli\nHi Thanks for posting this issue! It turns out that this duplicate output has been in kubectl, and all the other cli cmds, since . This should be fixed by , and hopefully we can cherry pick it back to at least ."}
{"_id":"doc-en-kubernetes-2f9c6224492589431ce544bf6b6edab10c3ff06174d612eb6875bebaba0cb87e","title":"","text":"IMO it feels like a bad practice to have mutating resource handlers. The deletePod handler is used for queueing a Recreate deployment once it has no pods running. It should merely look at the ReplicaSets that have owner refs as opposed to using dc.getReplicaSetsForDeployment. The orphaning/adoption dance is already happening inside the reconcilation loop.\nI'd like to work on this\nfeel free to grab it. I can't assign the issue to you so I'll add myself; ping me on the PR when you have one."}
{"_id":"doc-en-kubernetes-859a8b847337ca2cf9879726f3bdb71f0db5559ee2589591ceae612e41f71a1b","title":"","text":"There is a funny subtle behvaiour in get namespaces, in which if i call more than one time, it seems that the removal from Terminating state is accelarated. It seems this might indicate a bug, or maybe its not a bug and the eventual semantics of is a little funny sometimes. in any case, here's what we are seeing : 1) when i delete a namespace the first time usning kubectl, we get a state change to . that is correct, of course. 2) but ... if make the call a second time, then the same status line is immediately removed from . Expected behaviour : A second call to \"kubectl delete ...\" shouldn't have any effect on the timing of removal of a namespace from 'Terminating' state*... So i guess there is some \"eventual consistency\" subtle bug that is resolved by repeated calls to\nsuggests this may not be e bug, maybe its just the etcd eventual deleter thingy.\nLooks to definitely be a bug. The 2nd delete seems to have interrupted the delete process and artifacts (like events) from the namespace still exist in etcd.\ninvestigating...\noops, see the error, will submit a fix.\nI have a fix, but need a test case, will submit PR tomorrow AM"}
{"_id":"doc-en-kubernetes-496d6605c1619bb131a3d458a58442cd69b560a8ee63a4aeca6bff6594949a24","title":"","text":"The test image currently uses as its BASEIMAGE: Per , v3.12 reached end of support on 2022-05-01 and is no longer receiving regular bugfixes or security fixes. Current versions of the image appear vulnerable to known CVEs as a result. I don't know if a fix is quite as simple as just The image's base should be an in-support version which is maintained with at least security fixes. No response <details</details<details</details<details</details<details</details<details</details<details</details\n/sig testing\nis like that, but then there are 2 additional steps the changes the image that has been built with the postsubmit job the repo to use the new image More detail in If you want to try, please make sure you are going to follow up with all the steps, you can see other promotions PRs too /help\nThis request has been marked as needing help from a contributor. Please ensure that the issue body includes answers to the following questions: Why are we solving this issue? To address this issue, are there any code changes? If there are code changes, what needs to be done in the code and what places can the assignee treat as reference points? Does this issue have zero to low barrier of entry? How can the assignee reach out to you for help? For more details on the requirements of such an issue, please see and ensure that they are met. If this request no longer meets these requirements, the label can be removed by commenting with the command. <detailsIn response to : is like that, but then there are 2 additional steps promote the image that has been built with the postsubmit job update the repo to use the new image </details\nThank you for this information! Is the manual and of images described in that same document something that you would consider a prerequisite for a proposed change like this? Or is it safe to assume that the automated PR test bucket(s) would provide sufficient test coverage?\nI always trust the CI system\nGood enough for me! Leading off with\n/triage accepted"}
{"_id":"doc-en-kubernetes-3257ecff81fbc60fa2a8f982346c462089eb7c896f62cce1d4e02ff6df66d913","title":"","text":"We are using (and ) to interact with Kubernetes clusters behind a couple of proxy layers. One of these proxies completes the headers to include character encoding details. This was working fine until recently, when client-go/kubectl have started to fail when interacting with the discovery API. It seems that a recent change has made kubectl more restrictive of the content-type that it expects. I suspect that it is related to this commit: For example, this is an error from a kubectl command: We can of course try to fix it on our end, removing that charset information. However, it is a valid content-type and the way kubectl is handling the content-type there seems like it is relying on implementation details of kube-apiserver (that it does not add metadata to the content-type). I propose to fix this issue in , by properly parsing and handling the content-type.\ncc /kind regression /kind bug /sig api-machinery /priority important-soon looks like the switch statement in downloadLegacy() needs to be more tolerant of the variety of content-types it can get from a server (including text/plain or no content-type if old servers could have returned that...)\n/transfer kubernetes/kubernetes\nSomething went wrong or the destination repo kubernetes/kubernetes/kubernetes does not exist. <detailsIn response to : /transfer kubernetes/kubernetes Instructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\n/transfer kubernetes\nYep, the string comparison for the content-type is flimsy, let's address that. Thanks for the report!\n/assign\n/triage accepted\n/cc\nThe fix is being cherry-picked back to: : :"}
{"_id":"doc-en-kubernetes-93c0a615d6725c12b66abb2dead01e17ca700d451119b1467d3edce68f73a013","title":"","text":"<!-- Please use this template while reporting a bug and provide as much info as possible. Not doing so may result in your bug not being addressed in a timely manner. Thanks! If the matter is security related, please disclose it privately via --What happened: If computer name prefix does not match VMSS name, then Azure cloud provider will use VMSS API instead of using metadata service data. This leads to throttling, because every affected kubelet is making API requests. What you expected to happen: Azure cloud provider should use metadata service data instead of calling VMSS API when instance is current. How to reproduce it (as minimally and precisely as possible): Set computer name prefix of the VMSS to something different than default. You should observe Azure provider making requests to the VMSS API rather than using metadata service, which leads to throttling if you have multiple kubelets with this bug. Anything else we need to know?: This is bug was introduced by - returns false even if instance is current, because it expects node name to match VMSS name + instance id, however node name is derived from the computer-name-prefix which can be different than VMSS name. Node name should be matched against , not against . Environment: Kubernetes version (use ): 1.18.x and higher Cloud provider or hardware configuration: azure /sig cloud-provider\n/sig cloud-provider\n/area provider/azure\nCC\n/remove-sig cloud-provider /sig cloud-provider (trying to remove the label)\n/sig cloud-provider\nThanks for the reporting. Unfortunately, computer name prefix is not included in IMDS, so it can't be used without VMSS API for constructing the NodeName. Would check what could we do to workaround the issue.\nCan you memorize the computer name prefix in the first request to the Azure API and then use it for the isCurrentInstance check?\nopened to fix the issue. Would you like to have a look?\nI can confirm that your change should fix the problem I reported. Do you know if your fix will be to 1.18.x?\nis working on the cherry pick\nThank you for quick fix!"}
{"_id":"doc-en-kubernetes-59e10ac9f842c5137d16434532387dd09ba10f619a9264b3e368e8bd51bb44b7","title":"","text":"What happened: It's most visible in the test <img width=\"323\" alt=\"Screen Shot 2020-05-20 at 6 09 10 PM\" src=\"https://user-\"It shows up in kind, kube- and kops jobs The regression was introduced somewhere within , possible culprits are: (FYI (FYI (FYI There are more tests affected, just not as prominently (visually inspecting ) [sig-apps] StatefulSet [] Basic StatefulSet functionality [StatefulSetBasic] should perform rolling updates and roll backs of template modifications with PVCs [sig-apps] StatefulSet [] Basic StatefulSet functionality [StatefulSetBasic] should perform rolling updates and roll backs of template modifications [Conformance] [sig-storage] In-tree Volumes [Driver: nfs] [Testpattern: Inline-volume (default fs)] volumes should store data [sig-storage] In-tree Volumes [Driver: nfs] [Testpattern: Pre-provisioned PV (default fs)] volumes should store data I wasn't sure whether to tag this as kind/regression or not, I'll leave that up to or someone /sig node /sig api-machimery /sig scheduling\nThe label(s) cannot be applied, because the repository doesn't have them <detailsIn response to : What happened: It's most visible in the test (FYI (FYI (FYI StatefulSet [] Basic StatefulSet functionality [StatefulSetBasic] should perform rolling updates and roll backs of template modifications with PVCs [sig-apps] StatefulSet [] Basic StatefulSet functionality [StatefulSetBasic] should perform rolling updates and roll backs of template modifications [Conformance] [sig-storage] In-tree Volumes [Driver: nfs] [Testpattern: Inline-volume (default fs)] volumes should store data [sig-storage] In-tree Volumes [Driver: nfs] [Testpattern: Pre-provisioned PV (default fs)] volumes should store data /sig api-machimery /sig scheduling Instructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository."}
{"_id":"doc-en-kubernetes-aa1ca8a6c4f552108d5e4f5627cf5341a0bb7748f7b82e510f647cbc02b0a76b","title":"","text":"</details\n/cc\nwill increase the time for deletion of pod and it should be expected because currently pod is deleted from apiserver before removing the sandbox container but after this fix pod will be deleted from the apiserver only after deleting the sandbox container.\n/remove-sig scheduling doesn't affect scheduling running time at all.\nYeah, is almost definitely the culprit. Last time we changed that code path, it also caused a regression. The degradation in performance is expected, but a 5x slowdown is definitely larger than it should be.\ndo we know if that's because the status used by the path is stale, or if removing sandboxes is actually that slow?\nif there's not a quick fix, should we revert and bring it back in once the root issue is understood?\nI think a revert is appropriate in this case.\nok, I'll open it\nopen at\nI agree with that 5x is something which shouldn't happen and i can take a look on why thats happening and that might require changing the way how the sandbox is currently deleted which is only during GC."}
{"_id":"doc-en-kubernetes-397ec6d3ebe6b2748109f44339689e34d7c15d3a74b9e08d5b86ed108288f149","title":"","text":"Is this a BUG REPORT or FEATURE REQUEST?: What happened: shows: The use of \"REDACTED\" for a CA here is misleading, since it's a public certificate that is not a secret. What you expected to happen: I expected to see the full CA in pem format (like when you cat ) or something like \"SUPPRESSED\" or \"TRUNCATED\" How to reproduce it (as minimally and precisely as possible): Anything else we need to know?: Nope Environment: Kubernetes version (use ):Cloud provider or hardware configuration:OS (e.g. from /etc/os-release):Kernel (e.g. ):\n(for thoughts)\nReiterating the mentions to trigger a notification: <detailsIn response to : (for thoughts) Instructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\nuse you can see the complete certificate authority data\nHey Thank you - I have tracking that . The point of this issue is that at best it's a bad UX; at worst it's a miseducation on pki. \"REDACTED\" isn't the right word here\nIf there are multiple clusters displaying certificates, the configuration will be too long, so this switch is turned off by default. But we can add an example to help use. thanks\nHey I don't personally view truncation the responsibility of the CLI... users can pipe things to a pager if they really care. That being said, I'm not concerned about the length. The issue is purely about the word choice of \"REDACTED\". The term means: The use of the term \"REDACTED\" in the output implies that a CA public key is remove for security when (as you've pointed out) it is removed for truncation. Truncation and redaction are fundamentally different things, and k8s is misusing the word redacted for this use case.\nNot quite a bug, but I think you're right a better term could be used there as it is a public key, not a private key. /remove-kind bug\nIssues go stale after 90d of inactivity. Mark the issue as fresh with . Stale issues rot after an additional 30d of inactivity and eventually close."}
{"_id":"doc-en-kubernetes-be8602378a7d778cfbc76bff14bb786f501b2719d8c237c227cdd36827d468ae","title":"","text":"If this issue is safe to close now please do so with . Send feedback to sig-testing, kubernetes/test-infra and/or . /lifecycle stale\n/remove-lifecycle stale I'm not sure what I need to do here. This was marked as a feature. Do you need something more from me?\nWe are waiting for someone to fix this.\n/remove-good-first-issue I don't think this fits our definition of a :\nSo I have to keep replying to the comment bot every 30 days so it doesn't get auto-closed? That feels... odd\nthanks for the education :) unfortunately other issues are higher priority and no one has stepped up to fix this. The autocloser is a form of bug bankruptcy that keeps our issue load in check. The contribex special interest group is definitely open to feedback on the bot.\nThe biggest hurdle in this is that Kubernetes is using some base64 trickery to display the string (basically setting the certificate data to the decoded bytes from ). We can change this to or but not (it has to be a valid base64 encoded string). I tried exploring multiple options for the text, but I couldn't find a suitable one. Given that we already print the user tokens and other credentials this is a bit weird. Should we redact those too? Is it okay if we print the CA certificate instead of truncating it?\nI can't help but laugh at: I'm not wed to any particular name. I also have no problem displaying the entire certificate. The crux of the issue is that a CA public key is not secret or sensitive. It's a public key\nI agree. The issue is with the constraints imposed by how this is implemented currently. I tried to come up with a suitable string (given the constraint) and couldn't find anything acceptable... The only suitable option that worked for me is . I personally prefer to print the CA certificate since we are already printing user tokens and client certs that are actual secrets.\nI agree. Who do we need to get in agreement that printing the full CA is the best path forward?\nthe data is not omitted for security purposes, but to make the output more readable. see REDACTED is misleading... it should probably be \"elided\" or \"...\"\nyou can run to view full content\nthat's the crux of this issue - REDACTED is misleading."}
{"_id":"doc-en-kubernetes-0c20cfda2a4143b97ca34085e4ba4933a2256b4cc12b09ed01e77387d3b1e642","title":"","text":"However, pointed out that other \"long\" fields are still completely printed in plain text such as client certs, so I think we need to revisit the \"readability\" argument. What do you think?\nthey are not... client cert/key are elided in the same way for the same reason: something like \"++data+omitted++\" could be better If the intent of config view without --raw specified is to be readable (which it currently is), printing many extremely long fields isn't great\nHmm what about: I'm pretty sure that would give us a valid b64 string (8+12=20%4=0) and it would subtly hint that the data isn't a secret.\nOnly valid characters are . And adding a prefix from the certificate only complicates things. is a suitable usable option. I'll submit a PR to continue the discussion there."}
{"_id":"doc-en-kubernetes-2c5e5762130beb36422dea9bedde0f0f5452416204c98eb7971e6661a493007b","title":"","text":"Alpha environment tests are failing - The root cause of the tests failing in alpha environment appears to be: So basically kubelet is not allowed to update PVC. Which is weird. Need to investigate further. Also for multizone environment - This line looks weird from e2e: So the deployment is straight up not getting started. /sig storage /assign\nFor multizone failure it looks like the pod has a node selector that conflicts with the zone where the PV was provisioned\nI think failures on alpha cluster is related to generated api code: Obviously none of the PVC patch code sets this stuff and yet is showing up in object diff. I am wondering why. cc\n/wg apply /cc\nIt looks like we may have to reset field before comparing PVCs for admitting them as well.\nYeah, server side apply has an alpha feature to kube-apiserver that allows to track ownership of fields. This is what you see in the managedFields field. It’s automatically when objects are modified. If you want to compare your object against a staticly compiled expected object, you should be emptying/setting the managedField to nil.\nthis is the code in question: Thing is, we don't actually want the kubelet taking ownership of any fields other than the one it changed. If we nil out the managed field for the comparison, that allows the kubelet to mess with managedFields in any way it wants to\ncan you pick this up to fix failure in alpha clusters? Reading last comment - it doesn't seem like a storage issue.\n/unassign /assign\n/milestone v1.14 We're entering and this looks relevant to the 1.14 release. Please if I am incorrect\nWe don't have a good way to do what is suggesting now, so I suspect we will have to default to ignore this field specifically when comparing (by setting it to null on both sides), at least for now.\nwill you be able to submit a PR to fix this?\nAs per /remove-sig storage /sig api-machinery\nSure, I'll try to send a pr this morning"}
{"_id":"doc-en-kubernetes-f45c12318e35043c65362c27e361ee13adfef369e183f2b4f0557c11353bd9e9","title":"","text":"from https://k8s-\nReally suspect this is\nThis is putting us into test-go post-commit run is broken\ncc:\nmarked this p0 - if a fix isn't easy let's try to identify what introduced this and revert.\nThis flake on: 2016-10-30 07:19 PDT First instance: : It is flaking frequently, blocking PRs from getting merged, and slowing the submit queue ( at 75%). As on-call, I'm going to try and identify the cluprit and revert. Can use all help in doing so.\ncan you investigate whether your additional tests could cause this (by breaking the deployment of those pods, for instance) , Saad Ali wrote:\nPR looks most suspect to me. It mucked with , and the errors we are seeing in each of the failures is related to : That PR was merged at Oct 30, 2016, 6:18 AM PDT and the first flake in was at Oct 30, 2016, 7:19 AM PDT. Thoughts? CC\nI'm going to initiate a revert of to see if that fixes the issue\nIMO we can just change the flaky test to not get pods but some other resources, such as deployments. We can't really expect to clean up pods immediately.\nI'm going to proceed with revert (). Please feel free to send a follow up PR to handle this more gracefully.\nGoing to leave this opened until we have a few hours of runs with the revert. Want to verify, with some degree of certainty, that the revert took care of this flake.\nWe had 15 green runs of in a row. Considering this flake fixed."}
{"_id":"doc-en-kubernetes-57b8eab4723f670b6603d9bfb82377ab99cdb5a96c17a18501cc8d4284272435","title":"","text":"<!-- Please use this template while reporting a bug and provide as much info as possible. Not doing so may result in your bug not being addressed in a timely manner. Thanks! If the matter is security related, please disclose it privately via --What happened: I create a job with initcontainer with restartPolicy : when the initContainer exec failed and it will be restarted, the pod Phase will be pending all the time. but in JobController, the only care the running pod. So this pod will be always restarted. What you expected to happen: when restart count to backoffLimit, the job failed. How to reproduce it (as minimally and precisely as possible): Anything else we need to know?: Environment: Kubernetes version (use ): Cloud provider or hardware configuration: OS (e.g: ): Kernel (e.g. ): Install tools: Network plugin and version (if this is a network-related bug): Others:\n/sig apps\n/assign"}
{"_id":"doc-en-kubernetes-3a5f8f6fae70577f1cbce129e8b4c7178012c138aa12442a5305f077f7f55297","title":"","text":"xref: When running the GCE PD CSI Driver against all dynamic provisioning tests from this PR: We encounter the following test failures that need to be resolved: /kind bug /sig storage /cc\nSo it seems that the flag is getting passed all the way through to the driver and it is behaving as we would expect it to in terms of mounting. All the correct commands are being run: When exec-ing into the gce-pd-driver container we can observe that the mounted point is indeed read only. However, when viewed from the host node's perspective the mount is still read write. And therefore since the subpath folder is created by the host, the folder is still being created successfully which fails the test. This behavior suggests that is either not properly propagating the read only option on remount, or not properly remounting at all. I can't figure out how to \"view mount events\" if that's even possible, some help would be appreciated if anyone is aware of how to do that. That would help me narrrow down which of those two options is happening. I have also attempted (from inside the gce pd driver container) to unmount the point (successful) and bind mount it back with the readonly option directly (not using remount trick, still unsuccessful). Suggesting that maybe it is the that is not being mount propagated and not a problem with in particular. /cc\n/assign\n/milestone v1.13 /priority important-soon"}
{"_id":"doc-en-kubernetes-20ba07f969c78351406740c712da4b4d48c539c847c31f97f90729edba71a7c4","title":"","text":"From shippable, I noticed there are several failed cases, for example, E0413 17:34:38. ] watch of *api.Pod ended with: Stop requested --- FAIL: Example (0.61s) got: want: a-hello b-controller c-framework FAIL coverage: 85.9% of statements FAIL 7.652s cc/\nanother example:\nah-- this is a different thing. I will fix this, too.\nI'm seeing this from an unrelated change\nYeah I think that's the one I expected to see."}
{"_id":"doc-en-kubernetes-6d76348628664e8e1283688eb2571b9c9ddbeb679dd9d941b90f8deb0741d5d7","title":"","text":"Is this a request for help? (If yes, you should use our troubleshooting guide and community support channels, see ): No What keywords did you search in Kubernetes issues before filing this one? (If you have found any duplicates, you should instead reply there.): describe psp Is this a BUG REPORT or FEATURE REQUEST? (choose one): FEATURE REQUEST Kubernetes version (use ): Environment: Doesn't matter. What happened: doesn't show PSP's fields: What you expected to happen: All PSP fields should be shown. How to reproduce it (as minimally and precisely as possible):\nIf you're agree with that than I (or :) ) could implement it.\nyes, improve the describer. Look at the scc describer for formatting.\nany ideas which SIG should own PSP?\nIn this case it should be sig-cli /sig cli\nyeah, this is cli related. I'd go with sig-auth for actual behavior."}
{"_id":"doc-en-kubernetes-cdccab911562ccfffce851739652bcccdbeabc99e62d699c46baa58ef67f3bbe","title":"","text":"Up to 1.21, that field is immutable. That means in practice, a CSIDriver object has to be deleted and then re-created with a different value instead. Mutation via delete/created is unnecessarily complex and as the only consumer of that field (the kube-scheduler) can handle mutation just fine (it'll always ask the informer cache for the current value), it's better to allow mutation.\n/sig storage /assign\nThis is independent of allowing other fields to become mutable, but that might also make sense.\n/triage accepted"}
{"_id":"doc-en-kubernetes-b9774b02da3d7547dcc6ec82c6d1246d6586e28e6fe16ad58f3c28b2558ac905","title":"","text":"Resource quota example is missing readme. Was it once there and was accidentally deleted?\nI wrote something up in Origin repo I need to repurpose for Kubernetes. Sent from my iPhone"}
{"_id":"doc-en-kubernetes-a55cc9cf23a1aa66af6ba96638da8c1361b9b34938bf9fbe2ee98dcb858f1383","title":"","text":"Which jobs are flaking: periodics-kubernetes-bazel-test Which test(s) are flaking: //pkg/kubelet/cri/remote TestVersion Testgrid link: Reason for failure: Error: Received unexpected error: rpc error: code = Unavailable desc = transport is closing panic: runtime error: invalid memory address or nil pointer dereference [recovered] panic: runtime error: invalid memory address or nil pointer dereference Anything else we need to know: To reproduce with stress: go test -race -c ./pkg/kubelet/cri/remote stress - TestVersion Detailed logs\nThis issue is currently awaiting triage. If a SIG or subproject determines this is a relevant issue, they will accept it by applying the label and provide further guidance. The label can be by org members by writing in a comment. <detailsInstructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\n/sig testing\nI 'll give it a see and fix it :)\n/assign\n/remove-sig testing SIG node owns their unit tests. SIG testing owns the framework, infrastructure, etc.\n/sig node"}
{"_id":"doc-en-kubernetes-8b547b0918354692a878225113bae9a07e658398c400349efcb40fc715b52424","title":"","text":"A V(2) log line that includes preemption details (preemptee and preemptor), somewhere here We used to have most of this information as part of the logging for Events. But we removed the preemptor from it as a security measure Logs can have pod names.\nThis issue is currently awaiting triage. If a SIG or subproject determines this is a relevant issue, they will accept it by applying the label and provide further guidance. The label can be by org members by writing in a comment. <detailsInstructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\n/sig scheduling\n/help\nThis request has been marked as needing help from a contributor. Please ensure that the issue body includes answers to the following questions: Why are we solving this issue? To address this issue, are there any code changes? If there are code changes, what needs to be done in the code and what places can the assignee treat as reference points? Does this issue have zero to low barrier of entry? How can the assignee reach out to you for help? For more details on the requirements of such an issue, please see and ensure that they are met. If this request no longer meets these requirements, the label can be removed by commenting with the command. <detailsIn response to : /help Instructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\n/assign\n/help-remove"}
{"_id":"doc-en-kubernetes-f2910a5a7757744dd88700c551fb3f5fe81db92c4ae532e768f1444ef3ba7b10","title":"","text":"It's at line 111 When the server stopped before timeout, the first goroutine will be block on because the receiver in the second goroutine is not exist. Changing to a buffered one may fix it. no goroutine leaks run related test with leak checker No response <detailsdev </details<detailsunknown </details<details</details<details</details<details</details<details</details\n/sig api-machinery\n/assign /triage accepted\n/cc\nIt should close the channel rather than write to it, that's the easiest fix"}
{"_id":"doc-en-kubernetes-754763163ccf179fd132c393411aa2d267e649990c98a81bb38938ec6debc8b2","title":"","text":"The comments in says that : However there are several cases where this is false : When using a Service with a targetPort that references a name in the ports arrays of the pod. Some software (ab)uses this API to get which ports are used (See Although unfixable, my goal is to fix the documentation so other people don't have the same problem). The Azure Load Balancer implementation does this for instance. Some other network/security software might use this information. Also, is it possible to document/reference the above issue in the comment ? Since it is described as informational I expected for the network to work without this information. But it's definitely not the case. Not needed since it's mostly a doc issue. I'm interested in contributing this patch but would like some feedbacks before doing it. :) <details</details<details</details<details</details<details</details<details</details<details</details\n/sig network\n/triage accepted -- thoughts? I believe the doc are due for a refresh.\nI'd take this one. /assign\nI didn't see that you wanted to do this patch, I am very sorry and now I unassign it. /unassign\nNo worries But I'm still not sure where / what to add. Should I only fix the part saying it is only informational ? Is there anything else ? (cc\nThe Kubernetes project currently lacks enough contributors to adequately respond to all issues and PRs. This bot triages issues and PRs according to the following rules: After 90d of inactivity, is applied After 30d of inactivity since was applied, is applied After 30d of inactivity since was applied, the issue is closed You can: Mark this issue or PR as fresh with Mark this issue or PR as rotten with Close this issue or PR with Offer to help out with [Issue Triage][1] Please send feedback to sig-contributor-experience at . /lifecycle stale [1]:\n/remove-lifecycle stale"}
{"_id":"doc-en-kubernetes-739981494c57aaeb7141fd974134e4440eb950badca5a937fa4552d94ec1af8a","title":"","text":"In , we merged a work around for heapster panic We should get rid of the work around after heapster version is updated. /cc /cc\nIssues go stale after 90d of inactivity. Mark the issue as fresh with . Stale issues rot after an additional 30d of inactivity and eventually close. If this issue is safe to close now please do so with . Send feedback to sig-testing, kubernetes/test-infra and/or . /lifecycle stale\nStale issues rot after 30d of inactivity. Mark the issue as fresh with . Rotten issues close after an additional 30d of inactivity. If this issue is safe to close now please do so with . Send feedback to sig-testing, kubernetes/test-infra and/or . /lifecycle rotten /remove-lifecycle stale\nRotten issues close after 30d of inactivity. Reopen the issue with . Mark the issue as fresh with . Send feedback to sig-testing, kubernetes/test-infra and/or . /close"}
{"_id":"doc-en-kubernetes-2c477631a31f290ae271a7ca0a83a97399cbf63c4cfd883674b6ddce7387fc19","title":"","text":"What happened: Latest kubelet + Windows Docker version 1.14.0-dev refused to start any containers, error output: \"invalid option: Windows does not support MemorySwap\" How to reproduce it (as minimally and precisely as possible): It's always reproducible with Windows Docker 1.14.0-dev + latest kubelet code base. Anything else do we need to know: It didn't happen for Windows Docker 1.12.2-cs2-ws-beta. The reason behind the issue was because new code was for Windows Docker to check Resources parameters, it doesn't accept a non-zero value for MemorySwap: But kubelet was default to set an unlimited (-1) value. The issue was gone in my setup after applied a private fix to disable MemorySwap by set it to zero for Windows."}
{"_id":"doc-en-kubernetes-ca640d6faf4286189814f4ad1e01c7d7710200569bbe8ad0dc1acda128b91391","title":"","text":"During the startup process of kubelet, we have Since we have following Code segment 1 and 2 seems dup to me. Am i getting this wrong?"}
{"_id":"doc-en-kubernetes-6208b2974deee3332ab4020708350835523c141c61d79e5104e4c7dba961daa5","title":"","text":"ci-kubernetes-e2e-gce-device-plugin-gpu.Overall ci-kubernetes-e2e-gce-device-plugin-gpu.Pod kubetest.Up Intermittently since 02/23 No response No response /sig testing /sig release\nThis issue is currently awaiting triage. If a SIG or subproject determines this is a relevant issue, they will accept it by applying the label and provide further guidance. The label can be by org members by writing in a comment. <detailsInstructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\n/cc /sig node BTW, would you help to confirm if this is a release cut blocker? We plan to release 1.30.0-alpha.3 on Tuesday 27th February 2024.\nIt passed! Maybe it was some infra issues?\nFailed again. /sig k8s-infra\nMany recreating of nodes in log.\nThis is caused by infra flaky.\nThe problem is\nThis job only runs one test and GPU instances are hard to find we don’t need many of those instances for this job, can we please change this job to only use 2 workers?\n/cc\ntests failed in a recent run. It failed the first time. The Job failure now changed to test case failing today. It failed 4 times in a row.\n/assign\nI launched a test run, logged into the master node and took the kubeconfig and found the following: Relevant bit\nLink some related updates recently on this job.\nThe error above in is weird to me. It said No firmware image found . In successful run before, like (02-27), there are no such failure logs. failure in only kubelet log shows the pod crashed and restarted for many times. No pod log or containerd log.\nAfter , the CI is not always failing now and 4 in a row.\n/close\nClosing this issue. <detailsIn response to : /close Instructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\n/reopen as keeps flaking\nReopened this issue. <detailsIn response to : /reopen as keeps flaking Instructions for interacting with me using PR comments are available ."}
{"_id":"doc-en-kubernetes-93e6b458b8ba607e02020a8718331e5d8d90eb2c8daf949aa5e83a98cf741ed8","title":"","text":"If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\nThe problem is different with the failure after the revert. We(release signal team) plan to open an issue for the flake later. (Probably, we should open that one and then close on behalf of that new issue later.)\nFeel free to close it then. However, it failed 3 times out of last 4 runs. It's more a failure than a flake in my opinion. If doesn't help/ not appropriate, I'd suggest to revert installer and image changes\ncomment is clear about the root cause, we have a problem installing the drivers, can you redirect this issue if you think is a bug?\nAfter was merged, there are 6 failure in last 24 run. Tab stats: 8 of 10 (80.0%) recent columns passed (258 of 262 or 98.5% cells) The Flaky rate is lower."}
{"_id":"doc-en-kubernetes-cef38c378015a3c37c6ee1632d82c03243825979d637fc81c55115a5c8a6f392","title":"","text":"<!-- Please only use this template for submitting reports about continuously failing tests or jobs in Kubernetes CI --Which jobs are failing: Which test(s) are failing: Since when has it been failing: Testgrid link: Reason for failure: Anything else we need to know: Introduced in /cc /sig node /assign /milestone v1.20\nThis issue is currently awaiting triage. If a SIG or subproject determines this is a relevant issue, they will accept it by applying the label and provide further guidance. The label can be by org members by writing in a comment. <detailsInstructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details"}
{"_id":"doc-en-kubernetes-90b5ca2d1baaaaa2aa124277e8c3fdace970eae885ff81aa4062d924fde9dbb7","title":"","text":"PLEASE COORDINATE WITH SIG-SCHEDULING BEFORE WORKING ON THIS ISSUE. We want to avoid the situation where two people accidentally start work on the same fix. Main tracking issue: The method of is a simple wrapper around . It should be converted into a simple closure instead of being a method. i.e. we should have, (code is in ) /sig scheduling /kind cleanup /help\nI'd like to take this task. /assign /remove-help\nThanks, go for it! :smile:"}
{"_id":"doc-en-kubernetes-d71ca3fa6a0d8529c7a86525f4c578d8692c435e96ab7540a0643bb36519a584","title":"","text":"<!-- Please use this template while reporting a bug and provide as much info as possible. Not doing so may result in your bug not being addressed in a timely manner. Thanks! If the matter is security related, please disclose it privately via --I0726 15:05:12. ] UnmountDevice succeeded for volume \"volume-name\" %!(EXTRA string=UnmountDevice succeeded for volume \"volume-name\" (UniqueName: \"fake-plugin/fake-device1\") on node \"mynodename\" ) Here at the same time, simple, detail information is returned and formatting problems are caused. No format error. For example, only datail is returned. Found from running Kubernetes version (use ): Cloud provider or hardware configuration: OS (e.g: ): Kernel (e.g. ): Install tools: Network plugin and version (if this is a network-related bug): Others: /good-first-issue\nThis request has been marked as suitable for new contributors. Please ensure the request meets the requirements listed . If this request no longer meets these requirements, the label can be removed by commenting with the command. <detailsIn response to : <!-- Please use this template while reporting a bug and provide as much info as possible. Not doing so may result in your bug not being addressed in a timely manner. Thanks! --#### What happened: I0726 15:05:12. ] UnmountDevice succeeded for volume \"volume-name\" %!(EXTRA string=UnmountDevice succeeded for volume \"volume-name\" (UniqueName: \"fake-plugin/fake-device1\") on node \"mynodename\" ) #### What you expected to happen: No format error. For example, only datail is returned. #### How to reproduce it (as minimally and precisely as possible): Found from running Kubernetes version (use ): Cloud provider or hardware configuration: OS (e.g: ): Kernel (e.g. ): Install tools: Network plugin and version (if this is a network-related bug): Others: Instructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\n/triage accepted /sig instrumentation /wg structured-logging\nSeems like caused by , this line still not in structured logging, we need to handle these messages."}
{"_id":"doc-en-kubernetes-c360b40c66141d5646d0351669bbca5e7c3c6ba302e184cee03898eed6d86ac9","title":"","text":"I don’t know if this package will be accepted for structured log migration. It seems that only kubelet is required now.\n/assign\nI think it‘s only kubelet required too.\nHi, sorry about that I didn't realize that you assigned yourself to this issue. Hope is helpful for this issue.\nThe Kubernetes project currently lacks enough contributors to adequately respond to all issues and PRs. This bot triages issues and PRs according to the following rules: After 90d of inactivity, is applied After 30d of inactivity since was applied, is applied After 30d of inactivity since was applied, the issue is closed You can: Mark this issue or PR as fresh with Mark this issue or PR as rotten with Close this issue or PR with Offer to help out with [Issue Triage][1] Please send feedback to sig-contributor-experience at . /lifecycle stale [1]:\nThe Kubernetes project currently lacks enough active contributors to adequately respond to all issues and PRs. This bot triages issues and PRs according to the following rules: After 90d of inactivity, is applied After 30d of inactivity since was applied, is applied After 30d of inactivity since was applied, the issue is closed You can: Mark this issue or PR as fresh with Close this issue or PR with Offer to help out with [Issue Triage][1] Please send feedback to sig-contributor-experience at . /lifecycle rotten [1]:\n/cc\nInteresting. I'm running the same test locally: And I'm not finding the same log output issue there. Can anyone else see it on (master)?\nit's quite likely that the source code line changed. Search for \"EXTRA string\" instead.\nThe Kubernetes project currently lacks enough active contributors to adequately respond to all issues and PRs. This bot triages issues and PRs according to the following rules: After 90d of inactivity, is applied After 30d of inactivity since was applied, is applied After 30d of inactivity since was applied, the issue is closed You can: Reopen this issue or PR with Mark this issue or PR as fresh with Offer to help out with [Issue Triage][1] Please send feedback to sig-contributor-experience at . /close [1]:\nClosing this issue."}
{"_id":"doc-en-kubernetes-ce988bfc6ad2fcc73f6949053dcba493aabc2a4f61d6f09e0644b61d62fb522f","title":"","text":"<detailsIn response to : The Kubernetes project currently lacks enough active contributors to adequately respond to all issues and PRs. After 90d of inactivity, is applied After 30d of inactivity since was applied, is applied After 30d of inactivity since was applied, the issue is closedReopen this issue or PR with Mark this issue or PR as fresh with Offer to help out with [Issue Triage][1] Instructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details"}
{"_id":"doc-en-kubernetes-11cf454c594ad6f800f58a0212c358c898b7772cd85a5cbba9749540b6918b99","title":"","text":"In We should be using ConteaxtEval for evaluation. Use ContextEval for cel evaluation n/a No response <details</details<details</details<details</details<details</details<details</details<details</details\n/assign /triage accepted /sig api-machinery"}
{"_id":"doc-en-kubernetes-18a84ca90a3e633f3a3a09fc3ce20a2ee5a77d52c40aa5d987ef95b0afbf9b00","title":"","text":"E2eNode Suite.[sig-node] Node Performance Testing [Serial] [Slow] Run node performance testing with pre-defined workloads TensorFlow workload From the testgrid, it has been 1/2 flaky since the start of history (03/20/2022) and has been failing constantly (0/2) since 04/01/2022 No response /sig node\n/triage accepted /priority important-soon\nThe failing test seems to be expecting GPU on machine and the image seems to be broken. I tried the image manually: Logs:\n/assign as discussed at SIG Node CI meeting, can you please take a look? I see that pod is now running for couple minutes and then stops. I wonder if this is something environment related - maybe pod was evicted or something. I doesn't seem to be able to repro the failure of the pod locally\nGitHub didn't allow me to assign the following users: xmcqueen. Note that only , repo collaborators and people who have commented on this issue/PR can be assigned. Additionally, issues/PRs can only have 10 assignees at the same time. For more information please see <detailsIn response to : /assign Instructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\nThis is no longer a flake. It has failed every run since the start of June. Locally this test runs perfectly with a legit correct local run of the test suite code, doing the full tensorflow run of 300 epochs. The failure in the testgrid job seems to be fast (173 sec to build and fail, locally its 160 sec before it starts the tensorflow pod). I wonder if there's a blocked port or something from the data initializing python program. It needs access to: . Here's the command it runs: Is it feasible to add debugging code, like a pre-check for that http port? It would be good to at least force it to fail with a meaningful error message.\nNote that HEAD is now at image version 1.2, and it is still failing per . The successful local runs of the test mentioned above used image version 1.2."}
{"_id":"doc-en-kubernetes-924fdb64dda3a1d51367c5f2ba44327b61b51dcf8cc0ad56f75173d2dbe65d92","title":"","text":"Locally image version 1.1 does produce the cuda/GPU error shown above by :\nMore info is needed to find the RC for this test failure. I've put in a PR to get the test framework to preserve the container logs when the pod fails.\nThe pod logs show that it is timing out on an http connection, and it looks like it's the one where it tries to download the test data set for the tensorflow run. You can see it . Search down to . The data_download script in the image attempts to pull down a dataset from: . That url is live and healthy, so they might be blocking the gcloud.\nThis test depending on that external data doesn't look quite right :/\nI'm discussing this test with sig-testing. I'll post an update later.\nTalked to from sig-testing: he's contacting the remote to see if they are blocking us, because it looks like they are is pulling the script from a that is So I see three choices: upgrade this thing to a more recent edition delete it grab everything that's needed to run this and keep it in-house <- clear winner here what do you think: Maybe the folks over at sig-testing have a place to store the artifacts in-house? They are not massive (about 10MB see below) Size of data about 10MB: the tools and libs for the tensorflow run: the training and eval files for the testrun:\nHere's a relevant issue in a related k8s repo about using/not using docker registries: Registries used in k/k CI should be on Kubernetes Community infra\nlooks good for the subject of putting up a new docker image containing the deps. This seems feasible and effective. I'd prefer it to deleting the test. I'll discuss it a bit more with a few others, but this seems like the way to go forward with this. TBS, I plan to take the above listed test requirements, and build a docker image containing all that is needed, and reset the test to pull the new docker image.\nI've got some Dockerfile changes to fix this. I've tested it locally with no network and it does the full 300 epochs. PR coming soon.\nIs it worth pushing up 182 files to get this green? It is about 6M of python, and 5M of a test dataset."}
{"_id":"doc-en-kubernetes-1089cbe285cd80c5b6feae67c8fda7b5283b08a5d4cab211d1adc2743664a2e8","title":"","text":"Its all stashed away in the to get the docker image to contain the prereqs. It is selected subdirectories of the quite old archived version (v1.9.0) of the code from here: current version v2.7.2 and two small test data sets from here:\nOK I got it down to 43 files, including the 2 datasets. That's the drivers and the datasets and the libs they need. It all gets packed into the docker image.\nquick question - is there another repository that is under sig-node that can host this image?\nTo be clear, the test's docker image is currently in a repo, the : This PR adds artifacts to an existing docker image in the above repo.\nI've cancelled the above PR. I'll try to get something else going.\nThere has been a decent amount of discussion in slack relating to handling the test's required artifacts. The solution that seems best is to get some practical version of the test requirements into locally owned and controlled storage, so the endpoints that provide access to the assets will at least be under internal control. This is not code, and is not under development. It is a static dataset, that is now stored in an offsite archive, and will not require maintenance. The datasets are published with the \"\": (see bottom right) \"This allows for the sharing and adaptation of the datasets for any purpose, provided that the appropriate credit is given.\" Here are some links to the dataset and where its hosted: https://archive-\nI'd like to get the dataset that holds the test data (about 650k) hosted in GCS where I can get it via http. With that change it is expected that the test will go green. The failures show it is failing when it tries to download the artifact from the uncontrolled remote endpoint. Somebody, please help me get this artifact into an internally controlled server. Thanks\nWas the idea of downloading this only on image creation in dockerfile rejected?\nThanks for the reminder I did not fairly evaluate that proposal. I just jumped to my earlier proposal. Thinking about this proposal I see that it would mean downloading the test data for inclusion into the docker image, while the image's test driver tooling is built to pull down the dataset itself from a remote."}
{"_id":"doc-en-kubernetes-1e0a13eacbc2a35a9de3ea0888ffa655ef37373231443eb0a06d10729155f72f","title":"","text":"That is a simple change to handle, while simply re-hosting the artifact on GCS would require only changing the endpoint. Also building the docker would probably still require network access to the artifact, which would probably require us to put it somewhere anyway. I assume, putting it into the docker image would best be done by downloading the test data artifact from somewhere. Would we then be storing the artifact twice - once in the docker image, and once in a locally controlled endpoint? If the docker build server has unrestricted access to the artifact, this proposal would be good. I don't have access to the servers to run tests. However, I could put up an RB for the Dockerfile to try it out. This is probably worth the effort. I'll get an PR going for trying to pull down the remote test-data artifact from the image build server, and bunde it into the docker image, and I'll link it to this ticket.\nthank you. We do not change the image for a long time, so downloading it once would likely be enough. Once it is self-contained, it is even better for the continuous perf runs in tests.\n/reopen\nYou can't reopen an issue/PR unless you authored it or you are a collaborator. <detailsIn response to : /reopen Instructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\n/reopen\nReopened this issue. <detailsIn response to : /reopen Instructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\nCurrently the image builds for the first arch (amd64) but fails when it enters the cross compiling phase (arm64) and specifically when it tries to do apt installs from within the qemu-*-static emulator: The other images in node-perf fail to build in the same way. The jessie-dnsutils image build fails to build in the same way. I have tried pulling in newer base images up to buster, but it did not fix the problem. -Start focusing on the version of qemu. - qemu is version locked in the Makefile and has not changed in two years. There was an and the images were building at that time."}
{"_id":"doc-en-kubernetes-21583406d11cc50f0d960b70331655967f38b78788b3327204ea9b4d500f4be7","title":"","text":"Try to find clues over there.\nI tried pinning many relevant versions of from but none helped. Here's a clue: This ugly hack gets it past the observed error:\nNote the previous PR was missing an important change, but I do not see any feasible way that could explain the current problem with failing, so I have not felt any urgency to put up a PR for it. FTR I need to bump the VERSION. I did not bump the version in I thought maybe the CI would bump it automatically.\nThere's an image version in . Its still pulling the old version and failing. Bump it from 1.2 to 1.3. So one more commit is needed.\nThe new image is present and runs perfectly. Tested it on local linux and in google cloud: It ran to completion. Note the new image URL. Its NOT the same as the old one:\npodspec references the image Here's the . tf-wide-deep's url is built The above has staging in it. . Its already built and pushed to staging. It must be promoted when this new version number goes out.\nFTR: My \"testing\" above did not catch the current problem because I did not run the full e2e_test binary. I only tested the image via a . AND I should have read way back at the beginning.\nLocal run of : has to change too.\nLocal test using the published 1.3 image run the standard way was successful with the above: ran the test successfully. So promote the image per bump the image version in the e2e test manifest change the\nAfter the image promotion is merged, there's STILL one more commit coming: bump the image version in the e2e test manifest change the\nits promoted and is found at the expected url local containerd cleared out local test was rerun pulling from the new prod registry local test using that new registry passes\nits fixed. the test is green again."}
{"_id":"doc-en-kubernetes-8476bba8e23a7ed885cd34c1077ce17dbf12901ac7ae9725333264f6ea8cfdae","title":"","text":"What happened: A deployment is created that uses a after the cluster master is created but before the cluster worker nodes are fully initialized. During the initialization of the worker nodes, the nodes are temporarily available for scheduling without the necessary label to match the deployment's node selector. Depending on how long the nodes are available for scheduling without the necessary node labels for the deployment, the scheduler will start to spam with cluster with pods in status. This spamming stops once the worker nodes are fully initialized and the pods are scheduled successful. The exact timing of all of this is TBD. What you expected to happen: Pods in status are annoying and require manual cleanup. It would be preferred that the scheduler avoid this an simply issue warning events. How to reproduce it (as minimally and precisely as possible): See issue description. Anything else we need to know?: No. Environment: Kubernetes version (use ): 1.18.3 Cloud provider or hardware configuration: IBM Cloud Kubernetes Service OS (e.g: ): Ubuntu 18.04.4 LTS Kernel (e.g. ): 4.15.0-96-generic Install tools: Managed service install Network plugin and version (if this is a network-related bug): N/A Others: N/A\nRelated to and\n/sig scheduling\nby spamming are you referring to scheduler logs? also, why are the nodes not labelled from the get go?\nJust to confirm, all the pods in question are still in Pending state, yes? If the pods are in pending state, all of them will get scheduled eventually once the nodes get their labels.\nThe pods are not in state. They are actually in state. The pods eventually get scheduled but we are left with anywhere from a few to dozens of these state pods. Also, we are looking at labeling the nodes via kubelet flags to see if that helps.\n/remove-kind bug\nAfter we add some retry logic and wait time in this problem can be solved: As you can see this function returns initial node as the first request failed.\nI also get hit by this, when single controller node gets rebooted and scheduler starts as static container before kubelet becomes ready."}
{"_id":"doc-en-kubernetes-6c53201dbb91d52a8ade960092a5ed582f4d35e33a53b75d555d15ae815dc25c","title":"","text":"Can be cleaned up with:\n/sig node\nthe spam cycle is this loop: \"pod create, pod schedule, worker reject, pod failed\" -- we don't have a lot of protection in higher level workload controllers to handle pods that fail at kubelet admission time versus pods that just dont create due to quota/admission. in general, is the spam causing any other stability issues? is pod gc pruning the pods?\nI wonder if we can do a trick in node lister to mitigate any race condition here...\nwe could maybe do something similar to to ensure node lister has synced...\nhacking on option to ensure if kubelet has a node lister with a valid kube client, we wait for it to sync at least once here which should mitigate this issue. see:\nWe noticed the problem on the OLM pod which is managed by the OLM operator. Right or wrong, the operator does not like seeing the pod go to status and no longer reconciles the pod. The end result is that manual intervention (i.e. pod deletion) is required to fix the problem.\nsure... i would have thought the pod is failed in that instance and reached a terminal state. either way, i am looking at what i can do to mitigate this.\nThe problem persists in our environment as well, and the workaround cited above will not do anymore. Is there something in the works for this one?\nIssues go stale after 90d of inactivity. Mark the issue as fresh with . Stale issues rot after an additional 30d of inactivity and eventually close. If this issue is safe to close now please do so with . Send feedback to sig-testing, kubernetes/test-infra and/or . /lifecycle stale\n/remove-lifecycle stale"}
{"_id":"doc-en-kubernetes-fc07b6eddbb9d1c63170ff68b48220d761822f4a0e8799d7065a14dbb9d41309","title":"","text":"Hey Guys , I try to use the option --override with kubectl expose . First i did this and it work : Now i want to add another service with kubectl expose and this command does not work I have this error If i remove --overrides the command work . `\n/sig cli\nexpose tries to convert version from \"v1\" to \"_internal\" which is not registered. This is the reason that the function doesn't works. == \"\" 316 if inline := cmdutil.GetFlagString(cmd, \"overrides\"); len(inline) 0 { 317 object, err = cmdutil.Merge(scheme.Codecs.LegacyCodec(scheme.Scheme.PrioritizedVersionsAllGroups()...), object, inline) 318 if err != nil { 319 return err 320 } 321 } === 156 // LegacyCodec encodes output to a given API versions, and decodes output into the internal form from 157 // any recognized source. The returned codec will always encode output to JSON. If a type is not 158 // found in the list of versions an error will be returned. 159 // 160 // This method is deprecated - clients and servers should negotiate a serializer by mime-type and 161 // invoke CodecForVersions. Callers that need only to read data should use UniversalDecoder(). 162 // 163 // TODO: make this call exist only in pkg/api, and initialize it with the set of default versions. 164 // All other callers will be forced to request a Codec directly. 165 func (f CodecFactory) LegacyCodec(version ) runtime.Codec { 166 return versioning.NewDefaultingCodecForScheme(f.scheme, f.legacySerializer, f.universal, schema.GroupVersions(version), runtime.InternalGroupVersioner) 167 }\nkubectl help expose Expose a resource as a new Kubernetes service. Possible resources include (case insensitive): pod (po), service (svc), replicationcontroller (rc), deployment (deploy), replicaset (rs) In order to control the scope, I only register them as un-versioned type in the fix."}
{"_id":"doc-en-kubernetes-e021a9a686a92d83ba285392c70089f3a0813200ee010fb06825ed96dc15cfdb","title":"","text":"xref I think this may be one of the few cases where we can actually change the behavior in a \"breaking\" way. User creates service with , is empty Controller or user updates the status subresource, sets (not wiped or validated, should be) User updates in any way, is wiped It's clear that the is not meant to be set when is not \"LoadBalancer\". The question is: 1) Fail validation? 2) Silently discard the value? 3) Warn and then discard the value? 4) Just warn and leave it. Because we allowed it in the past, but it is so volatile, I lean towards 3. and - the people most familair with this type. Edit: PrepareForUpdate() runs before warnings. So: (1) is easy (2) is easy (4) is slightly less easy than (2) (3) is slightly less easy than (4)\nThis issue is currently awaiting triage. If a SIG or subproject determines this is a relevant issue, they will accept it by applying the label and provide further guidance. The label can be by org members by writing in a comment. <detailsInstructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\nOMG change it to clusterIP as expected Status is wiped update status as ClusterIP ... and if I edit the service back to loadbalancer it keeps the status??\n/kind bug\n/retitle Service can be set when !=\nWe talked about this one at the SIG Network meeting today. suggested we use the feature gate for deprecation option to phase the current (problematic) functionality out.\n/triage accept\nThe label(s) cannot be applied, because the repository doesn't have them. <detailsIn response to : /triage accept Instructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details"}
{"_id":"doc-en-kubernetes-ea06bb5e10f9c1a3c6e6bdd4b05e49a63cf7c6313178f3d65784c243007cbe65","title":"","text":"Subpath unix socket tests can flake because of long socket path. On Linux the path limit should be 108 chars or less. /sig storage"}
{"_id":"doc-en-kubernetes-bf4acf56a8324d5d9229417a25d2febc989c56201f9f4bdce8fffe7a46728546","title":"","text":"A CSI plugin that the 1.10 controller manager can't detect (which could occur in a number of scenarios, such as when a rollout is being performed or a node is evacuated) seems to retry without backoff. In this case, I have a pod referencing a PVC -PV that references my CSI plugin, but no controller plugin running. The controller is probing every 100ms continuously for a single volume attachment. The controller should backoff. Have not tested with higher numbers of pods, but in a busy system this could cause the controller manager to burn CPU. Also, there isn't actually a great error telling the user why the probing is happening (I knew in this case, but an admin likely has no idea).\n/assign\nIssues go stale after 90d of inactivity. Mark the issue as fresh with . Stale issues rot after an additional 30d of inactivity and eventually close. If this issue is safe to close now please do so with . Send feedback to sig-testing, kubernetes/test-infra and/or . /lifecycle stale\n/remove-lifecycle stale /milestone v1.14 adding it to next milestone as per if it doesn't make it then it can be further moved\nas KEP has been moved, unless an exception is made, /milestone v1.15\nthe linked kep above is unrelated to this issue. This is a bug and I don't think it needs an associated enhancement\nsorry, got diverted by the comments on that KEP. /milestone v1.14\nHello from Bug Triage Shadow . I'd like to remind that code freeze is coming in 3 days (Friday, 8th March). Is this still planned to be fixed for 1.14?\nWe have not been tracking this for 1.14. /assign can you help triage and see if this is still an issue and if there is a quick fix?\nI can't reproduce the issue, do you remember what you did? From the logs it seems that Watch channel consistently returns so new is called recursively (beware, 1.10 code, but it's very similar in master): Can it produce fairly stable 100ms delay between calls? Anyway, this is not usual code path, watcher does not return in most cases."}
{"_id":"doc-en-kubernetes-386907f979b2ab6cc28dd5590343cbf2830e874540a448be35232bf4673dff7e","title":"","text":"if I read the code correctly, such recursive must end up with timeout - caller does not care about return value and caller's watch probably won't return anything useful. Is there a reason why we don't return error? This will kick in exponential backoff in A/D controller.\nclearing the milestone for tracking purposes to let the fix land in its own time /milestone clear\nIssues go stale after 90d of inactivity. Mark the issue as fresh with . Stale issues rot after an additional 30d of inactivity and eventually close. If this issue is safe to close now please do so with . Send feedback to sig-testing, kubernetes/test-infra and/or . /lifecycle stale\nI found the error would happen when a pv is attahing(volumeattachment has created) and the csi-attacher is deleted at the same time. Then the log would print so frequently (because of recursion).\nis this fixed by ?\nStale issues rot after 30d of inactivity. Mark the issue as fresh with . Rotten issues close after an additional 30d of inactivity. If this issue is safe to close now please do so with . Send feedback to sig-testing, kubernetes/test-infra and/or . /lifecycle rotten\nRotten issues close after 30d of inactivity. Reopen the issue with . Mark the issue as fresh with . Send feedback to sig-testing, kubernetes/test-infra and/or . /close\nClosing this issue. <detailsIn response to : Rotten issues close after 30d of inactivity. Reopen the issue with . Mark the issue as fresh with . /close Instructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details"}
{"_id":"doc-en-kubernetes-a0c3ad6976551b855a45933e19f845be402e9fd035bccb2466a57a10d2f14bed","title":"","text":"<!-- Please use this template while reporting a bug and provide as much info as possible. Not doing so may result in your bug not being addressed in a timely manner. Thanks! If the matter is security related, please disclose it privately via --What happened: ! What you expected to happen: How to reproduce it (as minimally and precisely as possible): Anything else we need to know?: Environment: Kubernetes version (use ): v1.20.1 Cloud provider or hardware configuration: OS (e.g: ): 7 Kernel (e.g. ): Linux zsm-test-2 3.10.0-1160.11.1.el7.x8664 SMP Fri Dec 18 16:34:56 UTC 2020 x8664 x8664 x86_64 GNU/Linux Install tools: Network plugin and version (if this is a network-related bug): Others:\n/sig node\n/triage accepted /priority backlog"}
{"_id":"doc-en-kubernetes-bc75029de49571476e249812075ff7ffff85fb8014f5cfe5e32228962baa98b2","title":"","text":"Hi , I followed the instructions in examples/logging-demo, and there was no \"external IP\" field for elasticsearch-logging or kibana-logging, as shown below: I start up my cluster via cluster/kube- Did I miss something or is it a bug? Thanks.\nCorrect. This demo does not create an ELB so there is no external IP address. Access is via service proxy at the master. Type: and you should see the URLs for Elasticsearch and Kibana.\nplease let us know whether: a. Satnam's comment resolves your issue b. the demo should be updated to prevent this misunderstanding c. the issue isn't resolved Thank you!\nYes, Satnam's comment resolve the issue. I updated the demo in .\nAwesome, thanks for sending the patch! We'll discuss there. Always fantastic to have docs updated regularly by folks trying them."}
{"_id":"doc-en-kubernetes-59d4410d28b628c1be3d77bd9cd1cfdd4f523921f65910962a77fbec070a39e8","title":"","text":"https://k8s- Failed: TestNamespaceController {namespace}"}
{"_id":"doc-en-kubernetes-e2f4fe7b99012f8e7173e2f6c46ef5707967f19510033d4b6b03e3a31c7298f0","title":"","text":"<!-- Please only use this template for submitting reports about continuously failing tests or jobs in Kubernetes CI --ci-kubernetes-e2e-gci-gce-slow [sig-storage] Dynamic Provisioning DynamicProvisioner [Slow] should provision storage with different parameters 5/13 https://k8s- v1beta1.StorageClass got removed in We should remove the v1beta1 test.\n/assign can you help take a look?\n/triage accepted\n/sig storage\nI run the test with the latest from master without removing the beta test and it failed with a different error before getting to the beta test part: Do you have any idea what would've happened in my setup? I redeployed k/k using master HEAD\nFor this testcase: I saw that the provisioned PD has a size of 1.5Gi according to kubernetes: And in gcloud It's really interesting that the test was expecting 2Gi to be created, is this something that was changed recently?\nThe logs you posted are showing pd csi as the provisioner. Were you testing with the intree driver or csi driver? The CSI driver should to the nearest Gb. I believe intree does the .\nI think I did intree because I didn't send as an argument in my command: It's interesting that the PVC has this annotation too:\nmmm something interesting about my setup is that I created the cluster with CSIMigration turned on Do you think that might be forcing the usage of ?\nYes likely. Try with CSIMigration off and see if that helps. If so, then this potentially seems like a rounding bug with the csi driver\nThanks I tried with and it did indeed round up to 2Gi."}
{"_id":"doc-en-kubernetes-30e07950978ed617c12104c8342559dfe477718ba7eb61f98c21f95bb37c1439","title":"","text":"In kubectl v1.8.0+, schema info is cached in the location specified by (defaulting to ), written with world-writeable permissions (rw-rw-rw-). If is specified and pointed at a different location accessible to other users/groups, the written files may be modified by other users/groups and disrupt the kubectl invocation. CVSS score: CVSS:3.0/AV:L/AC:H/PR:L/UI:R/S:U/C:L/I:L/A:N (3.3, low) What versions are affected? kubectl v1.8.0+ What configurations are affected? Invocations that point at world-writeable locations Impact Malformed responses written to the cache directory can disrupt the kubectl invocation Workaround Use the default location in the $HOME directory or point it at a directory that is only accessible to desired users/groups. (original description follows) ==== What happened: The files inside of \".kube/http-cache\" are world writeable (rw-rw-rw-). While the default for these files appears to be the home directory, using the \"--cache-dir\" flag could put these files into a place where world writeable files would allow any user / process to modify the cache files. Modification of the cache files could influence the kubectl utility in a negative way for other users. What you expected to happen: Apply stricter file permissions to the http-cache files. How to reproduce it (as minimally and precisely as possible): Run any generic kubectl command which is successful and then list the cache directory $ kubectl get pods --all-namespaces $ ls -la Anything else we need to know?"}
{"_id":"doc-en-kubernetes-03c8940a04bf023e0d3632a105949b7e85efda694ffafea56ecb99998a766122","title":"","text":": I estimate this is a low severity security issue with a CVSS score of \"3.3 / CVSS:3.0/AV:L/AC:L/PR:L/UI:N/S:U/C:N/I:L/A:N\" - Environment: Linux Kubernetes version (use kubectl version):Client Version: version.Info{Major:\"1\", Minor:\"12\", GitVersion:\"v1.12.6\", GitCommit:\"\", GitTreeState:\"clean\", BuildDate:\"2019-02-26T12:49:28Z\", GoVersion:\"go1.10.8\", Compiler:\"gc\", Platform:\"linux/amd64\"} Server Version: version.Info{Major:\"1\", Minor:\"12\", GitVersion:\"v1.12.6\", GitCommit:\"\", GitTreeState:\"clean\", BuildDate:\"2019-02-26T12:49:28Z\", GoVersion:\"go1.10.8\", Compiler:\"gc\", Platform:\"linux/amd64\"} Cloud provider or hardware configuration: AWS. Running kube api server in hyperkube. OS (e.g: cat /etc/os-release): NAME=\"CentOS Linux\" VERSION=\"7.1808 (Core)\" ID=\"centos\" IDLIKE=\"rhel fedora\" VERSIONID=\"7\" PRETTYNAME=\"CentOS Linux 7.1808 (Core)\" ANSICOLOR=\"0;31\" CPENAME=\"cpe:/o:centos:centos:7\" HOMEURL=\"\" BUGREPORTURL=\"\" CENTOSMANTISBTPROJECT=\"CentOS-7\" CENTOSMANTISBTPROJECTVERSION=\"7\" REDHATSUPPORTPRODUCT=\"centos\" REDHATSUPPORTPRODUCTVERSION=\"7\" OSTREEVERSION=7.1808 Kernel (e.g. uname -a): Linux hackit.internal 3.10.0-862.11.6.el7.x8664 SMP Tue Aug 14 21:49:04 UTC 2018 x8664 x8664 x86_64 GNU/Linux Install tools: Manual installation. Others: n/a\n/sig product-security\nClosing this issue. <detailsIn response to : if you think this is serious, send an email thanks!"}
{"_id":"doc-en-kubernetes-5f4971cee99e57062ae7961121de35cc677214a3e6ca77682b093e74b00a9b30","title":"","text":"</details\nAdjusted the CVSS score to to indicate user interaction is required (since they must specify a non-default cache option pointed at a shared writeable location)\nThere are two places this should be done: PathPerm and FilePerm should be set here: the temp file and final file should have stricter permissions (probably 0660) here:\n/cc\n/help\nThis request has been marked as needing help from a contributor. Please ensure the request meets the requirements listed . If this request no longer meets these requirements, the label can be removed by commenting with the command. <detailsIn response to : /help Instructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\n/assign\n/assign I would like to work on release-1.9, as I just made a hot fix for our team's k8s version of 1.9.\nGitHub didn't allow me to assign the following users: yuchengwu. Note that only and repo collaborators can be assigned and that issues/PRs can only have 10 assignees at the same time. For more information please see <detailsIn response to : /assign I would like to work on release-1.9, as I just made a hot fix for our team's k8s version of 1.9. Instructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\nhelp fixing this is welcome, though it would only be fixed in releases 1.12+, per\n/label official-cve-feed (Related to kubernetes/sig-security)"}
{"_id":"doc-en-kubernetes-27809a3a2eb2b85924d4f4faf65a841d86c20af577cb16fb38fbd54d281dded1","title":"","text":"release-master-blocking ci-kubernetes- Sub-tests: - ! 2023-01-20 09:17:28 +0000 UTC TestRoundTripSocks5AndNewConnection/proxiedwithvalidauth-https(invalidhostname+InsecureSkipVerify)TestRoundTripSocks5AndNewConnection/proxiedwithvalidauth-http /sig api-machinery cc\n/assign /triage accepted Will be fixed by Thanks for noticing, great report by the way"}
{"_id":"doc-en-kubernetes-56cd908854c413b8fe52b6acbb170713ddc1a064a7bd9d42b4a3ba81ae7efcaf","title":"","text":"Found some error logs when sending series events to apiserver: seems the later eventTime differed from the first one in series events After investigation, I found the problem was in the marshal. MicroTime is supposed to work on microsecond level precision, but init with nanosecond level precision, in this case, generates microsecond level precision data and generates nanosecond level precision, so when unmarshalled, the new time is not considered the exactly same as the original time (they are different at nanosecond level) MicroTime marshal should work at same precision for json and proto, so unmarshal can generate the exactly same MicroTime result: This causes the series events won't be updated on api-server after created master <details</details<details</details<details</details<details</details<details</details\n/sig api-machinery\n/assign\nGood catch... we'd probably need to truncate to microseconds when marshaling and unmarshaling proto to ensure any already persisted data is fixed up when read. Something like this:\nThanks for advice, going to send a PR soon"}
{"_id":"doc-en-kubernetes-0ad427ca9c00c618017f4be9fb19a1df6f4ab39ddc68f128fc109eb4d0409cbf","title":"","text":"Cluster logging assumes kubelet stores container log at a specific directory with specific name. Currently the path is: After the CRI work, the path will be: However, we don't have test coverage for this. We don't even know whether the log files are there or not. We should add a node e2e test for this, basically: 1) Create a container which prints logs 2) Check whether the expected log file is created. 3) Check whether the logs are collected in the file as expected. And it would be better to launch a pod to do the check instead of accessing local file system directly, which makes containerizing easier. /cc because you are collecting missing test. /cc\nI will try to add a node-e2e test there\nI think it's ok to add this for CRI. But there maybe problems on today's interface for rkt. If I'm not wrong, rkt does't store logs at same location with docker. cc/ Is this true?\nThanks! :)\nAfter gets merged, we need to update the test to verify the cri log path."}
{"_id":"doc-en-kubernetes-89e8d863e20723ba6498e9abf897dd4e49e26cad02c0dde0a4957b2eb3151e69","title":"","text":"<!-- This form is for bug reports and feature requests ONLY! If you're looking for help check and the . --Is this a BUG REPORT or FEATURE REQUEST?: /kind bug What happened: Currently APIServiceRegistrationController watch service change and find the related APIServices related if any. Then it AddAPIService/RemoveAPIService for APIAggregator. I dig into the kube-aggregator and find it no need totally, because update APIServices status when related service or endpoint change. So if does not watch services, it will not miss any event for APIServices. What you expected to happen: How to reproduce it (as minimally and precisely as possible): Anything else we need to know?: Environment: Kubernetes version (use ): Cloud provider or hardware configuration: OS (e.g. from /etc/os-release): Kernel (e.g. ): Install tools: Others:\n/cc\n/kind enhancement /sig api-machinery\n/assign\nthat does not follow. AvailableConditionController modifies the APIService object. APIServiceRegistrationController must still watch the APIService objects to observe those changes.\noh, I misread the description. You're saying APIServiceRegistrationController doesn't need to watch v1.Service objects"}
{"_id":"doc-en-kubernetes-c082543f1889f90c5ee780d309eeb8f41ee57a7e9e8bb6b6fe7c2e4381ff48af","title":"","text":"What happened: After fixing golint failures under (now still in progress), we need to put for each and . Then common error checking code has become like: We can replace this code with which is more readable and easy to be understood what the code does. Now we have following e2e test directories which can be replaced: [x] test/e2e/apimachinery/ () [x] test/e2e/apps/ () [x] test/e2e/auth/ () [x] test/e2e/common/ () [x] test/e2e/node/ () [x] test/e2e/storage/ () [x] () [x] () [x] () [x] () [x] () [x] () [x] [x] test/e2e/storage/persistentvolumes- [x] () [x] () [x] [x] [x] [x] [x] ()\n/sig testing /kind cleanup\n/assign PR will be submitted soon.\nFix the following in\nFix in\nFix in\nHi, maybe I could take files beneath test/e2e/storage/vsphere/\noops, I'm finished and prepare for PR.\nThank you That's OK\nFix the following in\n/assign\nAll related PRs were merged. I think this ticket can be closed. BTW, should we also replace to ?\nThanks for your work. thing is managed with We are almost finishing the work also :-)\n/remove-kind bug\nI found in these files. I'm working on :\nI can work to the following item."}
{"_id":"doc-en-kubernetes-cea21dfbf74ea26aea8b8fcc0c8d9d3e2b3e4488cd9f6d30773b87652fe6e220","title":"","text":"[x] Original approval issue: [x] Test pr: [x] Two weeks soak end date: 19 November 2020 This issue is created to allow edit by /wip /hold According to this APIsnoop query, there are still some remaining Deployment endpoints which are untested. a Deployment with a static label the Deployment with a new Label and updated data the Deployment to ensure it's patched all Deployments in all Namespaces find the Deployment(1) ensure that the Deployment is found and is patched Namespaced Deployment(1) via a Collection with a LabelSelector Discover useragents: List endpoints hit by the test: Display endpoint coverage change: If a test with these calls gets merged, test coverage will go up by 5 points* This test is also created with the goal of conformance promotion.\n/sig testing /sig architecture /area conformance\nBobyMcbobs commented in original Issue: Unsure if the endpoint replaceAppsV1NamespacedDeploymentStatus is able to be hit. When this endpoint is attempted to be updated the data after retrieving the update resource doesn't reflect the values in the update\n/assign\nWhen using the appv1.Deployment structure instead of maps of interfaces to build the structure in a patch of the Deployment, this error comes up. Still in progress yment-static\":\"true\"}: does not match template , spec.selector: Invalid value: \"null\": field is immutable]\nreplaceAppsV1NamespacedDeploymentStatus to test. Increase cover by 1 Endpoint\n/reopen PR reverted by due to flaky test\nReopened this issue. <detailsIn response to : /reopen PR reverted by due to flaky test Instructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\nIssues go stale after 90d of inactivity. Mark the issue as fresh with . Stale issues rot after an additional 30d of inactivity and eventually close. If this issue is safe to close now please do so with . Send feedback to sig-testing, kubernetes/test-infra and/or . /lifecycle stale\n/reopen Soaking, ready for promotion 19 November 2020\nReopened this issue. <detailsIn response to : /reopen Soaking, ready for promotion 19 November 2020 Instructions for interacting with me using PR comments are available ."}
{"_id":"doc-en-kubernetes-ff9ccf83336a9bf28b0576a991dfda56392cc48f280c272644ff154877b8104d","title":"","text":"If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\n/remove-lifecycle stale\n/triage accepted\n/merged"}
{"_id":"doc-en-kubernetes-c66c9bab938b839459ac01dfd08140ea19c2b9a1199fe50a2b49f94fd9cbbbfb","title":"","text":"<!-- Please use this template while reporting a bug and provide as much info as possible. Not doing so may result in your bug not being addressed in a timely manner. Thanks! If the matter is security related, please disclose it privately via --What happened: I try to backport bookmark feature to our internal branch. When I enable bookmark feature, apiserver leaks memory. I spend some time to dig the root cause and find the bookmarkWatchers of service cacher is keep increasing. I am not sure if it exists at upstream code. Please let me know if it is ok at latest code. What you expected to happen: Avoid memory leaking How to reproduce it (as minimally and precisely as possible): a cluster. not update service. will find kube-apiserver's memory keep increasing. Anything else we need to know?: Kubelet will watch service with bookmark, and add a watcher to bookmarkWatchers. and service init resource version is 0 Without updating service, we cannot update \"lastProcessedResourceVersion\", it will keep 0, bookmarkTimer branch will always be continued. As a result, it cannot trigger popExpiredWatchers, so the service watcher will keep increasing every watching. Environment: Kubernetes version (use ): Cloud provider or hardware configuration: OS (e.g: ): Kernel (e.g. ): Install tools: Network plugin and version (if this is a network-related bug): Others:\n/sig scalability PLAT :)\nHow about making this change ?\n- can you also please take a look?\nAgree with u. That's same way I fix in our branch. But I wanna hear the suggestion of and .\n! With the same bugfix wrote above, we can see the memory not increasing any more.\nGiven that watch-bookmarks doesn't create any additional go-routines (well, creates 1, but cannot leak them), it has to leak the data. The only data that can be leaked is exactly bookmarkWatchers. So I looked into the code and got into exactly the same conclusion - that we must be leaking watchers for resources that aren't changing. The fix as above makes sense (though we also need a unit test for it).\nThanks a lot.\nhas the fix by\nThanks for reporting the bug, I will look at this day later. In addition, also need to think carefully about whether need to send a bookmark when there are no events at all.\ncc"}
{"_id":"doc-en-kubernetes-5702d542be386cac9e4333778515fd94620fcdb52c66d8dfc11487da78971dda","title":"","text":"I noticed while running e2e tests locally that some tests were apparently being incorrectly skipped: I checked, and this is also true for our Jenkins runs. Currently it looks like and have this bug, though only the former actually skips the test, I think. The issue appears to be that the code inside the top-level in these files is being run before the testContext struct is filled out (and before is even called), and so testContext.Provider is empty. I'm not immediately sure what the proper fix is; I'm also worried that there might be other things broken by the fact that these blocks are called before initialization is complete. We also probably investigate if there is a better way to mark these tests as skipped; right now, there's no indication in the test report (, Jenkins test report, etc) that these tests were skipped.\nI merged and which tried to disable these tests for non-{gke,gce} but apparently didn't follow-up and verify that they were still running on our platforms. /cc\nSeems easy enough to fix. I'll have a PR out to fix this before Monday."}
{"_id":"doc-en-kubernetes-07b454797aabeff00bf033fc34fa4c22bd06966ba8574ac0fa868459bacbceea","title":"","text":"<!-- This form is for bug reports and feature requests ONLY! If you're looking for help check and the . If the matter is security related, please disclose it privately via --Is this a BUG REPORT or FEATURE REQUEST?: What happened: Testgrid: https://k8s-\n/assign\nThe test's assumption about test parameter is wrong. Previously it assumed the group name is the same across all zones, but they are different in regional clusters. Looking into using node taints as the permanent fix, as suggested by the TODO comment in the test.\nHm I think taint based evictions is still an alpha feature.\nOK. I propose the following solution: Get the zone of Schedule the StatefulSet pod to that zone using node affinity; wait for the pod to run Remove the node affinity from StatefulSet Delete\nCan it be simplified? Schedule the pod Figure out which zone that pod got scheduled in Delete the instance group of that zone\nUnfortunately I don't know of a good way to look up the instance group given the zone, since there could be instance groups from multiple clusters in the same project\nHow about: Schedule the pod Figure out the zone Taint all the nodes in the zone Kill the pod\nTaint with NoSchedule? yeah that sounds good"}
{"_id":"doc-en-kubernetes-d1a3d40c65839848acfb179492aae28f1dc063c7e214fb748a3d51efaf88302e","title":"","text":"Given a cronjob created with a scedule of \"0 6 *\" that failed at last run (status.lastSuccessfulTime is now \"before\" status.lastScheduleTime). Then I trigger the cronjob by hand and this jobs ends successfully. However status.lastSuccessfulTime is not updated, thus making it difficult to monitor failed cronjobs. status.lastSuccessfulTime should be updated with a timestamp of the succeeded job endTime Install the attached files: Wait until cronjob has actually trigged, test with to see that lastScheduleTime and lastSuccessfulTime are populated ok (as expected). Then apply the fail config-map, now the job will not complete OK: Again wait until cronjob has trigged automatically, observe to see that lastScheduleTime is updated but lastSuccessfulTime is not. Then apply the \"OK\" configmap and then create the job by hand: when the job has finished and completed ok, observe that lastSuccessfulTime is still not updated. Wait until next schedule and observe that the field is now updated. No response <details</details<detailsTested on kind cluster on Mac </details<details</details<details</details<details</details<details</details\nHad to upload the files as .txt files as github wouldn't accept .yaml or .yml files\n(Double-check: I may read wrong messagess ) /sig apps\nThe label(s) cannot be applied, because the repository doesn't have them. <detailsIn response to : When I tried without the prefix in the job name, the status is updated. /sig apps /area job not sure if this is a bug or an expected behavior. Instructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\n/area workload-api/cronjob\nI have just tested in my kind cluster and neither with the same naming pattern as the normal jobs (which apparently isn't epoch) or with epoch - this will not trigger a update of the field. I have also tested it on another cluster (rke2 based) and here I'm not able to get the field updated no matter which naming pattern I use.\nSorry for the mistake. I can repro the problem you have and not related to the prefix. /cc\n/assign\n/remove-sig apps /sig cli\n/triage accepted"}
{"_id":"doc-en-kubernetes-f256e345b7b51b103065fba025a54804b2fa9969c4fe88ae4c8d8670e2ed79f1","title":"","text":"AFAIK, we don't have a documentation on how to add a new API group. Currently we have and that try to add a new API group. I will draft a guide to ease the future endeavor of adding groups. And probably by writing this guide, I will see how can we make the API group machinery easier to use. cc"}
{"_id":"doc-en-kubernetes-b73bb51cef5016ee2dcd80aae4248808fec811a4dad3a3b7abbba7673a446c48","title":"","text":"What happened: No endpoints created in the EndpointSlice for my service due to the port on the service being too long. Logs show: EndpointSlice created has no endpoint: What you expected to happen: Service that works with Endpoints should work with EndpointSlice. If this is intentional, it should be clearly documented How to reproduce it (as minimally and precisely as possible): Create a service with a long port name Environment: Kubernetes version (use ): 1.16 Cloud provider or hardware configuration: kind\n/sig networking\nThe label(s) cannot be applied. These labels are supported: <detailsIn response to : /sig networking Instructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\n/sig network These SIGs are my best guesses for this issue. Please comment if I am incorrect about one. I am a bot run by vllry. ‍\nFor a rough measure of impact, Istio had 5 ports that were over this limit (and many more that were right at 15 characters), so I suspect it may not be terribly rare.\n/triage unresolved Comment when the issue is assessed and confirmed. I am a bot run by vllry. ‍\nThis needs to be resolved before endpointslices graduate from alpha\n/assign\n/remove-triage unresolved"}
{"_id":"doc-en-kubernetes-e059cd0647305d70a174ce6bffb9584a993a023526be428075760ddef27cc3b0","title":"","text":"Which jobs are failing: pull-kubernetes-bazel-test Which test(s) are failing: bazel test //pkg/scheduler/internal/queue:godefaulttest Since when has it been failing: Testgrid link: Reason for failure: WARNING: DATA RACE Read at 0x00c000230db0 by goroutine 47: runtime.mapiterinit() +0x0 (PriorityQueue).flushUnschedulableQLeftover() +0x17a (PriorityQueue).flushUnschedulableQLeftover-fm() +0x41 () +0x61 () +0xce () +0x5a Previous write at 0x00c000230db0 by goroutine 45: runtime.mapassignfaststr() +0x0 (UnschedulablePodsMap).addOrUpdate() +0x30e () +0x601 testing.tRunner() +0x162 Goroutine 47 (running) created at: (PriorityQueue).run() +0x172 () +0xe0b () +0x50 () +0x5a5 testing.tRunner() +0x162 Goroutine 45 (running) created at: testing.(T).Run() +0x659 testing.runTests.func1() +0xa8 testing.tRunner() +0x162 testing.runTests() +0x4ee testing.(M).Run() +0x2ee () bazel-out/k8- +0x2fb Anything else we need to know: I'm very sorry about that. UnschedulablePodsMap is not thread-safe. In , lock will be acquired before operating UnschedulablePodsMap. But in , lock isn‘t be acquired. In the previous case, only a single thread runs so it's ok. But after add goroutine , there will be data race. So I need to acquire lock before operating UnschedulablePodsMap.\n/sig testing /sig scheduling"}
{"_id":"doc-en-kubernetes-4f192dd202c02f7ffbf9fec50f8a461695dfaeb54dc2c0d9c429922f335d948f","title":"","text":"<!-- Please only use this template for submitting enhancement requests --What would you like to be : Current constructors: Could we export a new constructor like to inject custom queue. Why is this needed: In my project, i want to create a with a custom queue based on workqueue.Interface but have no way to do that.\n/sig api-machinery\nthanks for the issue! We've reviewed it in the bug triage, and the consensus was that to ask you to send a PR with this and discuss it there. Would you be able to do it? thanks!\nYes, i will do that."}
{"_id":"doc-en-kubernetes-5881ca0980b851b5719338c7fae783dda8e5f4172ff43857090ef00a488ce809","title":"","text":"<!-- Please use this template while reporting a bug and provide as much info as possible. Not doing so may result in your bug not being addressed in a timely manner. Thanks! If the matter is security related, please disclose it privately via --What happened: when I run , found bound variable err. What you expected to happen: Build success How to reproduce it (as minimally and precisely as possible): Anything else we need to know?: Environment: Kubernetes version (use ): Cloud provider or hardware configuration: OS (e.g: ): Kernel (e.g. ):Linux paasn1 4.14.78-300.el7.Centos.x8664 SMP Sat Jan 5 15:49:43 CST 2019 x8664 x8664 x8664 GNU/Linux Install tools: Network plugin and version (if this is a network-related bug): Others: GNU bash version 4.2.46(2)-release (x8664-redhat-linux-gnu) Copyright (C) 2011 Free Software Foundation, Inc.\n/sig release"}
{"_id":"doc-en-kubernetes-c7797df648733fdf9e156246e323d8c03549ac9a3ca2ad30e283629a87abe7ee","title":"","text":"What happened: kubelet doesn't seem to be able to manage containers started with CRI-O any more I'm testing kubelet + CRI-O on an Ubuntu 18.04 server installation, inside a qemu/kvm VM managed by Vagrant (see ), in order to be able to test k8s + KubeVirt + KataContainers, which are all compatible with CRI-O kubelet will obsess on starting new containers, which looks like and endless process, and the cluster never gets ready. For instance here are some syslogs : What you expected to happen: It used to work (see asciinema recording at ) a few days ago, but I fear some recent changes broke it. How to reproduce it (as minimally and precisely as possible): Testing with vagrant up --provider=libvirt on Linux with a clone of that commit: vagrant ssh into the VM and ps -edf f or tail -f /var/log/syslog would do... Anything else we need to know?: Environment: Kubernetes version (use ):Cloud provider or hardware configuration: qemu/kvm VM +OS (e.g: ): Ubuntu 18.04.2 LTS Kernel (e.g. ): Linux kubernetes-vagrant-01 4.15.0-48-generic -Ubuntu SMP Wed Apr 3 08:28:49 UTC 2019 x8664 x8664 x8664 GNU/Linux Install tools: kubeadm as in Network plugin and version (if this is a network-related bug): Others:\n/sig node\nI can confirm this. If I start cri-o from a systemd service it fails with the above mentioned error message. Starting crio with without systemd works just fine. systemd file:\nThanks for your detailed report :) Just want to do some quick version confirming - you've been using k8s 1.14.1 both when it did and did not work? Any other version changes?\nNot sure for before, when it used to work, but according to the asciinema recording mentioned above (), it seems the same. However, not sure for CRI-O, underlying OS stuff (systemd, kernel, etc.). Sorry.\nI was looking for some information for days. I have exactly the same issue when I use podman + systemd on 18.04.2."}
{"_id":"doc-en-kubernetes-dca27e474f7f12153e878b1f27ed663eb17fd42e2832a567b0d652b26f30c566","title":"","text":"It is the same for both 14.15 and 14.18 kernels. Podman without systemd runs fine. Podman with systemd fails at runc with the same error podman[]: open /dev/stderr: no such device or address podman[]: NAME: runc - Open Container Initiative runtime podman[]: runc is a command line client for running applications packaged according to ... podman[]: time=\"2019-05-11T13:04:59-07:00\" level=error msg=\"open /dev/stderr: no such device or addressn\" podman[]: open /dev/stderr: no such device or address\nI've tried to reproduce it on 18.04.2 using Podman from but I am not able to reproduce it. Do you have any reproducer? How did you install Podman?\nI used ceph unit files to run ceph in podman rather than docker. Systemd unit successfully starts docker but fails with podman. I can successfully run the same podman container run command from the command line. How were you trying to replicate this? Can you post your unit file?\nHi, I'am having the same problem with Kubernetes with CRI-O. After joining a second worker node it fails to start flannel. gives the following error: What I've tried: Installing previous version of Kubernetes (from to ) Installing previous version of crio (from to ) Building runc myself from their github page and selecting this in the Running it on Ubunut 18.04.2 Issue: (the one giuseppe linked)\nI wonder if the issue is with systemd and runc. In both cases, crio and podman, I see runc help output which tells me that something is not being passed to runc that it needs.\nI've only used so it is different than your test and you confirmed it works from the CLI for you as well. Could you share your service file so I test it?\nI've tried systemd and cgroupfs but no luck so far.\nso it seems that runc fails to re-open . That seems like a runc issue, as it doesn't really need to reopen it. To fix the failure, you could use something like:\nJust an example."}
{"_id":"doc-en-kubernetes-1a33d9e403b3728be8b8de7e79372d00143f16de361084780a7a026e224a4dd2","title":"","text":"I use ceph-ansible to install cluster so it may need to install additional daemons like mon before running mgr but the error, in this case, should be different, [Unit] Description=Ceph Manager [Service] EnvironmentFile=-/etc/environment ExecStartPre=-/usr/bin/podman stop ceph-mgr-storage2n1-la ExecStartPre=-/usr/bin/podman rm ceph-mgr-storage2n1-la ExecStart=/usr/bin/podman run --rm --net=host  --memory=16G  --cpus=4  -v /var/lib/ceph:/var/lib/ceph:z  -v /etc/ceph:/etc/ceph:z  -v /var/run/ceph:/var/run/ceph:z  -v /etc/localtime:/etc/localtime:ro  -e CLUSTER=ceph  -e CEPHDAEMON=MGR  -e CONTAINERIMAGE=  {{ cephmgrdockerextraenv }}  --name=ceph-mgr-storage2n1-la  ExecStopPost=-/usr/bin/podman stop ceph-mgr-storage2n1-la Restart=always RestartSec=10s TimeoutStartSec=120 TimeoutStopSec=15 [Install] WantedBy=multi- delighted <:\nBTW, I projectatomic ppa and used apt-get -y podman to install. My podman version is 1.3.0 Sasha Litvak — You are receiving this because you commented. Reply to this email directly, view it on GitHub , or mute the thread\nCorrected file."}
{"_id":"doc-en-kubernetes-bae004f13a06f51a77aaeb8de64c7434d457cced5ee158b56a68659aa46bb958","title":"","text":"[Unit] Description=Ceph Manager [Service] EnvironmentFile=-/etc/environment ExecStartPre=-/usr/bin/podman stop ceph-mgr-storage2n1-la ExecStartPre=-/usr/bin/podman rm ceph-mgr-storage2n1-la ExecStart=/usr/bin/podman run --rm --net=host  --memory=16G  --cpus=4  -v /var/lib/ceph:/var/lib/ceph:z  -v /etc/ceph:/etc/ceph:z  -v /var/run/ceph:/var/run/ceph:z  -v /etc/localtime:/etc/localtime:ro  -e CLUSTER=ceph  -e CEPHDAEMON=MGR  -e CONTAINERIMAGE=  --name=ceph-mgr-storage2n1-la  ExecStopPost=-/usr/bin/podman stop ceph-mgr-storage2n1-la Restart=always RestartSec=10s TimeoutStartSec=120 TimeoutStopSec=15 [Install] WantedBy=multi- Giuseppe Scrivano <:\nWould it have an effect system wide? Can I add it to a unit file or unit override instead? Giuseppe Scrivano <:\nyes, I think it is better to add it only to the unit file\nWhy would runc work outside of systemd? And if it needs stderr, giving it Null in settings will redirect stderr to Null I presume? On Tue, May 14, 2019, 9:46 AM Giuseppe Scrivano <:\nRunning into same issues on Ubuntu 18.04.2 using cri-o-1.13 via PPA.\nPR here: opencontainers/runc\nUntil it is avaialble with project atomic and such, can I use a null workaround? Giuseppe Scrivano <:\nyes, but you will lose anything written to stderr. You could probably also use a file, like\nWhat about actually mounting dev and using /dev/stderr ? On Tue, May 14, 2019, 3:11 PM Giuseppe Scrivano <:\nis a symlink to . It is present, the issue here is that is a socket. The fix for runc got merged, could you use an updated runc?\nI tried apt update/upgrade and restarting crio, but I'm still getting the issue."}
{"_id":"doc-en-kubernetes-a9af1b8339c2af3d5a8899da2b8916e4bc8194e1229dc4a9dc77c862af07f4a9","title":"","text":"Could it be that crio's runc is not updated yet? Edit Build new runc from and it is working now. Think crio hasn't updated it yet with their repository.\nI'm running this version from PPA repo for ubuntu and it fails: could you point to version/commit or whatever helps identify the fixed version/snapshot ? Thanks in advance.\nat the moment it is only fixed upstream. You'd need to build runc manually\nIs there a version of the package on ppa we can downgrade to? On Wed, May 15, 2019, 4:35 AM Giuseppe Scrivano <:\ncould we update the runc version in the PPA repo?\nThank you. I seem them built but nut pushed to ppa yet.\nshould be available now.. let me know..\ncri-o-runc (1.0.0-rc6-1~dev~ubuntu18.04~ppa42) bionic; urgency=medium autobuilt -- Lokesh Mandvekar (Bot) <, 15 May 2019 16:54:40 +0000\nThank you, everyone, for a quick turn around. Quick testing showed that fix is working, but more testing is needed. Lokesh Mandvekar <:\nI confirm my initial tests with seem to confirm the regression is fixed. Many thanks\nI also confirm the fix working for podman. Olivier Berger <:"}
{"_id":"doc-en-kubernetes-c1c167f97d0d68a2d507d0d919d58cd680ff8ed576e3225c7b24d703a964bd61","title":"","text":"<!-- Thanks for filing an issue! Before hitting the button, please answer these questions.--Is this a request for help? (If yes, you should use our troubleshooting guide and community support channels, see ): No What keywords did you search in Kubernetes issues before filing this one? (If you have found any duplicates, you should instead reply there.): initContainers imagePullPolicy initContainers required value Is this a BUG REPORT or FEATURE REQUEST? (choose one): BUG REPORT <!-- If this is a BUG REPORT, please: Fill in as much of the template below as you can. If you leave out information, we can't help you as well. If this is a FEATURE REQUEST, please: Describe in detail the feature/behavior/change you'd like to see. In both cases, be ready for followup questions, and please respond in a timely manner. If we can't reproduce a bug or think a feature already exists, we might close your issue. If we're wrong, PLEASE feel free to reopen it and explain why. --Kubernetes version (use ): Environment: Cloud provider or hardware configuration: GCE OS (e.g. from /etc/os-release): Kernel (e.g. ): Install tools: cluster/kube- Others: What happened: When trying to create either a ReplicaSet or Replication Controller whose Pod template contains an Init Container, the validation fails saying is a required value. What you expected to happen: The create should succeed and the Init Container should receive a default . How to reproduce it (as minimally and precisely as possible): Try to create either of the following: ReplicaSet Replication Controller Anything else do we need to know: If I add to the Init Container in the ReplicaSet or Replication Controller Pod template, the create succeeds. When I create an Init Container without in a StatefulSet Pod template, it succeeds. When I create an Init Container without in a bare Pod, it succeeds:\nLooks like defaulting not being applied on server? That is probably due to removing defaulting from convert. Surprised we didn't catch this earlier\nSome values still not being defaulted\n, I'm going to create a PR if none works on it.\nWe decode from bytes, set defaults and then convert; but for init containers, we delay the decode from \"decode phase\" to \"convert phase\" at ."}
{"_id":"doc-en-kubernetes-69dbac4fd7f4ee2d4feb0af6994cc2f74e65b4d3465b8e0c7014cfb5909cea9b","title":"","text":"The defaults has been handled for initContainer at . /cc I'm working to find a solution for that; if any suggestion, please let me know.\nWould this bug affect the deletion of jobs with an init-container? I ran into this bug and adjusted my manifest to add the pull policy to the init-container. However, I now cannot delete said job claiming a field is immutable. Example:\n, not sure now. Let me check it,\nSame error here with version 1.5.1, 1.5.2 The Job \"grafana-import-dashboards\" is invalid: [0].imagePullPolicy: Required value With this setup: kubectl apply -f monitoring Not working even if I specify a pullpolicy in the yaml. If I roll back to 1.4.7, it is working. Thank you. Greg.\nany updates on a PR to address this issue? Is anyone from able to pick this issue up otherwise?\n, here's the latest status: there's is a PR to generate for each Object (), it works for Object (Pod, Job) now but List; I'm working to make it generating for List.\nWhy would you need List to work? No client side logic should be decoding lists except as a generic submission.\nThe List tests are failed in . There's no for List, so no code to decode init-container annotation, e.g. of should call 's recursively.\nThanks for the update\n+1\nAny update on this? I still getting this issue with the last version of 1.5.3 on GKE?\nping for the comments? should we merge the PR for 1.5?\n, there's a PR () that handled that issue; but we're discussing not to merge it because initContainers had been moved to fields, and fields is the recommended way to use since 1.6. But I'd like to get & 's comments on 1.5.\nThis is still an issue in 1.5.3 when using annotations: Without the deployment would not be accepted\nping & :).\ni dont think this is fixed in 1.5 I just encountered this bug on 1.5.3\n, yes, the PR is NOT merged to any branch, so the issue is still here. I'll create a PR for 1.5.\nHi Anyone help this issue."}
{"_id":"doc-en-kubernetes-cd1dc8216bffde40e45195a0aceb1bbd0c560296ad557f563dafcf3df89eeea5","title":"","text":"#cat mongo- apiVersion: v1 kind: Pod metadata: labels: name: mongo-svc-a name: mongo-deployment spec: replicas: 1 template: metadata: labels: name: mongo spec: containers: image: mongo name: mongo ports: name: mongo containerPort: hostPort: volumeMounts: mountPath: \"/data/db\" name: mongopv volumes: name: mongopv persistentVolumeClaim: claimName: mongopvc When i want create mongo db container by using this file folowing error is occured. The Pod \"mongo-deployment\" is invalid:spec.containers: Required value\n, it seems there're two here, please remove the first one and have a try.\nHi my kubectl binary is configured at ubuntu 16.04 machine."}
{"_id":"doc-en-kubernetes-7c77a1e7ea84d582b899fa6b8d7216504334c47e064fb8eaad5c523c3bc66a32","title":"","text":"NAME LABELS STATUS 10.1 Ready,SchedulingDisabled 10.2 10.1.10.2 Ready 10.3 Ready apiVersion: v1 kind: Node metadata: creationTimestamp: 2017-03-03T14:41:52Z labels: 10.1.10.1 name: 10.1.10.1 resourceVersion: \"\" selfLink: /api/v1/nodes/10.1.10.1 uid: -001f-11e7-8e83- spec: externalID: 10.1.10.1 status: addresses: address: 10.1.10.1 type: LegacyHostIP address: 10.1.10.1 type: InternalIP capacity: cpu: \"2\" memory: 4041352Ki pods: \"110\" conditions: lastHeartbeatTime: 2017-04-05T18:58:42Z lastTransitionTime: 2017-03-03T14:41:52Z message: kubelet has sufficient disk space available reason: KubeletHasSufficientDisk status: \"False\" type: OutOfDisk lastHeartbeatTime: 2017-04-05T18:58:42Z lastTransitionTime: 2017-04-05T15:27:52Z message: kubelet is posting ready status reason: KubeletReady status: \"True\" type: Ready nodeInfo: bootID: -6124-4ac7-ab96- containerRuntimeVersion: docker://1.12.6 kernelVersion: 4.9.16-coreos-r1 kubeProxyVersion: v1.2.4+coreos.1 kubeletVersion: v1.2.4+coreos.1 machineID: osImage: Container Linux by CoreOS 1298.7.0 (Ladybug) systemUUID: C3FD9852-2DFB-21F4-CE07-2597A746D99D When i create pv for creation pod is shows pv is pending. Please give any suggestions.....\nIs there solution of this problem with version 2.7. Still facing the same. what exactly expectation of helm upgrade --install relname chartsdirectory is is expected to work on first time installation in case with CI/CD Pipelines."}
{"_id":"doc-en-kubernetes-65bf469670180d69e57fdc1a820a8f2f435cec685a9975e90f226dd126851fba","title":"","text":"As far as I can tell, container liveness probes of type exec flat out don't work against clusters built at head. I had the kubelet always log the output of the exec'ed command, and it's always empty no matter what I exec, causing the probe to fail since it requires the output to be equal to the string \"ok\". If you want to test it, try adding a liveness probe of this form to any container: I also tried these lovely forms, to more closely match the liveness example under examples/: On the plus side, this may be an opportunistic time to switch over to\nThe reason that the e2e test for exec liveness probes hasn't caught this is that it only tests that containers are restarted when they fail liveness probes. It doesn't test that containers aren't restarted when their liveness checks should be succeeding. The e2e test's exec'ed commands also only ever produced empty output, rather than the expected \"fail\" output.\nIs it possible to include this in a new pre-release, since it fixes a bug in health checking? I see it merged in master and I'm wondering what the policy is of releasing new code in a separate (0.16.*) release."}
{"_id":"doc-en-kubernetes-65d9c28fb77fe0bb882df060209fc22bf399af9c3b9dfe5829b0e5a45dc20b3d","title":"","text":"<!-- Please only use this template for submitting enhancement requests --What would you like to be : Consolidate ) into Score plugin set in config API. Why is this needed: NormalizeScore extension point is implemented in . We decided to treat NormalizeScore as an \"optional\" step of Score plugin, see . However, we deferred the removal of due to API review delays. Now that PR was merged. We can start the actual removal from the API and start the API review process. /sig scheduling\n/assign"}
{"_id":"doc-en-kubernetes-2d8b9ac51e195a52880b5015da51ade673286a0fb623e49df0244ac4dfec5669","title":"","text":"PreStop in Lifecycle is a very excited feature that I want it so much. With it, I can do some clean work before killing the Pod/Container, for example, I need this container stop receiving requests first, then complete processing its requests already received before it is killed. However It does not work as PostStop now. And I have checked the latest code and found it is still not implemented. Could anyone give a time point on it?\nSorry, prestop hasn't been implemented yet. See discussion in .\nI'll implement it once is finished\nthanks, I will also subscribe\n/subscribe\nMoved to post-1.0\nPR is in flight.\nwas merged. We should add an e2e test, though.\nTried by E2E, it works! And it can be closed. Well done."}
{"_id":"doc-en-kubernetes-1de7ff63e7ce8b64bf52f2105a2efc7375b6d0f9c29fe56834b2956453e0280d","title":"","text":"<!-- Please only use this template for submitting enhancement requests --What would you like to be : Allow to have an additional flag (e.g. or ) to allow a to drain pods off a node, even if they violate PodDisruptionBudgets Why is this needed: There are times where you want to drain pods off a node even if violating a PDB. For example, if you are running a one-worker node kubernetes cluster (not minikube), and you want to upgrade it. Or, like me, you are running a test cluster and want to scale down to zero nodes when you're not going to be using the cluster for a while in order to save costs. If that worker node has a deployment with a PDB, then will get stuck as it cannot get rid of the pod while respecting the PDB, which is expected; but in this case, it is a hindrance. Adding the switch will allow users to explicitly acknowledge that they want to ignore the PDBs (and accept the corresponding downtime for an application/service)\n/sig node\n+1\n+1 Our company offers Kubernetes as PaaS. We manage Kubernetes, including IaaS OS updates and if a customer does not specify the PDB correctly, we are not able to drain the node for the OS updates. A flag would be werry helpfull for our use case.\n+1\n/active\nI have created to address this. This is a pretty simple implementation, but I think it achieves the desired results. A suitable workflow would be: 1) Configure you component to use drain as normal 2) After some timeout, run drain with --disable-eviction option.\nI'll wait until the PR is merged and released before testing, but if I understand right, it would require me do this, right? If so, I think it would be better if we did it a different way and have one call, something like this: Where implies that if the drain is still running after this time, switch to the mode without having to call kubectl again\nI think that would require a sizeable refactor or add a great deal of complexity to the code, the same could be achieved with In practical terms, I'm focused on automatic node remediation via destroying the node and getting a new one. The timeout would be very large, the drain library is blocking, so I want it to fail quickly and requeue."}
{"_id":"doc-en-kubernetes-b7bae32dd27973078cd1b23d1dbdce25c736eb34a86ee68bc51d78045bb08cfb","title":"","text":"I'm going to be performing the calculation of when it's appropriate to evict/delete outside of the drain library, and I suspect most will be as well, so there's not really a great incentive to do it in the drain library itself IMO.\nThat's all good, I have no issue using the flag as long as does error out and return a non-zero code as expected (there's been a few times where it seems to error out, but still return 0)"}
{"_id":"doc-en-kubernetes-0ebc3a18530206d54bf966306c4d4d56c4fb7ca61cedeffe6a0cab2bf1fe76ea","title":"","text":"In we discovered a bug where kube-apiserver metrics were being incorrectly reported (GET vs LIST problem). In this was addressed together with test exercising that proper metrics are exposed. However, this test isn't ideal because if params wouldn't be swapped, it wouldn't catch them. We should ideally develop a lower-level unit test for mapping url paths to exact values of labels/parameters. /sig apimachinery\nThe label(s) cannot be applied, because the repository doesn't have them <detailsIn response to : In we discovered a bug where kube-apiserver metrics were being incorrectly reported (GET vs LIST problem). In this was addressed together with test exercising that proper metrics are exposed. We should ideally develop a lower-level unit test for mapping url paths to exact values of labels/parameters. Instructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\n/sig api-machinery\nare you planning on sending a fix for this one too? Thanks!\nNot very soon - if someone else could help with it, it would be great.\n/assign\n/triage accepted\ncc"}
{"_id":"doc-en-kubernetes-540712d0b27ee7f52d7d02bd3719013304c1ce42c6b5d8b93c3ee3671a9c1c7c","title":"","text":"The \"Volume plugin streaming [Slow] NFS should write files of various sizes, verify size, validate content\" e2e test is failing on our 2k-node gce clusters (https://k8s-) with the following error: cc\nCould you take a look into it or reassign as apt? This is part of the correctness suite and large cluster tests are release-blocking.\n/assign\nis this run using the kubemark environment? I'll also take a look. It could be that the commands this test needs to execute may not work with kubemark. /assign\ncc\nNo, this is not kubemark. This is on a real 2000-node cluster.\nAlso all other volume-related e2e tests are passing, except this one.\nLooking at the log I see that the nfs test fails trying to verify the test file's content. The test exec's into the nfs client pod and executes: The return code from above is 127. The test then deletes the nfs client pod calling which works. The nfs-server pod is deleted in the AfterEach. For some reason the reported error is 137 but the returned err 127. The same command is exec'd for all of the tested storage plugins so not sure why it's failing just for nfs. Error 127 might be \"command not found\".\nThe other strange thing is that it's only failing in this suite and not the others.\nWhen I look at the kubelet log, I see the exec command coming through: I also see these suspicious kubelet soon afterwards: So I think the grep command is causing the node to run out of memory and get evicted. I think we can fix this either by: Making the grep command more memory efficient. I'm guessing it's reading the whole file into memory or, Make this test Serial so that it doesn't run in parallel with other pods, since it consumes a lot of memory.\nAn unrelated issue, is that the node allocatable feature didn't actually evict the pod. Instead this pod got a system OOM kill. /cc\nrelated to the memcg notifications, not node allocatable. Thanks!\nany theories why we see this only for the nfs-io tests, and not the glusterfs-io test for instance?\nthe gluster test didn't run on this in this suite."}
{"_id":"doc-en-kubernetes-fd4ca8cfd935058701597731533334687f2913bf0b9d409a83f3ac801f0182e1","title":"","text":"I'm suspecting that the reason why we didn't see a problem with this test on other suites is that the load could be higher here. I've asked to see if we can reduce the memory usage of the grep call.\npointed out that there are no newlines in the generated file, so when grep is called it could buffer the entire file in memory. Adding a newline at the end of the dd input file should do it.\n/priority critical-urgent\n[MILESTONENOTIFIER] Milestone Labels Complete Issue label settings: sig/scalability sig/storage: Issue will be escalated to these SIGs if needed. priority/critical-urgent: Never automatically move out of a release milestone; continually escalate to contributor and SIG through all available channels. kind/bug: Fixes a bug discovered during the current release. <detailsAdditional instructions available <a href=\"\"here</a</details\nIt seems like even if the generated file is broken up by newlines, grep could still use on the order of hundreds of MB of memory for a 100MiB file (even with the -m1 and -F options). We could check the file line by line, but it would take a long time. A better solution would be to check the MD5 file hash. It's fast and uses little memory. The expected hashes will be hard-coded into the test, and if different file sizes are in the future, a hash needs to be computed and as well.\nThanks a lot for fixing this!\nRef"}
{"_id":"doc-en-kubernetes-effe9bff1df2586252b385cbf4a7cacd04cb77f4739ff9c987e63818f9dc1c82","title":"","text":"Currently there is no way to run without any identity (be it service principal or E/MSI). In reality code running on a worker node can use via - and only via - without the need to access ARM. Here is the proposed change a new mode authN = this mode, is basically telling the cloud provider to work any identity, this means only IMDS calls is allowed. This mode can be detected from if no SPN set and both MSI settings are set to . to use a config map within the cluster. This config map can be loaded on the cluster statically. or created - if not existent - in that then only is used in The net new result of the above is users bootstrapping clusters can configure the worker node to use without any identity on said worker nodes.\n/sig azure These SIGs are my best guesses for this issue. Please comment if I am incorrect about one. I am a bot run by ‍\nIn-tree AzureDisk volume plugin is also depending on Azure credential to fetch lun id. we should figure out how to solve it without credential\nfetch disk lun id is also a dependency in kubelet code now, after my investigation, it could be removed on Linux node, while it cannot be removed on Windows node, need to change the k/k upstream code, there is a discussion on going here: , let' wait for the discussion result.\nFor Linux: 1- use IMDS my understanding is new api versions will have the disk/lun pairs. 2- don't use arm for the return of is the second parameter passed to . For Windows: I don't understand the dep on ARM for this. The above still apply, we just need to find a way to encode the value as ut return from that allow us to find the disk by on a windows machine. might be a good source for this. Ideally speaking. The value should be the same. But i will leave it for you to decide.\ncould be a good idea since currently it's a LUN num returned by Attach() and then it could be changed to the windows disk num, I will make the code change on v1.15 later, thanks for discussion.\nUpdate: PR would remove the VM API calls when getting the LUN number for AzureDisk."}
{"_id":"doc-en-kubernetes-1ae4374d89f67ace7ae0a669d00ed1569d20a542da5564faf73570670010978e","title":"","text":"What would you like to be : Support dynamically set log level for kube-scheduler, is blocked on node controller approval, but we can move forward with a scheduler only change. /sig scheduling\n/help\n/assign /remove-help\ndo you have a PR for this one?\nI submitted the PR."}
{"_id":"doc-en-kubernetes-4e19a56e8ccacbab4546f622baecdb0b8ef30250430c55f8bcdc4de99b100cd2","title":"","text":"Found in openshift, but I'm pretty sure the codepaths are the same. That means the token is getting ignored and in-cluster config is being used in preference to flags.\nIs a similar issue?\nYep, kind of regression.\nSigh.\nYes, most likely the same cause.\nI'm on it.\nUnit test :)\nProbably will want to cherrypick this into 1.5.\nWIP in"}
{"_id":"doc-en-kubernetes-671a12759d851f1b99028bc477360c418e1157a45a2e814ff1c5c1cdd09da36a","title":"","text":"This helps debugging scheduler's behavior, currently we don't have logs that indicate when the scheduler first saw a pod or a node. The log level should be v=3. /sig scheduling /cc\nDo you mean add logging for callbacks in ?\nyes, at least for add and delete events\nSure, I can also give this pr, before tomorrow\n/assign\nWe actually have one v=2 log line for a Node being to the cache's tree: But it's probably good to keep this one."}
{"_id":"doc-en-kubernetes-3871d24647a528f2ff7df25634241fd3cde165662602f0e95a0cba53fe987aee","title":"","text":"<!-- This form is for bug reports and feature requests ONLY! If you're looking for help check and the . --Is this a BUG REPORT or FEATURE REQUEST?: /kind bug /priority failing-test /area test-infra What happened: job is failing. See https://k8s- https://k8s- https://k8s- https://k8s- What you expected to happen: test should pass\nThere are no sig labels on this issue. Please by: a sig: e.g., to notify the contributor experience sig, OR the label manually: e.g., to apply the label Note: Method 1 will trigger an email to the group. You can find the group list and label list . The in the method 1 has to be replaced with one of these: bugs, feature-requests, pr-reviews, test-failures, proposals\n/assign\nReiterating the mentions to trigger a notification: <detailsIn response to : /assign Instructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\nI'm guessing that something got messed up in the bazel cache when reverting from 0.5.2 to 0.4.5. I'm not sure how best to fix.\nHow about updating everything to bazel 0.5.3 at once? I know you're trying to do a step-by-step approach to avoid breakages, but I suspect it will be cheaper to just deal with a brief outage while we update? (aka update the bazelbuild image to 0.5.3 and then merge your PR with updated and working BUILD rules?)\nthere's at least one other issue I've discovered with the latest go rules - the binary no longer works, due to changes introduced in that might be arguably better (at least everything still builds?), though it's frustrating that it makes things more broken in a different way."}
{"_id":"doc-en-kubernetes-dd58629e743b6e4eeaad6ce708921acac15734ca4e62ad8f92d2c296b406f940","title":"","text":"What happened: SCTP nodeport service is not working for all IPs present in the node when ipvs is enabled. It is working only for node's InternalIP. What you expected to happen: SCTP nodeport service must work for all ips present in the node when ipvs is enabled. How to reproduce it (as minimally and precisely as possible): Create a kubernetes cluster with at least one node having multiple interfaces with different IPs. Start an SCTP server deployment. Then create a nodeport service for the same. Now try to reach the service using all the IPs present in that node. The service will be accessible only via node's InternalIP not any other IPs. Environment: Kubernetes version: v1.15.2 OS: ubuntu 18.04 Kernel: 4.15.0-54-generic Network plugin and version (if this is a network-related bug): ipset version v6.34\n/sig network /area kube-proxy /area ipvs\n/triage unresolved Comment when the issue is assessed and confirmed. I am a bot run by vllry. ‍\n/remove-triage unresolved"}
{"_id":"doc-en-kubernetes-1e3d5f814d3d1cd0ae92f2f9adc262569d0e3418b95dbcb250b25bdb3821fc75","title":"","text":"This issue proposes the writing of a new test for the API endpoint with the intention of promoting it to Conformance. This issue contains a very masic mock test in order to start the conversation about what a good e2e test would look like for this endpoint. The code below is verified to be hitting the intended endpoint, show as per the queries in APIsnoop's live view of the cluster's audit logs. According to this query, there is only one API endpoint which is for Namespace that isn't being covered. We selected patchCoreV1Namespace to write a test for. Dependencies: The mock test: If a test with these calls is build and merged, conformance will go up by 1 endpoint.\n/sig testing /sig architecture /area conformance"}
{"_id":"doc-en-kubernetes-f39360e36dc13c7f3c135efe81c665373563851402d40604cca632c72aed6a7c","title":"","text":"<!-- Please use this template while reporting a bug and provide as much info as possible. Not doing so may result in your bug not being addressed in a timely manner. Thanks! If the matter is security related, please disclose it privately via --What happened: Default scheduler crash if scheduler extender filter returns a not found node What you expected to happen: Default scheduler just should consider the not found node as not an candidate, instead of crash itself (panic). How to reproduce it (as minimally and precisely as possible): cluster only have 1 node: 0.0.0.1 scheduler extender (with nodeCacheCapable enabled) filter accidentally returns a candidate node: 0.0.0.2 scheduler recieved the candidate nodes: 0.0.0.2 from scheduler extender, but got below error: The is SuggestedHost: filteredNodes[0].Name: Anything else we need to know?: Environment: Kubernetes version (use ): Cloud provider or hardware configuration: OS (e.g: ): Kernel (e.g. ): Install tools: Network plugin and version (if this is a network-related bug): Others:\n/sig scheduling"}
{"_id":"doc-en-kubernetes-43588246e28947764a0f8a5c9c5b1d76a5819a8eaa1c439f79abd429269822c0","title":"","text":"What happened: kube-scheduler () crashes due to a panic while running : The race condition is triggered when an out-of-tree plugin, that writes to during its Filter phase, is registered through . Access to in general happens from a single goroutine and there are no practical concurrency issues, with the exception of the phase, that parallelizes calls for each node across plugins: A quick glance at the tip of the branch seems to indicate the bug is still present there. What you expected to happen: Ideally access to should be synchronized everywhere for correctness, including by all in-tree plugins, since there are no guarantees about how will call multiple plugins concurrently. It seems to me it is something that was considered at some point, given that has methods. Sadly, almost no in-tree plugins are synchronizing access. For now, I will try to work around the problem by doing a single write to during and storing a data bag structure where I will write state into during (with proper synchronization), instead of writing to directly. How to reproduce it (as minimally and precisely as possible): Install a scheduler plugin that writes to during its phase. In-tree (builtin) plugins are not synchronizing their access to . Environment: Kubernetes version (use ): v1.18.8 Cloud provider or hardware configuration: linux-amd64 (AWS) OS (e.g: ): Ubuntu 18.04.5 LTS (Bionic Beaver) Others: custom kube-scheduler binary built with an out-of-tree plugin using\nThis issue is currently awaiting triage. If a SIG or subproject determines this is a relevant issue, they will accept it by applying the label and provide further guidance. The label can be by org members by writing in a comment. <detailsInstructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\n/sig scheduling\nAFAICT all in-tree plugins write to the cycle state in PreFilter and then only read from cycle state in Filter. So this shouldn't be a problem with in-tree plugins as concurrent read access should be fine during RunFilterPlugins. But since your out-of-tree plugin writes in Filter, I can see why this can be a problem."}
{"_id":"doc-en-kubernetes-278036efba637c2a6a49330142a64ca8f7b65a0d21755fde24824f9c473a00c9","title":"","text":"Aside from adding Lock/Unlock in in-tree plugin, offering mutual exclusion on each key as a built-in feature within CycleState could be an alternative solution. This way, reads from in-tree plugins during Filter wouldn't interfere with writes during Filter from out-of-tree plugins.\ncc\nshould we change the Read/Write CycleState methods to be thread safe?\nAdding locks without a proper justification would just increase the latency. could you clarify your use case? Why do you need to write anything during Filter?\nyeah, we should measure it, but I don't expect it to have significant effect...\nAdding locks without a proper justification would just increase the latency. Without contention (which is the case with in-tree plugins), I don't think this is a concern.\nAdapting source from : source: essentially, 11.3 ns/Lock-Unlock pair\nI remember locks showing up in profiling for PreScore/Score for certain plugins. Although I have to admit that the accesses were not properly implemented. I'd like to hear an answer for before doing any experiments.\n+1 to that I'd also like to understand the motivations / requirements of mutating CycleState in Filter().\nMy particular use-case is that we have resources being dynamically to/removed from Nodes, and those resources need to be allocated (and tracked) in the scheduling loop. Technically I could pre-compute all the available dynamic resources for all Nodes at the phase, and snapshot it, but that would require me to pre-compute everything for all Nodes on all scheduling cycles. Instead, what I'm currently doing is computing what dynamic resources can be utilized on-demand during the phase (kind of a lazy loading of dynamic resources for each pair), and saving state of what extra resources were considered during for each Node that was evaluated. Then, later in the scheduler plugin can lock-in the exact dynamic resources for the particular Node that was selected. I understand the concern with adding extra synchronization overhead without a real need. Even if my use-case is not good enough to justify it, there is another argument for it: technically, the lack of access synchronization on violates the . Even if in the end we decide synchronizing all access is not worth the overhead (for correctness), I would strongly encourage better documentation saying which plugin phases are allowed to modify , and perhaps even forcing a snapshot-read-only copy of to be used by plugins.\nUsing the scheduler to track resources is not a good idea."}
{"_id":"doc-en-kubernetes-b826aaa34ec054d9cc587ce8587a04002d4daa70694ce73071b413f1bf7ba0c6","title":"","text":"The scheduler should be stateless. Can't you track the resources with an external controller or with a device plugin and making it part of the Node spec?\nthere may be some misunderstanding in my poor choice of words. What I mean by \"tracking\" is that the scheduler needs to track usage and availability of these extra per-Node resources, to be able to allocate them during placement/scheduling. Exactly the same it does with volumes, cpus, memory, ephemeral disk, etc Doing computations in a controller outside of the scheduling loop would be similar to \"precomputing\" these resources. In my particular use case I chose a different way because: introduces another point of failure, if the external controller stops pre-computing extra resources and publishing them somewhere (CRD, Node spec, etc) scheduling halts (at least scheduling on these extra resources) Node resources in my use case are volatile (they are coming and going frequently), and the extra latency by an extra controller (vs doing all computations right in the scheduling loop) may be problematic for my use case. Even if you consider my use case not enough to add proper synchronization to , I'd still encourage making it explicit what scheduling phases are allowed to modify it (and even passing in a snapshot read-only version of to phases that are not expected to update it), to avoid confusion and race conditions.\nOk, if you are not tracking the existence of resources in the scheduler, I wouldn't suggest the external controller. Still, have you considered the device plugin framework? That way, the default kube-scheduler could keep track of resources usage. Otherwise, I still don't see why you would need to update the CycleState during Filter. As its name indicates, CycleState only lives during a scheduling cycle.\nRegardless of the correctness of usage of CycleState to do what wants, I still think we need to do at least one of the following: that writing to CycleState during Filter will lead to data races, a read-only CycleState to Filter plugins (whether we do this at the interface level or if we simply drop Write requests during Filter is an implementation detail), or make CycleState concurrency safe CycleState as it exists right now allows plugin authors to shoot themselves in the foot IMO.\nI'm leaning towards 1, but it's fair to continue the discussion before deciding.\nCan you run the \"controller\" as a goroutine along with your scheduler plugin?"}
{"_id":"doc-en-kubernetes-038a628d524f09c8393af798b6053a1fded9e84e2c01785ad258fb86c8d6d618","title":"","text":"So it's not an external controller binary, and hence don't introduce another point of failure. It's pretty like how the default scheduler manages its internal cache. But note that cache is only needed when you need a more efficient data structure. If a {key:value} map is good enough, spinning up a XYZInformer and obtain the latest XYZ resource on-demand is the most recommended way. I think it quite depends on how you design the internal XYZ cache. For example, it should only tweak the cache in an incremental manner - in other words, unchanged resource will be kept unmodified. You can check out the \"doubly-pointer snapshot\" of scheduler's cache. With above said, if I understand your case correctly, you may need to build a sophisticated XYZ cache, which runs in a separate goroutine and responds to your XYZ resource changes. And then upon each scheduling cycle, take a snapshot (may needs a specific snapshotting algorithm) in PreFilter() and WRITE to CycleState, and then in Filter() READ from CycleState. My personal option is to suggest all plugin developers to read from CycleState in Filter().\nnailed it. What you describe is almost exactly what I’m doing today. There are some details particular to my use case (e.g. just tracking current state with an informer and an incremental cache is not enough, because there’s some logic involved in deciding if currently available resources are usable by each pair, and I currently do that computation in as pairs are evaluated). In any case, I just wanted to share my use case to provide some color, but I don’t want what I’m doing distracting us from the more important questions, mentioned by above. My personal preference would be as a very short term next step, then when someone has cycles to do it. I would be OK with punting on until a more strong need shows up.\nNote that limiting to only reading from works for me, as long as I can still read my own data structure from it and do writes on it (i.e. don’t deepclone for the phase).\nalso adds overhead (from copying the state, which can be very expensive).\nCould implement just by changing the interface passed to (e.g. ). That wouldn’t be backwards compatible, but very cheap."}
{"_id":"doc-en-kubernetes-73e2408e2bea0ec3fe3177e8a87291d57a052eb7c05f546c204ec27aefee3293","title":"","text":"I guess that would work in the sense that you cannot change the high level values of the CycleState map, but you could still override what's inside them, if they have pointers. But maybe that's ok.\nbut you could still override what's inside them, if they have pointers. But maybe that's ok. yeah, if it was pointers and if the underlying data needed to be modified, I'd imagine it would have its own Lock/Unlock semantics\nfor such implementation, the only remaining concern is backwards compatibility. So I would leave that decision to who leads the scheduling-plugins repo.\nI rethink 's original problem a little bit: the original racing issue occurs when reading CycleState in in-tree plugins, this indicates one of the (out-of-tree) plugins is (over)writing the most high-level value of the CycleState map. But if we control the locking inside the out-of-tree plugin, and mutate that particular in-place, that would be good enough? Case 3 in the above program demonstrates the correct way to manipulate the specific out-of-tree state. Let me know your thoughts. If it resolves 's concern, we don't need to make any change on the Filter() interface.\nCase 3 is how I am working around the problem today to avoid the race. I write a \"data bag\" once into top level in , then only modify my own inner structure (with proper locking) in , without doing any further modifications to directly (top level) in . That is acceptable, but I'd encourage you to consider making the interface (and documentation) more explicit for plugins, so they avoid the mistake of writing directly into . Without changing anything (no documentation, no readonly interface), it is pretty easy for a plugin to use in a way that you don't want to support.\nThanks for confirming that. Yes, documenting it, or(and) publishing examples would be quite helpful to clarify the semantics. As a wrap-up: a keyed value can be read from CycleState, literally , in any phase. It's always race-free."}
{"_id":"doc-en-kubernetes-3be32b87589c297c7c2842b086ef786e82de3563d7c057c1ca18b1f858ccd170","title":"","text":"if you want to mutate the top-level keyed value in CycleState, if it's a phase that CycleState is accessed consecutively, such as PreFilter, PreScore, it's safe to mutate CycleState in any manner if it's a phase that CycleState is accessed parallelly, to be race-free, (for now) you only have 1 option: keep the keyed value's pointer unchanged, and use an internal lock to mutate that pointer obj [TBD] we may come up with changes to add read locks to all in-tree plugins when reading values from CycleState, so that the out-of-tree plugin can surround its mutating logic between and - th mutating logic thereby can be any style Any thoughts on the TBD item? I would assume read locks wouldn't cost too much.\nFor now, I think the documentation is correct, it clearly says Read() and Write() are not thread-safe: The key thing here is whether in-tree plugin should use RLock/RUnlock or not. If yes, out-of-tree plugins can be freed to use Lock/Unlock. (see the TBD item in the previous comment)\nI like the TBD, and optimizing for correctness in how in-tree plugins are accessing . I understand the concern with synchronization overhead, but I would be really surprised if it is the bottleneck in scheduling code.\nIt can be, when you have 5k-nodes clusters, or bigger.\nI don't think it will be expensive, a rlock should be equivalent to an atomic add operation (basically it increments the number of readers). As a comparison, looking up the state from the map is likely to be more expensive.\nOf course we are assuming no contention in the default plugins.\nThis is my impression as well. A vanilla version of scheduler doesn't impose write lock in Filter() and Score(), so should be good.\nWe can always see with the benchmarks and revert if its not negligible.\nI will run the benchmarks to see if there is perf downgrade. /assign\nIssues go stale after 90d of inactivity. Mark the issue as fresh with . Stale issues rot after an additional 30d of inactivity and eventually close. If this issue is safe to close now please do so with . Send feedback to sig-contributor-experience at . /lifecycle stale\n/remove-lifecycle stale This is still relevant, yes?"}
{"_id":"doc-en-kubernetes-bcac7b47c31a3bdcd9ddd71bfacab7dca4a98e08ad1188f8a89f61b6d0a63d38","title":"","text":"IIUC the volumebinding plugin has been fixed, but external plugins can still shoot themselves in the foot.\n/priority backlog\nYes. I'm working on this, late this week I will send out a PR along with perf test result."}
{"_id":"doc-en-kubernetes-4bcfc015d2b9d9fa1e84475a3e08fb9ef8299b4e5f821f790188ad7bdae6ecbc","title":"","text":"<!-- Please use this template while reporting a bug and provide as much info as possible. Not doing so may result in your bug not being addressed in a timely manner. Thanks!--What happened: There's , and . The former is used by local and nfs volume plugins, while the latter is used by , which is also used by local and nfs volume plugins. What you expected to happen: Can we consolidate these? This is really confusing and could easily be the source of bugs. <!-- DO NOT EDIT BELOW THIS LINE --/kind cleanup\ncc\n/assign I'd like to clean up code."}
{"_id":"doc-en-kubernetes-ef45ff070b2c05a069c3f77b9a64e0bfff59d1f717362ac04203c879e5a7b68c","title":"","text":"feature request Enables easier testing for pv/pvc, and all derivatives like pet sets. The only alternative is to use manually created pv of local type. I have time to work on it myself and created this feature to collect feedback/ideas on what provider should be enabled by default.\nSince this was merged should we close this issue?"}
{"_id":"doc-en-kubernetes-01cc291fb1d6afeae9b284d60dd407b578a72d26e4c50043cbdf654c92184c45","title":"","text":"What would you like to be : Add support to the kube-apiserver so that we can disable health-checks via query params. Why is this needed: Currently, the healthz endpoint is inflexible in that any and all health checks are currently executed when the endpoint is hit. Since the kube-apiserver only currently supports a liveness check, this means all persistent health check failures will eventually result in a kubelet driven restart. However, this may not be desirable behavior if you have additional monitoring services layered on top of your control plane. For instance, take the etcd failure from the perspective of the kube-apiserver. Currently, persistent failures during the etcd health check from the api-server healthz endpoint will cause the kube-apiserver to be restarted. This sometimes helps (in the case of a faulty etcd-client connection) but also can be detrimental (the kube-apiserver restarts until the etcd cluster is functioning, then the kube-apiserver bombards the etcd cluster with requests to update local caches, saturating the etcd cluster with requests, causing it to timeout on its own liveness probes and trigger a kubelet-driven restart). Let's say that we could now disable etcd from our liveness health check. Then if we had an external monitoring service, we could ask independently ask etcd if it was healthy and we could also ask the kube-apiserver if it thought etcd was unhealthy. Based on the combination of those answers, we could enable more intelligent responses to certain situation, like restarting the kube-apiserver when it thought etcd was unhealthy but etcd was actually reporting as healthy but not doing so when etcd was actually unhealthy. This feature is backwards-compatible and opt-in. There would be no impact to existing behavior unless these query params are actually used. /sig api-machinery /kind feature\n/cc\nSGTM\n/assign"}
{"_id":"doc-en-kubernetes-3229e3f28f67afa9ddea85d9d31a50803c261e0ae30f76f950060d1790215e64","title":"","text":"I just noticed you forgot to register the image locality priority function in in I also missed it when I reviewed your PR. I have fixed this in . But it would be great if you could figure out a mechanism that would ensure that all predicate/priority functions that are tested (you can assume they will always live under ) are registered, and fail the test if not. That will allow us to prevent this mistake in the future.\nThanks! Will fix this soon!\nSeems need to wait for in before I send PR? Otherwise this test will always fail ..."}
{"_id":"doc-en-kubernetes-e853f94b17b31c823d86645df008663c9942e2f2dd039e3926420121cf6b1093","title":"","text":"This is really annoying because it requires me to interact with my shell while running up / down whereas before the commands just ran to completion on their own."}
{"_id":"doc-en-kubernetes-b9a7a79272b4f05e986ec052f9f8348c60cffedf2d99a44ddafb98b4101b558f","title":"","text":"I am enabling CSI migration on a vSphere cluster. A node was fully drained, all its Pods are deleted, but just before KCM detaches in-tree vSphere volume volumes from it, KCM restarts. A new KCM starts, sees attached volumes in that are not needed by any Pod on the node and calls vSphere in-tree volume plugin . The in-tree vSphere volume plugin + cloud provider marks the volume as detached right away, without reaching to vCenter API. The volume remains attached forever. I don't think it's related to CSI migration, IMO a regular node drain + KCM restart at the same time can cause the same issue. KCM logs: The in-tree cloud provider has a and from it. It apparently has not processed all nodes at the time A/D controller tries to detach the node. The in-tree vSphere volume plugin + cloud provider detaches the volume in vCenter. That's really tricky, it's very hard / impossible to reproduce on a real cluster. I had to inject errors to Detach() in-tree volume call, so they fail until KCM is restarted. Newly started KCM would detach the volume just fine (no error would be injected there), but marks the volume as detached right away. For the record (and search-ability), the CSI driver that tries to attach the volume to another node complains either: Or: cc It could be somewhat related to , however, not all cases of can be related to KCM restart. As I mentioned above, it's very hard/impossible to synchronize KCM restart with draining a node in a real cluster. <details</details<detailsvSphere 7.0.3 </details<details</details<detailsOpenShift </details<detailsCRI-O </details<details</details\n/sig storage\nTried just moving a Pod between nodes without CSI migration involved + restarting KCM when the volume is still attached to the old node. In-tree volume plugin on the new node says:\n/assign\nWhat CSI driver version and K8s version are you testing with?\nKubernetes is 1.25 and CSI driver is 2.7, still, I think it's not releated to the CSI driver, it's KCM who marks the in-tree volume as detached due to not populated node cache."}
{"_id":"doc-en-kubernetes-27c84c169fad17ff032040ca08c7bb73c7215213dd29c5491da1ee415abb2a33","title":"","text":"/assign\n/triage accepted\nI found an easier way how to reproduce this without extra code in KCM: a StatefulSet with 1 Pod that uses 1 PVC. the node where the Pod runs. down kube-controller-manager to zero (not really possible for me) or kill all kube-controller-manager Pods. This will cause new KCM processes to wait for leader election. In my env. this gives me ~30 seconds for the next step. Delete the Pod. A new one will be scheduled on a different node (because the current node is cordoned) + there is no KCM running at this time that would detach the volume from the old node. KCM wins its leader election, sees a volume attached to a node where it is not needed -it calls of the volume plugin. VSphere cloud provider does not have its caches populated yet (= this bug) -the volume is not detached, but returns success.\ndangling attach does not help here. The volume is still attached to the old node even after ~10 minutes. fails with and not with . I can see there is a lot of dangling attach code in the cloud provider, but I did not debug further."}
{"_id":"doc-en-kubernetes-fc09a9ef1f5e957552c4803f2a8d5445e8992bd48fceb2f6d692294e4122c44c","title":"","text":"says \"If the deletionTimestamp of the object is non-nil, entries in this list can only be removed.\". can you elucidate if I am assuming this does something different than reality\nThis is garbage collection finalizer field, not the namespace finalizer field that that new field was inspired from On Thursday, September 8, 2016, Tim Hockin wrote:\nI don't parse that - is this not a general-purpose \"refcount\" thing? It seems to work that way, except not as documented\nThis is a general purpose field and it should work like Tim thinks it should work. can you take a look at this when you don't have anything pressing for 1.4?\nThe namespace finalizer should switch to using this field if it is currently doing something else.\nI'm adding validation rule to prevent adding finalizers if deletionTimestamps is set. I'm still investigating why it can't be removed. This bug is strange.\nActually I guess it's a bug. It uses patch, and patch doesn't work on list of simple type, is a list of string.\nYep, it's a bug. I used to remove one finalizer and it worked as expected. I think this is a known bug of , we don't have a way to specify the mergekey of a list of simple type.\nIs there a bug open for this? Frankly, if is going to fail for things like this, we should not be offering it as a feature, or it should use PUT rather than PATCH , Chao Xu wrote:\nCreated cannot use PUT, because it's doing a three way merge (current state, last configuration, current configuration).\nThese are intended for automated use, not human use, so kubectl edit not working is just unfortunate, not a blocker. Although it's really unfortunate that PATCH is so broken. , Chao Xu wrote:\nWhy does need to be 3-way? I don't think that is the semantic a user would expect from an edit operation AT ALL. It certainly surprised ME. WRT automated vs human - if PATCH doesn't work (worse - says it succeeded but didn't) for some class of fields, then PATCH doesn't work, and we should not support it. Full stop. It's going to end up causing someone an outage. These are not necessarily related to this bug, but I'd call them pretty serious flaws."}
{"_id":"doc-en-kubernetes-9d31dc444c003a67666d57a34035c70818d4ce333ea6367245ecb1f620bc17fd","title":"","text":", Daniel Smith wrote:\nDiscussed in real life. PATCH works, it's just that there are certain operations you cannot express using the strategicmerge patch syntax. I was wrong. is building . So perhaps it uses PATCH over PUT due to the efficiency.\nFWIW, I printed out the patch generated by when I removed one finalizer. The patch is empty, i.e., \"{}\".\n, Chao Xu wrote: Well, no, if there are operations that can not be represented, it really DOESN'T work. I still think it should not use PUT, but I bet Brian can prove me wrong."}
{"_id":"doc-en-kubernetes-e89d62bf4b636217c9a329c55c70299ddc99e1e546e8bf63873df452671a65c2","title":"","text":"<!-- Please only use this template for submitting reports about continuously failing tests or jobs in Kubernetes CI --Which jobs are failing: aks-engine-azure-windows-master-containerd-serial-slow Which test(s) are failing: [sig-windows] [Feature:Windows] Kubelet-Stats [Serial] Kubelet stats collection for Windows nodes when running 10 pods should return within 10 seconds Since when has it been failing: Unknown. Testgrid link: Reason for failure: The test spawns 10 pods with the same pod name, which contains multiple containers with the same container name. Because of this, the test fails. Logs: Anything else we need to know:\n/sig windows These SIGs are my best guesses for this issue. Please comment if I am incorrect about one. I am a bot run by vllry. ‍"}
{"_id":"doc-en-kubernetes-7b39b114e3f8b25c65861c9c397d758891e478151ec492017dfbfbc72cee16e3","title":"","text":"[X] Test PR: kuberenetes/kubernetes [x] Two weeks soak end date: 1 July 2021 [x] Test promotion PR: kubernetes/kubernetes According to this APIsnoop query, there are still some remaining StatefulSet endpoints which are untested. Create a watch to track deployment events. Create a deployment with a static label. Confirm that the pods are running. Get the deployment status. Parse the response and confirm that the deployment status conditions can be listed. Update the deployment status. Confirm via the watch that the status has been updated. Patch the deployment status. Confirm via the watch that the status has been patched. Using an existing as a template for a new for deployment lifecycle test. This query shows the endpoints hit within a short period of running the e2e test If a test with these calls gets merged, test coverage will go up by 1 point This test is also created with the goal of conformance promotion. /sig apps /sig testing /sig architecture /area conformance\nThis issue is currently awaiting triage. If a SIG or subproject determines this is a relevant issue, they will accept it by applying the label and provide further guidance. The label can be by org members by writing in a comment. <detailsInstructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\n/reopen\nReopened this issue. <detailsIn response to : /reopen Instructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details"}
{"_id":"doc-en-kubernetes-0f3fba1dceb9c94d1a19ea07536eb93382a47fa8d74b43be2f3310b5be7e4348","title":"","text":"<!-- This form is for bug reports and feature requests ONLY! If you're looking for help check and the . --Is this a BUG REPORT or FEATURE REQUEST?: /kind feature What happened: In certain large cluster scenarios we would like to anticipate kubelet behavior hitting request limits, and backoff in response. E.g., What you expected to happen: This is a request for add'l backoff coverage, so nothing surprising. How to reproduce it (as minimally and precisely as possible): Build a cluster with as many nodes as your subscription will support on azure until kubelet behavior contributes to backend API request limits. Environment: Kubernetes version (use ): and Cloud provider or hardware configuration**: azure OS (e.g. from /etc/os-release): Ubuntu 16.04.2 LTS (Xenial Xerus) Kernel (e.g. ): Linux k8s-pool7--7 4.4.0-83-generic -Ubuntu SMP Mon Jun 26 17:54:43 UTC 2017 x8664 x8664 x86_64 GNU/Linux Install tools: acs-engine Others:\nThere are no sig labels on this issue. Please by: a sig: e.g., to notify the contributor experience sig, OR the label manually: e.g., to apply the label Note: Method 1 will trigger an email to the group. You can find the group list and label list . The in the method 1 has to be replaced with one of these: bugs, feature-requests, pr-reviews, test-failures, proposals\n/sig azure"}
{"_id":"doc-en-kubernetes-b7b38d259f42f0ebfbd04619ac113fee642398eea52ecd5ecf726aec11aad321","title":"","text":"https://perf- /kind failing-test <!-- If this is a flake, please add: /kind flake --/sig scheduling <!-- Please assign a SIG using: /sig SIG-NAME --\nThis issue is currently awaiting triage. If a SIG or subproject determines this is a relevant issue, they will accept it by applying the label and provide further guidance. The label can be by org members by writing in a comment. <detailsInstructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\n/cc\nThe goleak check was in\ncc\ncc\nAt least didn't fail with a goroutine leak. It failed because of:\nGood to see it's not a regression. Let me raise a PR to see whether is the only culprit.\nEven if it's not the only culprit, I wouldn't think this is a regression, but rather a consequence of the fact that we didn't use goleak before.\nit's just coincidental gets merged at the same timing of the goleak check PR. In other words, the scheduler perf test would also fail if the goleak PR didn't go in.\nThis can be solved by removing this line"}
{"_id":"doc-en-kubernetes-51464d59b43813ee63ecd6a7cc1e9bc3f70587ec972f984ff4e8cbc7ca42cbef","title":"","text":"When trying to use against a containerized kubernetes cluster (i.e. deployed following ) I get the following errors: This is simply because the command is missing in the hyperkube container image. I will be creating a PR with a fix shortly.\nmayukh istioctl manifest apply --set profile=demo Detected that your cluster does not support third party JWT authentication. Falling back to less secure first party JWT. See for details. Istio core installed Istiod installed Egress gateways installed Ingress gateways installed Addons installed Installation complete mayukh istio version istio: command not found mayukh kubectl -n istio-system get svc prometheus NAME TYPE CLUSTER-IP EXTERNAL-IP PORT(S) AGE prometheus ClusterIP 10.101.216.106 <none9090/TCP 20m mayukh kubectl -n istio-system get svc grafana NAME TYPE CLUSTER-IP EXTERNAL-IP PORT(S) AGE grafana ClusterIP 10.97.41.187 <none3000/TCP 21m mayukh -n istio-system get pod -l app=grafana -o jsonpath='{.items[0]}') 3000:3000 & [1] mayukh Forwarding from 127.0.0.1:3000 -3000 Forwarding from [::1]:3000 -3000 Handling connection for 3000 Handling connection for 3000 E0601 00:12:20. ] an error occurred forwarding 3000 -3000: error forwarding port 3000 to pod 6277818d8d58b0b6d52d07c776f1e930860a60466790efbaf9460a380ea497ad, uid : unable to do port forwarding: socat not found E0601 00:12:20. ] an error occurred forwarding 3000 -3000: error forwarding port 3000 to pod 6277818d8d58b0b6d52d07c776f1e930860a60466790efbaf9460a380ea497ad, uid : unable to do port forwarding: socat not found Handling connection for 3000 E0601 00:12:20. ]"}
{"_id":"doc-en-kubernetes-9a13ac1022f69644bf5dad66fc5e8ceee3107a26196b6aa329c239c5fd026072","title":"","text":"an error occurred forwarding 3000 -3000: error forwarding port 3000 to pod 6277818d8d58b0b6d52d07c776f1e930860a60466790efbaf9460a380ea497ad, uid : unable to do port forwarding: socat not found Handling connection for 3000 Handling connection for 3000"}
{"_id":"doc-en-kubernetes-9ef34f91f24cf25c8ee57899ba28766272335d7f286e3495efa44b9480fef2d3","title":"","text":"Right now it is possible to add a cluster for which it won't be possible to create a client. For example client-related code does: while validation checks only: cc:\nswitching this to 1.6 as it's too late for 1.5. ok? (please switch it right back if you disagree)\nNot a release blocker for 1.7.\nare you interested in fixing this?\nof course, will do that.\n[MILESTONENOTIFIER] Milestone Removed Important: This issue was missing labels required for the v1.8 milestone for more than 7 days: kind: Must specify exactly one of [, , ]. priority: Must specify exactly one of [, , ]. Removing it from the milestone. <detailsAdditional instructions available <a href=\"\"here</aThe commands available for adding these labels are documented <a href=\"\"here</a</details"}
{"_id":"doc-en-kubernetes-0f2f59a4098799df231d0f4ddd6804f8b020392d4e4aa17c98fa7a6b4bb4eccd","title":"","text":"The heapster needs to be installed if end user want to use autoscaler feature, we should mention this in document."}
{"_id":"doc-en-kubernetes-f8a45a347ca0a58053d8639059bb6be0a8ec66e1196dbd0290c02922abc9dcd2","title":"","text":"/kind cleanup /area conformance /area test /sig node /sig testing /sig architecture /assign /assign We have a number of tests that only have the tag that may make sense as tests, we should identify whether these tests cases meet (or could meet) the . We should open PR's to promote these if so. I would advise starting with the test/common cases\nGitHub didn't allow me to assign the following users: mgdevstack, mkurylec. Note that only and repo collaborators can be assigned. For more information please see <detailsIn response to : /kind cleanup /area conformance /area test /sig node /sig testing /sig architecture /assign Instructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\n/assign\nThe tests in test/e2enode don't actually get run as part of Conformance as near as I can tell. If we want to run them as part of conformance, they need to be migrated into test/e2e/common. Some of the tests in e2enode may not be able to be ported over because they assume a level access we don't have in a conformance environment.\nregarding this testcase: [sig-storage] HostPath should support r/w [NodeConformance]: It uses hostPath and privileged=true, to read/write from /tmp folder. Which would you suggest is the best orientation to overcome the privilege mode setting and allow its promotion?\n/assign\nRegarding , the suite runs the test only if the kubelet configuration was required to have the QOS (cgroups-per-qos) enabled. This configuration for the kubelet is selected at the time the cluster is set up. Looking for where this might be configured found: where is set to true. For other settings it is also configured to true. where is set to true. BUT also found where it is mentioned that the flag is no longer available. Questions: would it be ok to promote this test to be conformance, even it can be skipped based on configuration of the cluster? And not supported in the containerized kubelet?\nI would consider cgroups to be an implementation detail that Conformance shouldn't be verifying\nIt uses hostPath and privileged=true, to read/write from /tmp folder."}
{"_id":"doc-en-kubernetes-b0f7522fd47a7c9f3d30009b5f18d7b37ba2aa8723a4c0ef112b90078054f0fc","title":"","text":"Which would you suggest is the best orientation to overcome the privilege mode setting and allow its promotion? Michelle Au If my understanding is correct, on some platforms, writing to /tmp requires privileged. I don't see a way to to remove that requirement.\nIf my understanding is correct, on some platforms, writing to /tmp requires privileged. I don't see a way to to remove that requirement. Yeah writing to host doesn't seem like a thing we can expect of conformance tests. I wonder if it makes sense to tag this test with something to denote the requirement, ie: I see some tests in test/e2e_node use\nThe hostpath tests in already have a comment explaining this. Maybe the comment should be updated to be more clear that that is the reason why we can't add it to conformance.\nThat's /tmp on the host or in the container filesystem?\n/tmp on the host\nping what else remains to be done before we close this out? I am guessing that we want to close , is there anything beyond that?\nOnce tests under (issue- ) and are promoted we are good to close this issue. Short update for reference- None of [NodeConformance] tests from \"test/e2e_node\" seems eligible for [Conformance] promotion at this moment due to node level dependencies. Remaining [NodeConformance] tests from \"test/e2e\" has following status - Waiting for to be closed and then below tests would be promoted to Conformance Needs to be Promoted Can not be promoted to Conformance as //Slow by design Can not be promoted due to Node level dependency\nIssues go stale after 90d of inactivity. Mark the issue as fresh with . Stale issues rot after an additional 30d of inactivity and eventually close. If this issue is safe to close now please do so with . Send feedback to sig-testing, kubernetes/test-infra and/or . /lifecycle stale\n/remove-lifecycle stale\n/lifecycle frozen\n/assign"}
{"_id":"doc-en-kubernetes-f0769863fab39b633cda95f4d162dbe6bf42638a791c88a2a3acb41bb68633b3","title":"","text":"<!-- This form is for bug reports and feature requests ONLY! If you're looking for help check and the . --Is this a BUG REPORT or FEATURE REQUEST?: /kind feature /sig aws What happened: In k8s 1.7, when creating a service of type LoadBalancer, K8S creates the ELB and a Security Group for it. For these resources, it adds a tag, constructed dynamically, containing the cluster id: \"${cluster_id}\" with the value \"true\" What you expected to happen: We'd like to have also a statically named tag , as it was in 1.5 (\"KubernetesCluster\") with the value of the cluster id. How to reproduce it (as minimally and precisely as possible): Running on AWS cloud provider, create a service of type LoadBalancer. Anything else we need to know?: The reason for what we expect is that we have tight security constraints that require us to fine tune the IAM policies for the Instance profiles that the controllers use. With a statically named tag we were able, in 1.5, to restrict the permissions in the IAM policy to act only on the k8s managed resources (ELBs, security groups). Here is the relevant part of the IAM policy: With a dynamically named tag we are not able to do this any more as AWS IAM doesn't support something like: We are sharing the Instance Profiles among different k8s clusters, hence the \"\" in the value of the tag. Environment: Kubernetes version (use ):Cloud provider or hardware configuration*: AWS OS (e.g. from /etc/os-release):Kernel (e.g. ):Install tools: kube-aws Others:\nReiterating the mentions to trigger a notification: <detailsIn response to : <!-- This form is for bug reports and feature requests ONLY! -- /sig aws In k8s 1.7, when creating a service of type LoadBalancer, K8S creates the ELB and a Security Group for it. For these resources, it adds a tag, constructed dynamically, containing the cluster id: \"${cluster_id}\" with the value \"true\" We'd like to have also a statically named tag , as it was in 1.5 (\"KubernetesCluster\") with the value of the cluster id. Running on AWS cloud provider, create a service of type LoadBalancer."}
{"_id":"doc-en-kubernetes-a01a3c41e631760f28b69af73c7e0688234151a20f7dd4b67642a3c81f6b4dfa","title":"","text":"The reason for what we expect is that we have tight security constraints that require us to fine tune the IAM policies for the Instance profiles that the controllers use. With a statically named tag we were able, in 1.5, to restrict the permissions in the IAM policy to act only on the k8s managed resources (ELBs, security groups). Here is the relevant part of the IAM policy: Environment: Kubernetes version (use ):OS (e.g. from /etc/os-release):Others: </details\nI believe they are moving away from the tag since there are many resources that can be shared by multiple clusters. could you use an annotation?\nWe removed the condition in roles in the end. Thank you for your suggestion.\nThat annotation affects the ELB but not any security groups that are created. can you please reopen this? You reported a real issue, and even if it no longer affects you, it needs to be tracked and fixed.\nIndeed, this might be a feature that someone else would need.\nIn fact, this is a feature I need!\nFood for thought: should we propagate the tags on the ELB to the SG, or should we introduce an annotation for the SG specifically?"}
{"_id":"doc-en-kubernetes-cbff279a5a2209254c6527f41e3183737e33b6607a7e6fb4945b04f53611a022","title":"","text":"My application is clustered and I have 2 pods running. I am using a rolling upgrade strategy and my partitioning number is 1. First I need to upgrade the pod with the ordinal number 1 so I triggered that upgrade with partitioning number 1 and my upgrade gets completed. After this, my pod having the ordinal number 0 gets restarted and it's automatically upgraded to a new image. This is not an expected behavior because still, my partition number is 1 and pod-0 should come up with an older image only. I am facing this issue for the v1.27 cluster. Prior to this, it was working fine. pod-0 should not upgrade to new image until I upgrade the partition number to 0.a Do helm install and upgrade pod-1 only keeping partition number to 1. No response <detailsconsole $ cat /etc/os-release $ uname -a C:wmic os get Caption, Version, BuildNumber, OSArchitecture </details<details</details<details</details<details</details\nThis issue is currently awaiting triage. If a SIG or subproject determines this is a relevant issue, they will accept it by applying the label and provide further guidance. The label can be by org members by writing in a comment. <detailsInstructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\n/sig/api-machinery\n/sig apps /sig storage\nHi, Can someone suggest here? As this is introduced in my setup after the v1.27 upgrade? Is there any change in partition logic?\nThis can be reproduced by using the said STS. 1 - Deploy STS using attached that has a replica count is 2 and updatingstrategy.rolling.partition as 1. 2 - Do edit STS and update the image to 1.15. This will upgrade, pod-1 with image 1.15. 3 - Next delete the pod pod-0 and check that POD-0 came up with a new image which is not the expected behavior. POD 0 should not be updated to a new image until the partition is changed to 0. We have noticed this behavior change in v1.27."}
{"_id":"doc-en-kubernetes-78482ca8aa8474337677389890aeefe9bf7c97cdeab333d3d2b330616683b55b","title":"","text":"Pla suggest : apiVersion: v1 kind: Service metadata: name: nginx labels: app: nginx spec: ports: port: 80 name: web clusterIP: None selector: app: nginx apiVersion: apps/v1 kind: StatefulSet metadata: name: web-am spec: updateStrategy: type: RollingUpdate rollingUpdate: partition: 1 serviceName: \"nginx\" replicas: 2 selector: matchLabels: app: nginx template: metadata: labels: app: nginx spec: containers: name: nginx args: /bin/sh -c sleep 30;touch /tmp/healthy; sleep 3600 image: nginx:1.14.2 ports: containerPort: 80 name: web"}
{"_id":"doc-en-kubernetes-29db4d0749071d4e54a9083e12a93fb8b6b43cd54c28626ec2efc45861483a16","title":"","text":"Smoke tests run by publishing-bot on client-go master are failing Ref for failure logs: Tests defined for client-go, after publishing, Since commit No response As part of , when module is used from within the staging directory, can be used by traversing to the parent directory. But once the module is made as a separate repository as part of publishing, then traversing to repository using path will not work. No response /sig-api-machinery\n/sig api-machinery\nThe label(s) cannot be applied, because the repository doesn't have them. <detailsIn response to : /sig api-machinery Instructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\nLooking\nI'll remove the swagger file dependency, I don't think we strictly need it.\n/triage accepted"}
{"_id":"doc-en-kubernetes-3e8f9b25d1f3d96c25ea3e7b882f872baf67d41cad6a3bbcf2df0e192dbdfda5","title":"","text":"This is not 100% reproducable in my cluster. There is 20% chance of failing in my environment.\n/area hw-accelerators mentioned a root cause that might be due to unit tests sharing the same Unix socket and interfering with each others. cc\nCan you please share latest commit id of your code base using ?\nThanks very much, can you try the follow command please:\n/assign\nThe latest commit id of my code base:\nI found this is not 100% reproducable in my cluster. There is 20% chance of failing in my environment.\nResult of go test command:\n/sig scheduling\nThanks!\nHi, does these unit tests pass with PR\nI run 100 times without failure.\nGood to here that, Nice work! Thanks!"}
{"_id":"doc-en-kubernetes-2aac55d80f27da70e775c33573d217f00c9becbbc11daa530f1576640fc1582a","title":"","text":"The test flakes very occasionally. Below are the findings of the circumstances of the flake. During two pods are created in quick succession. The test flakes when the second Pod fails to come up. The second pod fails to come up with error: Upon exploring the logs we discovered that there was only 1 Attach call made at the beginning, and one Detach call upon timeout of the test [See timestamps]: We tracked down the error to the likely culprit in func : based on the error message returned. This is likely caused by (string) being empty [uncomfirmed], which is what the Kubelet would get from the DevicePath field on the Node object. This DevicePath field is normally on Attach and removed on Detach of the volume. The DevicePath is clearly present on the Node when the first pod is being brought up. But it is possible that it is somehow missing or not able to be retrieved by the time the second Pod is starting up (and continues to be unavailable exhausting the 15m timeout on the test) even though no Detach operation was performed on the volume in the meantime (according to logs). We are currently unable to reproduce this flake on command and did all analysis based on logs from the two flakes we saw naturally occur in the 100s of runs performed of the test. This flake was originally seen in and the test logs can be found here: https://k8s- Environment: GCE Kubernetes version (use ):Cloud provider or hardware configuration: GCP OS (e.g. from /etc/os-release):Kernel (e.g. ): /kind flake /kind bug /sig storage /cc\nIssues go stale after 90d of inactivity. Mark the issue as fresh with . Stale issues rot after an additional 30d of inactivity and eventually close. If this issue is safe to close now please do so with . Send feedback to sig-testing, kubernetes/test-infra and/or . /lifecycle stale\n/remove-lifecycle stale A check in triage shows that this failing test is still a problem, and in fact has become more of a problem since October 4th. Has there been any progress on a fix?"}
{"_id":"doc-en-kubernetes-2e50b6735500133e652390e8c295dea37c7d3b52bec7ef89ae580efb33be52fd","title":"","text":"/priority important-soon\n/assign for triage\nWe should find the root cause and fix for 1.13\n/assign Patrick can you PTAL at what is causing flakiness in this test.\nI will have a look. Right now the log output of the sidecar containers and the CSI driver are not captured, which makes debugging a bit hard. In PR I am adding code which includes those message in the test log. Let me finish that PR and once it is merged, debugging the next test flake might provide some better insight into what went wrong.\nHey, this issue is considered a blocker for 1.13beta. Is there anything we can do to get it resolved sooner, so that it doesn't delay beta next week?\nis trying to instrument the e2e tests to get logs from the csi driver containers so that we can debug. He has tried running the test locally ~100 times and didn't see it flake.\nhere's one recent one you can check out:\nMichelle Au <: I had a look, but something is odd about that one. The error for the second pod is there: But under there is no further information about node bootstrap-e2e-gce-mz-minion-group-3kn6. Shouldn't that be available?\n\"Patrick Ohly\" Here's a better example where the kublet log with the error is available: That one also has the log of the external-attacher, but that doesn't tell me much. I'm now comparing a \"bad\" run with a \"good\" local run. One observation is that in the \"good\" run, operationExecutor.UnmountDevice is completed before the second usage, including the \"Volume detached for volume\" step in Then operationExecutor.VerifyControllerAttachedVolume is done again for the second pod, exactly as it was done for the first: In the \"bad\" run, operationExecutor.MountVolume is invoked before that \"Volume detached for volume\" and then runs with an empty device path, without going through VerifyControllerAttachedVolume again: So to me it looks like a race between two conflicting operations. Does that make sense? How to fix it is less obvious. The relevant code seems to be this in Any suggestions?\nThis seems inline with the observations that made initially. We should trace where DevicePath is getting set."}
{"_id":"doc-en-kubernetes-50ecdb83cd1167eaf5e0774965d35fc15510d148142317b954cbbba2a849431a","title":"","text":"Could there be a race with AD controller or external attacher if they try to detach the volume? is another race around this area. Unclear if it's the same yet.\nMichelle Au <: No idea. Even if it isn't, it's at least related to the same piece of code. Probably worth fixing both together. Can we get someone who has worked in this area to fix these issues? Today is the first day that I've ever seen this code, so I'm a bit afraid that it'll take me a long time to get it right, and I'm worried about regressions.\nI've asked to help investigate\nIt is not a blocker for beta or freeze. Can be addressed in stabilization. Let's keep the 1.13 tag on it. And based on progress by stabilization we can reassess if it is even a 1.13 blocker (at this point I don't believe it is).\nThe in the actual state of the world gets reset during unmounting:\nThat removes the device path looks wrong to me. I've submitted PR which changes that. Let's see how much that breaks...\n/assign The conclusion in the CSI working group was that while my analysis and the resulting fix might be correct, further debugging and a second pair of eyes would be good. Let me reassign, because I have done what I could for this issue.\nThe RT considers this a blocker now, because flakes are happening often enough to interfere with getting clear signal on several jobs. Even if the underlying problem isn't that serious.\n/assign I will help triage as well.\nAck Thanks\nInstead of which may have negative side effects for non-CSI volumes, let's do this instead: Modify the CSI method in to ignore the passed in (which is populated from devicePath. Instead let's have that method generate the from scratch, e.g. attachID := getAttachmentName(volID, driverName, string(nodeName)) This will remove any CSI dependency on devicePath and not effect other volume plugins.\nI am trying to twist my head around the code. IMO it affects also AWS that uses devicePath stored in and it should be reproducible there too. So far I was unsuccessful. In the logs it's clearly visible that UnmountDevice succeeded. So why kubelet continues with MountVolume instead of MountDevice? That's wrong."}
{"_id":"doc-en-kubernetes-d22ef691d33dfaf1e5b5dad591a3014f6fd8d0e62064752134a5c7ffdbfdc8ce","title":"","text":"A workaround in CSI volume plugin won't really help here, it will work only for drivers without .\nI finally reproduced it on AWS with a long sleep in : Steps to reproduce (with the hack above): a pod with a volume the pod. Wait until is called. a new pod with the same volume. - checks that the volume is still attached and the device is still mounted and enqueues an operation to mount the volume. When finishes, starts and 1) uses empty and 2) calls on volume that's not DeviceMounted.\nFix: Any idea how to test it for real? Fake CSI plugin with slow NodeUnstage would be the best, however we don't have one.\nI'm going to work on a targeted fix for CSI. We can continue to pursue fixing aws in-tree separately.\nPR to fix csi is here:\nwhat are the final set of PRs pending for this? , and ?\nOnly is needed to fix this particular test failure.\ncan you please confirm if we are going to fix this using or ?\nI discussed with offline, he has some suggestions for that I am investigating. I think we will want both PRs eventually, although only one of them will be needed to fix the CSI test for now. Since it's his off hours, I will continue working on today.\nAre we talking 1.13 for as well?\nReopening this since we are seeing consistent failures in latest runs. We also see one more test fail in the job with same error. Failing job https://k8s- Failing Tests [sig-storage] [Serial] CSI Volumes CSI plugin test using CSI driver: gcePD [Serial] should provision storage [sig-storage] [Serial] CSI Volumes CSI plugin test using CSI driver: hostPath [Serial] should provision storage Error This is now 1.13 release blocking.\n/priority critical-urgent /kind failing-test\n/remove-priority important-soon\nI think this is a different issue and could be due to the fact that CSI is intentionally not backwards compatible between 1.12 and 1.13. We may need to enhance the test to figure out what version the cluster is on and launch appropriate versions of the CSI driver.\n\"intentionally not backwards compatible\" -- is there a writeup for the release notes and release announcement for this change?"}
{"_id":"doc-en-kubernetes-532bee311f9ca5785c50bc17dbae5c7650ea95f95b647a2552dbb2ae250d0a88","title":"","text":"Yes has the release note\nthanks, I'll also call it to the Release Notes lead's attention\nwhat does it mean from the user impact perspective for CSI to not be backward compatible with 1.12? can you please point me to any discussions on this behavior for 1.13?\nKubernetes v1.13 speaks CSI v1.0 and no longer understands CSI v0.x. Any CSI drivers deployed on previous versions of Kubernetes speaking CSI v0.x must removed, updated to be CSI 1.0 compatible, and redeployed on the on Kubernetes v1.13. If only the master is upgraded to 1.13 (and nodes are 1.12 or earlier) older drivers will continue to work. When nodes are upgraded to 1.13, then older drivers will stop working. I will capture this in detail in the \"Action Required\" section.\nIm tracking these upgrade test failures elsewhere: They are not related to this original issue, the flaky test referenced by this issue should have been solved by and for non CSI /close\nClosing this issue. <detailsIn response to : Im tracking these upgrade test failures elsewhere: They are not related to this original issue, the flaky test referenced by this issue should have been solved by /close Instructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\nas FYI to call this out clearly in the Release notes."}
{"_id":"doc-en-kubernetes-d28453dfc6bcf829133f35627fac482c86b9caf8a20943cea67f030688a8c436","title":"","text":"Runs k8s with master golang: test Failure was observed on Aug 25th, it is passing after that. However it can be reproduced with stress tool. No response No response It is failing on both ppc and x86 architectures. Steps to reproduce: up golang tip version. clone kubernetes master /sig testing\nThis issue is currently awaiting triage. If a SIG or subproject determines this is a relevant issue, they will accept it by applying the label and provide further guidance. The label can be by org members by writing in a comment. <detailsInstructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\n/assign\nit is a sig node test, sig testing charter is not about doing or maintain tests, we need to remove that old mindset of having a QA department, that is not how software development works or should work :) Besides, thanks for spotting it and submittn a patch /sig node /remove-sig testing\nFlaky in\nmerging this PR will fix the issue"}
{"_id":"doc-en-kubernetes-aa91440105d9e0a880514623ff11a0ea2fc9d1181a696540d8af19c86c926acb","title":"","text":"It is possible to express a Service where multiple front-side ports target the same back-side port. This is not very useful and may cause trouble for implementations of Services. `\njust for the purpose of clarity, are you requesting that duplicate targetPorts should be rejected or the opposite? PS: I am new to Kubernetes so please pardon the dumb question :)\nI think that should probably be rejected at validation unless we have a good reason WHY we need that.\nAye aye, thanks for the clarificiation!\n/assign\nGitHub didn't allow me to assign the following users: xiangpengzhao. Note that only can be assigned. <detailsIn response to : /assign Instructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\nwhat problems did this cause with services? This change broke backwards compatibility for services that depended on this feature. For example, lets take the load balancer service The load balancer here terminates SSL so the backend is always port 80; however, you want to make sure you support both protocols so having the same targetPort here is a valid use case.\nThis is an interesting (ab)use of Service. Not really what I had in mind - the LB in a Service is generally assumed to be L4. I'll try to dredge up the issues that this was causing.\nFYI, the does non layer4 things as well.\nOne issue is that this REQUIRES connection tracking when we otherwise might not need it. Consider two ServicePorts: my-service:80 mys-service:443 Let's say we map both services to a backend on port 8080, but we pass-through the VIP, so the pod sees packets as dest = my-service:8080. We MUST conntrack this VIP because we don't know whether to map the src port to 80 or 443. If service backends have unique TargetPorts, then we wouldn't have this issue. Port rewrites are at worst stateless, and at best can engage direct-return mode (no rewrites). It's worth noting that the PR which fixes this, which is proposed for revert, doesn't quite go far enough. It prevents a Service from using a given number twice and from using a given name twice, but not from using a number and a name that resolves to that number."}
{"_id":"doc-en-kubernetes-4a483a3c82b1bc41aa37a9f41461962211ab2500ff27fab076c9dbd88f68a578","title":"","text":"That is a much trickier (impossible to do statically) proposition. :( Kind of a mess for getting high-perf Service networking.\nwas this an issue that came up during performance testing or IRL?\nAnyone has a workaround to deal with this? I think there is a lot of people using AWS ELBs on this way\nif your use-case is Ingress, then you can set . This should avoid generating the invalid Service spec. Obviously, you will lose the redirect behavior.\nThis is not an abuse of Service, but a quite common and wide-adopted pattern, especially in the AWS world.\nyou can't re-open an issue/PR unless you authored it or you are assigned to it. <detailsIn response to : /reopen Instructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\nyou can't re-open an issue/PR unless you authored it or you are assigned to it. <detailsIn response to : /assign /reopen for further discussion Instructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\n/reopen for further discussion.\nI suspect we should probably revert the original PR. While I understand it might not have been the intention for Services, if I understand this correctly then it's a breaking change to a v1 API. Can we at least point to any documentation that indicates this was a code bug rather than a supported behavior?\nSentiment on today's sig call was that we should revert the original commit, since it broke lots of users, and then revisit to see if there's another solution.\nFYI: the revert PR is\nIssues go stale after 90d of inactivity. Mark the issue as fresh with . Stale issues rot after an additional 30d of inactivity and eventually close. If this issue is safe to close now please do so with . Send feedback to sig-testing, kubernetes/test-infra and/or . /lifecycle stale\nStale issues rot after 30d of inactivity. Mark the issue as fresh with . Rotten issues close after an additional 30d of inactivity. If this issue is safe to close now please do so with ."}
{"_id":"doc-en-kubernetes-c9520d012b388262634de2c95cda13e5277a086c4973e0f847d3a162dd04c237","title":"","text":"Send feedback to sig-testing, kubernetes/test-infra and/or . /lifecycle rotten /remove-lifecycle stale\nRotten issues close after 30d of inactivity. Reopen the issue with . Mark the issue as fresh with . Send feedback to sig-testing, kubernetes/test-infra and/or . /close"}
{"_id":"doc-en-kubernetes-09aa0bcb31564b73405d3dff3c57f03e717ee20452470641ef888656e6997574","title":"","text":"Reboot tests have been failing since 10/12 Dashboard: https://k8s- Sample Failure: https://k8s- This test failure will block 1.13 beta, so please let's fix it before then. Thanks! /sig cluster-lifecycle /kind failing-test /priority important-soon\nNote: this failure is similar to but is not the same failure, according to This is a consistent failure rather than a flake, and all nodes are failing to reboot. It's also been failing consistently since 10/11. /cc\nCopying my notes here: Looking at one recent test run: The test failed because kubelet never became \"not ready\": When looking at one of the kubelet logs, I do see that kubelet had problems making Node updates to the api server: Errors continue until kubelet is restarted 8 minutes later (the test times out at 2 minutes):\nSo it seems like even though kubelet is unable to update its Node status, the Node is still showing up as Ready\nAdding sig-node to this discussion: /sig node\nI'm trying to repro on a dev 1.11/1.12 cluster so we can get master logs for the node controller\n/cc adding from sig-node\n/assign\nwas the only PR that went in during that time (if we don't count infra or gke changes). This PR touches the node controller and changes the node heartbeat, so it could be the culprit. I was able to grab the logs from the mater was it was still up. Snippet from the kube-controller- showed a lot of error: The PR author is out for a couple of days, so he cannot investigate. I can look at it later today maybe. If anyone from GKE wants to help, I can send you all the logs.\nThe node controller doesn't seem to be working at all as it wasn't marking the nodes not ready. suggested that the api group may not have been whitelisted in GKE, so I checked and that seemed to be true. We can whitelist it, but for an alpha feature, it should not be properly gated and not affect production. We can revert PR to unblock release, and wait for to re-submit a patched PR. /cc\nThanks for debugging this! Just got time to take a quick look. Does it mean should also be put behind the feature gate?"}
{"_id":"doc-en-kubernetes-7034e0c54621f512a943c26e586b687e71244f1a626e4d5c092896b001f68abd","title":"","text":"I think reverting is a good option to unblock stuff. - creating listed may not be necessary, but setting HasSynced is (so if we have to put it after feature gate, we should put both).\nIs there any guideline on how to revert my original PR ? I haven't done it before. Just revert in my fork and submit a new PR? I also just put up a fix PR . I am fine with either submitting the fix PR or revert my original PR first.\n/reopen /assign I will manually close this after verifying the bots.\nReopening this issue. <detailsIn response to : /reopen /assign I will manually close this after verifying the bots. Instructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\nThis is passing in the latest test run. Thanks for the quick turnaround\nNice! Closing this now. /close\nClosing this issue. <detailsIn response to : Nice! Closing this now. /close Instructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details"}
{"_id":"doc-en-kubernetes-51cd929448b3f8fa92237306f83461a030cd5dfe6702e988f5001cdfbd724eae","title":"","text":"<!-- Please use this template while reporting a bug and provide as much info as possible. Not doing so may result in your bug not being addressed in a timely manner. Thanks! If the matter is security related, please disclose it privately via --What happened: On running kubetest with external storage driver option with driver name \"\", encountered error of flag redefinition. The issue happens because: As part of this patch , a flag is defined for number of inline volumes. The driver name is used to construct the flag name. Thus, when running tests on an externally provided driver (in this instance the Persistent Disk CSI driver) with driver name \"\", a redefinition error happens since the ephemeral test suite code path is executed once for and once for External PD CSI driver. [example run command: kubetest --test --check-version-skew=false --testargs=\"-- --storage.testdriver=/path/to/gcp-compute-persistent-disk-csi-driver/test/k8s-integration/config/test-\" Test config looks something like this: { StorageClass: FromFile: ... SnapshotClass: FromFile: ... DriverInfo: Name: ... } ] What you expected to happen: Test should run and succeed. How to reproduce it (as minimally and precisely as possible): Use an test-, with driver name as \"\", and run kubetest with focus as \"External.Storage\" kubetest --test --check-version-skew=false --testargs=\"-- --storage.testdriver=/path/to/gcp-compute-persistent-disk-csi-driver/test/k8s-integration/config/test-\" Anything else we need to know?: Possible fixes options: , (), when using a FromFile or FromExistingClassName option to pass in SnapshotClass information, pickup the provisioner name from the class definition's \"driver\" field. the flag which has a dependency on driver names. (need to understand the impact of removing the command line flag) Environment: Kubernetes version (use ): latest master Cloud provider or hardware configuration: OS (e.g: ): Kernel (e.g. ): Install tools: Network plugin and version (if this is a network-related bug): Others:\n/sig storage\n/assign Do you know why this parameter is set as a flag and not a part of the ?\nTo me, is something that defines what the driver can do. This flag is about what the test does and not related to driver capabilities."}
{"_id":"doc-en-kubernetes-270b31a5e196f61611ca75afa5bbde1923e891fb5ac161d33aa5adc26b5fc154","title":"","text":"Making it depend on the driver name was probably unnecessary. A single flag for all incarnations of that test probably would have been sufficient and thus better (\"keep it simple, stupid!\" - the stupid being me here...). It also shouldn't have used the global flag set. In other words:\n/reopen I don't really want to get into the habit of defining per test case flags in There are thousands of test cases and it would be unwieldy to discover all the flags available and understand how to use them. What factors would influence how many volumes you want the test to create? Can we just hardcode a value in the test case?\nReopened this issue. <detailsIn response to : /reopen Instructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\nThe defaults must be usable so that users do not need to understand them. The flags can be discovered with - not ideal, but at least they show up. That's effectively what the default value for the flag does. This is a way to to run a custom stress test: the default is so low, it just tests that the feature works. By increasing it as high as the underlying CSI driver and storage system supports, one can run this (and potentially other tests) as part of stress testing. Not sure whether it's worth supporting this. It seemed worthwhile at the time, and better than hard-coding just one value.\nThe bigger issue is: how do users of the test suite actually learn what tests they run and what those tests do? I suspect the answer to that is simply \"read the source\", in which case they'll also find the knobs that can be used to tweak testing.\nSince increasing the parameter depends on the csi driver/storage system, then I would recommending splitting this up into 2 test cases: 1) test basic functionality 2) test stress scenario, and add the parameter to the driver test config similarly to other stress tests:\nAt least for the storage tests, because we are intending that they be run externally, having all the tunables in one place I think makes it easier to discover.\nIssues go stale after 90d of inactivity. Mark the issue as fresh with . Stale issues rot after an additional 30d of inactivity and eventually close."}
{"_id":"doc-en-kubernetes-25193f95a984c83e59d3bf61132d542700dd22032bd8c1cb3f7d9def5d988f52","title":"","text":"If this issue is safe to close now please do so with . Send feedback to sig-testing, kubernetes/test-infra and/or . /lifecycle stale\nStale issues rot after 30d of inactivity. Mark the issue as fresh with . Rotten issues close after an additional 30d of inactivity. If this issue is safe to close now please do so with . Send feedback to sig-testing, kubernetes/test-infra and/or . /lifecycle rotten\n/remove-lifecycle rotten\nIssues go stale after 90d of inactivity. Mark the issue as fresh with . Stale issues rot after an additional 30d of inactivity and eventually close. If this issue is safe to close now please do so with . Send feedback to sig-testing, kubernetes/test-infra and/or . /lifecycle stale\n/remove-lifecycle stale /help\nThis request has been marked as needing help from a contributor. Please ensure the request meets the requirements listed . If this request no longer meets these requirements, the label can be removed by commenting with the command. <detailsIn response to : /remove-lifecycle stale /help Instructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\nIssues go stale after 90d of inactivity. Mark the issue as fresh with . Stale issues rot after an additional 30d of inactivity and eventually close. If this issue is safe to close now please do so with . Send feedback to sig-contributor-experience at . /lifecycle stale\n/remove-lifecycle stale\nI saw that the flag was removed in , I think we can close this one if doesn't apply anymore\n/close The flag was removed in #diff-20517805986874f69e3254bed93d59996d5a6fd571a70ab8120736ded5aafa24\nYou can't close an active issue/PR unless you authored it or you are a collaborator. <detailsIn response to : /close The flag was removed in #diff-20517805986874f69e3254bed93d59996d5a6fd571a70ab8120736ded5aafa24 </details\n/close\nClosing this issue."}
{"_id":"doc-en-kubernetes-51e228a10e330c9cdfe0846a8842785d5f11851661581fda232b614aedbd5649","title":"","text":"<detailsIn response to : /close Instructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details"}
{"_id":"doc-en-kubernetes-0786e42c64621bc4d095c4e71177424c568ccfa7fdb0651a37698d341e1aa584","title":"","text":"<!-- Please use this template while reporting a bug and provide as much info as possible. Not doing so may result in your bug not being addressed in a timely manner. Thanks! If the matter is security related, please disclose it privately via --What happened: This requires 1.16 alpha feature: Endpoint Slices being enabled. After deleting a , EndpointSliceController kept enqueuing the service after syncing it successfully, until its retry budget exceeded. And the logs are misleading by indicating something wrong happened while the cleanup has been executed successfully: What you expected to happen: The deletion should be triggered only once and there should be no warning and error logs if the cleanup is successfully. How to reproduce it (as minimally and precisely as possible): Endpoint Slices feature a then delete it. kube-controller-manager logs, there would be repeated logs. Anything else we need to know?: Environment: Kubernetes version (use ):Cloud provider or hardware configuration: OS (e.g: ): Kernel (e.g. ): Install tools: Network plugin and version (if this is a network-related bug): Others:\nThere are no sig labels on this issue. Please add a sig label by either: a sig: e.g., to notify the contributor experience sig, OR the label manually: e.g., to apply the label Note: Method 1 will trigger an email to the group. See the . The in method 1 has to be replaced with one of these: bugs, feature-requests, pr-reviews, test-failures, proposals. <detailsInstructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\n/assign\n/assign"}
{"_id":"doc-en-kubernetes-008ca2b69bf6f84855aa7fc74e613d0733365e6cd47dcc25c8f18519fbaaf463","title":"","text":"Scheduler throughput drops significantly in http://perf- ! After , seems to be the curlspirit.\nThis issue is currently awaiting triage. If a SIG or subproject determines this is a relevant issue, they will accept it by applying the label and provide further guidance. The label can be by org members by writing in a comment. <detailsInstructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\n/priority critical-urgent /sig scheduling /sig scalability"}
{"_id":"doc-en-kubernetes-886e866c35ce4d0e6b72ae097ad54b4a91c03df05db27917345e0b92fff76268","title":"","text":"<!-- Please use this template while reporting a bug and provide as much info as possible. Not doing so may result in your bug not being addressed in a timely manner. Thanks! If the matter is security related, please disclose it privately via --AppendReason panic when status is nil. Some plugins return nil to mean success. (Or, shouldn't be used on success status? I use field as a reason for why it is success.) not panic or describe this behaviour on the document\nThis issue is currently awaiting triage. If a SIG or subproject determines this is a relevant issue, they will accept it by applying the label and provide further guidance. The label can be by org members by writing in a comment. <detailsInstructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\n/sig scheduling /assign I think there are some solutions we can take. just add this behaviour to AppendReason document. (most easy solution) Status when status is nil, like below. all plugins not to return nil when success (hard way..) I support the second idea.\nI'd expect (and in fact it is) the caller to be responsible to ensure the status is not nil: The 2nd option may work, but semantically should be called by a non-success status.\nHi OK, I see :) I created the small PR to describe that reason field is for non-success status. Please take a look"}
{"_id":"doc-en-kubernetes-504dcfb92cdb541009bacf4c1b680d2ebe053bb62fdf81ca10ac728d32628e43","title":"","text":"<!-- This form is for bug reports and feature requests ONLY! If you're looking for help check and the . --Is this a BUG REPORT or FEATURE REQUEST?: What happened: Pod was deleted successfully but there is still a lock on the ceph image. I find two logs,one for that lock not removed and another for that lock removed. I find that there are little difference between them. the log that lock not removed: and the log that lock removed: What you expected to happen: How to reproduce it (as minimally and precisely as possible): Anything else we need to know?: Environment: Kubernetes version (use ): v1.6.7 Cloud provider or hardware configuration**: OS (e.g. from /etc/os-release): centos7 Kernel (e.g. ): 3.10.0-327.el7.x86_64 Install tools: Others:\n/sig storage\nError 16 is EBUSY. similar to\nwere you running kubelet with cadvisor on?\n/assign\nyes, every kubelet run with cadvisor,and i know what code 16 represent for,but i still don't know the reason, is this a bug? How can i avoid this kind of situation from happening,thank you\nIt is probably related to\nWe see some other case may cause , kubelet run with nodeexporter and nodeexporter is deployed by k8s daemonset with -v /:/rootfs:ro (mount kubelet host rootfs into nodeexporter container). rbd device is mounted before nodeexporter started, because k8s 1.6.7 not support mount propagation, when the rbd plugin unmount /dev/rbd0 on the host namespace, the mount info will still leave over in node_exporter container namespace, may cause . hope this would help\nthanks, does the problem persist if node_exporter stops collecting filesystem metrics?\nIssues go stale after 90d of inactivity. Mark the issue as fresh with . Stale issues rot after an additional 30d of inactivity and eventually close. If this issue is safe to close now please do so with . Send feedback to sig-testing, kubernetes/test-infra and/or . /lifecycle stale\nStale issues rot after 30d of inactivity. Mark the issue as fresh with ."}
{"_id":"doc-en-kubernetes-e58e0280341e2b39cd2fb25afc5193b55cba235066305ff633e2c34889e5eefe","title":"","text":"Rotten issues close after an additional 30d of inactivity. If this issue is safe to close now please do so with . Send feedback to sig-testing, kubernetes/test-infra and/or . /lifecycle rotten /remove-lifecycle stale\nRotten issues close after 30d of inactivity. Reopen the issue with . Mark the issue as fresh with . Send feedback to sig-testing, kubernetes/test-infra and/or . /close\nI'm experiencing this issue. From what I can read above, if the rbd mount happens before the node_exporter starts; we'll always experience status 16?\nWe're getting this issue. This is because the mounted volume is mounted on Cadvisor container too. So the /dev/rbdX file is still in use. But still didn't find any solution to fix.\nrbd volume will stuck when failed as will always returen\nMy solution: container use a volumeMount. make a script file to be executed every a minute:"}
{"_id":"doc-en-kubernetes-fdf6c8e5598e80992f62cef52eca112e7ba1d532fe8ace89dc932a5edac53dea","title":"","text":"<!-- This form is for bug reports and feature requests ONLY! If you're looking for help check and the . --Is this a BUG REPORT or FEATURE REQUEST?: /kind feature What happened: Currently, IPVS proxier does not run any e2e test cases. What you expected to happen: I think it's important to create test suite for IPVS in upstream CI which will help a lot in moving IPVS existing alpha. Anything else we need to know?: Let's work together. /sig network\nSelf /assign\nThere are to ways to do this work: , but doesn't support any kube-proxy config right now, so I had this pr , and also, need support it too, see this pr: kubernetes/kubernetes-anywhere build cluster. See this pr: kubernetes/test-infra The second way is much easier and faster, but I was told they gonna deprecated the , so in the long term, we still need and 's support. If anybody have better way or have some suggestion about this, please let me know, I'll be really appreciate, Thanks."}
{"_id":"doc-en-kubernetes-01f8727bc339d7af9aee90f0167a90af1aad001fdd8330c2abd8f77d4ae97abe","title":"","text":"When the master dials nodes, it is mutating the tlsConfig held by the master kubelet client, to add a server name. That means if two different nodes are dialed from a master started with , x509 verification will fail for the second one. Introduced in\nthis is a regression from 1.3 and breaks pod logs/exec/attach and node proxying on installations that use\nWe probably need to review that PR and look to see if any other deliberate copies were flagged incorrectly by govet.\nInstallations that depend on secure trust from externally provided kubelet server certs are the highest security rings. If we don't fix this we need to tell anyone using this flag not to upgrade until we release 1.4.1.\nthe rest of the changes in the govet cleanup PR looked fine to me. fix for this issue is in\nSince we are telling folks they have to upgrade to kubectl 1.4.0 for rollingupdate to work it would definitely be a confusing message to say other folks should not upgrade to 1.4.0 at all.\nIf the fix is merged into master by this afternoon, lets discuss whether or not we should cherrypick it - e.g. risks involved, how we can manually qa to verify nothing else was broken, etc.\nRisk involved: If you are using a secure cluster (defined as ensuring no one can MITM your nodes and masters) and you upgrade to 1.4.0 and it does not include this fix - you are broken.\nWhat is the likelihood of interactions with other code? Is it possible this fix could introduce a breaking change to something else? Has the fix been manually verified? Are we confident that this will fully fixes the issue (vs other edge cases exist for secure clusters that this doesn't catch). Ideally I would like to see this cherrypicked into 1.4 this evening and then cut a new beta that includes it.\nNone. The single line that is currently broken was changed. The rest is a new function (with no side effects), and tests. No, it restores the original intent of the TLS config copy, and tests cover the cases we are using this with (secure, insecure, custom dialer) Yes. I swept for the same issue elsewhere and didn't find any issues. We test with a fully secured master-kubelet setup in Openshift and got a clean test run with this fix applied.\nSGTM. Thanks for fixing this."}
{"_id":"doc-en-kubernetes-4d6e11fe929d479b5781c3745eba5f47af0a9520890604cfcfab06d0bbf7fee5","title":"","text":"What happened: has a typo, (should be ) This also shows up in the rendered documentation. What you expected to happen: Documentation has correct path for readyz endpoint. How to reproduce it (as minimally and precisely as possible): Easiest to see this wrong in the source code, I think. Anything else we need to know?: Environment: Kubernetes version (use ): v1.16 Cloud provider or hardware configuration: not applicable OS (e.g: ): not applicable Kernel (e.g. ): not applicable Install tools: Network plugin and version (if this is a network-related bug): Others: /sig docs /kind bug /priority backlog"}
{"_id":"doc-en-kubernetes-d87f35a9862b77934b500147b1ec4bb0e958c531e2fdcae282b21499e2f51131","title":"","text":"pointed out to me that a CEL x-kubernetes-validations rule in a CRD was hitting the estimated cost limit when performing a simple \"string has prefix\" check on an en enum field. Putting a limit in the OpenAPI of the enum string field fixed the problem, but having to add a limit like this is counter-intuitive and unnecessary, since the allowed enum values already make the max allowed length obvious. The apiserver automatically computes the max length of an enum as the length of the longest allowed enum value. I believe we already refuse evaluate CEL rules on values that fail OpenAPI value validation, since we rely in the value validations to establish all length limits. I haven't tested it locally, but something like this should work (given enough nested arrays): As a workaround, one can manually set on the string enum, e.g.: 25+ All No response No response No response No response\n/sig api-machinery /triage accepted /assign\n/assign\nI wasn't sure if we special-cased the maxLength / maxItems / maxProperties failures, or if all openapi validation had to pass before we started messing with CEL evaluation\nit looks like we currently special-case specific openapi errors that short-circuit CEL validation, I'm not sure an enum validation failure would currently prevent evaluation:\nThe enum err is converted to () error which is not short-circuit here. But for defaulting case, all error would be short-circuit. I would suggest to add error into the blockingErr list to short-circuit CEL validation. Would this make sense?\nyeah, that sounds good if we're going to rely on it to catch things that are longer than the maximum length enum value"}
{"_id":"doc-en-kubernetes-f35278f93af61e186544e4223e8223254ed1c96c2a29c22b72803675d1a55703","title":"","text":"/kind bug /sig node /area hw-accelerators What happened: In devices are allocated from a function called . This function inspects only container resource limits, ignoring container resource requests. This is a usability problem because it is valid to omit limits for extended resources. If limit is omitted for a device resource, then the scheduler properly places the pod with respect to available node resources. However, the device manager never dispatches to the device plugin for that resource. The result is that the user application cannot access the requested device from inside the container. What you expected to happen: Following the documented behavior of extended resources, I expected that specifying a container resource request for a device would cause the device subsystem to properly configure access for my workload. How to reproduce it (as minimally and precisely as possible): Launch an alpha GKE cluster with GPU support enabled. Attempt to run any program that accesses the GPU from a container where the resource is specified in requests but omitted from limits. Environment: Kubernetes version (use ): v1.8+ Cloud provider or hardware configuration: device plugins enabled, devices available.\nRelated issue:\ncc\n/assign\nFYI. The said issue has been fixed in . There are several valid cases when requesting extended resources: The user can specify 'limits' only. The API server will automatically set 'requests' to be equal to 'limits'. The user specify 'limits' and 'requests', and they are equal. The user specify 'requests' only. We can either reject the request saying that 'limits' must be set, or implicitly treat 'limits' as equal to 'requests'.\nHi went with to begin since it was ready to go and smaller (easier to revert if necessary.) Hopefully it's easy to rebase your work on top of that. Sorry for the inconvenience.\nI feel the documented behavior is incorrect. The original intention was to always require for extended resources until we have better APIs to express bursting scenarios. I vote for fixing the documentation and fixing the validation logic and having kubelet use limits only.\n/reopen"}
{"_id":"doc-en-kubernetes-1ecb677d90f8210e5dc6b2a237f898f5f4f944a4f983b5bfcd84eaf55dd52c98","title":"","text":"Recently I was experimenting with changing for a Golang pre-purposed faux editor when editing resources when I noticed that it couldn't be marshaled. Upon inspecting the output JSON I noticed it had the same comments that the default yaml version has. Valid JSON cannot have comments. Since the comment also gives warning / error information upon saving it must be decided where to put that in the JSON output. Perhaps a new JSON property?\nMaybe just not save the comments as part of the json to get rid of the warning, but still should them?\nIs this a 1.2 thing? I do not see any errors when running the two commands. I made an edit in one run, and did not make an edit in a separate run. Maybe I don't understand the problem that is seen.\nthere isn't a kubernetes error that is shown, the issue is that the output JSON that kubernetes generates is actually invalid JSON due to the comments at the top. This caused errors when I was trying to unmarshal that JSON in another project.\nOkay. Got it. The editor is being very strict about json having no comments.\nIt's not only about editor, it's simply part of language specification. Explanation for this: IMO if we want to use format, we should comply with it's standard. Also currently used sign for comments is unusual in javascript world, from which json evolved. In this world comments are marked in \"C++ like\" style (what in early days of JSON was also supported), i mean using and . So IMO in time of converting into JSON - we should strip such comments (I would prefer that), or they should be in some documented and specific to kubernetes way into JSON object."}
{"_id":"doc-en-kubernetes-72f2121d9c00ea31717a3620ac148f7960937c4f1fc6a8c75433d9055d46cc48","title":"","text":"Checked in: Not checked in: Can anyone explain why?\nones published outside k/k that are in libraries/modules expected to be usable without running custom code generation steps are checked in\nexample issue: there were similar issues with other instances\n/triage accepted\nI guess I am asking why not check in\noh, I'd be in favor of that :)\ndownstream users very often end up realizing that we warn folks not to use ( ) because they go looking for why this file is not checked in. This paper cut makes people realize that they are doing something that we don't want them to do. (not arguing for or against, just stating where this was effective before)\nI think the v0.0.0 local replace directives in accomplish similar things w.r.t. making people go digging about friction using k/k as a library"}
{"_id":"doc-en-kubernetes-c343f6e7b091e9b4529e6227acb6329e7535006eb05bcc59008c8d5d37213826","title":"","text":"While running one of my soak tests I noticed the error message: I don't see why I should get this error message. Why should it matter if the namespace already exists? It was created as part of the service creation call just beforehand. And the second attempt to create this pod in exactly the same namespace works. Is this a bug?\nRunning a second time did not produce this error.\nI agree it should not matter. The error is produced there, and I suspect there were multiple concurrent calls being made that try to create a resource in a namespace that had not yet existed. We need to update the code in question to swallow the error if the namespace already exists.\nFYI - longer term, I would like to eliminate the use of autoprovision of namespace and require you to first create a namespace manually. what is there now is more to preserve existing behavior prior to namespace as object was introduced."}
{"_id":"doc-en-kubernetes-4be11de3bdb9e2a86ad3a19c4d6f5342db422d48e5979c118fca9ce077f64900","title":"","text":"We allow CRD authors to write . Currently we don't restrict writing validation for , which means people can have . For example, people can have additive name constraints (require CR to be ) with the following schema: This brings us two questions: do we if \"metadata\" is included in the schema: simplest option is to publish standard metav1.ObjectMeta schema info, and use any CRD-specified \"metadata\" properties as additive validation (just like in-tree types do, publishing standard ObjectMeta schema info, and implementing type-specific name validation internally) alternatively, we can try merging user-specified metadata schema with the standard one. Technically people should only add extra validation to existing metadata properties (adding new fields to metadata won't work since they will be pruned by built-in validation first), but it's unclear to me how arbitrary people may want the metadata validation to be term question is whether to allow people specifying metadata validation at all, or just allow a subset (like name validation) in CRD v1 /sig api-machinery /area custom-resources cc\nWhat is the difference between these two? I mean the end result is the same right? The only thing that can be is validation. Nothing else works unless you are considering extensions?\nThe former means we drop/hide user-specified metadata schema when we publish CRD OpenAPI. We still validate user-specified constraints on server side, but client-side validation won't see them. The latter enables client-side user-specified metadata validation.\nIf it's possible to merge with standard metadata schema info, or at the cost of not publishing the standard schema. My inclination would be to publish the complete standard schema for metadata/apiVersion/kind, which we know is complete and accurate, and use and user-specified schema for those fields as addition validations server side"}
{"_id":"doc-en-kubernetes-b8cb95fef1c3c524a6b804cd28cd2236820a76e473144ffd27898d08f0344824","title":"","text":"Need to finish so the client allows you to select an init container, and verify the kubelet works correctly.\n(this is a nice intersection of new in 1.3, node and usability, if anyone wants to give it a shot. Pretty sure Clayton is busier than I am and I don't think I'll get to it)\nWill be happy to review any changes - should be fairly straightforward.\nYes, init containers are \"run to completion\" always. , harryz wrote:\nIIUC when the init containers of pod have not complete, the pod should be in status, which means will fail, if init containers run to completion, they will exit, still fail. I don't know what you mean exec should work on init containers?\nthe init container is still up and running even when the pod is in pending, we should be able to exec and debug why it's hung, for example.\nadd a straightforward exec implementation ptal."}
{"_id":"doc-en-kubernetes-a4e08d4660e59e77b869eee3e8c161e5e01175b90d68e1077fd3e84d4f4d567b","title":"","text":"[x] Original approval issue: - Reverted due to flakes [x] Two weeks soak end date: 20 November 2020 This issue is created to allow edit by /wip /hold According to this APIsnoop query, there are still some remaining Pod endpoints which are untested. a Pod with a static label the Pod with a new Label and updated data the Pod to ensure it's patched the Pod's status the Pod's status Ready condition to False the Pod's status to check if Ready is False all Pods in all Namespaces find the Pod(1) ensure that the Pod is found and is patched Namespaced Pod(1) via a Collection with a LabelSelector the Pod to check that it's deleted Discover useragents: List endpoints hit by the test: Display endpoint coverage change: If a test with these calls gets merged, test coverage will go up by 4 points This test is also created with the goal of conformance promotion. /sig testing /sig architecture /area conformance\n/reopen Test PR Merge Promotion PR to be created next\nReopened this issue. <detailsIn response to : /reopen Test PR Merge Promotion PR to be created next Instructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\nreverted by due to flaky test\nIssues go stale after 90d of inactivity. Mark the issue as fresh with . Stale issues rot after an additional 30d of inactivity and eventually close. If this issue is safe to close now please do so with . Send feedback to sig-testing, kubernetes/test-infra and/or . /lifecycle stale\n/remove-lifecycle stale\n/reopen PR on test grid for 2 weeks before promotion to conformance\nReopened this issue. <detailsIn response to : /reopen PR on test grid for 2 weeks before promotion to conformance Instructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\n/triage accepted"}
{"_id":"doc-en-kubernetes-23c92439a8a6a56c88d16c6d832940b868e99c5934b866ad407565ebf4fae31b","title":"","text":"https://k8s-\nI have a similar problem when I try to make release on my local desktop:\nRelated to .\nSimilar problem for :\nCan you open a new issue for your problem with information about your set up. Mac/Linux? Docker version? etc.\nPlease add details to (to be opened) issue. I suspect that is a different issue.\nNew issue:"}
{"_id":"doc-en-kubernetes-c2ab3144aef8ccfdc5bfa24e077aa0687e2812de3da5460311eb2f2561c37ff3","title":"","text":"The arg large-cluster-size-threshold for kube-controller-manager works as zone size threshold when determining node eviction rate. If there are multiple zones, the description for arg stating \"cluster\" threshold would be missleading. ARG: Matched arg description Multiple zones cluster No response <details</details<details</details<details</details<details</details<details</details<details</details\nThis issue is currently awaiting triage. If a SIG or subproject determines this is a relevant issue, they will accept it by applying the label and provide further guidance. The label can be by org members by writing in a comment. <detailsInstructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\n/sig docs"}
{"_id":"doc-en-kubernetes-00113afbb73b199bafe3ba83166412a7273f0dfa544fc7d4bd6466795076d1a7","title":"","text":"<!-- Please only use this template for submitting reports about flaky tests or jobs (pass or fail with no underlying change in code) in Kubernetes CI --Which jobs are flaking: bazel-test-master Which test(s) are flaking: gotest: Testgrid link: Reason for failure: Anything else we need to know: links to appreciated [Will update shortly] links to specific failures in spyglass appreciated <!-- Please see the deflaking doc () for more guidance! --/sig api-machinery\nThe label(s) cannot be applied, because the repository doesn't have them <detailsIn response to : Which jobs are flaking: bazel-test-master Which test(s) are flaking: gotest: Testgrid link: Reason for failure: links to appreciatedlinks to specific failures in spyglass appreciated /sig api-machinary Instructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\n/sig api-machinery\n/assign /triage accepted\nFlaked twice ~2 weeks ago but now looking good: Since it's been relatively quiet, I'm going to go ahead and close this for now. /close\nClosing this issue. <detailsIn response to : Flaked twice ~2 weeks ago but now looking good: Instructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\n/reopen /assign I meet this test failure in And I can reproduce it with local stress testing. The reason is the same with\nReopened this issue. <detailsIn response to : /reopen /assign Instructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details"}
{"_id":"doc-en-kubernetes-aa932f22e6eeb52ad965f613ba00464d5c8654efa2d7218c3ea52d951e8f2500","title":"","text":"<!-- Please only use this template for submitting reports about failing tests in Kubernetes CI jobs --Which jobs are failing: pull-kubernetes-integration Which test(s) are failing: test-cmd runkubectlrun_tests Since when has it been failing: According to triage, the first failure was 2019-02-21 12:00 UTC Testgrid link: Reason for failure: Unknown Anything else we need to know: This may be a flake? I can't tell yet. The testgrid link is not a continuous failure, but it could be that not enough PR's have whatever change is causing the failure Triage link: It looks like it may be happening in CI as well\n/sig cli /priority critical-urgent /milestone v1.14\n/remove-kind failing-test /kind flake I can't see it consistently happening in CI, so it must be a flake?\nWhat I got when running locally...\nAnd I see the same thing in the build- for\nTrying to track down the first occurrence. I think there was an XML parsing error that was obscuring this that was fixed recently. Backtracking through pull-kubernetes-integration failures and visually inspecting them. - runkubectlruntests - 2019-02-22 15:45 PST - runkubectlruntests - 2019-02-21 01:13 PST - runkubectlrun_tests - 2019-02-20 16:50 PST ... and then I don't see this particular test failing during 02-19 or 02-18\nSo going by that, let's cast a net in commits that landed just prior to the earliest PR failure, and N hours prior\nSuspects - ~only because it touches nope, rolled back to right after this landed, no repro after 30 attempts - ~enables server side apply <-- this is my guess~ nope, rolled back to right after this landed, no repro after 30 attempts - reduce lock contention in watchcache\nthe test might be racy in the first place, maybe enabling the feature changes the time profile of the test.\nWelp, now that I have what I think is a local repro loop for this, and I'm bisecting, it's looking like the whole commit range I suspected is fine Looking closer, this particular test case is overly optimistic, it deletes a job and immediately gets pods expecting there to be none."}
{"_id":"doc-en-kubernetes-c5cfc8410c7669697d22e2ec552cdc8ebab9bc4a4cb7d08b2ecda8ea1d48026b","title":"","text":"I'd still like to nail down what timing change surface this before I go change the to a or using a different set of flags for the call\nWell, for some reason seems to be when the flakes started showing up\nI realize that the flake is hard to reproduce especially since it's hard to run specific tests in . You can comment a lot of the tests by playing with some of these shell scripts so that they mostly just run the test you want to run. if you do so, you might be able to run 10 or 20 times the test in a row at this PR and right before. I'm curious what the difference would be in term of flake rate.\nYeah that's what I've been doing. It's and specifically looks to be . Which has nothing directly to do with pods, so I guess it's just a matter of this slowing the apiserver down just enough to start tripping a poorly written test.\nFYI for the above. My plan for now is just to rewrite the test so it's more forgiving. But I wanted to confirm that it's expected / known that apiserver performance changed a bit as a result of this.\n/assign /assign\nWhat metric are you seeing suffer with that commit? There should be no impact.\nA racy test goes from flaking 0% of the time to flaking 10-20% of the time Jordan Liggitt <:\nflakes: It should probably use instead. The thing is, I can't get it to flake before. I hacked up test/cmd/legacy- to run just this, and none of the other test cases\nIt definitely should. I'll take a look at job-pod deletion propagation time before/after that commit, but there's nothing I can think of in that path a new networking API group would impact."}
{"_id":"doc-en-kubernetes-59b9ae5b52e0b40199c236ff5490b403bebb933bd0751cd293a630848c9a0b07","title":"","text":"<!-- Please use this template while reporting a bug and provide as much info as possible. Not doing so may result in your bug not being addressed in a timely manner. Thanks! If the matter is security related, please disclose it privately via --What happened: What you expected to happen: No data races How to reproduce it (as minimally and precisely as possible): Anything else we need to know?: Environment: Kubernetes version (use ): Cloud provider or hardware configuration: OS (e.g: ): Kernel (e.g. ): Install tools: Network plugin and version (if this is a network-related bug): Others:\nThis issue is currently awaiting triage. If a SIG or subproject determines this is a relevant issue, they will accept it by applying the label and provide further guidance. The label can be by org members by writing in a comment. <detailsInstructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\n/assign\nfor the record as we tend to lose the logs in the gcs bucket\n/sig instrumentation"}
{"_id":"doc-en-kubernetes-75719ce40debdc24665e0bd0c466c638c232f756949a78511459bd94dde06b58","title":"","text":"<!-- Please use this template while reporting a bug and provide as much info as possible. Not doing so may result in your bug not being addressed in a timely manner. Thanks! If the matter is security related, please disclose it privately via --here is supposed to be set through , but is not called from anywhere except tests, so always returns . should be set through here. Am I missing something? N/A None Kubernetes version (use ): master () Cloud provider or hardware configuration: N/A OS (e.g: ): N/A Kernel (e.g. ): N/A Install tools: N/A Network plugin and version (if this is a network-related bug): N/A Others: None\nThis issue is currently awaiting triage. If a SIG or subproject determines this is a relevant issue, they will accept it by applying the label and provide further guidance. The label can be by org members by writing in a comment. <detailsInstructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\n/sig scheduling\nThanks for pointing that out. The plumbing of extenders to the framework impact calling extenders on preemption. That seems to be a path not covered by tests end to end. Can you submit a PR to fix this? as you mentioned we need to pass the extenders on profile creation. We need to patch the fix back to v1.21 and v1.20 (the release where this was refactored). /cc\nI have submitted a PR for a fix."}
{"_id":"doc-en-kubernetes-88c095cfe151193749adeb925825c9856381f9f8544c8822d93b8eb731e64566","title":"","text":"ci-kubernetes-node-kubelet-serial-hugepages when start the pod with the resources requests that contain multiple hugepages resources should set correct hugetlb mount and limit under the container cgroup Unexpected error: \"pod ran to completion\", Multiple sizes huge pages is going to be graduated to GA, so having this test job fixed is quite important.\n/sig node\n/assign\n/triage accepted\n/cc\nI did a small analysis and I can see that the pod failed to start, I can not find any failure messages under the kubelet log, so it really hard to say why it failed.\nwere you able to reproduce it locally? btw, there is a \"right way\" to fix this - to fix dynamic update of a hugepage resources. This would eliminate need in a kubelet restart. There is even a PR for that: , we just need to make it working again.\nafter conversation with on slack I'd say we should stick to the current procedure - restarting kubelet when hugepages allocation changes. So, we should investigate why kubelet doesn't restart properly in this test case and fix it.\nNope, I failed to reproduce it locally, also under the CI, it does not appear often. The best thing will be to describe the pod in case if it failed, so we will see the reason.\nunder the code we have so from the log , I can see that the kubelet received the pod before the kubelet stopped, so the problem that pod dropped to Failed state because of the kubelet stop.\nSo some additional details, the one who responsible to restart the kubelet again, is the server service: restart the kubelet service are waiting for hugepages to be available are starting the pod server service restart the kubelet again because it was not healthy for too long - failed to start because of the kubelet restart\ngreat findings! Should we just stop the kubelet and let the server service to start it again? We can also check if the kubelet is restarted successfully before starting the pod.\nI tried it under the PR , also I unsure how can we check if the kubelet restarted successfully. All tests that I wrote before I checked if the node has NotReady state and waited until it will be Ready, but in case of the restart probably the node will always have Ready state because the time period is very short. I will continue to investigate and will check locally how can I fix it."}
{"_id":"doc-en-kubernetes-df5fd4d114576094d5b1a71b893d21a3e13b2c561e9313633a9dfbe243849634","title":"","text":"So the only reliable solution that I found, it just trying to start the pod number of times. I tested it 5 times and the test passed. Can you please review\nCan we access kubelet service from hugepage test cases? If we can we can use its APIs like or to check if kubelet is ready after the restart. Overall I have a feeling of duplication test/e2enode/services functionality in This is a main reason for this and similar bugs I believe. It would be nice to get rid of this duplication if possible. That would nicely isolate kubelet service controls in one place and eliminate conflicts like this one. wdyt?\nnot really, the used mostly to start all needed components to emulate the cluster(not only the kubelet) and the test does not have access to it. I think you can separate the readiness check under the importable method, but it worth the separate PR.\nI updated the PR to check the healthiness of the kubelet and looks like it works just fine.\nthank you. it looks good to me so far. I'll add lgtm label when is merged and tests are all green.\nthe job is still flaky. can you reopen this issue?\nI opened a different one - , the issue can affect any job, not only the hugepages one.\nI can open a fix to W/A an issue(start the pod until it will succeed), it should fix all problems, WDYT?\ni think it's ok as a workaround, until is fixed\nwould it be better to wait until api server reports correct resource value before running the pod?\nWe are waiting for it, the problem that the scheduler and the kubelet see a different picture regarding resources requested by pods.\nok. In this case restarting the pod until it succeeds seems reasonable to me to fix the flakiness until is fixed.\nany progress on the workaround fix?"}
{"_id":"doc-en-kubernetes-38d52777560a5c6fb71d98b8b0953e61c4be2731ef4a377191b06cc53fcf6655","title":"","text":"https://k8s- Failed: [] Network Partition [Disruptive] [Slow] [] [ReplicationController] should eagerly create replacement pod during network partition when termination grace is non-zero {Kubernetes e2e suite}\nhttps://k8s- seems to be the first failure here. It seems to have gone in with a bunch of RC/RS changes () cc/"}
{"_id":"doc-en-kubernetes-68fed35f8c13d4bb380046d0fa704085775461d797cc59667c2cdfcb32483bea","title":"","text":"<!-- Please use this template while reporting a bug and provide as much info as possible. Not doing so may result in your bug not being addressed in a timely manner. Thanks! If the matter is security related, please disclose it privately via --Since upgrading to Kubernetes v1.19.11 (from v1.19.10) we see occasional panics in kubelet related to portforward streams: This issue seem to be introduced with should not panic and crash. We have not been able to successfully force this to happen, but it has happened 41 times in the last 2 weeks across 7 clusters (~100 nodes). Kubernetes version (use ):Cloud provider or hardware configuration: Self managed clusters on AWS. OS (e.g: ): Kernel (e.g. ): Install tools: Network plugin and version (if this is a network-related bug): Others: /sig node /cc\nHey thank you for the report! :pray: Uhh, thats unfortunate. The panic occurs there on L101: It looks like because works. The stack trace indicates the same: A on is a no-op, so would not be harmful. does also not seem to be , a crash in would have been visible in the stack trace. It may be possible that something assigned to the stream interface. Which means a check for will return , but accessing its fields is causing a segmentation fault. This does not explain why the crash is not visible in the trace.\nWith a quick look at the streaming code it's likely to be triggered by a timeout in waiting for streams: If the above case is hit then it goes down to , and if the timeout occurs before then it means that either or And then it attempts to remove the streams where at least one is : So it probably makes sense to check for either before passing to or inside the method where it crashes. This theory lines up with our kubelet logs, where I can see this just before the panic (didn't notice before):\nThank you for the summary working on a fix in\n/triage accepted /assign\ndo you use dockershim within your cluster?\nYes, we still use dockershim.\nGood, that's also why the panic happens within the kubelet.\nJust to make sure I understand the scope of this issue, this only affects things like or other things that use the API, correct?\nExactly :+1:"}
{"_id":"doc-en-kubernetes-d25af00145ce02f273723fdbada2e8cd6b506acebe2a7cb4f99e67bf0c6cc038","title":"","text":"Start service with a config file without specifying the argument in the config file, instead, pass the value via the CLI args, the value is just ignored, for example, I have a config file as this, NOTE: I have not specified the in the config file, but give a value for the via the CLI, the proxy mode is still set as after the service is started. The reason for this is that CLI args is parsed before the loading of the config file, if the config file haven't specified the or leave it as empty, then it will be use the default value instead. Two choices from my side, Add a note somewhere, maybe where the option is defined, just say that the CLI args will be ignored when the config file is provided regardless the option is specified in the config file or not (this should be noted for lots of options). Merge the value passed from CLI with the config file if the option has an empty value, but we need to consider the backward compatibility. and check which mode is effective. No response_ vanilla deployment with code in the upstream. <details</details<details</details<details</details<details</details\nThis issue is currently awaiting triage. If a SIG or subproject determines this is a relevant issue, they will accept it by applying the label and provide further guidance. The label can be by org members by writing in a comment. <detailsInstructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\n/sig network Hey Tim, WDYT? /assign\nHeh, this is a recurrent topic :smile: , you did a great analysis: I think that explains here why merging flags and config overriding defaults is not a desirable option (the whole thread is worth reading) I'm in favor of documenting the behavior\nthanks for showing the earlier thread, I have created a simple PR to document this in the CLI option."}
{"_id":"doc-en-kubernetes-cafbbd3cd664270b9fbd0ce7e8faa35ad0e8800a7c5d7cdeb5164eec0f1b24f1","title":"","text":"For some reason they stopped working and we even don't create manifest file for cluster autoscaler. I used the following command to start the cluster:\nIt's possible it's related to"}
{"_id":"doc-en-kubernetes-7bfb1f284722c2f05aa84834ae56662d5ec19fb895a91cc774b04328b8cdc59d","title":"","text":"Which jobs are flaking: ci-kubernetes-e2e-gce-scale-correctness Which test(s) are flaking: [sig-network] Services should be able to preserve UDP traffic when server pod cycles for a NodePort service Testgrid link: https://k8s- Reason for failure: /sig network /assign\n/triage unresolved Comment when the issue is assessed and confirmed. I am a bot run by vllry. ‍\nCRI-O CI is also hitting this failure. We have been unable to reproduce locally. our CI runs on aws\nhmm, I only see 1 failure out of 4 and seems that the connectivity didn't work at all\nIf this issue has been triaged, please comment . If you aren't able to handle this issue, consider unassigning yourself and/or adding the label. I am a bot run by vllry. ‍\n/assign\nGitHub didn't allow me to assign the following users: JacobTanenbaum. Note that only , repo collaborators and people who have commented on this issue/PR can be assigned. Additionally, issues/PRs can only have 10 assignees at the same time. For more information please see <detailsIn response to : /assign Instructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\n/cc Saw the test flaking recently: Wondering if this is just a timeout issue."}
{"_id":"doc-en-kubernetes-7ebfc2af23ebf8477a19d5b139a7e0f62f200d8158a5532a965c9eb49bf73a96","title":"","text":"Maybe we can increase the timeout and/or switch to a TCP based test:\nthe test should use UDP, and UDP is unreliable by definition ... however is curious that most of the failures happen because it is not able to communicate with the first pod I think that is something that should be investigated, I've analyzed some occurences and it works fine for the second pod, however, the UDP losses should be totally random\nHi everyone, yeah so.. collecting some data on this... using a two node kind cluster, locally, with the antrea plugin, it passes using a 4 node AWS cluster with the antrea CNI plugin, it failed using a two node kidn cluster locally with calico, it passes So I guess the best reproducer is to run these tests, for now, in the cloud.... Now that I have a reproducer for this maybe can live debug it some time next wk with one of yall :)\nbear in mind that this test is connecting to a nodePort in the node IP, the test is functionally correct, it may be flaky due to the unreliable UDP nature or due to the pod scheduling, but if is failing constantly it has to be environmental.\nSecond failure after two weeks of green 5k runs:\nyep, :/ https://k8s-|UDP /assign\n/reopen I'm afraid this is going to be challenging :smile: The difference now is that we are testing NodePort and ClusterIP If both fail, we can blame the test for sure, If only NodePort fails, we need to check the job, if GCE or KIND If only ClusterIP fails, ... :thinking:\nReopened this issue. <detailsIn response to : /reopen I'm afraid this is going to be challenging :smile: Instructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\nWe're hitting this really frequently in 5k-node tests. So I started looking a little bit into that."}
{"_id":"doc-en-kubernetes-5846160d0947d31469e9f6af93fd879e379d05716d3adf8932316babb4b3ae6c","title":"","text":"I don't have full understanding of what exactly is happening, but I already see a couple things that may be causing issues: when creating the first pod, you wait for corresponding endpoint for that being created; but that doesn't mean that kube-proxy was actually programmed correctly at that time (in general it isn't); given iptables programming is rate-limitted, it may easily happen that actually programming happens 10s+ later That explains some delays, but it doesn't explain e.g. this run: where iptables were programmed at but the test didn't respond until If someone more familiar with networking stuff could look into it, it would be great. - can you help triaging ^^ for deletions it's even worse, because you don't even wait for endpoints to be there (i.e. Endpoints can still contain the old pod when the test start pinging to look for the second one) So I claim that timeouts are visibly to low in this test, but I suspect it's not everything. I will send out the first PR that tries to improve it in couple minutes.\nthanks wojtek, have been slammed but Ill try to collect more data on it this wknd as well.\nIt seems that my PR didn't really help much. We should dig deeper, but I would need someone with more networking expertise to look a bit into\nActually - it seems only failures from one group are on non . So maybe that's actually related. The non-kind related ones look like:\nOK - for this one, I actually know what is happening - I will send out a PR with a fix for this too.\nfor the KIND jobs already increased the resource requests on some of the jobs, there are issues on KIND jobs with kube-proxy not able to take the iptables lock and not being able to create the iptables rules for the services until the next resync\nis to fix the flakes I mentioned above; all other flakes seem to be kind-oriented to me (which I'm not super familiar with).\nWe did not adjust the CPU on all jobs yet, pending how to manage that on prow ... I do see just a few other flakes at the bottom of the triage dashboard that look like timeouts instead.\nActually - almost all failures are effectively timeout. The top one that is is actually also a timeout. Didn't look into kind failures, but this may easily be related to iptables issues."}
{"_id":"doc-en-kubernetes-0543f3a3993360bbed011d4060e35a21f3f79414d82ea25f247261ac2b705804","title":"","text":"After my second deflaking PR, we seem to be in the stage where: tests are failing on dozens of different things (probably iptables-issues related) tests are failing only on NodePort tests on \"Failed to connecto to backend 1\" which it would be great if someone else could look into (kube-proxy seemed to be programmed correctly based on logs)\nIf this issue has been triaged, please comment . If you aren't able to handle this issue, consider unassigning yourself and/or adding the label. I am a bot run by vllry. ‍\nThis has failed again - Linking to specific part of log where timeout occurred .... we see this\nSo reviewing the above logs note how in STEP: checking client pod connected to the backend 1 on Node IP 10.138.0.5 28 attempts are made all in the 25th second of the 2 minute at 11am Should this Step back off and wait a few seconds in between tries? So looking at this code we see the following\nFrom reading ... the intention expressed in the test code looks good to me. It is not clear to me why there is a mismatch between the 28 tries in the POD Client Logs and the expressed intention running the logContainsFn(podBackend1) every 5 seconds still digging ....\nwe are trying to catch one failure in this PR that has additional debugging, it dumps the conntrack table of the nodes on a failure\nfor those following the issue, we've extra debugging to the test so it dumps the whole conntrack table once it finish. We should look that there is NO stale entries, connections from the pod client to the deleted endpoint) and there is conntrack entries to the new endpoint from the pod client.\n- we have runs with the conntrack dump now, e.g. this one: would you be able to take a look into it?\nnot yet :/ , but I think tomorrow can have some time\ncc\nFor anyone reproducing this issue btw, you need a multinode k8s cluster (it isnt doable on kind i guess bc less likeliehood of network failures when running local) -\nCurious about two potential hypotheses to fix this, have we looked into either of these? 1) can we confirm that is turned on/off? 2) is it possible that scaling the number of coredns replicas could help ?"}
{"_id":"doc-en-kubernetes-66f25b05b1c90d93aad713cb7c943585d55afc690fe7a4f7e9fa0b7a5a5f23b1","title":"","text":"3) Would simulating this in kind by overloading coredns be a possibility as a way to make a more reproducible bug model?\nregarding 1) kubeadm does not ship nodeLocalDNSCache I don't think GCE kube-up does by default either\nok, so now im curious about (2) - i.e. what if we just bumped coredns replica count?\ninteresting, :thinking: why do you think is a DNS thing? the netcat command that polls the services use the IP addresses not the hostname. I think that it has to be related to some iptables/conntrack race or similar,at least in KIND jobs we know that we have an iptables problem that affects network tests, that has to be fixed before this because it affect this test You can instrument the test locally with this and dump the state of iptables and conntrack on failure if you are able to reproduce it\ni figured DNS bc i believe its known that conntrack and DNS at scale can get very messy, right ?\none other curious thing about this test is that we only create 2 pods in the recycling event for the svc. that likely means that any real error is hidden by chance..... If we created a 3rd pod, then we might potentially double the failure rate thus making the flakyness go away and expose the real underlying issue .\nThere's an invocation of dumpConntrack(cs) here [sig-network] Conntrack should be able to preserve UDP traffic when server pod cycles for a ClusterIP service\nis updating the logging to run in an infinite loop ..\nThis is improving in KIND with the new changes we rolled out yesterday, let's see how it evolves this week\n- which changes ? Are you saying this is no longer reproducible on Kind?\nthere was a lot of noise in the CI signal, especially in KIND jobs due to the CNI portmap bug we finally solved yesterday now we should have a better signal and we can go to the bottom because probably the failure will be \"legit\" and not a consequence, so this comment should apply\nHi ! I ran this a few times on kind and it passed, but I see , so I'm going to try the loop approach you suggested. Here's what I'm using to test:\nquick note, <- potential motivation to possibly have runnable on some subset of tests in upstream eventually, i.e."}
{"_id":"doc-en-kubernetes-deebb69aabcaf9005201c68e52f1f862e88c30f5231beaa0034b1d576abe9a4a","title":"","text":"so we can do things like to reproduce this sort of thing. matt did you get around to the 'try-till-it-fails' scenario?\nI haven't been able to get it to fail on kind yet, -- 100 cycles and 0 failures on a three-node kind cluster! I'm letting the loop run for a few hours, will check back in when I have that data. Is it worth running this test on a non-kind environment?\nI ran for several hours on my mac, through 1144 iterations no error yet. My conclusion: this test isn't flaky on KinD. Possible next step: run a large number of iterations of the same test on a GKE cluster.\nFlakes on kind are also happening:\nnot many since the portmap fix on the 14th though ( / ) <img width=\"1111\" alt=\"Screen Shot 2020-07-23 at 11 45 46 PM\" src=\"https://user-\"\nAgree - it's much better than it was before.\nI was analyzing this failure, the problem is that kind jobs does not have verbosity enough in kube-proxy We can see that the packet was never replied, this is the NAT entry from client -nodePort in the node with the client This is the same flow, but in the node with the server and the pod-server-1 seems to be healthy we need more logs :eyeglasses:\nWe have an open ticket to bump those in CI, it's just been lost in the noise. We should still do that. Antonio Ojea <:\nIf you are busy just assign it to me, I'll need this verbosity soon :)\nIt's in the kind repo, shouldn't be too hard to find, we have ~60 open atm. Antonio Ojea <:\nIf this issue has been triaged, please comment . If you aren't able to handle this issue, consider unassigning yourself and/or adding the label. I am a bot run by vllry. ‍\nI made another attempt to reproduce these failures locally on KinD, but still no success. Here's what I did: create a KinD cluster run two instances of sonobuoy at the same time This took ~4 hours to run; all the kube tests passed and sonobuoy reported all successes."}
{"_id":"doc-en-kubernetes-738ad9cb1373dead6b2f556ac241cd0c8a6cc7a0f09950111e765d9eb1dab68b","title":"","text":"Would love to help out in reproing this, I'm at a loss for good next step so any suggestions welcome!\nthat's the problem with these flakes, they are not easy to repro, and with this low rate of failures it may be caused for combination of running tests, environmental problems (cpu, memory, ...) , ... I always start for checking the current failures and try to find patterns\nLet's take a look to one occurrence The tests fails because can't connect to the first pod If we check the code of the tests, we can see that it always uses the port as source port, we can use that to identify the nodePort assigned interesting thing here, we are source nating the pod ip, and we don't see any other conntrack entry The test verifies it has endpoints however, kube-proxy is using endpointlsices since this version but it has no entries so the iptables rules to forward the traffic are not there Seems we have to adapt this function in the e2e framework to use endpointslices cc: EDIT seems that this function is being used by all the network tests that can cause flakes if the endpoint is updated before the endpointslice (we switched recently kube-proxy to use the later)\nGreat catch. Ideally this function should validate if both Endpoint and appropriate EndpointSlice objects(s) exist.\nActually - let me take it and send a PR for it.\nis out for review\nInteresting failures here It fails to connect to the first pod\nHmm - I took a quick look at kube-proxy logs and these look fine to me....\nI clicked too fast, sorry ... I was looking at the wrong service and I though I've found something\nIf this issue has been triaged, please comment . If you aren't able to handle this issue, consider unassigning yourself and/or adding the label. I am a bot run by vllry."}
{"_id":"doc-en-kubernetes-7b1ee5acca525c6b88a5b6825056ab7c4df586768503c885715c9a87d944744a","title":"","text":"‍\njust saw it as well\ncan you please add links to the job with the failures, I just merged new code to improve the logging 2 days ago\nok, here there is a new occurrence and with the new debug logs is easier to follow has unreplied conntrack entries: looking at the other side The iptables rules are there but no conntrack entries related to 10.244.2.32\nIn I revert PR with debug info as I think that it likely caused significant memory growth in scalability tests, but we have one run that has finished without being killed by OOMKiller (I increased memory limit for that jobs as a mitigation). Maybe logs from this run will be useful for debugging:\nIf this issue has been triaged, please comment . If you aren't able to handle this issue, consider unassigning yourself and/or adding the label. I am a bot run by vllry. ‍\nTwo more flakes today:\nIt seems that the packet never gets DNATed in the receiving host, so it does not reach the pod in the node. There is the same test that is using ClusterIP instead of NodePort that is more reliable. This is being a tough one\n/triage accepted\nYesterday's flake - I don't think I'm seeing the log messages from the latest log PR you made ? Or are the prow logs not logging at a high-enough level?\nLogs are there, please bear with me: The error on the test is: click on the job failure and you'll see a larger output where you can see the server pod name and get the node where the pod is running: and the client-pod name and its node: and the flow we are looking for: with the node-name we can get the kube-proxy logs there, clicking in artifacts: and you can see the \"new\" conntrack log entries : and in the client: wow, there is a kernel bug there , can it be related? :thinking:\nwohoo - some progress :) I'm not an expert here, but I think it can be? Adding here - they may be able to confirm/reject the hypothesis.\nThis kernel panic may be a coincidence, maybe this is the reason\nAdding some recent failures:\nSpotted another flake today:\nAnother flake in the 5k job:"}
{"_id":"doc-en-kubernetes-f7dbb3a7bd33f39ed155e3c1ad6b8c55c94743d34cf8c863dee73b5ba5ef4403","title":"","text":"spec.variables should be marked omitempty. I'm trying to create a validating admission policy that uses in v1.27 as follows: I'm setting : to \"Strict\" in CreateOptions so that the server will reject requests with unknown fields but it complains that the don't exist in v1.27 and it does make sense because this field was introduced in v1.28: I'm using a 1.28 client.\n/sig api-machinery\n/assign\n/assign /triage accepted\nThanks for raising this! /cc"}
{"_id":"doc-en-kubernetes-1338c1265dc2da957b27c9fcf90bf88eaa9473c2ffb7e29c3dbb2785dec4e5f0","title":"","text":"From another issue, said: DockerManager.SyncPod() calls the network Plugin and then sets hairpin mode unilaterally. This is wrong. We should either pass the bool down to here or else we should get rid of the default no-op plugin and use the non-existence of a plugin as a flag to control this. This should be a relatively easy one for people who want to contribute but don't know where to start.\nI volunteer myself to do this. Actually I have done something about kubernetes network plugin so guess I can do this. I will start to do this when I am on the office tomorrow.\nAwesome. I'd advise starting with a sketch so we can all make sure it is a good approach, before jumping in with both feet. , TonyAdo wrote:\nok, I will start with a sketch as soon as possible.\nI thought about this for a while: if we pass the bool down to here, of course we can check this flag, if it is set, we should set veth hairpin mode, but problem is that if user set this flag, but still use a custom network plugin just like you mentioned the ovs plugin, I am afraid we will still enable hairpin mode in all interfaces on this host, since we cannot find a peer interface. Is that correct?\nYeah, I think the better fix is to use the existence of a network plugin as an indicator. It sort of makes --configure-cbr0 redundant , TonyAdo wrote:\nok, I will try to get rid of the default no-op plugin and use the non-existence of a plugin as a flag to control this.\nI decided to use the existence of a network plugin as an indicator, also I will get ride of the no-op network plugin, I am not going to add new flag to kubelet structure, if the user don't specify network plugin name, the networkPlugin within kubelet will be which just means non-existence of a plugin, and we have to check the existence in the docker manager to avoid nil pointer exception also for setting, I think if a custom network plugin exists, kubelet should not do this. How about this? Should I open a PR?\nfyi, saw a hairpin pr today that might be related. ymmv.\nwhat do you think about this?\nWhat you described sounds right to me."}
{"_id":"doc-en-kubernetes-63d22d436297e1e341b6c3c9db0f8b916cd9a8fc746e2ad0128076b35b0cb5ac","title":"","text":"I think the only way to see if there is some detail I am missing is to try it and see if anything pops up that would make this not work. , TonyAdo wrote:\nI need to check which part is responsible of what, but what if the hairpin package was just a utility package that should be called by the plugin if it needs to set this up? As long as it is single and simple function call, doing so would avoid adding extra logic here and fix the issue. This is valid only if the aim is to have a \"linux bridge plugin\", an \"OVS plugin\", etc.\nactually kubelet will call after get called, I think to make the network plugin can call hairpin package is not that easy, of course, I agree that network plugin should setup hairpin mode if it really needs, kubelet should not set hairpin mode unilaterally. Besides, kubelet is not going to provide an \"linux bridge plugin\" or \"OVS plugin\".\nthen the validity conditions of my comment are not met, so I see any simpler way. Thanks for checking! :)\nstill what do you think about disable hairpin completely and leave this to network plugin?\nor if there is a custom network plugin, we don't setup hairpin mode, only do this in default.\nCompletely? It will break my use case (and anyone using Apache Kafka).\nOnly by default is what I understood from the previous comments, and LGTM.\nReally sorry about that, but I really want to know are you using a kubelet network plugin in your use case? If no, I think we will still setup hairpin, as in default we will do this.\nI don't use any plugin, so it is ok if hairpin is set up by default.\nok, I think this will not break your use case.\nIt's not really the network plugin that needs hairpinmode though, it's the pure-iptables-based kube-proxy code. It's possible the proxying code could change again in the future in such a way that hairpinmode wouldn't be needed any more. So it seems wrong to put that code into the plugin. makes it so that we set hairpinmode on the veth if it's attached to a bridge, but just silently do nothing if it's not."}
{"_id":"doc-en-kubernetes-3bf0b78c82438b679aab4fecf6e8905cfec517728c795c679dfaefb3dd282eea","title":"","text":"(Yes, we currently set hairpinmode even if you're using the userspace proxier, but AFAIK it's not needed in that case.)\nIMHO... The proxy has the responsibility of service objects, while the kubelet is responsible of pod objects, including their networking. Also, the responsibility of hairpin must be at the same place as the code that creates the network interface, otherwise you have code managing a single object (the pod interface) spread over multiple programs, which will eventually lead to complex problems and a system harder to understand even for the end-user. Thus, it doesn't seems wrong to me that the kubelet does whatever is required to make the pod work in the environment of the cluster. While its true that the hairpin requirement is only for the proxy in iptables mode, a switch to this mode can happen at any time in the life of the cluster. And when it happens, its probably because the userland proxy becomes unresponsive, which means in a situation where the administrator is in a hurry to find a solution. Currently, you just change the proxy mode and restart the proxies. There's little chance the administrator will think about hairpin or restarting the kubelet in this case, so the kubelet should be reacting to this; which means the proxy and the kubelet of a given host should be able to communicate (AFAIK, this doesn't exist now). And all this is currently only to avoid setting up hairpin on the pod's port of standard Linux bridge.\nFeel sure I'm missing something but... The CNI bridge plugin now has a flag, so kubenet can use that. Having done that, we can set false when the plugin name is non-blank.\nIs this fixed now?\nI apologise for leaving a somewhat vague answer, but I have been in that code recently. I don't think it's fixed.\n, Bryan Boreham wrote:\nThanks for the PR. CNI Bridge plugin is expecting . Not . Hence the bridge plugin did not recognize it.\nEDIT 2016-12-09: The chain below was broken by , so the issue can be closed again (remain closed) I suggest this should be re-opened. The chain of calls is: sets to , e.g. when using CNI calls something goes wrong and it sets the mode on everything. An example of \"something goes wrong\":"}
{"_id":"doc-en-kubernetes-f4d4b7c473a266ffc48b535d9080ecbdaed52af3a9bd64395678a7e260a39c7f","title":"","text":"for a PodFailurePolicyRule. However, when trying to create a JobSet programmatically with a JobTemplate that defines a PodFailurePolicy rule for failing a job immediately if the failed pod had condition DisruptionTarget, I get a validation error indicating that cannot be null. I've tried using both and (with matching versions for package as well). Error: JobSet spec: JobTemplate passes validation. JobSet: applying example JobSet in the description of this issue. JobSet: and applying example JobSet in the description of this issue. No response 30 (tests on a kind cluster using kindest/node:v1.30.0) <details</details<details</details<details</details<details</details<details</details\nThere are no sig labels on this issue. Please add an appropriate label by using one of the following commands: - Please see the for a listing of the SIGs, working groups, and committees available. <detailsInstructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\nThis issue is currently awaiting triage. If a SIG or subproject determines this is a relevant issue, they will accept it by applying the label and provide further guidance. The label can be by org members by writing in a comment. <detailsInstructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\ncc\n/assign\nI confirmed the issue, moreover, I see that when the JobSet webhook is disabled then it works ok, I think it must be the JobSet webhook which adds the \"onExitCodes\" field during deserializion /serialization step. I can also see that when I rebuild JobSet using kubernetes field \"onExitCodes\" with \"omitempty\", then it the JobSet creates ok. I will open a PR."}
{"_id":"doc-en-kubernetes-da00f1cf1a24475bca3717ed24d8bd1cf5079c51cd54d6d4d01b160ed7cae851","title":"","text":"Is this a BUG REPORT or FEATURE REQUEST?: /kind bug What happened: DaemonSet has an e2e test running that has dubious purpose. What you expected to happen: The e2e test should be refactored and commented clearer as it will be part of the conformance.\nTo elaborate, we're concerned that the following may not actually ensure that DaemonSet recreates Failed Pods: For example, it seems possible that the Node might simply overwrite the Pod status with its own information if it hasn't yet seen the change to the Phase. Ideally, we should verify that the Pod UID we force to Failed no longer shows up in the DaemonSet's list of Pods after the DaemonSet becomes happy again."}
{"_id":"doc-en-kubernetes-ee80683e4fc8d1f9267feeed119de250fd6d0c3f47227cc07d302319c50cf3e7","title":"","text":"What happened: is a subdirectory of the btrfs subvolume . kubelet fails to start with the following error: What you expected to happen: kubelet starts normally and doesn't fail with an error message How to reproduce it (as minimally and precisely as possible): Install a system using btrfs as file system for / Create a sub-volume for /var Start kubelet (even in disconnected mode) Anything else we need to know?: This is a follow-up issue of and has been promoted to fatal from onward, before it was a warning PR to fix the issue is Environment: Kubernetes version (use ): forward Cloud provider or hardware configuration: both bare metal and virtual OS (e.g. from /etc/os-release): SUSE 12+ or openSUSE (Tumbleweed/Leap) Kernel (e.g. ): Install tools: SUSE/openSUSE rpm Others:\n/sig storage\nfixed by /close\nIf I may - adding some notes - it might be helpful for others, until I get the time to do some more detailed tests and report a bug. I still had issues with kubelet not starting, with the same error - (version 1.13.4) My btrfs setup is that btrfs root is mounted at /, then I have also a subvolume /var and in it a subvolume /var/lib. The issue for me I think comes from the fact that I do not have a flat btrfs setup () but a nested one, and as such, I have as a mountpoint ONLY the root directory. Manually adding a mount entry for /var (heh - I should probably have mounted /var/lib) in /etc/fstab made it work: /var btrfs rw,relatime,ssd,space_cache,subvolid=259,subvol=/var 0 0` Also, even without this additional \"intermediate\" mount point I think it should have been working. I wonder if in , the loop 'for !found {' isn't exiting a tiny bit too fast (without testing for the root mountpoint itself) . Maybe it should be something like:\nHas something changed in the code? Kubelet from kubernetes 1.20.1 is failing. Doing a mount as I described above fixes it.\nsee (open) and (closed) - apparently this was recently fixed (but isn't released yet)."}
{"_id":"doc-en-kubernetes-097cec69e6f696ee556282ba1b462f036510b26986c7b0a8c41c6e03be57a385","title":"","text":"<!-- Please use this template while reporting a bug and provide as much info as possible. Not doing so may result in your bug not being addressed in a timely manner. Thanks! If the matter is security related, please disclose it privately via --Hello, Found a potential data race in testing. There is a test If we update it like this (we actually can do it ): If watcher updates received object we get a data race. No data race My suggestion: send copy of object here:\n/sig api-machinery\n/cc /triage accepted Do you remember if we fixed similar issues before?\nHi ,The code avoiding data race seems to already exist If you commented the code on above, the following test case will pass (because of data race), so I think there is no data race in the code : )\nHi Did you run my test with flag? Of course if I comment there will be data race. But let's take a look at: We store in and do with the same . So, watcher receives that is stored in map. we call we take the same from : Therefore, if watcher modifies received in another goroutine we have data race. Output:\nHi You're right ,thank you for solving my confusion. One solution could be to copy of , I will try to fix it."}
{"_id":"doc-en-kubernetes-c7d12b68f0527cb4211e1e8a31ccfe4349122b3159e0a706002c6b019cbf11bb","title":"","text":"What happened: When running and no resource is found, kubectl returns , which is confusing, since we're trying to list instances of a cluster-scoped resource. What you expected to happen: When listing cluster-scoped resources, the message should either be or How to reproduce it (as minimally and precisely as possible): (assuming there are no ; can also try or similar) Environment: Kubernetes version (use ):\n/sig cli\n/reopen Reopening, because did not fix this issue. It only fixed , which also produced a misleading message, but that's not what this issue is about. This issue is about cluster-scoped resources, not about listing namespaced resources in all namespaces. Output of kubectl get with the changes from included:\nReopened this issue. <detailsIn response to : /reopen </details\n/assign\nIssues go stale after 90d of inactivity. Mark the issue as fresh with . Stale issues rot after an additional 30d of inactivity and eventually close. If this issue is safe to close now please do so with . Send feedback to sig-testing, kubernetes/test-infra and/or . /lifecycle stale\nStale issues rot after 30d of inactivity. Mark the issue as fresh with . Rotten issues close after an additional 30d of inactivity. If this issue is safe to close now please do so with . Send feedback to sig-testing, kubernetes/test-infra and/or . /lifecycle rotten\nRotten issues close after 30d of inactivity. Reopen the issue with . Mark the issue as fresh with . Send feedback to sig-testing, kubernetes/test-infra and/or . /close\nClosing this issue. <detailsIn response to : Rotten issues close after 30d of inactivity. Reopen the issue with . Mark the issue as fresh with . /close Instructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\n/reopen This seems to be an error in the get code for kubectl, it looks like it tries to figure out if all the requested resources are namespaced and only then return specify the namespace in the output. This seems to be happening when we have only cluster-wide resources."}
{"_id":"doc-en-kubernetes-1027112ca48a1952e66a5a235f8ecc4ced7e393fc603d44340ff46ba509184a5","title":"","text":"You can't reopen an issue/PR unless you authored it or you are a collaborator. <detailsIn response to : /reopen Instructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\n/reopen\nReopened this issue. <detailsIn response to : /reopen Instructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\nRotten issues close after 30d of inactivity. Reopen the issue with . Mark the issue as fresh with . Send feedback to sig-testing, kubernetes/test-infra and/or . /close\nClosing this issue. <detailsIn response to : Rotten issues close after 30d of inactivity. Reopen the issue with . Mark the issue as fresh with . /close Instructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details"}
{"_id":"doc-en-kubernetes-0c3eca7956c73a24f455c71a58b712ce183e874d4a3e3eb9b913120c8a8ac10c","title":"","text":"<!-- Please only use this template for submitting reports about continuously failing tests or jobs in Kubernetes CI --Which jobs are failing: Conformance - GCE - master (ci-kubernetes-gce-conformance-latest) Which test(s) are failing: [sig-scheduling] SchedulerPreemption [Serial] PreemptionExecutionPath runs ReplicaSets to verify preemption running path [Conformance]Changes Since when has it been failing: 2020-06-08 22:27 PDT Testgrid link: Reason for failure: Regression by Anything else we need to know: /sig scheduling /priority critical-urgent\ncc/ /assign"}
{"_id":"doc-en-kubernetes-ff400dc6754846fd0992db1028d6e5a3aef2be3cc27ba2c2792aa3874bf81494","title":"","text":"Something like: cc\ncc\n- Can you please look at this issue since you're recently working on documentation?\nthanks for the reminder. I will take care of it."}
{"_id":"doc-en-kubernetes-0c07e91eb7d66ee92c3b6d05f523cece58039dffcf1123c44ba752c30922deab","title":"","text":"N/A [sig-node] NodeProblemDetector: N/A https://prow- Not cgroupv2 compatible when access cpu and memory data\n/sig node\nHi, can i work on this ?\nSure. Thanks\n/assign\n/triage accepted\n/priority important-soon\n/cc\nare you still working on this? Were you able to setup environment to repro the test failure? Let us know if you need any help. Thanks!\nSorry for the delay, I followed the wrong guide and ended up getting stuck. But now I have managed to set it up. Please give me few more days to work on it.\nThanks Feel free to reach out here or on slack if you need any help, thanks for working on this!\n/retitle Make nodeproblemdetector test cgroupv2 compatible\nFailing job:\nTestgrid:"}
{"_id":"doc-en-kubernetes-cf37b0ab7a08d0e2457507bc271b1cb7d1e13fbaee3773971363bc32ae71b35b","title":"","text":"We are running heapster, dns server, kube-proxy etc. in kube-system namespace. Some e2e tests try to make sure all those kube-system pods are at ready state before firing the real tests, others just simply ingore them. But both cases might cause e2e flakiness when some of those kube-system containers are in crashloop, for example, for heapster crashloop, for influxdb crashloop, etc. When such cases happen, the developer doesn't have much information on why since the nodes might be teared down. We should watch those kube-system pods, and onContainerFailure, we should collect the related logs. One simple way is using kubectl logs <Pod<Container--previous cc/\nis this related enough to and that you can bundle it in?\nThe logs from the pods should be retrieved via the api server. This is different from the kubelet logs. I'll upload a PR for this.\nGreat job It will be very helpful for debugging."}
{"_id":"doc-en-kubernetes-f29982137c07dadb3e257d7391e0d9b7bfd5f1ce8fdca72f798301cc159728a0","title":"","text":"Is this a BUG REPORT or FEATURE REQUEST?: /kind bug What happened: After labeling clusters in a Federation, when you type the field is empty. What you expected to happen: The fiels should contain information about the labels in the cluster How to reproduce it (as minimally and precisely as possible): a federation control pane in any Cloud Provider and any DNS provider your cluster: info aboyut the cluster with the flag: Note: if you describe it, you see the labels being : Environment: Kubernetes version (use ): 1.8.0 Cloud provider or hardware configuration**: anyone OS (e.g. from /etc/os-release): Ubuntu 16.04 Kernel (e.g. ): Linux Install tools: curl Others:\n/sig multicluster\nI'm working on it."}
{"_id":"doc-en-kubernetes-d6a2e132f72678770d8d41e08acf9f7359a4b887e33d930dbbc18ea7ecc9db7e","title":"","text":"/kind bug What happened: When setting an elbSecurityGroup in for AWS, and deploying multiple services with differing ports, then the security group rules flap in AWS, causing services to become inaccessible (only one will be correctly configured at a time). What you expected to happen: I expect that the configured elbSecurityGroup's rules will not be touched at all. It must be pre-configured to permit access. Alternatively, all ports for all services are queried and managed together. How to reproduce it (as minimally and precisely as possible): See above. Anything else we need to know?: Most people are not running large clusters which may hit the AWS limits, so they won't use this option, nor see this issue. Environment: Kubernetes version (use ): 1.7.1 Cloud provider or hardware configuration**: AWS OS (e.g. from /etc/os-release): CentOS 7 Kernel (e.g. ): 4.10.13-1-ARCH Install tools: Custom Others:\nThere are no sig labels on this issue. Please by: a sig: e.g., to notify the contributor experience sig, OR the label manually: e.g., to apply the label Note: Method 1 will trigger an email to the group. You can find the group list and label list . The in the method 1 has to be replaced with one of these: bugs, feature-requests, pr-reviews, test-failures, proposals\n/sig aws\nThis is fixed by"}
{"_id":"doc-en-kubernetes-984aaf2876c4d4651f6765b291626bccfb7d6364087d7ebb11c97e2832c59ee5","title":"","text":"<!-- Please use this template while reporting a bug and provide as much info as possible. Not doing so may result in your bug not being addressed in a timely manner. Thanks! If the matter is security related, please disclose it privately via --What happened: Below k8s UT fail when run with master-golang. TestCheckpointStateRestore/RestorecheckpointwithinvalidJSON error:=== RUN TestCheckpointStateRestore/RestorecheckpointwithinvalidJSON I1124 05:38:14. ] [cpumanager] initializing new in-memory state store unexpected error while creatng checkpointState: could not restore state from checkpoint: json: unexpected end of JSON input, please drain this node and delete the CPU manager checkpoint file \"/tmp/cpumanagerstatetest479660663/cpumanagercheckpointtest\" before restarting Kubelet --- FAIL: TestCheckpointStateRestore/RestorecheckpointwithinvalidJSON (0.00s) TestStrategicMergePatchInvalid error:=== RUN TestStrategicMergePatchInvalid expected \"invalid character 'b' looking for beginning of value\", got \"json: invalid character 'b' looking for beginning of value\" --- FAIL: TestStrategicMergePatchInvalid (0.00s) TestJSONPatch error:=== RUN TestJSONPatch invalid-syntax: expected error invalid character 'i' looking for beginning of value, but got json: invalid character 'i' looking for beginning of value --- FAIL: TestJSONPatch (0.00s) TestV4ErrorDecoder expandless error:=== RUN TestV4ErrorDecoder wrong error for message \"{\": want=\"error stream protocol error: unexpected end of JSON input in \"{\"\", got=\"error stream protocol error: json: unexpected end of JSON input in \"{\"\" --- FAIL: TestV4ErrorDecoder (0.00s What you expected to happen: The tests should pass with no error. How to reproduce it (as minimally and precisely as possible): a stable version of golang installed master go from upstream git clone cd go/src cd ../.. cp -r ./go /usr/local export PATH=/usr/local/go/bin:$PATH kubernetes repo and run the unit tests. () Anything else we need to know?: These failures are occurring due to mismatch of error messages between upstream go and the k8s tests."}
{"_id":"doc-en-kubernetes-d960907779eeb813df14a2ea4a00c97f52784ded937dbf1c71e5b05391843c78","title":"","text":"There is an addition \"json:\" prefix to SyntaxError messages after this below commit: Environment: Kubernetes version (use ): latest source code from upstream Cloud provider or hardware configuration: OS (e.g: ): \"CentOS Linux 8 (Core)\" Kernel (e.g. ): Linux rajalakshmi-workspace 4.18.0-193.19.1.el82.ppc64le SMP Mon Sep 14 14:48:04 UTC 2020 ppc64le ppc64le ppc64le GNU/Linux Install tools: [root go]# go version go version devel + Fri Nov 20 00:54:43 2020 +0000 linux/ppc64le [root go]# Network plugin and version (if this is a network-related bug): Others:\nThis issue is currently awaiting triage. If a SIG or subproject determines this is a relevant issue, they will accept it by applying the label and provide further guidance. The label can be by org members by writing in a comment. <detailsInstructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\n/sig testing\nwill fix it in Go 1.16 as release-blocker.\nI would fix the test here to be more tolerant. exact string matches on error messages aren't really an API… I would not expect that to block the golang release\nAgree too, we can think about having a substring match to avoid these issues instead!\n/assign Lets see if we can fix this in the k8s tests"}
{"_id":"doc-en-kubernetes-c00c4002c1ceb0f27a0d12e9f8008c8e3b435d9e5dbb0a6b5f62beaf7aea3677","title":"","text":"I am doing v1.28.3 kube-apiserver testing in the test environment, and I often encounter kube-apiserver panic, especially when kube-apiserver call webhook fails. The following are the captured panic logs: and: kube-apiserver does not panic even if calling webhook times out or fails Deploy a webhook, remove the service corresponding to the webhook, and then create pods in batches to trigger webhook calls No response <details</details<detailsprivate cloud </details<details</details<details</details<details</details<details</details\n/sig api-machinery\nsomewhat related to\nThe scenarios that trigger panic are similar(the apiserver fails to call webhook), but the log shows that the location of panic is different.\n/assign For context of  Thank you. /triage accepted\nagain, full stack logs:\nI'll try to test the mr:\nLet me try to explain the panic - both goroutines are trying to write to the of the first goroutine: This request has timed out, and consequently the filter is trying to send a response to the caller. This is the client facing goroutine from the timeout filter. Note that the call to is protected by a mutex, that means that if the other goroutine is using the delegated object via the adapter, then this panic should not happen. second goroutine: let's check the other goroutine, it is also trying to write to the of the object: Now uses an unprotected version of the ResponseWriter object: The filter precedes the filter in the execution order, so the in is not using the thread safe implementation. This is the order today: one quick resolution is to move the after the timeout filter: , but there could be code that precedes that needs to write warning? Also note that the panic mentioned in is not as severe, it does not cause the apiserver to crash. For a more proper solution, i think we would want any thoughts?"}
{"_id":"doc-en-kubernetes-f6276625b0df7e8b951a42b4ce484e48eb7715b4546658f042b200999787970a","title":"","text":"/cc\nI tend to remove the timeout filter, the sooner the better, and I think the Apiserver panic is the P0-level bug, directly affecting the availability of the cluster I have cherry-picked and tested it in my cluster, and I still need to continue to observe for 1-2 days\nI opened a minimal WIP fix for the warning writer in since the timeout filter removal is a much bigger change to reason through\nthanks a lot, I will try to test and cherry-pick the above minimal WIP fix in\nsorry, I completely overlooked , I only saw and that looked too big to take safely right now I prefer to\nCan we backport the fix to 1.28?\nWhy 1.28 specifically? I'm not aware of anything that changed in 1.28 that made this more or less prevalent. If this meets backport criteria, going to all active branches could make sense (1.27+)"}
{"_id":"doc-en-kubernetes-4cc03be8a4c78b1eefdd458c17e35e10d17eae1288a6e1bcb08b9d7d585d4633","title":"","text":"07-02 and 13-02 No response No response /sig\nThis issue is currently awaiting triage. If a SIG or subproject determines this is a relevant issue, they will accept it by applying the label and provide further guidance. The label can be by org members by writing in a comment. <detailsInstructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\n/sig scalability\nI might be related to the npd.\nThe same test failures happening again on 25.02:\n/sig node\n/sig node /cc The restart of caused the flake. I also opened to track it in NPD repo.\nI missed this log. It is caused by oom. Adding some memory may help. I submitted to fix it. Does mean killing metrics-server?\n/remove-sig node Sorry, I misread the log. The NPD found the metric server is OOM. It is about the metrics server, not the NPD. Sorry for disturb. I may add some memory for metric-server and check why metric server memory changed. I heard that metrics server recently released a new version. I will check if it is related.\n/sig node\nOOM metrics shows the metric server is OOMed. This is odd, that only metrics-server-nanny is with a memory limit. It uses whose command is . metrics-server container has no memory limit, but it got killed by OOM. Does this indicate that the node is OOM? (edited)\nThe last run failed in node-problem-detector ContainerStateTerminated. I0227 05:03:31. ] PodStartupLatency: perc50: 14m29.018886827s, perc90: 28m44.625518807s, perc99: 32m2.19303496s; threshold 1h0m0s I0227 05:03:32. ]"}
{"_id":"doc-en-kubernetes-a11072f45f81b280dea24734413173747c87bdd472ce068bad905ce856d1cb81","title":"","text":"Loaded restart count threshold overrides: map[coredns:2 fluentd-gcp:999 konnectivity-agent:2 kube-controller-manager:1 kube-proxy:2 kube-scheduler:1 l7-lb-controller:1 metadata-proxy:2 prometheus-to-sd-exporter:2] I0227 05:03:32. ] ClusterOOMsTracker: gathering cluster OOMs tracking measurement E0227 05:03:32. ] TestMetrics: [action gather failed for SystemPodMetrics measurement: restart counts violation: RestartCount(node-problem-detector-tgl8h, node-problem-detector)=1, want <= 0; RestartCount(node-problem-detector-vpsmz, node-problem-detector)=1, want <= 0] I0227 05:03:32. ] Step \"[step: 29] gathering measurements\" ended W0227 05:03:32. ] Got errors during step execution: [measurement call TestMetrics - TestMetrics error: [action gather failed for SystemPodMetrics measurement: restart counts violation: RestartCount(node-problem-detector-tgl8h, node-problem-detector)=1, want <= 0; RestartCount(node-problem-detector-vpsmz, node-problem-detector)=1, want <= 0]] I0227 05:03:32. ] Waiting for the chaos monkey subroutine to end... I0227 05:03:32. ] Chaos monkey ended. I0227 05:03:32. ] I0227 05:03:32. ] Stopping DnsLookupLatency probe... I0227 05:03:47. ] Stopping InClusterNetworkLatency probe... I0227 05:04:22. ] Resources cleanup time: 50.539700215s E0227 05:04:22. ]\nE0227 05:04:22. ] Test Finished E0227 05:04:22. ] Test: E0227 05:04:22. ] Status: Fail E0227 05:04:22. ]"}
{"_id":"doc-en-kubernetes-8045aaa7c39478c40c58e17441fd375a472ca56f31885311d26bd40c11b2233d","title":"","text":"Errors: [measurement call TestMetrics - TestMetrics error: [action gather failed for SystemPodMetrics measurement: restart counts violation: RestartCount(node-problem-detector-tgl8h, node-problem-detector)=1, want <= 0; RestartCount(node-problem-detector-vpsmz, node-problem-detector)=1, want <= 0]] E0227 05:04:22. ]\nI0227 05:04:22. ]\nI0227 05:04:22. ] Running testing/huge- I0227 05:04:22. ]"}
{"_id":"doc-en-kubernetes-672dc953e52a460d81d453213d4e2e634446d10e93c8bdf3e045ab9c2943abdc","title":"","text":"This is a sub issue for\ncc can you please assign this to me? Thanks.\n, please create PR directly and cc to Tim for review.\n/sig node\nIssues go stale after 90d of inactivity. Mark the issue as fresh with . Stale issues rot after an additional 30d of inactivity and eventually close. Prevent issues from auto-closing with an comment. If this issue is safe to close now please do so with . Send feedback to sig-testing, kubernetes/test-infra and/or . /lifecycle stale\nStale issues rot after 30d of inactivity. Mark the issue as fresh with . Rotten issues close after an additional 30d of inactivity. If this issue is safe to close now please do so with . Send feedback to sig-testing, kubernetes/test-infra and/or . /lifecycle rotten /remove-lifecycle stale\nRotten issues close after 30d of inactivity. Reopen the issue with . Mark the issue as fresh with . Send feedback to sig-testing, kubernetes/test-infra and/or . /close"}
{"_id":"doc-en-kubernetes-9f1156fd8ea5be22a99955ab2508557bd1a6742b9b29a9f12fac2e60dab6df47","title":"","text":"Is this a BUG REPORT or FEATURE REQUEST?: /kind bug What happened: The section for my was stripped, but left for all other resource types What you expected to happen: should put my changes into the cluster as requested How to reproduce it (as minimally and precisely as possible): Here are the resources (Note you will need to change the ownerReference UID and name to some object that exists in your cluster) Observe applying misses out the for the : Observe doing this again updates it correctly: Anything else we need to know?: Environment: Kubernetes version (use ):Cloud provider or hardware configuration: minikube\nReiterating the mentions to trigger a notification: <detailsIn response to : Instructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\nTurns out, shutting down stops this from happening. After creating the object, it has the right ownerReference. On starting , it immediately removed the\n/assign\nfixed in\nThat was impressively fast! Thanks for the fix :)\ncc"}
{"_id":"doc-en-kubernetes-49b5f1148571079e399a01da886ccfbbffa2370f5f5a5844a5ac847b772993f9","title":"","text":"What would you like to be : We want to start migrating metrics which people depend on for operations to stable status. etcdobjectcounts was mentioned in a recent SIG meeting as a viable candidate. However, the 'etcd' prefix is not in line with stated metric conventions. SIG consensus determined that creating a new version of the metric with the metric name 'apiserverstorageobjectcounts' would be sufficient allow this metric to be promoted. We'd want to simultaneously deprecate 'etcdobjectcounts' (since it would be a duplicate) and remove it after the deprecation grace period window. /assign /cc\nGitHub didn't allow me to assign the following users: erain. Note that only , repo collaborators and people who have commented on this issue/PR can be assigned. Additionally, issues/PRs can only have 10 assignees at the same time. For more information please see <detailsIn response to : What would you like to be : /cc Instructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\n/sig instrumentation api-machinery /cc\nAck. I am going to work on this.\n/assign /triage accepted\nIn the SIG api-machinery meeting today, we agreed that the stable version of this metric should be named 'apiserverstorageobject_count'\nThanks and will use in the PR. (And sorry for missing today's meeting - due to a conflict)"}
{"_id":"doc-en-kubernetes-9d5ff2ec60db4b621aa04c218edf9e88ebba3dc541b6528b009593e479ccf9f2","title":"","text":"This was broken by , and therefore broken in as well. Any fix should be offered as a cherry pick so \"make release\" is clean there.\nI went ahead and worked something up for this this morning, to get Shippable builds back on the rails."}
{"_id":"doc-en-kubernetes-e2a306502c13953a6abed9b24a426f5781d6f68d679664b0ba9d46cde514bf5f","title":"","text":"<!-- Please use this template while reporting a bug and provide as much info as possible. Not doing so may result in your bug not being addressed in a timely manner. Thanks!--What happened: These test cases are very hard to follow because they keep toggling the feature gates on and off, and subsequent test cases depend on the behavior of the previous. <!-- DO NOT EDIT BELOW THIS LINE --/kind bug\n/assign\nIssues go stale after 90d of inactivity. Mark the issue as fresh with . Stale issues rot after an additional 30d of inactivity and eventually close. If this issue is safe to close now please do so with . Send feedback to sig-testing, kubernetes/test-infra and/or . /lifecycle stale\n/remove-lifecycle-stale\nStale issues rot after 30d of inactivity. Mark the issue as fresh with . Rotten issues close after an additional 30d of inactivity. If this issue is safe to close now please do so with . Send feedback to sig-testing, kubernetes/test-infra and/or . /lifecycle rotten\nRotten issues close after 30d of inactivity. Reopen the issue with . Mark the issue as fresh with . Send feedback to sig-testing, kubernetes/test-infra and/or . /close\nClosing this issue. <detailsIn response to : Rotten issues close after 30d of inactivity. Reopen the issue with . Mark the issue as fresh with . /close Instructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\n/reopen\nReopened this issue. <detailsIn response to : /reopen Instructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\n/remove-lifecycle rotten\n/help\nThis request has been marked as needing help from a contributor. Please ensure the request meets the requirements listed . If this request no longer meets these requirements, the label can be removed by commenting with the command. <detailsIn response to : /help Instructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\nHey, I'd like to take this if nobody is working on it\nHi thanks for volunteering! Let us know if you have any questions"}
{"_id":"doc-en-kubernetes-10e0800334eabc101ba3e719bde6addc76496fd00734e09bb571fcc5b6be7d48","title":"","text":"What happened: In a dual-stack cluster when specifying both ipv6 and ipv4 addresses in the entries are inserted in ipvs for the ipv6 nodeports but with an invalid ipv4 address; The example shows the ipv4 loopback address but any ipv4 address in will cause a faulty entry. What you expected to happen: No ipv4 entry shall be inserted for the ipv6 nodeport. How to reproduce it (as minimally and precisely as possible): Configure with both ipv4 and ipv6 addresses Bring up a dual-stack cluster with proxy-mode Create a service with ipFamily:IPv6 Do Anything else we need to know?: Proxy-mode=iptables seems to work. With no ipv4 entries are inserted for ipv6 nodeports. In it is recommended to use but doing so makes kube-proxy enter rules for address (in addition to \"127.0.0.1\"). While these entries cause no harm they can't be used and should not be there. These entries are created for both proxy-mode \"iptables\" and \"ipvs\". This should not be handled in this issue but in another (new) issue if needed. Example; Environment: Kubernetes version (use ):Cloud provider or hardware configuration: Kvm's on Ubuntu OS (e.g: ): xcluster Kernel (e.g. ):Install tools: \"from scratch\" Network plugin and version (if this is a network-related bug): calico/cni:v3.12.0 Others: /sig network\n/triage unresolved Comment when the issue is assessed and confirmed. I am a bot run by vllry. ‍\n/area ipvs /area ipv6\n/remove-triage unresolved /assign I take this myself if no one objects.\nThe ipv4 address is obviously a ipv4-converted ipv6 from .\nAccording to the code here (), we may need to filter the before passing it to .\nLooks that that's it. Please go ahead and make a PR, or I can do it if you prefer. I will help test the PR of course. Here are some review comments right away; Use v1.IPFamily as param to the filter function rather than The filter function seem usable and generic and may be placed in some \"util\" package, but leave it as-is and let the maintainers decide that.\nHi feel free to send a PR please, as you have took this already :) I'd like to help review it!\nThe \"addresses\" are really CIDRs which make the fix even simpler."}
{"_id":"doc-en-kubernetes-3bb3543c0323a70faea91ccc43a3eaf950d3e30c811cd93e021b94a126351841","title":"","text":"and provide for configuring Flannel's parameter but none of the others listed in"}
{"_id":"doc-en-kubernetes-349c839c0314e1386e949c881af8735bbe6497abcf3a15697139e75deb1d1835","title":"","text":"Github automatically adds a artifact to every release, which sounds very similar to the artifact that we generate as the release artifact. Assuming we don't have control over the naming or presence of the autogenerated source code artifact, I think we should consider renaming our published artifact to better disambiguate between the two.\nWe distribute most of our artifacts outside of GitHub. We should probably stop uploading this one entirely.\ncc\nWe've got a thread going on w/ Tim. /assign\ncan you please link so folks can follow the bread crumbs? ( cc )\n-- It's currently sensitive. Will share once there's an outcome.\nACK. thanks for the heads up\nI think it's safe to discuss openly at this point. We had a report that the sha512 hashes on our release artifacts were incorrect. We have confirmed that the mismatch was due to a bug in the release pipeline, and not malicious. This issue here was just a point of confusion that came out of that discussion.\nthanks !\n/assign\nGoing one step back since I think the exists for historical reasons: The client and server binaries were part of this bundle, which is not the case any more: So it seems that is mostly a collection of licenses and scripts. Do we really need it or may we cleanup some code paths and remove it completely?\n+1 to put the word out and remove it completely if no one depends on it.\nI'm not confident that CI is not using these versioned scripts, in fact i'm fairly sure that it is. However it would not be pulling it from github.\nThe Kubernetes project currently lacks enough contributors to adequately respond to all issues and PRs. This bot triages issues and PRs according to the following rules: After 90d of inactivity, is applied After 30d of inactivity since was applied, is applied After 30d of inactivity since was applied, the issue is closed You can: Mark this issue or PR as fresh with Mark this issue or PR as rotten with Close this issue or PR with Offer to help out with [Issue Triage][1] Please send feedback to sig-contributor-experience at . /lifecycle stale [1]:\n/remove-lifecycle stale I think this remains as relevant as ever.\n/milestone v1.23\n/triage accepted\nHi and k8s 1.23 bug triage shadow here."}
{"_id":"doc-en-kubernetes-dbd15f43daca5c062443bfb889f914327019b692ac2d45e387a4f7451b197946","title":"","text":"There hasn't been any activity or linked PRs for this issue. Just wanted to check in to see if this is still on track for 1.23.\n/milestone v1.24\nHi I am part of the bug triage team for v1.24. We noticed that this issue hasn't been updated for a while. Just wanted to check if there is any work planned that would go into this.\nthis one keeps moving down the milestones. i don't see any activity since november. So kick the can down the road /milestone v1.25\nHi My name Hossein Salahi from Bug-Triage team. I am just checking to ensure if everything is on track for K8s 1.25. Thanks!\nThat's a question for or\nany remarks on that?\nthe release 1.25 bug triage Lead. I'm reaching out to see the status of this issue and if we are still targeting the current release. Also, kindly if you can update the priority so we can track it in the future.\nthis one keeps moving down the milestones yet again!. i don't see any activity since i moved this down the line. So kick the can down the road: /milestone clear (instead of v1.26 as this has been moving down the line for a few releases now)\nThe Kubernetes project currently lacks enough contributors to adequately respond to all issues and PRs. This bot triages issues and PRs according to the following rules: After 90d of inactivity, is applied After 30d of inactivity since was applied, is applied After 30d of inactivity since was applied, the issue is closed You can: Mark this issue or PR as fresh with Mark this issue or PR as rotten with Close this issue or PR with Offer to help out with [Issue Triage][1] Please send feedback to sig-contributor-experience at . /lifecycle stale [1]:\n/remove-lifecycle stale\nWhat's the current status of this issue? Is there anything that we can do to move this forward?\nWe got new conversations going on during Kubecon on whether we should publish our release binaries on GitHub. Given that and I think we should rethink now what we publish on the release pages. I think we should start publishing all binaries for one. It is a common practice everywhere and I think users largely expect them to be there."}
{"_id":"doc-en-kubernetes-63377ea8eda65aaaac09acc9ed7dd8461429e8729a6daebbb93d9c0997493c68","title":"","text":"let's scope the issue to the actual \"Kubernetes Source Code\" tarball. We can still remove it from what I can see from the discussion.\nI agree with Removing the \"Kubernetes Source Code\" tarball should be as easy as removing this from krel: However, there are a few questions: is it safe to do so? I believe yes, because this \"Kubernetes Source Code\" tarball is pointing to the GCS bucket, and we'll still continue publishing that tarball to the bucket The link is also available in the changelog, so it's easy to find do we need any deprecation period? Considering the above two points, I don't think so. If you folks agree with that, I can go ahead and create a PR to update krel.\nI agree on that, let's drop an additional note on k-dev that others are aware of the change.\n/assign\nUpdate: An announcement to proceed with the removal of the \"Kubernetes Source Code\" artifact has been sent to k-dev: PR to remove this artifact from being published: Removal will go into effect starting the January patch releases."}
{"_id":"doc-en-kubernetes-4f983686ce6cef6162dd147e0e53be05aced0afa9196140d2e8041aafeaab5cd","title":"","text":"As discussed, similar patterns occur in many of the controller for federated objects. Lift this into a common library that can be used by all of the controllers. Open separate issues for each of the above, and track there. This can be the tracking issue over those.\nHi, , we have been starting develop work about cluster-controller and RS controller, so would you like to share me the plan about this work? Thanks.\nTracked in PR's referenced above. Closing this tracking issue."}
{"_id":"doc-en-kubernetes-747d699b97976a530bdb275e803e2be1bb68ac7168a6ff8b59806b000dbdf33f","title":"","text":"<!-- Please only use this template for submitting enhancement requests --What would you like to be : Change scheduler metric from Gauge to Histogram Why is this needed: is recorded for each preemption attempt. A histogram is better. /sig scheduling /help /good-first-issue\n/help /good-first-issue\nThis request has been marked as suitable for new contributors. Please ensure the request meets the requirements listed . If this request no longer meets these requirements, the label can be removed by commenting with the command. <detailsIn response to : /help /good-first-issue Instructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\nI can work on this issue.\n/assign\nI could work on this as well if doesn't have time to work on this.\nI am working on it, I will create the initial PR in less than a week. Sorry for the delay."}
{"_id":"doc-en-kubernetes-d5f405b98ac26e68bb52a403345cda18a0b690bbcb8385b5ec06372b302c6e2d","title":"","text":"<!-- Please use this template while reporting a bug and provide as much info as possible. Not doing so may result in your bug not being addressed in a timely manner. Thanks! If the matter is security related, please disclose it privately via --What happened: Data race in client-go with below sample code: Data race: What you expected to happen: No data race How to reproduce it (as minimally and precisely as possible): Using above given sample code Kubernetes version (use ):Cloud provider or hardware configuration:OS (e.g: ): OSX\n/sig api-machinery\nI'd like to take it. /assign\n/remove-sig api-machinery /sig auth cli\nBug triage for release 1.17 here. Since this issue has been tagged for the milestone, we would like to check it's status. Code freeze is on Nov. 14. Will this issue be resolved before then?\nAnother reminder here that code freeze is in less than two weeks (Nov. 14). We will need to move this issue out of the milestone if isn't going to be resolved before then."}
{"_id":"doc-en-kubernetes-76a1437b2e9419a405205d59af74d9b368790a08551fbd85b287efde2a056dcd","title":"","text":"During debugging , I noticed that one of services pods: monitoring and logging pods stuck at pending sometimes forever: I logged into the node, and found the container is running happily on the node: Looks like status is never reported from kubelet. I checked kubelet log, then I found: Why we cannot delete mirror pod? and Created a new one? Because this regression, once we run into such state, the rest of test won't be triggered, and eventually timeout / failed.\nchecks those service pods in ready state, and prevent the test to run once it is pending. cc/ I think this is the real issue I talked to you earlier about PodStatus.\nI spoke to offline. For some reason, kubelet thinks the spec of the mirror pod does not match the static pod locally. It then decides to delete the mirror pod and recreate it, repeatedly. As a result of that, it slows down (due to the QPS limit?) reporting statuses of other pods. This can be reproduced by running the e2e \"Shell\" test. I am looking into why the specs would suddenly mismatch. Did any recent PR modified the pod spec in apiserver?\nThe recently introduced default cpu share limit was not there in the static pod, causing the difference between its spec and the mirror pod's.\nSeeing this as well. Makes the whole cluster slow and unusable for me. Sent from my iPhone\nIs there a desired fix for this?\nsaid she would add default resource limits to the static pod manifest files. If we go for that, I think it's reasonable to fail (not run the static pod) if resource limit is not specified. Also, the resource limits set in the mirror pods are not meaningful because kubelet creates containers based on the spec of the static pod. In general, static pods do not subject to any defaulting values applied at the apiserver.\nPretty sure that this is fixed by . Certainly testing against head since then works well for me. What version are you testing?\nwas fixing node status, and the problem had been around for a while. This issue is about pod status, which is still unfixed on head. These two are very different, and my cluster did include the fix :) This issue will slow down kubelet at times, but if you get lucky and the pod status (running) get sent out in time, the tests will still pass."}
{"_id":"doc-en-kubernetes-dbb34576173a9a8fefb5aa440567e5b808bc06b80f759ff3da54bf129fc77991","title":"","text":"There is no way the issue is fixed by that PR. I am still seeing the issue on HEAD. The system gets worse the longer its up. I agree with the analysis of the issue in source PR Sent from my iPhone\nI am on this one, and here is proposed solution in my mind so far (not final yet since I haven't put too much thoughts into it yet): 1) For all our static pods I am going to update them with some initial cpu limit for now. This should quickly stop flakiness caused by this particular issue. 2) Next step is that for static pods, if there is no initial cpu limit, kubelet should refuse to run. Before I do this, I need to send an announcement to our container-vm users about this potential breakage for the in the future. Another potential fix is we can ignore resource changes when comparing if a mirror pod is updated or not. I think is proposed a similiar one earlier. But I had a concern with it: this is defeat the initial purpose we created mirror pods and sync back to master. So if one made some resource changes on the static pods, you have to always ssh-ed into the node to see if the new value applied or not. Also eventually (post 1.0) those mirrored pod's resource limit is going to be used by scheduler when calculating if user pods are fit or not, we need them to be accurated. Another concern is in the code implementation itself to distinuguish mirror pods future from rest of pods. I guess this shouldn't be the big issue here. Any more suggestions here? One puzzle I had is that why master component pods doesn't have issue at all. I manually restart kube-apiserver pod, still doesn't have the issue.\nDepending on time we could just remove the default limit from the limit range as a stop gap. Sent from my iPhone\nCan you just bypass defaulting at the apiserver when you create mirror pods? I can't think of any other solution that works in general and is easily maintainable. Approach 2 you described in the previous entry works if we are defaulting cpu limit, but breaks as soon as we default anything else."}
{"_id":"doc-en-kubernetes-f6117564d0d1387afeb4e79ddb55d47325739be1eed6d26971dac3a1b60079fd","title":"","text":"Same problem with ignoring resource differences when comparing mirror pod to static pod -- as long as defaulting only touches resources, that works, but if we change it to touch other stuff, then it breaks.\nCommit I had can quickly fix current bleeding issue. Currently we only have master component pods and fluentd pod are file based by default. Master component pods doesn't create any issue now because mirror pods aren't created at the first place (due to recent change made by Also file-based pods will have issue only when its spec doesn't have initial value, and need to update (delete miss matched mirror pod, and recreate a new one). Please note that with my commit, everything is working even I manually update its cpu limit from 100m to 200m, there is no issue at all, and query pod shows accurate result with the new limit too: How badly we request all users using file-based pod feature providing default cpu limit for now? Eventually we want to get rid of file-based pods completely, also encouraging the user provides the initial resource limit is aligning with our direction in a long run anyway.\nJust to clarify, what I was suggesting was for a \"real fix\" not the temporary measure to stop the bleeding (which you already have in your PR). But what do you think of what I suggested as the longer-term fix? My point is that today, we default CPU, but in the future, we may default anything. It seems that the most maintainable fix is to just not do defaulting when creating mirror pod from static pod. Is that possible? I wonder if the reason manual update works (what you described in your last comment) is because we don't go through the defaulting when you do an update, only a create? (Though that sounds like a bug.) IMO we don't need CPU default right now, especially since we already have kubelet flag for limiting pods per machine (though unfortunately it's currently set to 100 because we haven't reconfigured the tests that intentionally run lots of pods per node, to override the flag. we can change the default to 30 and add an override mechanism if this is important.)\nComparing a config file with the live state and expecting them to match is doomed. The recommended approach is described by . Kubelet could store the config as an annotation in the mirror pod and compare with that."}
{"_id":"doc-en-kubernetes-335d12ba313aa19619a2c4b18ee4ddf7703e123f21840f5e77167fa2ece19bed","title":"","text":"Does the scheduler favor nodes with the fewest pods even if they don't specify resources? LimitRange does cause that to happen.\nStoring the config as an annotation sounds good to me. Filed\nAlthough this issues was closed, it still appears in my environment\nSame here, this happens from time to time in our environments, is there a way to debug to find the root cause?"}
{"_id":"doc-en-kubernetes-db8b10ac218e6b31cf143cad82d1c00f690bf029d07656bb83a6568343d76936","title":"","text":"It's useful to have a helm chart that expose a value like . However, the annotation simply ignores , as seen in It's just checking that the value is not empty, so even if it's false - it'll try to create internal LB.\nReiterating the mentions to trigger a notification: <detailsIn response to : Instructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\n/assign\nGitHub didn't allow me to assign the following users: mcrute. Note that only and repo collaborators can be assigned. For more information please see <detailsIn response to : /assign Instructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details"}
{"_id":"doc-en-kubernetes-4b95e73f98faca5940a16ee2fccddf37ae93437213e073578ff34d76b732271f","title":"","text":"Is this a BUG REPORT or FEATURE REQUEST?: /kind bug What happened: Service loadbalancer source range don't work in ipvs proxy mode. Which result in gci-gce-ipvs CI failed. http://k8s- How to reproduce it (as minimally and precisely as possible): In gce environment. a service with backend. service type to service util service external IP is specified. the service from pod inside and pod out of the . Environment: Kubernetes version (use ): master Cloud provider or hardware configuration: gce OS (e.g. from /etc/os-release): ubuntu Kernel (e.g. ): Install tools: cluster/kube- Others:\n/kind bug /sig network /label ipvs\n/area ipvs\nThe test seems to be flaky now. Can you take a look please?\nI don't know why, but I think this flaky have the similiar reason with other normal service-test cases, like Those flaky are pretty annoying, problem is, I have run those cases in my own environment many times, never fail... : ( But still, I'll do my best to find the reason."}
{"_id":"doc-en-kubernetes-1983b5ea5659683f3a4ba96803f2792373389d2c0ef10ffa2f957e3f76064a9d","title":"","text":"I'm using K8S 1.10 and have a specific use case where I'm trying to use an emptyDir to share mounts between two containers in a pod. I'm using the MountPropagation feature successfully, but while debugging another issue I noticed that the node was running a very long running du command. Looking into the source I noted: Which runs du on any emptyDir, do measure disk usage. I recommend that in this case we should add the -x flag that will skip any mounted directories as they do not count against the file usage and would make these scans a lot faster. I could of course add a simple pull request for this but first wanted to see if this was known and on purpose. To reproduce bind mount/loop mount/device mount any large/slow directory into an emptyDir volume and watch as the du process slogs away Environment: Kubernetes 1.10 GKE <!-- DO NOT EDIT BELOW THIS LINE --/kind bug\nReiterating the mentions to trigger a notification: <detailsIn response to : Instructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\nI think it makes sense to ignore submounts.\nIssues go stale after 90d of inactivity. Mark the issue as fresh with . Stale issues rot after an additional 30d of inactivity and eventually close. If this issue is safe to close now please do so with . Send feedback to sig-testing, kubernetes/test-infra and/or . /lifecycle stale\n/remove-lifecycle stale\nIssues go stale after 90d of inactivity. Mark the issue as fresh with . Stale issues rot after an additional 30d of inactivity and eventually close. If this issue is safe to close now please do so with . Send feedback to sig-testing, kubernetes/test-infra and/or . /lifecycle stale\ndo you still want to submit a PR to fix this? Thanks!\nStale issues rot after 30d of inactivity. Mark the issue as fresh with . Rotten issues close after an additional 30d of inactivity. If this issue is safe to close now please do so with . Send feedback to sig-testing, kubernetes/test-infra and/or ."}
{"_id":"doc-en-kubernetes-eea01b18436aabcf24011fc8d3531d9a102c009dd1b30b027c1434bcedaabd88","title":"","text":"/lifecycle rotten\nRotten issues close after 30d of inactivity. Reopen the issue with . Mark the issue as fresh with . Send feedback to sig-testing, kubernetes/test-infra and/or . /close\nClosing this issue. <detailsIn response to : Rotten issues close after 30d of inactivity. Reopen the issue with . Mark the issue as fresh with . /close Instructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details"}
{"_id":"doc-en-kubernetes-5e1e446ffee113a0fecf68600791d8062befcc769908a1fdd16ec9abcfdd9c85","title":"","text":"[X] Two weeks soak end date: Nov 18 According to this APIsnoop query, there are still a remaining Service endpoint which is untested. A service resource didn&rsquo;t support endpoint until was merged. Create a watch to track service events Create a service with a static label. Confirm via the watch that the service was created. Delete the service via the service with the dynamic client with . Confirm via the watch that the service was deleted. The above steps have been implemented in the follwing . This query shows all the endpoints hit within a short period of running the above e2e test If a test with these calls gets merged, test coverage will go up by 1 point This test is also created with the goal of conformance promotion. /sig testing /sig network /area conformance\n/triage accepted\nWaiting on 2 week soak before promoting the test to conformance /reopen\nReopened this issue. <detailsIn response to : Waiting on 2 week soak before promoting the test to conformance /reopen Instructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details"}
{"_id":"doc-en-kubernetes-21b704a68514d2b400f1a4f479b12f48c6dc4bd890b9d4b14cffa5ee4c13d7d4","title":"","text":"During\nany chance this is related to pass: https://k8s- fail: https://k8s- https://k8s-\nThis is because for some reason the submit queue rebuild didn't catch the flake before merging. Protobuf landed first, merging a new api object like petset should have failed.\nQuick fix is to rebuild the proto marshalers and push.\nAlso, hack/verify-generated- is broken, which probably is the root cause.\nLooking at it now\nThree problems: -generated-protobuf wasn't updated when we made update-generated-protobuf use docker, so it didn't fail out script was verifying the old API groups, not any of the new API groups that have been merged without protobuf serializations (although pkg/api/serialization_test should have failed, possibly the job didn't build with the latest master?), which caused the tests to fail\nSomething about the merge bot is broken - it merged without all the tests being green: . That's why this broke -\nSpawned since there is something wrong with update-generated-protobuf - will merge that to fix the failing test and then debug the generator in the morning.\nThe bot can race with pushes between when it starts testing and when it merges..."}
{"_id":"doc-en-kubernetes-2eefef1a83ddd783604f0989f7bcbb86ee11f3e15f09ced1fa06f0ad1ee322e6","title":"","text":"What happened: states that Container.SecurityContext applies to a Pod, which can be misleading since we already have PodSecurityContext. What you expected to happen: Documentation states that it applies to container How to reproduce it (as minimally and precisely as possible): Anything else we need to know?: Environment:\n/sig api-machinery /kind documentation\n/cc\nIssues go stale after 90d of inactivity. Mark the issue as fresh with . Stale issues rot after an additional 30d of inactivity and eventually close. If this issue is safe to close now please do so with . Send feedback to sig-testing, kubernetes/test-infra and/or . /lifecycle stale\nNot stale.\nStale issues rot after 30d of inactivity. Mark the issue as fresh with . Rotten issues close after an additional 30d of inactivity. If this issue is safe to close now please do so with . Send feedback to sig-testing, kubernetes/test-infra and/or . /lifecycle rotten\n/remove-lifecycle rotten\nIssues go stale after 90d of inactivity. Mark the issue as fresh with . Stale issues rot after an additional 30d of inactivity and eventually close. If this issue is safe to close now please do so with . Send feedback to sig-testing, kubernetes/test-infra and/or . /lifecycle stale\n/remove-lifecycle rotten\nStale issues rot after 30d of inactivity. Mark the issue as fresh with . Rotten issues close after an additional 30d of inactivity. If this issue is safe to close now please do so with . Send feedback to sig-testing, kubernetes/test-infra and/or . /lifecycle rotten\nRotten issues close after 30d of inactivity. Reopen the issue with . Mark the issue as fresh with . Send feedback to sig-testing, kubernetes/test-infra and/or . /close\nClosing this issue. <detailsIn response to : Rotten issues close after 30d of inactivity. Reopen the issue with . Mark the issue as fresh with . /close Instructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details"}
{"_id":"doc-en-kubernetes-cc8f7dd240ce03e902a3a6988aef93cbe00593e0f5081bd8380b06a0a728c5de","title":"","text":"Currently watches only log when they complete, which can make analyzing kube-apiserver log files a bit tricky. (did component x never start watching after that list or did apiserver shut down?)"}
{"_id":"doc-en-kubernetes-f20fc01164fcc89f452d8525b342f756ca92b86e0d081c91856f4194b395bf42","title":"","text":"on the end of script there is a line: It produces high CPU load when it is run without tty. However it can be replaced with something like this:\nLooks like this has been fixed."}
{"_id":"doc-en-kubernetes-0ee4ddd459987abc7d27aa8e986e315d7cf0d398b8e3fe4f5bf18ea9cc8fa8dc","title":"","text":"https://k8s- Failed: [] Pods should support retrieving logs from the container over websockets {E2eNode Suite} Happened on a presubmit run in .\nhttps://k8s- Failed: [] Pods should support retrieving logs from the container over websockets {E2eNode Suite}\nhttps://k8s- Failed: [] Pods should support retrieving logs from the container over websockets {E2eNode Suite} Happened on a presubmit run in .\nhttps://k8s- Failed: [] Pods should support retrieving logs from the container over websockets {E2eNode Suite}\nhttps://k8s- Failed: [] Pods should support retrieving logs from the container over websockets {E2eNode Suite} Happened on a presubmit run in .\nhttps://k8s- Failed: [] Pods should support retrieving logs from the container over websockets {E2eNode Suite} Happened on a presubmit run in .\nhttps://k8s- Failed: [] Pods should support retrieving logs from the container over websockets {E2eNode Suite} Happened on a presubmit run in .\nhttps://k8s- Failed: [] Pods should support retrieving logs from the container over websockets {E2eNode Suite}\nhttps://k8s- Failed: [] Pods should support retrieving logs from the container over websockets {E2eNode Suite} Happened on a presubmit run in .\nhttps://k8s- Failed: [] Pods should support retrieving logs from the container over websockets {E2eNode Suite} Happened on a presubmit run in .\nhttps://k8s- Failed: [] Pods should support retrieving logs from the container over websockets {E2eNode Suite} Happened on a presubmit run in .\nhttps://k8s- Failed: [] Pods should support retrieving logs from the container over websockets {E2eNode Suite} Happened on a presubmit run in .\nhttps://k8s- Failed: [] Pods should support retrieving logs from the container over websockets {E2eNode Suite} Happened on a presubmit run in ."}
{"_id":"doc-en-kubernetes-fb627aaedb9842e8bb093b3b041ba36363dfd64edbe1bdb335141eb36555e82f","title":"","text":"https://k8s- Failed: [] Pods should support retrieving logs from the container over websockets {E2eNode Suite}\nhttps://k8s- Failed: [] Pods should support retrieving logs from the container over websockets {E2eNode Suite} Happened on a presubmit run in .\nhttps://k8s- Failed: [] Pods should support retrieving logs from the container over websockets {E2eNode Suite} Happened on a presubmit run in .\nhttps://k8s- Failed: [] Pods should support retrieving logs from the container over websockets {E2eNode Suite} Happened on a presubmit run in .\nhttps://k8s- Failed: [] Pods should support retrieving logs from the container over websockets {E2eNode Suite} Happened on a presubmit run in .\nhttps://k8s- Failed: [] Pods should support retrieving logs from the container over websockets {E2eNode Suite} Happened on a presubmit run in .\nhttps://k8s- Failed: [] Pods should support retrieving logs from the container over websockets {E2eNode Suite} Happened on a presubmit run in .\nhttps://k8s- Failed: [] Pods should support retrieving logs from the container over websockets {E2eNode Suite} Happened on a presubmit run in .\nhttps://k8s- Failed: [] Pods should support retrieving logs from the container over websockets {E2eNode Suite} Happened on a presubmit run in .\nhttps://k8s- Failed: [] Pods should support retrieving logs from the container over websockets {E2eNode Suite} Happened on a presubmit run in .\nhttps://k8s- Failed: [] Pods should support retrieving logs from the container over websockets {E2eNode Suite}\nThe test pod gets deleted by the test upon failure and so there is not enough information about the state of the test pod. I'm changing the test to not delete the pod and instead let the namespace deletion logic in the test framework handle deletions. I'm also increasing the pod lifetime from to a longer duration to avoid any possible race conditions in the test. The test will also retry if the logs output from the test pod does not contain non-space characters."}
{"_id":"doc-en-kubernetes-eece27d155d201a0e15709c787b1cd71b8915e6d11b38f7ce2b38d06c62ffea6","title":"","text":"I'm lowering the priority of this test once my changes go in and see if the flake re-appears.\nhttps://k8s- Failed: [] Pods should support retrieving logs from the container over websockets {E2eNode Suite} Happened on a presubmit run in .\nhttps://k8s- Failed: [] Pods should support retrieving logs from the container over websockets {E2eNode Suite} Happened on a presubmit run in .\nhttps://k8s- Failed: [] Pods should support retrieving logs from the container over websockets {E2eNode Suite} Happened on a presubmit run in .\nI suspect a race between container starting up and the test querying the logs. I'm adding a retry in the test to avoid the race. Lowering the priority since I still don't suspect any kubelet bugs.\nhttps://k8s- Failed: [] Pods should support retrieving logs from the container over websockets {E2eNode Suite} Happened on a presubmit run in .\nhttps://k8s- Failed: [] Pods should support retrieving logs from the container over websockets {E2eNode Suite} Happened on a presubmit run in .\nhttps://k8s- Failed: [] Pods should support retrieving logs from the container over websockets {E2eNode Suite}\nhttps://k8s- Failed: [] Pods should support retrieving logs from the container over websockets {E2eNode Suite} Happened on a presubmit run in ."}
{"_id":"doc-en-kubernetes-bc1f56f76d806f780c64417014c4b072addf1f4df6d254dd55319774f349a24c","title":"","text":"Flaked a couple of times on serial lane with the same error. Test is relatively new. This run only has the single test flake: /sig node /kind flake /cc\n/assign\nNew recent failures: /triage accepted"}
{"_id":"doc-en-kubernetes-ff21ac8c2a6c2ea4821e58492f0a8553d8a302a4afa86e4805bd139290e21fea","title":"","text":"The pod security policy capability test is using wrong CAP format with prefix. In our document, we explicitly say: It fails for cri-containerd. I'll find out why it doesn't fail for Docker, maybe docker has handled it. But we should fix the test anyway.\nThere are no sig labels on this issue. Please add a sig label. <detailsA sig label can be by either: a sig: e.g., to notify the contributor experience sig, OR the label manually: e.g., to apply the label Note: Method 1 will trigger an email to the group. See the . The in method 1 has to be replaced with one of these: bugs, feature-requests, pr-reviews, test-failures, proposals Instructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details"}
{"_id":"doc-en-kubernetes-2a143cea40d48fac7382a327617b071e2fea640f42b4cdcb181da1800ffddc19","title":"","text":"[ ] OWNERSALIASES:581: # conformance aliases [x] // More info: ~~ [x] // See [x] fs.MarkDeprecated(\"redirect-container-streaming\", \"Container streaming redirection has been removed from the kubelet as of v1.20, and this flag will be removed in v1.22. For more details, see \") [x] // More info: [x] // More info: ~~ [x] [x] [x] CHANGELOG/CHANGELOG-[scalability validation report]: ~~ [x] ALPHA stability level ) [x] ALPHA stability level ) [x] ALPHA stability level ) [x] ALPHA stability level ) [x] ALPHA stability level ) [x] ALPHA stability level ) [x] ALPHA stability level ) [x] ALPHA stability level ) [x] ALPHA stability level ) [x] ALPHA stability level ) [x] ALPHA stability level ) [x] ALPHA stability level ) [ ] link: [ ] link: Some needs to be regenerated\nThis issue is currently awaiting triage. If a SIG or subproject determines this is a relevant issue, they will accept it by applying the label and provide further guidance. The label can be by org members by writing in a comment. <detailsInstructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\n/help\n/kind documentation\n/remove-kind bug\n/good-first-issue\n/assign I shall fix .\n/assign i'm working in\ni have create a pr for\ni see , you need\n/assign fixing\n/assign\n/assign CHANGELOG/CHANGELOG-\n/assign CHANGELOG/CHANGELOG- Edit : Link is already resolved, unassigning\n/assign CHANGELOG/CHANGELOG-\n/assign ,111\nSorry, already assigned\n/assign i will take\n/reopen\nReopened this issue. <detailsIn response to : /reopen Instructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\n  is merged. Most KEP links are fixed. You may scan it again. fixed most of them.\n/sig architecture\n/assign I will work on and\n/remove-help"}
{"_id":"doc-en-kubernetes-9989a24ad22361651b9b153ae59f4b05a7f4d39865c45b98092f106c43ae9e01","title":"","text":"pull-kubernetes-unit TestDifferentWidths As far back as triage board history shows No response /sig api-machinery\n/cc /triage accepted\nThe Kubernetes project currently lacks enough contributors to adequately respond to all issues and PRs. This bot triages issues and PRs according to the following rules: After 90d of inactivity, is applied After 30d of inactivity since was applied, is applied After 30d of inactivity since was applied, the issue is closed You can: Mark this issue or PR as fresh with Mark this issue or PR as rotten with Close this issue or PR with Offer to help out with [Issue Triage][1] Please send feedback to sig-contributor-experience at . /lifecycle stale [1]:\nThe Kubernetes project currently lacks enough active contributors to adequately respond to all issues and PRs. This bot triages issues and PRs according to the following rules: After 90d of inactivity, is applied After 30d of inactivity since was applied, is applied After 30d of inactivity since was applied, the issue is closed You can: Mark this issue or PR as fresh with Close this issue or PR with Offer to help out with [Issue Triage][1] Please send feedback to sig-contributor-experience at . /lifecycle rotten [1]:\nThe Kubernetes project currently lacks enough active contributors to adequately respond to all issues and PRs. This bot triages issues and PRs according to the following rules: After 90d of inactivity, is applied After 30d of inactivity since was applied, is applied After 30d of inactivity since was applied, the issue is closed You can: Reopen this issue or PR with Mark this issue or PR as fresh with Offer to help out with [Issue Triage][1] Please send feedback to sig-contributor-experience at . /close [1]:\nClosing this issue. <detailsIn response to : The Kubernetes project currently lacks enough active contributors to adequately respond to all issues and PRs. After 90d of inactivity, is applied After 30d of inactivity since was applied, is applied After 30d of inactivity since was applied, the issue is closedReopen this issue or PR with Mark this issue or PR as fresh with Offer to help out with [Issue Triage][1] Instructions for interacting with me using PR comments are available ."}
{"_id":"doc-en-kubernetes-c8c180b3e6dad8fab04f5f208d5a3b3aff8479bb3125ba5ea96ea97ccc2bab4f","title":"","text":"If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\nRunning into this one a fair bit in my PRs, opened as a possible fix\n/reopen\nReopened this issue. <detailsIn response to : /reopen Instructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\nThe Kubernetes project currently lacks enough active contributors to adequately respond to all issues and PRs. This bot triages issues according to the following rules: After 90d of inactivity, is applied After 30d of inactivity since was applied, is applied After 30d of inactivity since was applied, the issue is closed You can: Reopen this issue with Mark this issue as fresh with Offer to help out with [Issue Triage][1] Please send feedback to sig-contributor-experience at . /close not-planned [1]:\nClosing this issue, marking it as \"Not Planned\". <detailsIn response to : The Kubernetes project currently lacks enough active contributors to adequately respond to all issues and PRs. After 90d of inactivity, is applied After 30d of inactivity since was applied, is applied After 30d of inactivity since was applied, the issue is closedReopen this issue with Mark this issue as fresh with Offer to help out with [Issue Triage][1] Instructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\n/reopen I got this in --\nReopened this issue. <detailsIn response to : /reopen Instructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\nThe Kubernetes project currently lacks enough active contributors to adequately respond to all issues and PRs. This bot triages issues according to the following rules: After 90d of inactivity, is applied After 30d of inactivity since was applied, is applied After 30d of inactivity since was applied, the issue is closed You can: Reopen this issue with Mark this issue as fresh with Offer to help out with [Issue Triage][1] Please send feedback to sig-contributor-experience at ."}
{"_id":"doc-en-kubernetes-6804092b9f9f291bd1b3f24d906e3547bfaa841487790436e20f51d6b5e5c17d","title":"","text":"/close not-planned [1]:\nClosing this issue, marking it as \"Not Planned\". <detailsIn response to : The Kubernetes project currently lacks enough active contributors to adequately respond to all issues and PRs. After 90d of inactivity, is applied After 30d of inactivity since was applied, is applied After 30d of inactivity since was applied, the issue is closedReopen this issue with Mark this issue as fresh with Offer to help out with [Issue Triage][1] Instructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\n/reopen Got this in\nReopened this issue. <detailsIn response to : /reopen Got this in </details\n/remove-lifecycle rotten"}
{"_id":"doc-en-kubernetes-ee0826757468ea334f95c77509b257c3d392b6713b4311ba7afa125d642dcec9","title":"","text":"The based container images for are shipped with the wrong architecture within the manifest if not amd64: Refers to We updated the images a few minutes ago to buster-v1.3.0, which are now ready to be used as base for the multi-architecture images. If the images contain the right architectures, then it may magically happen that the kube-proxy images are fixed, too. This is because we already use the build syntax for it: /cc\nThis issue is currently awaiting triage. If a SIG or subproject determines this is a relevant issue, they will accept it by applying the label and provide further guidance. The label can be by org members by writing in a comment. <detailsInstructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\n/sig release"}
{"_id":"doc-en-kubernetes-38c129cdce20cdb2ede95770ab1d3483b461da6e2945915e074275e9575e4c1c","title":"","text":"Similar to issue , building kubernetes1.23 resulted in 1 variation: builds should be fully determined by inputs on openSUSE or Debian, This bug was found while working on . <details23.0 </details<details</details<details</details<details</details<details</details<details</details\n/sig release /area build-release\n/assign\n/sig cli\n/triage accepted /assign"}
{"_id":"doc-en-kubernetes-d654f914f6cdd0bd3bce5db6eb3c4ebef631bd8d2558ac9d591cab12ba724061","title":"","text":"What happened: e2e tests detect whether PSP is enabled based on presence of PSP API objects. An addon that proactively includes a PSP in a manifest can trigger that detection and enable PSP-specific e2e tests that should not be run. What you expected to happen: In addition to detecting presence of PSP objects, consider dry-run creating a pod and verifying the annotation indicating the selected PSP policy is , indicating the admission plugin is active. /sig auth /assign"}
{"_id":"doc-en-kubernetes-7efc22ed3a8fb290fc16a7b40edf13445187015d16976624a8ed8ba9d9def220","title":"","text":"I ran conformance tests on a cluster that had a controller enabled that syncs a configmap across all clusters. It failed: Appears to be due to a 2 configmap limit: These clusters should pass conformance. This is a very common pattern (even Kubernetes itself does it!). I didn't run it myself, but apparently just install Istio and run the conformance suite. See for details Kubernetes version (use ): v1.20.2\n/assign\nmy guess is the CA configmap created by default in 1.20 exacerbated this issue. I agree the conformance test should make fewer assumptions about the cluster it is running in and should accomodate default configmaps. left a comment on the PR in flight for this. I'd like to see the change picked to 1.20 as well.\n/triage accepted /priority important-soon"}
{"_id":"doc-en-kubernetes-2d8398c9fce5250ef4305561c3c57f1d1339a0f7afa5a1e82fbf81794df0228c","title":"","text":"https://k8s- It seems that when it takes a while, it tries to collect debug info. And that collection is flawed because kubectl needs a container for log collection when the pod has more than one container."}
{"_id":"doc-en-kubernetes-74ce4fea7378dc634ccb83a22893b6255fa243ed3e0b4f0e57320154ecc66caf","title":"","text":"removed this test. We need to continue testing this. This wasn't about testing upgrades, the prefix has nothing to do with upgrades. It's just there so people can share etcd. We maybe don't need to double everything but we do need to make sure that keeps working.\nTo be honest - I think that what we was testing before didn't give us completely any information. We didn't testing what information we are storing in etcd and where. That said, if we had a hardcoded prefix somewhere in our code, we wouldn't discover it at all, and those tests would still be passing. We should have a test that is directly checking etcd contents. I think it should be possible to write a simple unit test for it. Will try to put something together next week."}
{"_id":"doc-en-kubernetes-70ca5b9393197b9c681e23f75e1621b021d66b70b0941bb143c6bb4d3bdbccb3","title":"","text":"I0921 20:36:33. ] Could not connect to D-Bus system bus: %sdial unix /var/run/dbus/systembussocket: connect: no such file or directory"}
{"_id":"doc-en-kubernetes-764cede5d9efbc9aa0896bda8776a6a1986b2e142264cb7af9fd4363856aed51","title":"","text":"Is this a BUG REPORT or FEATURE REQUEST?: /kind bug What happened: In https://k8s- failed due to . Afterwards crashed as follows: What you expected to happen: Graceful exit.\n/assign"}
{"_id":"doc-en-kubernetes-b57bf0f1b2dd71c1293d1dafb95c6e51debc5c1d26878053edb8d335b76a3a11","title":"","text":"What would you like to be : Once a metric be deprecated, we should update the unit test accordingly if there is any. We need to work out a way for a unit test freely select codebase version, and also should make sure the unit test won't fail in the subsequent release. Why is this needed: Deprecated also needs to be protected. /sig instrumentation /assign"}
{"_id":"doc-en-kubernetes-8a699a775e3a74244069706b35bb820e7902d476b6f1b85f7c2dea06c82b6838","title":"","text":"appears to have been trying to improve the logging in e2e test runs, but it seems to have affected much more than that, and now when a build fails, it logs so many useless messages that you have to scroll back to see why your compile actually failed. before: after: /assign cc (re )\nGitHub didn't allow me to assign the following users: oxxenix. Note that only with read permissions, repo collaborators and people who have commented on this issue/PR can be assigned. Additionally, issues/PRs can only have 10 assignees at the same time. For more information please see <detailsIn response to : appears to have been trying to improve the logging in e2e test runs, but it seems to have affected much more than that, and now when a build fails, it logs so many useless messages that you have to scroll back to see why your compile actually failed. after: /assign cc (re ) Instructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\nThis issue is currently awaiting triage. If a SIG or subproject determines this is a relevant issue, they will accept it by applying the label and provide further guidance. The label can be by org members by writing in a comment. <detailsInstructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\nthanks for tracking this down. I agree that the additional error message is distracting and would like to see implemented differently so that it's not shown for manual invocations.\n/sig release\ndo you have any tips or recommendations how to implement it differently? Thanks\nIt looks like that extra ouput is a from . I'd suggest to either revert or strip BASHCOMMAND from the error message if it's not that useful.\nIt looks like what you wanted in is the location and time stamp (\"The update now displays both the specific location of the error in the code and a timestamp.\"), which is what we also get from the calltree after \"make build\". I would revert and then investigate why that shorter calltree is not shown after an E2E test failure. Keeping the calltree around as in Dan's \"before\" seems reasonable to me."}
{"_id":"doc-en-kubernetes-fa3ce91d012c93a689f31707fcd49f25a383e5fcb317370a874f6097508c39c9","title":"","text":"It shows a little bit more location info with the change. Here is how failure looks like with the change (but without BASH_COMMAND: and without the change:\nI mean, the call tree is pretty useless in the \"compile error\" case... if your problem is that the code doesn't compile, you don't need to be told what part of the build scripts it is that invokes . It's just that before it was a small amount of useless output that was easy to ignore, whereas now it's a lot of useless output, which completely pushes the useful output offscreen. But IMO, \"no backtrace\" would be nicer than \"short backtrace\"... (I realize there are probably other cases where the backtrace is useful, and I'm not saying we should get rid of it in those cases, whatever they are.)\nI agree, stack trace for compilation errors is quite useless. Unfortunately a lot of targets 'make test-e2e-node' included, build binaries, so we can't easily show traceback in one case and don't show in another. Removing BASH_COMMAND from the output looks like a good trade-off to me, though. It has a little bit more info(sometimes usefull) about callstack and less verbose than currently. I'm also ok with reverting and investigating later.\n/assign"}
{"_id":"doc-en-kubernetes-f24559fc73906a9ab29b78f7d758564488c283b5a15c64d8dd22134579e26576","title":"","text":"<!-- This form is for bug reports and feature requests ONLY! If you're looking for help check and the . If the matter is security related, please disclose it privately via --Is this a BUG REPORT or FEATURE REQUEST?: /sig azure What happened: The internal IP address is got from instance metadata. I suspect the instance metadata may not be stable, e.g. it has undocumented rate limits (refer ). Cross refer What you expected to happen: How to reproduce it (as minimally and precisely as possible): Anything else we need to know?: Environment: Kubernetes version (use ): Cloud provider or hardware configuration: OS (e.g. from /etc/os-release): Kernel (e.g. ): Install tools: Others:"}
{"_id":"doc-en-kubernetes-d16ca614c05197aa8b2652e3bf008c3a9a5ce41d960188a2ba3459b13d8b1102","title":"","text":"Looks like some kubelet unit tests perform a lookupIp which ends up depending on the health of the local resolver. Specifically has this failing: Which leads to the test in: And we recently swapped out the node names from testnode to 127.0.0.1, so probably the culprit?\n/ref\nInteresting. I did swap out the node names in the unit test as part of to be 127.0.0.1, party because I thought that it wouldn't have any problems resolving during unit tests.\n/cc"}
{"_id":"doc-en-kubernetes-4a4c75dd985aaeba56d487bbde17d68f0425afc5d7e1167fdf1a1abcb97cd0b6","title":"","text":"cc\nWhat is the environment where you are running the tests? They pass for me on my mac, a linux box, and shippable seems to be passing them as well.\nI couldn 't reproduce the failure either.\nI'm out right now; i'll try them when i get back. For the record, I run hack/test- Paul Morie wrote:\nAhem FTR This acronym is going to go quad-plat Paul Morie wrote:\nI ran on both platforms and saw the tests pass.\nI'm still seeing it. Won't have bandwidth to dig in personally until tomorrow.\nseeing a similar failure after a git pull\nI knew I wasn't crazy\nenvironment: fedora 21, golang: 1.4, docker: 1.6\nI think I see the problem. When I set the node name to in the test I was assuming it would parse as a valid ip address and not end up at the LookupIP call. But only the node name (and not the kubelet hostname) is being set (except in my new test where I ) so it's trying to do a lookup. Do you think you could test just changing the default hostname from \"testnode\" to \"127.0.0.1\" and see if that fixes it for you?\nTrying now\nI still get failures but no panic / threaddump.\ngist coming\nthis looks like again?\nGood that it isn't panicing any longer. I can dig into the test failures tomorrow (I think it should be relatively simple to resolve).\ncool, thanks for looking\nPasses in master now for me. Closing.\nFYI, they will break again once goes in, and then we can merge my PR to see if that fixes it for you.\n:+1:\nis in -- are these tests breaking again?\nI just rebased which I think will fix them.\nThey are breaking again.\nis on it."}
{"_id":"doc-en-kubernetes-30529ae066bc3d8742ebaf2b5d541b47e2e9d65d08ef4d2580a588e88b337906","title":"","text":"apiserver reports creation failure if a node is not ready (unhealthy, or kubelet is not ready), but the api object is already stored in master. extracted from and A short term solution is to have health_registry check node health upon creation, not just on retrieving. Or ignore unhealthy node. A long term solution might fall into the interface revamp\nWhy can't apiserver accept an unhealthy minion.\nWhat's the point of having an unhealthy minion that we'll never be able to use? One reason I can think of is that the minion can later becomes healthy and thus a valid node in k8s. But for now, as we don't have the ability to handle unhealthy node, delegate the responsibility to user makes more sense. However, I admit the title is misleading. What I really what to convey is that kubernetes currently allow registrying unhealthy node, by storing it in RESTStorage. But the create operation reports fail because the node can't pass health checking. I don't think this is the right behavior. We can either return success and do health check later, or return failure and do not store it. (i prefer the latter one for the above reason).\nI tend to think that we should accept unhealthy minions - minions may go in and out of health, or go up and down at any time - so trying ensure health at registration time seems like a fruitless effort since we have to handle the error cases anyway.\nI've no problem with accepting unhealthy minions, and I agree with your point. The real problem here is that the apiserver returns an inconsistent status. For example, if I have an unhealthy minion that I want to registry. From user's perspective, the registration fails, because apiserver returns an error (resource not found). So I retry, but apiserver returns failure again; this time, the error is resource already exists, no matter the minion is healthy or not.\n- yes, we should remove the health check as part of the registration and just report success. agree there is inconsistency.\n+1 On Nov 7, 2014 8:48 AM, \"Derek Carr\" wrote:\nSeems like it's a clients responsibility to warn the user that a node might not be available prior to it being , but the the master's responsibility to assume that all properly specified nodes are reachable before making decisions regarding using a node."}
{"_id":"doc-en-kubernetes-7568b4abf7e2f804d9652dc70d160bdc940baf27110dcb02262716f8b93b0df2","title":"","text":"There's very little difference between an admin stopping a VM without clearing Kube and adding a node to the cluster prior to it being fully reachable (I think).\n\"clients responsibility\" ? if we enable kubectl to create minions, it would be kubectl who should check if the minion is reachable and complain? I'm not sure what 'client' means. (but I agree 100%, don't report failure adding a minion just because it isn't reachable)\nnode controller plugin should also give warning"}
{"_id":"doc-en-kubernetes-dde8d0812f4d036029fc33418cfa6f164f8f025662aa48e3db09b87dd15763e4","title":"","text":"There are no sig labels on this issue. Please by: a sig: e.g., to notify the contributor experience sig, OR the label manually: e.g., to apply the label Note: Method 1 will trigger an email to the group. You can find the group list and label list . The in the method 1 has to be replaced with one of these: bugs, feature-requests, pr-reviews, test-failures, proposals\n/sig api-machinery"}
{"_id":"doc-en-kubernetes-6cfc50824280daf2f66f5e84f9a8c720c316cfa724f1e90fa02f306c798e5010","title":"","text":"<!-- This form is for bug reports and feature requests ONLY! If you're looking for help check and the . --Is this a BUG REPORT or FEATURE REQUEST?: /kind cleanup What happened: VirtualMachineScaleSetVM and VirtualMachine are different data structures (because of different API versions), so the disk attach operation seems duplicate between vm and vmss. That means although disk operations are in same logic, some duplication are still required now. They should use same api version in the future and should be merged into same code base, e.g. move sharing logic to Refer What you expected to happen: vmss and vm disk operations should share same code logic. How to reproduce it (as minimally and precisely as possible): Anything else we need to know?: Environment: Kubernetes version (use ): Cloud provider or hardware configuration: OS (e.g. from /etc/os-release): Kernel (e.g. ): Install tools: Others:\nI have checked with currently we can’t merge, e.g. following two structs are different since they using different sdk-for-azure versions. While in the end when sdk-for-azure version is upgraded, above two versions should be same and at that time, we should merge functions like ‘AttachDisk’, ‘DetachDiskByName’, 'GetDiskLun', GetNextDiskLun, DisksAreAttached"}
{"_id":"doc-en-kubernetes-95820dc44dfb57216fa4466fe0b08e35d101b3576d175fcc0793312a1c18dfad","title":"","text":"<!-- Please use this template while reporting a bug and provide as much info as possible. Not doing so may result in your bug not being addressed in a timely manner. Thanks! If the matter is security related, please disclose it privately via --We have even encountered such case that pod continuously reports “Unable to attach or mount volumes for pod;...” when kubelet restarted. The reasons are listed in details as below: , as result of apiserver instantaneous unstablility, dswp get pvc‘s pv spec failed, then the pvc volume does not be to dsw. , syncStates() lists all the volume directories locally, and recontructs the volume from local volume, if the volume isn't in dsw, it will put the recontructed volume to asw, but now the volume.OuterVolumeSpecName isn't correct. chance, at this time, In , the fail to pvc volume is to dsw again normally in reconcile() -mountAttachVolumes(), it compares dsw and asw by volume's innerVolumename, it's all the same. but it could not sense the OuterVolumeSpecName dismatch. so it will not re-mount again to update asw cache , it checks pod's volume attach or mount success by comparing volume.OuterVolumeSpecName and , but this pv'c OuterVolumeSpecName isn't correct in asw. so pod continuously reports “Unable to attach or mount volumes for pod;...” reconcile() could sense the reconstruct volume's dismatch to the dsw, then re-mount again to make volume cache consistency Kubernetes version (use ): 1.19.4 Cloud provider or hardware configuration: OS (e.g: ): Kernel (e.g. ): Install tools: Network plugin and version (if this is a network-related bug): Others:\n/sig node\n/triage accepted\n/assign\nCC\nI think this is a point that can be optimized. /cc\nI will give a fix for this\nThanks for reporting the issue, this innerSpecName and outerSpecName are confusing and cause issues. Please ping me for review when you have a fix. Thanks!\nPlease don't triage your own issues immediately, give someone else a chance to review them :( /sig storage\nokay, I know"}
{"_id":"doc-en-kubernetes-d2133e059f55329e352d657a74769e784cffa355877b91f0fede5bbe2083c4fd","title":"","text":"<!-- Please only use this template for submitting reports about flaky tests or jobs (pass or fail with no underlying change in code) in Kubernetes CI --Which jobs are flaking: [sig-scheduling] SchedulerPreemption [Serial] PodTopologySpread Preemption validates proper pods are preempted Which test(s) are flaking: Testgrid link: Reason for failure: It's likely when the preemptor preempts low-priority pods, and gets running, the low-priority pods are still terminating. tries to wait for all low-priority pods to be totally deleted and do the final check. Anything else we need to know: links to appreciated links to specific failures in spyglass appreciated\n/sig scheduling /assign"}
{"_id":"doc-en-kubernetes-ae59c93f5ca20c2f0ee388defa733f31196c86b878735630d5ec3d27a9cb2b0a","title":"","text":"<!-- Please only use this template for submitting reports about failing tests in Kubernetes CI jobs --Which jobs are failing: Which test(s) are failing: kubernetes/test/integration/scheduler.TestPreemptionRaces Since when has it been failing: Not exactly known. The flake rate is not high. That's probably why it has not received attention. Testgrid link: Reason for failure: Logs are currently unaccessible. Test grid returns an error when clicking on the failed instances. Anything else we need to know: /sig scheduling /label help-wanted\nLet me try to fix it. /assign /remove-help\nThese are the last 4 failure log in . Although The flake rate is not high, but the fail point is same. I will try to solve this error based on this information.\nThis is currently the flakiest test for the flakiest PR job for kubernetes/kubernetes\n/remove-kind failing-test /kind flake /priority important-soon /milestone v1.15\nI will try to work out a PR if is not working on it.\nSure, you can take over it if you want. And It is strange about the flake in . I ran about 3,000 times on a local machine, but never failed. :joy: I also can help you to do someting ~\nIndeed, flaky tests are usually difficult to repro and debug.. A heads-up: I seem to have found the root cause. I'm working on a solution and will raise a PR after thorough testing.\nthis is still flaking (e.g. )\nI will take another look.\nThere are two additional flakes since the one reported on May 20: It's no longer that flaky , but the PR I gave earlier proved not to be a thorough solution but mitigation. I may not be able to continue looking into it in the next couple of weeks. Anyone is in interested, feel free to take it.\nLast failure was 4 days ago. will continue observing and see if it clears out /milestone v1.16\nBeen observing this one for a bit and it seems like its getting better Other than some failures a couple days ago it has been stable. Will keep an eye for a couple more days to make sure.\nSeems to have stabilized. /close\nClosing this issue. <detailsIn response to : Seems to have stabilized."}
{"_id":"doc-en-kubernetes-e3284e7ec3b2451465766c8140397c9c53c5b3fcbbfbb2dd0841fe99eeb55acf","title":"","text":"/close Instructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details"}
{"_id":"doc-en-kubernetes-42a2dcc81f7ae4dc8763c9cd41f1b3c90a02e5ad103b4424eb3978e5cae8304c","title":"","text":"Heapster pod: And logs: cc p0 for triage\nDid the required node scope change?\nIts either the API on the console or monitoring write scopes on the node. , Mike Danese wrote:\nI think this affects only GCE\nI will fix this issue since is on vacation.\nNeed to cherry pick so i'll leave open\nis out for review.\nhas been closed and the cherrypick here () has been merged so I'm closing this issue. Thanks everyone for the debugging."}
{"_id":"doc-en-kubernetes-2598e11fb35075705c3572da278160109a38df575ea6022e29df9bb46aaee489","title":"","text":"<!-- Please only use this template for submitting reports about continuously failing tests or jobs in Kubernetes CI --ci-kubernetes-e2e-gci-gce-serial (gce-cos-master-serial) Four tests: Kubernetes e2e suite.[sig-storage] CSI Volumes [Driver: csi-hostpath] [Testpattern: Dynamic PV (default fs)] subPath should unmount if pod is gracefully deleted while kubelet is down [Disruptive][Slow][LinuxOnly] Kubernetes e2e suite.[sig-storage] In-tree Volumes [Driver: hostPath] [Testpattern: Inline-volume (default fs)] subPath should unmount if pod is gracefully deleted while kubelet is down [Disruptive][Slow][LinuxOnly] Kubernetes e2e suite.[sig-storage] In-tree Volumes [Driver: hostPathSymlink] [Testpattern: Inline-volume (default fs)] subPath should unmount if pod is gracefully deleted while kubelet is down [Disruptive][Slow][LinuxOnly] Kubernetes e2e suite.[sig-storage] CSI Volumes [Driver: csi-hostpath] [Testpattern: Dynamic PV (default fs)] subPath should unmount if pod is force deleted while kubelet is down [Disruptive][Slow][LinuxOnly] is failing since June 29th 2021 (but the first failure was different than more recent ones). Commit , infra commit , . Other three tests are failing since July 9th 2021. Commit , infra commit , . Testgrid: https://k8s- Prow history: For CSI Volumes: For In-tree Volumes: There is another open issues for failing tests in this job: Failures reported there are about missing metrics, and were occuring before unmounting problems. So I opened a new issue, even though it's the same SIG. These tests don't look related to me, but I don't have enough context, maybe they somehow are. /sig storage /kind failing-test /milestone 1.22\nYou must be a member of the GitHub team to set the milestone. If you believe you should be able to issue the /milestone command, please contact your and have them propose you as an additional delegate for this responsibility."}
{"_id":"doc-en-kubernetes-fc6065ba1a9951ec296be8e3d5eed3592b7f58a8cbf0f7d83f44e7e6b4d2f288","title":"","text":"<detailsIn response to : <!-- Please only use this template for submitting reports about continuously failing tests or jobs in Kubernetes CI --ci-kubernetes-e2e-gci-gce-serial (gce-cos-master-serial) Four tests: Testgrid: https://k8s- Prow history: For CSI Volumes: For In-tree Volumes: #### Anything else we need to know: /kind failing-test /milestone 1.22 Instructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\n/milestone v1.22\n/cc\n/cc (my close/reopen yesterday was accidental, sorry)\nI can take a look /assign\nThe non-subpath counterpart of this test is passing: These tests seem to use the default storageclass, whereas failing tests use in-tree hostpath and csi-hostpath\nThe CSI ones are failing because the hostpath driver never received a NodeUnstage (whereas it did in passing tests), causing the driver to error on volume deletion.\nIn the failed run kubelet has this error: which seems to have prevented kubelet from triggering a UnmountDevice. This error didn't exist in a successful run.\nHostpath driver can run with and without support for attach. Do you know which mode is enabled for this test?\nBetween the last passing time and when the test started failing, are there any PRs merged that look suspicious?\nAccording to csi-attacher logs, attach is disabled for both the good run and the failing run. Haven't spotted any obvious PRs so far, but I did find this: in the successful run, the log message for the successful is this: and the failed run: so the reconstructed volume names are somehow different. Checking reconstruction logic\nThis merged 12 days ago and was a pretty big change to pod lifecycle and could have trickling impact to volume manager. Does the timing of test failure match?\nThere is also this reconstruction fix that merged yesterday and seems related to outer vs inner spec name. I don't know if that will help fix this issue:\ncc\nLast successful run started at 7/8 17:49 PT, first failing run started at 7/9 2:23 PT."}
{"_id":"doc-en-kubernetes-9f1333b0e960fd7acd243a96166fc71cf5883a52e7407e5f9400e1f08c54d3cf","title":"","text":"was merged at 7/8 16:48 PT, so it's possible (if there's a delay in picking up master) Also this is in the failing kubelet logs but not in the successful one: I0720 18:20:42. ] \"Reconciler sync states: could not find volume information in desired state, clean up the mount points\" podName=-e508-45c2-8509- volumeSpecName=\"pvc--a615-4e89-afe8-\"\nFrom , the newly introduced log message here exists in the failing kubelet logs but not the successful one:\nFailures starting 07/08: /sig node for visibility\nI think caused the failures. Before after kubelet restarts, pod will still be in desired state (even through it is in terminating state), so that clean up mounts can be done completely. I think now, terminating pods will not be back to desired state, but reconstruction to actual state cannot completely clean up mounts. I am testing it about this.\nIs there a reason why reconstruction to actual state cannot clean up mounts? Note that a pod can be force deleted at any time, so a terminating pod that is force deleted would still have to be capable of being cleaned up (it won't be in desired state)? (note too that the expectation of the changes in is that terminating and fully terminated and \"never seen before\" should all completely equivalent from the perspective of other loops in the kubelet) If I understand you correctly, is this an existing bug that we're now triggering consistently?\nWe have both \"grace pod deletion\" and \"force pod deletion\" test cases that are failing.\nThere were a couple of key changes to volume loops in : Roughly before was \"it looks like it's got no running containers\", and after my change is \"we are going to be stopping all the containers from now on\". So this change meant that once a pod starts terminating, we are in \"tearing down mode\" and so my understanding of this method was that \"any pod where containers are being torn down is not in the desired state of the world\"."}
{"_id":"doc-en-kubernetes-015fa24d1e5eab79f6c5c1e5c14379be6e8185c1f089267f0bfc29a8792fef5f","title":"","text":"After a restart, we would definitely not add a pod back that was in terminating state into desired state of the world, but \"restart + this change\" should be roughly equivalent to \"restart and force delete while restart is processing\" The other change was in which changed a live CRI read (which was trying to enforce that the dsowp is synchronized with the pod worker) into a threadsafe check against the pod worker directly (the pod worker can provide authoritative answers to \"this pod has no running containers\" now). The latter change seems less likely to impact you.\nThank you for the details! I think your change is making the right change, but unfortunately some of volume management still have issues, and so the tests rely on some behaviors that is not very right (pod volume is back to desired state even after it is in terminating) The cleanup mount currently does not have the logic to clean up global mount point if the volume plugin sets it up. I am still testing more and will share more details.\nis this a release blocker?\nYes, this impacts volume cleanup if kubelet restarts for any reason.\nOn a kube-up cluster running the latest master against the latest e2e binary, I was able to reproduce the in-tree failures but not the CSI ones for some reason. Will focus on in-tree failures for the time being\n/triage accepted\nmarking release-blocker per note on slack:\n/priority critical-urgent (once upon a time we used to not ship a release if any priority/critical-urgent issues were open, not sure if we decided to stop doing that)\nWe were able to find the trigger for the CSI failures - on 6/29 there was a change in the hostpath driver test manifest, and the driver's is enabled. This makes the driver return an error instead of only logging a warning message on if the volume isn't unstaged properly. IIUC there's a long standing issue that reconstruction doesn't handle node unstage. could you confirm? If that's the case, then these CSI failures are not release blocking. Furthermore, the CSI test with graceful delete started to fail on 7/9 because of since graceful and force deletes are now considered the same from the point of view of the volume manager. cc\nWhen that check was enabled, I don't think we realized it would impact the reconstruction path; it was enabled to validate a different issue."}
{"_id":"doc-en-kubernetes-f86a5554f8df185db14256e7a7df5d445e052035a06405a03d3d05e494e0b68d","title":"","text":"Are we running into the issue that reconstruction of staging directories isn't supported () OR is it that our reconstruction path requires drivers to have real mount points ()? Either way since both are known gaps with reconstruction, I think it makes sense to to disable the flag until we can address those issues.\nI was looking at this today too, trying to understand the underlying cause. That is why the failures are being reported now, but the part that I didn't understand is why NodeUnstageVolume is not being called for these volumes. In this instance: The DeleteVolume request fails because a staged path is still present: And in the kubelet log, NodeUnpublishVolume is called Then we hit this error: And NodeUnstageVolume is never called for this volume (same for other failures I looked at) so DeleteVolume keeps failing. So I think you're right that it's triggered by reconstruction, and it looks similar to , except in this case the restart happened before NodeUnpublishVolume.\nAnd it's probably not since the test uses a PVC\nWill look deeper into why UnmountDevice wasn't issued as well, but some update on the in-tree tests: the force delete test is skipped for the in-tree hostpath CSI driver because of Graceful tests are now failing as well because they behave like force tests due to\nI don't think the issue is related to PVC or not. I think it's related to drivers that do not actually mount devices as a separate filesystem than the root.\nThere are three possible flows during kubelet restart pod gracefully delete case, the pod still exists (it is not considered as terminated in function isPodTerminated(); note that this logic has some changes after , see , so pod volume will be back to desired state, and reconciler will try to set up volume and fill in actual state. When pod volume is finally deleted from desired state, reconciler will trigger teardown with information from actual state to unpublish and unstage volume. pod is forcefully deleted, pod volume will not be in desired state. If volume reconstruction succeeds (be able to reconstruct volume spec), volume will be back into actual state. Then reconciler will trigger teardown with information from actual state to unpublish and unstage volume (this part is similar to case 1)."}
{"_id":"doc-en-kubernetes-0adb5aef0ac6855d53f98480f83c42da417641735fd51e124c8667d192575ced","title":"","text":"last case is reconstruction failed (in this csi-host-path test, reconstruction will fail due to issue related to , it fail to check mount point for csi hostpath driver because it is using a bind mount of the same device ). In this case, reconciler will try to clean up mounts directly, but only published volume path, not the staged volume path (this logic is not implemented due to missing information). As a workaround, we can now disable check-volume-lifecycle if it does not cause other issues. Then we will work on to make sure csi-host-path driver can pass volume mountpoint check during reconstruction.\nThis issue is now mitigated with merged. There are two longer-term issues that are isolated to hostpath plugins (and maybe other plugins that are similar in behavior, perhaps emptydir and similar?): related to IsLikelyNotMountPoint as mentioned. Because these don't impact the core volume machinery and are not regressions, this issue doesn't need to be release blocking. We'll close this issue once (2) is tracked.\nis the IsLikelyNotMountPoint issue\nafter is merged, two csi-hostpath driver subpath test can pass, and two in-tree driver gracefully delete tests were skipped. https://k8s-\ncan you please cherry-pick to release-1.22 branch?\nOpened cherry-pick, please review:\n- what is the plan for reenabling them?\ncherry-pick merged!\nCheng and I will work on solving IsLikelyNotMountPoint issue for hostpath drivers. That should allow test to pass and we can enable them.\nCan this be removed from the milestone now that the workarounds have been merged, while leaving this issue open to track further work?\nWe can use existing issue to track the real fix. /close\nClosing this issue. <detailsIn response to : We can use existing issue to track the real fix. Instructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details"}
{"_id":"doc-en-kubernetes-15fd873e113c2b87c10057fb05584532c0b0add34bb5504a54b7699060ebd74c","title":"","text":"<!-- Please only use this template for submitting reports about flaky tests or jobs (pass or fail with no underlying change in code) in Kubernetes CI --ci-kubernetes-e2e-gce-cos-k8sbeta-default Networking Granular Checks: Services should function for service endpoints using hostNetwork Triage looks like it's mostly flaking on this job only: Examples /sig network <!-- Please see the deflaking doc () for more guidance! --\n/milestone v1.21 (marking in the current milestone until we get confirmation whether it is relevant to the release or not)\nSlack Thread:\n/assign\nThis is fixed by , let's wait some runs to confirm before closing\n/milestone clear (since this has been deemed to be non release-blocking)\n/triage accepted\n/close\nClosing this issue. <detailsIn response to : This is fixed by , let's wait some runs to confirm before closing Instructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details"}
{"_id":"doc-en-kubernetes-125faf475ef5c9fadc127cee08d3d538bc2825eff7d3fd81be9f088e2d4a0dbf","title":"","text":"Is this a request for help? (If yes, you should use our troubleshooting guide and community support channels, see ): No What keywords did you search in Kubernetes issues before filing this one? (If you have found any duplicates, you should instead reply there.): \"LRUExpireCache\" Is this a BUG REPORT or FEATURE REQUEST? (choose one): BUG REPORT What happened: requires write lock since needs to manipulate its list to track recently used item. Currently it uses read lock so it may introduce race condition. What you expected to happen: How to reproduce it (as minimally and precisely as possible): Anything else do we need to know:\nThis was fixed in"}
{"_id":"doc-en-kubernetes-07f06081dfd955758ffddc873b3e99f4b39044981ac813b4bb5583f8c899d069","title":"","text":"/kind failing-test <!-- If this is a flake, please add: /kind flake --/sig node\nThis issue is currently awaiting triage. If a SIG or subproject determines this is a relevant issue, they will accept it by applying the label and provide further guidance. The label can be by org members by writing in a comment. <detailsInstructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\n<img width=\"1312\" alt=\"image\" src=\"\"\nThis looks like a test you a few days ago: /assign Can you look into these failures?\nGitHub didn't allow me to assign the following users: chengjoey. Note that only with read permissions, repo collaborators and people who have commented on this issue/PR can be assigned. Additionally, issues/PRs can only have 10 assignees at the same time. For more information please see <detailsIn response to : This looks like a test you a few days ago: Instructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\n/reopen\nYou can't reopen an issue/PR unless you authored it or you are a collaborator. <detailsIn response to : /reopen Instructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\nAs mentioned in the previous comments, there should be only one expected state in the end. From the logs of these failed tests, after stopping the container, the phase changes to Failed. I think this does not meet expectations.\nThanks to for revert my previous PR. I think the pod final state should always be in one state. From the current e2e test results, the final state may be Succeed or Failed, seems pod.Status.Phase is empty. I think this seems to be wrong. I will study whether the node status (sufficient memory) and other factors interfere with the final state. Thanks again to for his help."}
{"_id":"doc-en-kubernetes-6e565094763832d3a5d1ad0e1b54421c02a01afeb800549a53740a06212edac2","title":"","text":"Add the log with phase \"Succeed\": The normal update status steps are 1-2-3-4, but the result of failure tests is that step 2 is executed last, which seems a bit strange. Because phase is empty\nfeel free to reopen the PR. But let's make sure to run the tests that this reported as failing before we merge. Thank you for your help on improving test coverage!\nThe links to the failed E2E tests in the overview of this issue all show the Pod phase as , but is there a link available somewhere for cases where it's empty string?\nIn fact, these failed tests are all empty strings, because not terminated: should be followed by phase Because my original test was written like this as I commented earlier, different order of updating status may lead to different final state\nThe Pod's final state is shown as in the dumped result immediately after the test fails. The referenced inside - is that coming from the Pod information fetched within ? Could it be that it's showing up as an empty string because it's not able to access the value externally?\nIn the tests submitted previously, and in my local(minnikube) tests, the pod status will change from to again see this log step 3 So I think the order of state changes affects the final result, and this is not very stable\nI think this is a separate issue, so as I mentioned in , I thought it might be fine to wait for either the or phase. However, after looking at , it seems that we may need to resolve this unstable final result first. WDYT?\nIf we can explain different results, it is fine. I don't want unexplained check for either condition in our tests"}
{"_id":"doc-en-kubernetes-8d17c7d91073b6e9465f606fb5c09d3baa0515bf8daa4489f4a84c8a2afde410","title":"","text":"<!-- Please only use this template for submitting reports about failing tests in Kubernetes CI jobs --Which jobs are failing: N/A Which test(s) are failing: [sig-auth] [Feature:NodeAuthorizer] A node shouldn't be able to create another node Since when has it been failing: N/A Testgrid link: N/A Reason for failure: The test creates a new ` node, but it does not delete it after the test ends: kubectl get nodes NAME STATUS ROLES AGE VERSION 1954k8s000 Ready agent 24d v1.14.1 1954k8s001 Ready agent 24d v1.14.1 foo NotReady <none4m3s k8s-master--0 Ready master 24d v1.14.1 Anything else we need to know: /sig testing\nIssues go stale after 90d of inactivity. Mark the issue as fresh with . Stale issues rot after an additional 30d of inactivity and eventually close. If this issue is safe to close now please do so with . Send feedback to sig-testing, kubernetes/test-infra and/or . /lifecycle stale\n/remove-lifecycle stale\nIssues go stale after 90d of inactivity. Mark the issue as fresh with . Stale issues rot after an additional 30d of inactivity and eventually close. If this issue is safe to close now please do so with . Send feedback to sig-testing, kubernetes/test-infra and/or . /lifecycle stale\n/remove-lifecycle stale\nIssues go stale after 90d of inactivity. Mark the issue as fresh with . Stale issues rot after an additional 30d of inactivity and eventually close. If this issue is safe to close now please do so with . Send feedback to sig-testing, kubernetes/test-infra and/or . /lifecycle stale\nStale issues rot after 30d of inactivity. Mark the issue as fresh with . Rotten issues close after an additional 30d of inactivity. If this issue is safe to close now please do so with . Send feedback to sig-testing, kubernetes/test-infra and/or . /lifecycle rotten\nRotten issues close after 30d of inactivity. Reopen the issue with . Mark the issue as fresh with ."}
{"_id":"doc-en-kubernetes-c380167759ca032d9ef1dc01b836b766174a5602fdd6643d76ef74667022d120","title":"","text":"Send feedback to sig-testing, kubernetes/test-infra and/or . /close\nClosing this issue. <detailsIn response to : Rotten issues close after 30d of inactivity. Reopen the issue with . Mark the issue as fresh with . /close Instructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details"}
{"_id":"doc-en-kubernetes-799a55cc839cd2ff4d6d16e3e37eb2f6d3baf0c870165e5d680f29079df92f3e","title":"","text":"and show that we run this job every few minutes, but it never passes. I'm going to send out a PR to disable this job until someone has the time to investigate and resolve.\nIt's failing before tests even start:\nAt this point I don't think it's a flake, we needed flannel for a release and it's less important now so I'm not investigating why the cluster isn't coming up soon. I do believe if the cluster comes up a majority if not all of the tests will pass, and the ones that don't will flake for known reasons."}
{"_id":"doc-en-kubernetes-5e2660950cda0f785263613d5dae344eb16f84b548bf3bbe07f6c94efec5b686","title":"","text":"The \"service\" file in this document contains a trailing comma on the line ' \"name\": \"my-service\", ` and thus kubectl fails to parse it properly."}
{"_id":"doc-en-kubernetes-ffd505027484e14846d7eca4dd46da75100eeb1e560f155a9f5f3a05b8321559","title":"","text":"<!-- Please use this template while reporting a bug and provide as much info as possible. Not doing so may result in your bug not being addressed in a timely manner. Thanks! If the matter is security related, please disclose it privately via --This is probably just a documentation / example issue. I'm a real noob so I can just tell this is an issue, but I can't fix the docs through a PR, otherwise I'd do it happily. Hope it's not a dupe. Following the guide at with IPVS. Using the following script: This generates the following YAML (only the interesting part): You see that the line has an extra comma. It's instead of (extra comma after second argument) This generates the following log: Removing the extra comma fixes everything. Setup a cluster with Kube-DNS IPVS mode Generate the nodelocaldns Deploy the manifest Pods are crashing Kubernetes version v1.21.0 Cloud provider or hardware configuration: Bare Metal with HA setup (behind an external LB, Kube-proxy as IPVS) OS (e.g: ): CentOS Linux 8 Kernel (e.g. ): Linux 4.18.0-240.10.1.el83.x8664 SMP Mon Jan 18 17:05:51 UTC 2021 x8664 x8664 x86_64 GNU/LinuxInstall tools: none, a classic kubeadm initNetwork plugin and version (if this is a network-related bug): Not relevant, but Canal (Calico + Flannel) Others: nothing. Love you all for making K8S. Thank you a billion times.\nThis issue is currently awaiting triage. If a SIG or subproject determines this is a relevant issue, they will accept it by applying the label and provide further guidance. The label can be by org members by writing in a comment. <detailsInstructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\n/sig docs\n, it seems like that don't get a value return. i saw command in Can you try this again and check if can kubedns get a value?\n/area dns /sig network /assign Documentation here is not changed since Sep 2019 and needs to be updated as well."}
{"_id":"doc-en-kubernetes-d83335fcac537a45eafa3dfd42630478844385174c0958db0174f56a0ffa2f73","title":"","text":"Dual-stack and ipv6 are used more widely now.\nI do get a value, but is replaced with as the extra comma doesn't seem to be replaced.\n/assign"}
{"_id":"doc-en-kubernetes-57d83ef880aa3c3668480a2a99a63d94dfed9404bcfb6c9db559f5ee8b331861","title":"","text":"More detail can be seen in e2e test\ncc\n, Marek Grabowski wrote:\nyou are running an older version of the test. can you get the latest and try again ? This test had not failed at all yesterday in our automated test runs yesterday. Closing issue for now.\nThis failure depends on if the pod is really deleted and kube-proxy (the test container on) catch that the endpoint is deleted in 5s . After sending \"delete pod\" request for 5s, if the pod was still there and kube-proxy still contains this endpoints(this may happen under pressure), hit test can still detect 5 endpoints. At least we should ensure the pod was deleted. Anyway, I think will be helpful to this test. What do you think?\nIt doesn't matter if the pod is actually deleted or not. We just care that kube-proxy thinks that it should not be part of the service endpoint. Having said that, it still doesn't hurt to do the check that you are doing. I will assign your PR to myself and review it. In the meantime, can you rebase your PR ?\nThanks for your review. The PR is up to date.\nFailed twice today in kubernetes-e2e-gce-parallel\nOne more today. The error message is not very helpful.\n18:28:37 STEP: dialing(http) endpointPodIP:endpointHttpPort from node1 18:28:37 STEP: Dialing from node. command:for i in ' wc -l This is the point where the test fails. Am looking into it. We need to have more tries, we just do seq (1 1), instead of seq (6 6). this is causing flakiness.\nFixed by PR"}
{"_id":"doc-en-kubernetes-a9637d9c1e2cf914512c87528401900f91ad84dba7ec541c905d0670e6dfc824","title":"","text":"Blocking merges, P0\nLooks like this e2e code hasn't changed since .\nUgh. What changed is that we moved it off teh flaky list. Perhaps wrongly so. , Stephen Milner wrote:\nAh, I see. removed a number of e2e service items from variables. Relevant here:\nthe test is actually still running \" Services should serve identically named services in different namespaces on different load-balancers\" is running green.. Saad's fix where we correctly wait for GCE operations to complete, is checked in. Closing this issue.\nLooks like this test is failing forever in and\nSomehow I lost this one. Will look at it ASAP , Janet Kuo wrote:\nThe problem is that this test depends on the api proxy which doesn't do UDP, as far as I can tell, so it could never have worked when UDP support was . I will see what I can do to rectify.\nFixed test is - it's a doozy.\nI've asked oncall to move this to the head of the queue. It's hard to figure out existing flakes when the test driving them has been totally changed. I want to re-examine all flakes after tghis\nis in. I am going to close a few issues and wait for reoccurence, which is to be handled P(-1)"}
{"_id":"doc-en-kubernetes-17394de51a0297c3b5009d62fc651309bc0945a0e851cb08a887385f3706124d","title":"","text":"<!-- Please only use this template for submitting reports about flaky tests or jobs (pass or fail with no underlying change in code) in Kubernetes CI --Seems to be impacting a few KIND jobs on the master branch the most, but also assorted other jobs see the triage link above Reported by Sample failure: /sig node <!-- Please see the deflaking doc () for more guidance! --\nxref probing tests\n/assign ref /triage accepted\nThanks Elana, still trying today... Does anyone know if we can use the kubelet logs in e2e tests?\nNot very portably in cluster e2e tests at least. Maybe in node_e2e instead.\nHave you considered that the issue is not with the test but with the underlying code in kubelet like detailed in this issue Like detailed in the issue I can reliably reproduce (on k8s 1.18) the startup probe passing within a few seconds, but the container is only marked as ready after about a minute.\nthere is already a fix in the 1.21 release for that (part of a bigger PR): And actually this test was to verify that part of the code, but unfortunately sometimes it doesn't respond as quick as it should (but never on my PC, only in ci/cd tests)...\nSome recent flakes:\none of the most common flakes in kind jobs currently failures like: Mar 25 17:03:42.508: Pod became ready in 42.007286524s, more than 5s after startupProbe succeeded. It means that the delay readiness probes were not initiated immediately after startup finished.\nMaybe we could speed up the review on\nThank you I will take a look. I'm making a pass through flakes tonight trying to identify if anything is on kind's end\nI don't think it's kind... since running on my PC with kind it works 100%, I think the issue is either the resources... or contention in the queue as mentioned to in the comment\n/milestone v1.22\ncommented on PR with the suggested fix\n/reopen\nReopened this issue. <detailsIn response to : /reopen Instructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository."}
{"_id":"doc-en-kubernetes-e379199dcc10ef545e6719d9fd62c8045ba08caeee0b4e30e1a8cba90abc8ed8","title":"","text":"</details\nYeah...\nThanks for your great work in keeping things in check\nFYI\nI have (finally) understood what's happening, and I can even explain why the deflake made things worse... Please have a look at the sleep before the worker starts to process probes: So, you have to remember that we have one worker per pod and per probe, and since we have these probe settings for the test: We have (PeriodSeconds=10) starting sometimes immediately (if the Kubelet has started since more than 10s) or after a quick sleep of rand10s. On the contrary, will likely sleep most of the time for rand60s. This causes a delay in the processing of notifications: If we are lucky, this delay is less than 5s and the test passes: Now you can understand why the deflaked settings were actually worse:\nThe reason I didn't spot these was that I usually start my cluster once for all and iterate on the tests... so the kubelet had plenty of time to start and I never ended in that sleep.\nFor the solution, I suggest we either remove the sleep completely, or we find something that would make all 3 probe workers start at the same time to avoid sync issues... For the record, I was able to spot this issue thanks to the following log error (because the channel was overflowing): What do you think ?\nI have found another issue in that code: If has already ran (and failed), we go to L245 and don't probe via and therefore wait for the next tick, making the test fail.\nThis seems consistent with what I found when I was adding e2es for the new probe-level termination grace period and why I did my timing math the way I did.\n/reopen I don't think this is fixed, I just hit it again and gubernator is still showing plenty of flakes:\nReopened this issue. <detailsIn response to : /reopen Instructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\nhow to check the exact commit used for each test?"}
{"_id":"doc-en-kubernetes-1225e4b3fb518feb838b0b4898217ee9b18190e72a1e6684a1b960954bc011fd","title":"","text":"For instance, I am not sure (from the name) that should include my latest fix...\nIn your PR the fix wasn't there:\nyou may send cherry-pick to 1.21 for which is using kubelet 1.21, not 1.22.\nis using kubelet. Cherry-pick to 1.20 may be needed as well.\nI have found this one that failed even with the fix: corresponding to:\nI don't know what could explain this atm... but it seems we still have delay propagating container statuses.\nIn the job artifacts there are multiple places with this info including the clone records/logs, and In testgrid it's a row.\nyeah afaict since the tests ran on my PR after your fix was merged into master, it showed the flake was still present :( Prow will always use the latest master when running CI.\nSorry for doubting what you said... I was probably in deny and hoped this annoyance was fixed once for all :(\nPlan: we'll see if Clayton's pod lifecycle refactor and/or the jitter refactor for probes affects this.\nClayton's PR jitter refactor\nsince isn't (likely) coming to 1.22 we could cut off 90% of flakes by increasing to 15s delay:\nit seems it fails even after merging so let's continue our discussion on slack\nthis is not a regression, our suggestion is not block release on it. feel free to remove the milestone\n/milestone v1.23\nIn Kubelet maybe, but in the flakiness of the e2e tests it was. Unfortunately we can't easily explore the current state :/\nIn the release blocking jobs the frequency is not horrible now, so unblocking is probably fine, it is not as bad as it was. Outside of the release cycle this is still the flakiest test in some jobs, see e.g. where it's the only flake source we have on record currently.\nif we end up needing it, we have a timing/period adjustment PR we can take that should help with the flakiness with"}
{"_id":"doc-en-kubernetes-8eb466904f09832d9690a82d180fa467ae255966eba3f5f112ba1d2e4e87f745","title":"","text":"Looks like similar issue with\nThat's the PV controller this is the attach controller, thus separate issues."}
{"_id":"doc-en-kubernetes-6b435dc960bb767c2bbe1807b1f897013b6c97adbc3485d977e3c4a83f87c11f","title":"","text":"Is this a request for help? (If yes, you should use our troubleshooting guide and community support channels, see ): No. Is this a BUG REPORT or FEATURE REQUEST? (choose one): FEATURE REQUEST in we have that perform unnecessary shallow copy on pods () : we get the pods from the cache i think we do not mutate the pods (i.e only reading values) so we can avoid this deepCopy i think.\n/assign\n/remove-kind enhancement /kind technical-debt\nThat's actually a shallow copy\n- I'm sorry i was thinking about another issue while filing this one.. I will update the issue, Thanks.\n/sig network"}
{"_id":"doc-en-kubernetes-6e9547d1b77f5f5e67275f7358d2b648055755d378e63a99679bc29fa4f8a944","title":"","text":"1 tests failed, 11 jobs failed, 241 builds failed. Failure stats cover 1 day time range '17 Aug 2017 22:57 UTC' to '18 Aug 2017 22:57 UTC'.\nThere are no sig labels on this issue. Please by: a sig: e.g., to notify the contributor experience sig, OR the label manually: e.g., to apply the label Note: Method 1 will trigger an email to the group. You can find the group list and label list . The in the method 1 has to be replaced with one of these: bugs, feature-requests, pr-reviews, test-failures, proposals\nSeeing this failure on the e2e release-master-blocking https://k8s- Example failure https://k8s- Looks like an RBAC issue cc\nthis is currently blocking the alpha.3 release\nlooks like this new test was by in which was merged about 6 days ago and is failing consistently\nThis has started failing again on our GKE test suite https://k8s- https://k8s- cc\n[MILESTONENOTIFIER] Milestone Labels Complete Issue label settings: sig/api-machinery: Issue will be escalated to these SIGs if needed. priority/critical-urgent: Never automatically move out of a release milestone; continually escalate to contributor and SIG through all available channels. kind/bug: Fixes a bug discovered during the current release. <detailsAdditional instructions available <a href=\"\"here</aThe commands available for adding these labels are documented <a href=\"\"here</a</details\n/assign\nSeems to be flaking now with the error: https://k8s-\n- we need to determine (today if possible) if this is truly release blocking. If so, please add the release-blocker label. And, if not, how do we best continue work on this for 1.8.x/1.9.0?\nthis test has two passes and four failures on the same commit I'm seeing gke-specific authz grants in that test that are incorrect: That commit removed the wait for the authz grant to become effective (which can lead to flakes), and granted superuser permissions to all users, which is incorrect and invalidates any other authz-related tests run in parallel with this test\ncc\nI can add the wait back."}
{"_id":"doc-en-kubernetes-44c35af259d5583c472df6fa7f370aa48aa2c399176d09a89c981e3feb3444e5","title":"","text":"edit: and fix the test given admin to all authenticated users.\nActually after staring at this test for about half an hour I can't figure out what different users exist or what permissions they're being granted. ClientSet, InternalClientset, and AggregatorClient are all initialized from the same config so I don't see how one would be able to create an RBAC binding but another would fail later. any thoughts here?\nI honestly think the gke specific BindClusterRole is a red-herring. It is needed so the client has permission to perform one of the setup steps. (I think it was either to create the wardler cluster role or to bind that role to the anonymous user) Once that setup step is complete we no longer need the that cluster role bound and so I don't think its related.\nThe gke authorizer allows the “bind” verb, so the client can create a binding to the cluster-admin. It cannot create a role directly unless it has permissions via RBAC. Since we don’t have a way to determine the username associated with , binding to all authenticated users is what was done as a workaround.\nI agree that the point at which the tests are failing indicate that the previous authorization issues are not the cause.\n/open\n/reopen\nSo a lot more information to work with now but the error is still occurring. I am still looking into this.\nany update on the investigation?\nFriendly v1.8 release team ping. This failure still seems to be happening, is this actively being worked? Does this need to be in the v1.8 milestone?"}
{"_id":"doc-en-kubernetes-6be393fdfa676676e2c2a919450eea6e64e7299e61f3fdbea996cf6d1af38cd8","title":"","text":"We've seen this a few times on GCE now, and I saw it flake on now as well (cc\ncc Victor integrated cadvisor into the kubelet. I wonder if a fake version of cadvisor gets integrated into kubelet in the case of builds.\nI noticed this first on GCE. (FWIW, GKE uses the same binaries as GCE.)\nJenkins shows the following four changes for the GCE build where this test started consistently failing: Commit by mattmoor Enable usage of a \"json key\" for authenticating with Commit by dbsmith Add a system modeler to scheduler Commit by dawnchen Convert RestartPolicy to string for v1beta3. Commit by abshah Build statically linked binaryies. With the change to go 1.4, we probably were generating dynamically linked binaries accidentally.\nIf I had to guess, I would take a stab and say it was , . cc\nFeel free to revert by 5pm if no fix.\nI believe and are correct, makes the build not use cgo which places a fake cAdvisor in the binary and that fails. Looking into the fix.\nis out, it is a revert of the old one along with a plan for not needing static binaries."}
{"_id":"doc-en-kubernetes-cf7ef9edd7e31f4b8286dd6738f8e55df1658c21b56557082b0f1b4070aed850","title":"","text":"First, the formatting. Second, \"init-ctr-demo_demos\" is an internal artifact, not fit for human-facing errors. Third, the container DOES exist, it's just in a terminating state.\nThe formatting issue should be fixed by . Regarding to the pod name, I think is better. As for the pod name, the error message is from the HTTP response of kubelet server. I've looked at the code, functions defined in interface are using ($podName_podNamespace) to identify a pod. Without spiting the parameter into two separated parameters, it could be tricky to change the output format of an error message. Personally I would prefer to change the signatures, but I don't know if we are planing to do it. In order to improve the readability, I a client-side check for pod status. See ."}
{"_id":"doc-en-kubernetes-d1e994e5749edcd07edbd0f19c7df9185dabd2833b25fd1731e4d47ccf8dce07","title":"","text":"<!-- Please only use this template for submitting reports about flaky tests or jobs (pass or fail with no underlying change in code) in Kubernetes CI --Which jobs are flaking: pull-kubernetes-e2e-gce Which test(s) are flaking: Discovery should validate PreferredVersion for each APIGroup Testgrid link: Reason for failure: The test does not handle API groups which are or removed dynamically during the run of the test. Just because an API group is present in at the beginning of the test does not mean or will be present by the end of the test. This is particularly true for the dynamic/custom-resource API groups registered/unregistered during the e2e test run /sig api-machinery /assign\nmitigation opened at\nSaw this today in It seems to happen relatively often Likely due to the CRD created in: Should the ValidatingAdmissionPolicy test CRD have its group name template changed to match the existing test exclusion for groups ending in /cc"}
{"_id":"doc-en-kubernetes-04298a57bb6bbadab2ae389c253044418da070e519876e8a987add6c4a3d1714","title":"","text":"[x] Original approval issue: Start 9 June 2020 3 Flakes No Flakes for 2 weeks According to this APIsnoop query, there are still some remaining ReplicationController endpoints which are untested. a ReplicationController with a static label the ReplicationController with a new Label and updated data the ReplicationController's status /Scale the ReplicationController's Replica count to 2 Replicas the ReplicationController the ReplicationController's Status all ReplicationControllers in all Namespaces find the ReplicationControllers(1) ensure that the ReplicationController is found and is patched Namespaced ReplicationControllers(1) via a Collection with a LabelSelector Discover useragents: See all endpoints hit from this test Display test coverage change If a test with these calls gets merged, test coverage will go up by 7 points This test is also created with the goal of conformance promotion.\n/sig testing /sig architecture /area conformance\nCan you please add in PR\n/reopen Merged. Promotion to be created\nReopened this issue. <detailsIn response to : /reopen Instructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\nIssues go stale after 90d of inactivity. Mark the issue as fresh with . Stale issues rot after an additional 30d of inactivity and eventually close. If this issue is safe to close now please do so with . Send feedback to sig-testing, kubernetes/test-infra and/or . /lifecycle stale\n/remove-lifecycle stale"}
{"_id":"doc-en-kubernetes-dc3be6ce7b42d453edf0753a2cf7546d061a63c2e475384cc9c94a48a05e7f4a","title":"","text":"Is this a request for help? No What keywords did you search in Kubernetes issues before filing this one? http proxy, httpproxy, httpsproxy, AWS, kubeadm, kubectl, cloud-provider Is this a BUG REPORT or FEATURE REQUEST? (choose one): BUG REPORT Kubernetes version (use ): v1.4.4 Environment: Cloud provider or hardware configuration: kubeadm on EC2 instances on AWS OS (e.g. from /etc/os-release): Ubuntu 16.04.1 LTS Kernel (e.g. ): Linux ip-10-0-133-99 4.4.0-45-generic -Ubuntu SMP Wed Oct 19 14:12:37 UTC 2016 x8664 x8664 x8664 GNU/Linux Install tools: kubeadm Others: httpproxy, httpsproxy and noproxy set in /etc/environment, Docker config as well on startup scripts that run What happened: won't complete and gets stuck on it's first deployment. When one looks through the logs it shows that the pods are downloading fine but we are stuck due to the kube-controller-manager continually crashing as it can't contact AWS. Elsewhere in kubernetes I see references to and if the AWS cloud provider could use that it would be able to connect as well. From the POD if I manually set httpproxy and httpsproxy and no_proxy I am able to use the AWS CLI without any issues. What you expected to happen: I would expect to successfully complete and use the environment proxies set up on the system so as not to depend on a default Internet Gateway being connected to the VPC just for AWS access. How to reproduce it (as minimally and precisely as possible): Set up a public VPC with Squid set up as a proxy Set up a private VPC peered to the public VPC Set up a master node and run Set up a regular node and run will never fully initialize. If you remove the --cloud-provider aws` then it will complete and nodes will register. Anything else do we need to know:\nwhat do you have in proxy variables ? (env | grep -i proxy) it tries to connect to (tcp 176.32.118.53:443) directly. You're expecting to connect there via proxy ? then do you have something like in your no_proxy ?"}
{"_id":"doc-en-kubernetes-984a2b369276e3156e0990106838472387b6975b0768d694bc54d8f6e19402d4","title":"","text":"Unfortunately on the pod itself for the the environment variables aren't passed through. As this is created using I don't have control over the pod definition. So basically none of the *_proxy variables are set inside the pod in the docker image, however if I and export the variables and try to wget that URL it works. Basically I'm hoping we can update to respect and pass the environment variables down to the system pods if they exist, eg, add: Without it I can't stand up a cluster with AWS integration on a completely private VPC and for such a small change it would make a rather big difference.\nok, so request is actually to pass proxy variables into containers. interesting. I'll check how it can be done. BTW, for noproxy value in your example: things like 10.0.0.0/8 would not work on 99% of software that respects noproxy variable. it must be always exact IP match.\nyou're right that what I'm actually asking for is to pass proxy variables through either from a commandline setting on kubeadm or pick it up from /etc/environment or shell environment that launches kubeadm. I didn't realize it till I dug in to try and think about how one might go about fixing this. At first I thought we would just pick up the variables from the system environment but then I realized kube-controller-manager is deployed in a container and so we'd need to update the yaml file that launches it to contain the environment variables. As I'd like to go to production with this soon hopefully this is a small change that can be .\nThis issue was moved to kubernetes/kubeadm"}
{"_id":"doc-en-kubernetes-b8a87f76b1813a17eca6bab9ea27ea64d02dfa7ec03a7ba6217cccf0eba27cf6","title":"","text":"What would you like to be : Deprecate the flag which enables exposing logs of apiserver. Why is this needed: As per , in order to disable accessing apiserver logs in the future, we need to deprecate the current default and wait for the requisite time period.\n/sig api-machinery\n/assign\nI'm in favor of deprecating this. It's potentially disruptive, I think an email blast to kubernetes-dev would be a good idea.\nThanks , I have posted in kubernetes-dev mailing list.\nI'm also in favor of removing this functionality from the apiserver. Since this is GA behavior, the deprecation period would be 12 months.\ncc"}
{"_id":"doc-en-kubernetes-a8f749d2105324b41eb33d6f0298d70067479a2b892424afb06eaf9d08e1bf7a","title":"","text":"<!-- This form is for bug reports and feature requests ONLY! If you're looking for help check and the . --Today there are many obstacles preventing users from setting up FlexVolume on a GCE COS image easily. The default FlexVolume driver directory is not writable. Custom mount utilities can't be installed. Solutions exist for these problems but they are time-consuming to execute manually. On top of these, there are also the common FlexVolume pain points such as manual driver and mount utilities installation, and manually restarting kubelet. Ideally all these tasks should be automated, making it a quick and simple experience to have custom out-of-tree volume plugins. Is this a BUG REPORT or FEATURE REQUEST?: FEATURE REQUEST /kind feature\nThere are no sig labels on this issue. Please by:<br(1) mentioning a sig: <bre.g., for API Machinery<br(2) specifying the label manually: <bre.g., for sig/scalability<br<brNote: method (1) will trigger a notification to the team. You can find the team list and label list\n/sig storage\n/assign"}
{"_id":"doc-en-kubernetes-5b657ca63c8d65ac5d765363aee55755b3df3fe0d7e8c3dbff40d3d4f941f728","title":"","text":"The path to the server tarball resolves to: is hardcoded, and does not exist in release builds (v1.3.0-beta.2) The file in question is located at: Correcting the path is a workaround for release builds\ncc\nHey, sorry about that. I think I got a fix, but I need your help to test it. I'm traveling and my cellular upload speeds are not so great. Please take a look at this and let me know if it solves the issue:"}
{"_id":"doc-en-kubernetes-c5ab0587303441aa15d1f99cf54443dd34979e6a47a9667d6d352ed63897253b","title":"","text":"What happened: The stackdriver logging agent is installed and start running no matter logging is enabled/disabled in the cluster or not. What you expected to happen: Do not install and start the logging agent if logging is disabled.\n/sig windows\ncc"}
{"_id":"doc-en-kubernetes-848285cc888dabd416033408eedbcc3ccb37536969c7e28abf4d6f0620c17a3f","title":"","text":"Is this a BUG REPORT or FEATURE REQUEST?: /kind bug What happened: When using IPVS and not natting (no cluster-cidr specified) requests from the host to service IPs are send with the service ip as the source ip. Let be the kubernetes API IP. Kubeproxy will set up with . This means that traffic from the host (which sees the ) will originate from (best matching ip/network). What you expected to happen: The source ip should be natted to the host ip. How to reproduce it (as minimally and precisely as possible): a ipvs based cluster any cluster-cidr from kube-proxy traffic to a service ip (e.g. http to the kubernetes api server) from the node / host network pod the outgoing interface looking for the service ip () Anything else we need to know?: Environment: Kubernetes version (use ):Cloud provider or hardware configuration: baremetal OS (e.g. from /etc/os-release): Kernel (e.g. ): (based on the Ubuntu 18.04 kernel) Install tools: custom repackaging of kubernetes release tarballs Others: /sig network\nWe have a similar issue in kube-router FWIW resolution is included in the issue\nAh nice, I reproduced the source IP selection on my local Ubuntu 18.04 without any kubernetes installed (use ip link add / ip addr add and then ping the new dummy IP / tcpdump lo). Might be related to the kernel version (4.15) as you couldn't reproduce it.... Changing the routing table would avoid an extra NAT rule. That sounds interesting but not doable without changing the kubernetes sources Murali Reddy < am Fr., 27. Apr. 2018, 14:06:\nlet me try with to reproduce. but its no brainer, this issue must be fixed.\ngetting closer... -CLUSTER-IP is only populated if masquerade-all or cluster-cidr is used: is no OUTPUT masquerading for service IPs if cluster-cidr is not specified (no branch matches): With cluster-cidr enabled I've saved the ipset.... I've disabled cluster-cidr and rebooted (to reset all iptables/ipset/...)."}
{"_id":"doc-en-kubernetes-0a7db03e41de00a95766c0a4c19610d15de5901707836c1c2d0b18155d4ac469","title":"","text":"I think the code should: Always populate Set a masquerade rule for service IPs (src == service IP should not happen on OUTPUT)\nHow to test this with a local cluster.... the node (install , & modprobe the modules , , , ). a new shell: define a service with manual external endpoints tcpdump on the outgoing interface of the node the service ip the tcpdump, it will show outgoing traffic from_ the service IP:\nCan you please take a look at this?\n/assign\nI assume this issue is related to some specific kernel versions? I just neither can't reproduce it nor can't see the failures in our test grid.\nI was aware of there is a PR() in flight, my team will evaluate the effects of that PR. But, please rebase the PR first, thanks!"}
{"_id":"doc-en-kubernetes-086f2ac7dcab9c158ca481b0b58fae22976bc98ae146a7d10d5bef748052b70d","title":"","text":"Is this a BUG REPORT or FEATURE REQUEST?: /kind bug What happened: e2e tests which use RBD server image fail with What you expected to happen: These tests succeed. How to reproduce it (as minimally and precisely as possible): Run e2e tests which use RBD server image. Anything else we need to know?: /sig storage This issue is introduced by below commit: It change the base image from fedora 26 to fedora 28. In fedora 26, hostname command is installed as a dependency for  However, in fedora 28, it is not installed, as a result, this issue happens due to lack of hostname command. Environment: Kubernetes version (use ): k8s 1.12 alpha\nPlease see the difference of the behaviour between fedora 26 and 28. [Fedora26] [Fedora28]\nCould you help solving this issue? I assume the fix for the code would be as below, but I'm not familiar with the build and push process for the test image. ( in doesn't work.) So, I need your help to make further progress.\n(or) Both need to be run from directory\nThank you for letting me know it. It should be s/all-containers/all-container/, but it seems to work. I will test with above patch and above build commands, then make a PR, once I confirm that it works well.\nsorry typing from memory :)\nI need to ask you two more things: Do we need to increment the version for this image, like 1.0.1, even for this small change? (I guess that and test/images/volume/rbd/VERSION will need to be updated in this case.) How the image push process will be? Does it pushed by CI, if there are a change in the code or docker file, or do someone who have permission will push it manually?\ncreates below three docker images: Is my understanding right that to make each test access to each arch image, needs to be modified like below? (Without doing so, each test seems to keep trying to access to , without arch.)\nFor fedora28 image, just installing hostname doesn't seem to work well. Please see the log below and attached . I guess that this is due to the difference of ceph version, which will be beyond my beyond my expertise."}
{"_id":"doc-en-kubernetes-77a7b6220df65aff10095b84f9a73f01dc0b39abd15f0db81cc810c0f75ed2d0","title":"","text":"So, could you help debugging this issue? Meanwhile, I will try testing with the image by using fedora26 for workaround.\nI confirmed that below commit, which change the base image back to fedora 26, fix this issue. Please see the log . After I checked the image-, I come to think that by pushing the image, each image will be able to be pulled without -arch, so I guess the last parameter for ImageConfig should be false. (I couldn't test it in my environment.) I will fix the commit as so and make a PR once I confirm it. Also, if it will easy to fix with Fedora 28, I will wait for the actual fix and quit making a PR for the workaround.\nthanks yes, a quick fix is welcome, looking forward\nThank you for your comment. I will continue with the workaround. I saw a commit for deleting multi arch consideration. I rebased the patch to the commit and made a PR. Please review it."}
{"_id":"doc-en-kubernetes-98264fae63e9d149f0171dd13218170b598f5e330429a04234e4c80f3d1b23d2","title":"","text":"<!-- This form is for bug reports and feature requests ONLY! If you're looking for help check and the . --Is this a BUG REPORT or FEATURE REQUEST?: What happened: VMAS nodes are reporting error in vmss cluster (vmType set to vmss). What you expected to happen: Node should register correctly in the kubernetes cluster. How to reproduce it (as minimally and precisely as possible): Create a VMAS nodes inside vmss cluster. Anything else we need to know?: Environment: Kubernetes version (use ): master Cloud provider or hardware configuration: azure OS (e.g. from /etc/os-release): Kernel (e.g. ): Install tools: Others:\n/sig azure"}
{"_id":"doc-en-kubernetes-142e8decd8d12f564814b0a08026ec0c2500ddcd8040f0182edc6a7f51b222f8","title":"","text":"0/8 is with a company's intranet, so it can be insecure. This is similiar to \"For GCE, allow insecure registries anywhere in 10.0.0.0/8\" Thanks!\nLinking to and .\nThis was"}
{"_id":"doc-en-kubernetes-82cb26acad9fc9df916ca73a6711809a93da9752a7947ed537b4183fa019f2e5","title":"","text":"What would you like to be : There is a tab in the k8s perf dashaboard for the scheduler perf benchmarks, but it is showing nothing, we should fix that: http://perf- What this dashboard should be showing is multiple runs per day of We can run it every hour, and summarize the 24 runs of each day to show 50%, 90% and 99% percentile. Why is this needed: Having daily runs of the benchmarks is important to track performance regressions. /sig scheduling /assign /priority important-soon\nRef\n/cc\nThe ci job's config is at And one of failed logs could be found at I tried to reproduce it on my local machine(because I do not have Gloud credentials, I run the task directly in the container): However it works well on my local machine. It is so weird. It will be very appreciated if anyone could help reproduce it with Gloud credentials.\ncan you try and reach out to the test-infra people? ping them on the slack channel perhaps.\nyes, I have tried to get help from , he is busy now. I'll try more.\nWhile working on reproducing this in order to file a bug against test-infra folks I think I found a solution. Will send PRs shortly. /assign\n/retitle sig-scalability-benchmarks#scheduler fails consistently due to build issues\n/sig scalability Testgrid link: https://k8s- Error logs:\nA word about the solution, the job uses the deprecated script which is not supported anymore by test-infra oncall / folks. I've rewritten the job definition to use pod utilities and it started working (will require to get merged first).\nThanks\nThanks !"}
{"_id":"doc-en-kubernetes-3d30f80339d7c67274e12c6190891fb695bbdad2c822e929f8e060cdf1647787","title":"","text":"/kind cleanup /sig testing /priority important-longterm This is a followup issue to address the \"standard build globals\" nit on At the very least I think we shouldn't capture our custom env var. Also think it'd be worth documenting the assumed variables in a list, similar to\ncc\n/assign"}
{"_id":"doc-en-kubernetes-295b16ebdca2097490863d761fa466b150fe770891272091eba588ee367ef882","title":"","text":"<!-- This form is for bug reports and feature requests ONLY! If you're looking for help check and the . --Is this a BUG REPORT or FEATURE REQUEST?: /kind bug What happened: I have tried FlexVolume on a Windows node, it does not work. I am now working on this feature now. Below is the FlexVolume initialization error What you expected to happen: How to reproduce it (as minimally and precisely as possible): Anything else we need to know?: Environment: Kubernetes version (use ): v1.7.7 Cloud provider or hardware configuration: OS (e.g. from /etc/os-release): Kernel (e.g. ): Install tools: Others: /sig windows /assign\nDetailed steps about how to use flexvolume on Windows with this PR:\ncc:"}
{"_id":"doc-en-kubernetes-9e98aee5fef5afc48ea03df9966a56409c15767acad0c30722b60251c7b487a9","title":"","text":"With the move to running the kube-apiserver in a pod, I am getting errors of the form: The root null pointer is from this line: The line numbers in the stack are from my continuing debug of the original null pointer. We need to protect against a null pointer in this path, which appears to only be invoked when doing a list call that matches a single item based on field selector. Will submit a PR to address.\nWith proper line numbers:"}
{"_id":"doc-en-kubernetes-f554e53a2b0895170400ceeeb3f9d1f2002f8788d2419fe2262e5a5418c69b9b","title":"","text":"I am tracking a changed behavior in networking with kubernetes 1.26. If I create a service of type externalTrafficPolicy: Local, and there are no local endpoints in a node, kube-proxy does not create hns loadbalancers for the clusterIP of that service. With kubernetes 1.25, it creates a loadbalancer for clusterIP (but not externalIP). This is the relevant log message: Looks like this comes with a result from a new function isAllEndpointsTerminating introduced in kubernetes 1.26. In Linux, we can still access clusterIP even if there are no local endpoints for the service. Thoughts? The end result is that in Windows, clients cannot access the clusterIP of such a service with there are no local endpoints.\nThis issue is currently awaiting triage. If a SIG or subproject determines this is a relevant issue, they will accept it by applying the label and provide further guidance. The label can be by org members by writing in a comment. <detailsInstructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\n/sig windows\ncc\nDoes this issue occur when there are no local endpoints for a given service? If there is a local endpoint, then you can reach the service again, right?\n, Also let us know, do you have any InternalTrafficPolicy configured ?\nDoes this issue occur when there are no local endpoints for a given service? If there is a local endpoint, then you can reach the service again, right? That is correct , Also let us know, do you have any InternalTrafficPolicy configured ? We do not have that configured"}
{"_id":"doc-en-kubernetes-c9a7a8de199b532837929cf19816bb31dfb902dab306b212a976e292dbb5b33e","title":"","text":"download- refers to release 1.1.4 but the latest is 1.1.7\nResolved by PR"}
{"_id":"doc-en-kubernetes-1f9c3e5cc71a2eb982031bdc2376fc7ce9ffe73bb0100bc8df9e6d81af2040c4","title":"","text":"Looks like CSI volume limit test is flaking. /sig storage /assign\nThis flake isn't happening in normal clusters, appears to be alpha only issue."}
{"_id":"doc-en-kubernetes-2d86227263004d2306fadc99355cf0a88aac1d38d7e2576116201cd81e3141b9","title":"","text":"What happened: In handler#handleSummary, return value from request.Request.ParseForm() is ignored. What you expected to happen: The error return should be checked.\n/sig node\nHi this seems to be easy to fix. Could I solve this?\nPlease go ahead."}
{"_id":"doc-en-kubernetes-e8aabcf3e5e99db1922665cc12fe15914953073d88b0bc68ff14545d188d9dd5","title":"","text":"BUG REPORT When using with multiple files (like when referencing a directory in ), the json or yaml output is not a valid list format, and when fed to just the first file gets created. Kubernetes version (use ): latest master What happened: only the first file in directory gets created. What you expected to happen: all files in directory must be created. How to reproduce it (as minimally and precisely as possible):"}
{"_id":"doc-en-kubernetes-857a05f0056981db8f074277daa3916820903c2001d424633751ac4f757a1787","title":"","text":"needs a test to ensure that clauses are implemented correctly. Specifically, that they are not implemented as \"deny\" rules. eg, given a client pod with IP , if you add a NetworkPolicy with: spec: ingress: from: ipBlock: cidr: 0.0.0.0/0 except: A.B.C.0/24 then it should block ingress from that pod. But if you then add a second policy with: spec: ingress: from: ipBlock: cidr: A.B.C.D/32 then it should reallow ingress from that pod. Then if you delete the first policy, ingress should still be allowed, and if you create the first policy again, ingress should still be allowed. (ie, the result does not depend on which order the two policies were created in). /sig network /priority important-longterm\nI can work on this as my first issue related to sig-network.\njust noticed this issue while working on a similar effort. since there was no activity for the past 2 months I pushed a PR for this scenario"}
{"_id":"doc-en-kubernetes-43fa70370952068e2f27033c41492113945a1e0b0ce47a43a47f5065969f2aa7","title":"","text":"After the cluster was updated to enable PD CSI Driver, there was a race condition where AD Controller (kube-controller-manager) started to BEFORE the PD CSI Driver was up, thus AD Controller gave us the error Then AD Controller triggered the disk detach because the disk/volume was not populated in the desired state of world. Unfortunately the disk was in use during this detach and got corrupted. Since cluster upgrades can cause the controller manager and CSI drivers to start at different times, the AD controller needs to be more conservative when detaching We have not been able to reproduce this, it came up from a customer. /sig storage /assign <detailsk8s 1.23, CSI migration enabled </details<detailsGKE </details<details</details<details</details<details</details<details</details\n/triage accepted\nI think there's potentially two changes we should do: 1) for the initial DSW population, we should skip the check that if the csi driver is not installed. 2) in general, we should not skip volumes during dsw initial population if an error is returned. If we had some temporary api server error then that would cause volumes to not get to dsw and similar behavior happen. cc\nfor a repro, maybe you can try something like: a cluster with csi migration enabled (control plane + nodes) a pod with intree-volume and get it running the csi driver node daemonset. verify the driver is removed from CSINode kube-controller-manager\ncc\n/cc was trying to repro this, but on GKE with the managed driver which made it complicated.\nYeah maybe gce kube-up would be easier for a repro\nIf CSINode says that it has migrated GCP PD volume plugin to CSI, then IMO A/D controller should always attach GCP PD volumes to it using CSI, regardless if the CSI driver is present or not. In-tree volume plugin in the kubelet is disabled and can't mount any in-tree volumes anyway. To flip the migration feature gate / upgrade / downgrade kubelet, the cluster admin must drain the node first to make sure everything is detached. A/D controller should not see change of CSINode migration annotations with any volume attached to it.\n/priority important-soon\nAnother option could be to process these volumes anyways but add them to DSOW with migration status as \"uncertain\", so as they can be re-processed later."}
{"_id":"doc-en-kubernetes-8286ec57fa1e024ad026b2342f466cffd200140a0a95c9b64ae688331c52eb45","title":"","text":"But surely it looks like might be easiest unless it has any downsides.\nThere wasn't a node update in this case---there was a control plane update which restarted the PD CSI driver, and it looks like the CSIDriver resource was updated or changed in some way which raced with the kube-controller-manager restart.\nIs there anyway we can know if the csi driver is up and running(like a health check ping) before populating DSW?\nWe use the CSINode registration to determine that. The discussion here I think is concluding that we should not use that as a criteria for adding the volume to DSW. We only need to check that CSI migration is enabled on the node.\n+1 we need to be very resilient against races, I think we have to assume a driver is in process of being installed at any point in the AD controller reconcilliation.\n/assign sunnylovestiramisu\nReproduced with the steps: In /var/log/kube-controller-\nTested with the and doing the reproduce steps above. The error of does not show up anymore. The error message still get logged in the kube-controller- for debugging purpose: Ran Ran On the GCP console, the disk is used by node: Ran I redeploy the pd-csi-driver and then make sure it is up and running. Ran"}
{"_id":"doc-en-kubernetes-30eea2f2891480afb1ab3afc376d0efcc64938c80ac22e3ecb844909d4b7762a","title":"","text":"We need to fix this potentially flaky test and add it back to conformance suite. Please see discussion in Also waiting on:\ncc /sig testing\n/area conformance\n/cc\nIssues go stale after 90d of inactivity. Mark the issue as fresh with . Stale issues rot after an additional 30d of inactivity and eventually close. If this issue is safe to close now please do so with . Send feedback to sig-testing, kubernetes/test-infra and/or . /lifecycle stale\n/remove-lifecycle stale\ncan we convert it to conformance now??\nOnce 1.15 opens up"}
{"_id":"doc-en-kubernetes-1ef380d0c9eaabcc31aac489962d126bda361dc7bef4fb0787ab2e17f956bc1c","title":"","text":"Which jobs are failing: pull-kubernetes-integration Which test(s) are failing: TestPreemptionRaces Testgrid link: /sig scheduling /help\n/reopen I just saw this again\nReopened this issue. <detailsIn response to : /reopen Instructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\nIssues go stale after 90d of inactivity. Mark the issue as fresh with . Stale issues rot after an additional 30d of inactivity and eventually close. If this issue is safe to close now please do so with . Send feedback to sig-testing, kubernetes/test-infra and/or . /lifecycle stale\nSeems to be green: https://k8s-\nIt's still flaky on both master and PR testing: master: https://k8s- () PR testing: (). /reopen\nReopened this issue. <detailsIn response to : It's still flaky on both master and PR testing: PR testing: (). </details\nFrom the above two logs, I'm feeling that there is a \"phantom\" event (very likely of type Update by enforced resync) triggering the pod in the completed N<subth</subtest, to be still alive in the N+1<subth</subtest.\n/triage accepted working on a fix.\nIt would be great if we assign a different name to the Nth round test's preemptor pod. Currently, whatever this is Nth test, the preemptor pod is always called 'preemptor-pod'."}
{"_id":"doc-en-kubernetes-9e39524d0e4c89f1ddf9c516fe4e1c471a25826ee6b222abfe1f6c0cc213064c","title":"","text":"<!-- This form is for bug reports and feature requests ONLY! If you're looking for help check and the . --Is this a BUG REPORT or FEATURE REQUEST?: BUG REPORT What happened: When deleting/recreating pods backing a UDP service, traffic from external senders stops flowing. What you expected to happen: Traffic would continue to flow to new pods. How to reproduce it (as minimally and precisely as possible): Based on reading of similar issues and commit messages it seems like this would be most noticable on high-volume UDP applications, statsd and syslog are good examples. a deployment and service for a UDP app. In my test case, this would be the container. an external node, create a small test script utilizing a statsd library such as pystatsd. The most important part is that the sender uses the same source-port for every packet sent. Ex: the sender. Install tcpdump into the pod and watch for incoming packets, example from the test case: the pod tcpdump in the newly created pod. Observe that no new matching packets are flowing. existing conntrack rules on all nodes in the cluster. You should see traffic start flowing to the pod again. example command for clearing conntrack rules on the nodes: Anything else we need to know?: I have tested this only on GKE-managed clusters that utilize google forwarding-rules for external LBs, but I believe the issue may be present on any platform where iptables mode for kube-proxy is used. Tested and verified on 1.8.6-gke.0 and 1.7.11-gke.1. Also tested with both Google Container-optimized OS and the Ubuntu-based node images. I have only test this on GKE It appears this was a known issue and should have been cleared up by: From my reading of , it seems as if the func should exec() to clear up UDP entries on service endpoint change, but the only place I see that func called is from syncProxyRules(). That func looks complex and possibly like it is only run occasionally and not on service/endpoint changes. Environment: Kubernetes version (use ): (Also tested on 1.7.11-gke.1) Cloud provider or hardware configuration: GKE-managed clusters, google cloud. OS (e.g."}
{"_id":"doc-en-kubernetes-27a4c6cdabc82c95dde932bb7bf1e1865b5f691c3b3e7bf1ed0c71bfc361fddc","title":"","text":"from /etc/os-release): tried both: and Kernel (e.g. ): and Install tools: Others:\n/sig network\nWe also ran into this in kubernetes v1.7.6 using OpenShift 3.7:\n~I believe this is the same as ~If so, it is fixed in 1.10: These are about HostPort, not NodePort, sorry for the noise.\nIssues go stale after 90d of inactivity. Mark the issue as fresh with . Stale issues rot after an additional 30d of inactivity and eventually close. If this issue is safe to close now please do so with . Send feedback to sig-testing, kubernetes/test-infra and/or . /lifecycle stale\nIs there any other known workaround for this besides updating conntrack rules on the kube nodes? I tried deleting and recreating the service, but this didn't update the rules. When I restart the stream on the client side data flows correctly again (perhaps it's perceived as a new connection), but I have to restart the stream for each sender.\n/remove-lifecycle stale\nFWIW, this is not fixed in 1.10 - that fix was for HostPort, and this is still a bug for NodePort.\nI am seeing this issue in 1.11. I am using LoadBalancer type.\nhas been working on a fix for this, FYI.\nThis issue is extremely impactful for telephony protocols that rely on static source ip and source port. Any update on a possible workaround? I was thinking a potential workaround might be to reduce the UDP timeout to something really short so the stale conntrack entries go away faster. Or maybe there is a way to flush conntrack on the event of pod deletion? Edit: Removed stupid idea of a workaround to disable conntrack for UDP. Can't disable conntrack for UDP because it's used for DNAT from node to container.\nThe workaround is to not use NodePorts. Other kinds of services are already correct.\nnot sure what you mean. I am seeing the issue with a LoadBalancer type so it does not affect NodePorts only. As a workaround I have created a small python script that watches the K8s API for pod deletions and clears the related conntrack entries when that happens. I can share it when I get to work on Monday if you want to try it out."}
{"_id":"doc-en-kubernetes-956c9b1270ab4e0eb744fb5e7c207e8e7b7c0a49e83c634550198bfe9b333eeb","title":"","text":"I've my script . Note that it's not tested much so don't use it in a production environment. Also it cleans up all conntrack entries for a deleted pod, not only the UDP related ones.\nWe are also seeing this issue with type loadbalancer. We are on 1.11. Our use-case is the following: multiple devices are sending to an aggreagator machine. This machine forwards the UDP packages to a loadBalancer in gcp. When debugging the service, I noticed the following: The traffic is assigned to one pod, and the packages are not loadbalanced (client and server are the same cfr use-case) When redeploying a new cluster, Conntrack is cleared on termination signal of the old pod. (client receives connection refused) red/black and highlander deployment strategies have the same effect. How would I expect this to be working then? New cluster is . Conntrack is cleared Data is routed to NEW pods old pods are terminated Best Regards, Tim\nI am seeing this exact same issue on our cluster which has k8s v1.14.5 . We have endpoint nginx pods backing a UDP service with nodeport. Suddenly one day we see traffic not hitting into nginx pod but reaching the node hosting the nginx pod. We could simulate the problem by restarting the nginx pods. I am assuming with this issue marked as closed; and having made its way into v1.14; we should not be seeing this. Would also like to point out; the one thing, in addition we have in the service is ExternalTrafficPolicy set to Local. Can anyone please guide/help on why we are seeing this in v.1.14.5 of our cluster.\nSeeing this with k8s v1.16.5"}
{"_id":"doc-en-kubernetes-ab39f39ee12e84e818aa9c92c6877733f255a4fad832176aa0525d5bb0051663","title":"","text":"What happened: Visited It shows the release notes for v1.17 What you expected to happen: See the release notes for v1.18 How to reproduce it (as minimally and precisely as possible): Open in any browser or with curl\n/sig docs\n/assign"}
{"_id":"doc-en-kubernetes-b0a441a371f8db864bebc11c7303837af71a86244bb0f50ec81fe1d8a09d5c71","title":"","text":"The test is in flaky suite and no one seems to care. We need an owner for this test. cc Last failure log:\ndo you think you can take a look at this?\nThis is happening because of a bad assumption. The assumption is: if the reflector's store is consistent with etcd, the store given back to clients of the framework.Controller is too. I'd really like a general fix applicable to all controllers for this problem, but fixing it just for the RCs is pretty easy. In more detail: This function reports when the reflector has listed from apiserver. But the store given to the reflector is the fifo of the framework.Controller. The store given back to the caller is still not consistent, and will not be till the framework.Controller puts all the objects from the fifo into the client state store (). The rc specific fix: Before starting the rc manager, seed the expectations store with an expectation like {rc:foo, adds: foo.Status.Replicas, deletes:0}. This will wait for status.Replicas number of ADD events till a timeout. The general fix: Somehow expose this through framework.Controller.HasSynced (clients don't care about the fifo anyway, shouldn't it just be an implementation detail?) ideas? cc'ing people who have written controllers that might be interested I don't think this (and generally anything flaky for 1 release) is a p0, I can fix it if you don't mind dealing with the flakes till I've cleared things off my plate. I'd be more than happy to review a fix.\nWell, it's a P0 because test is flaky. E.g last run:\n- did you have chance to look into it?\nThe \"DaemonRestart\" tests are still labeled at . I will send a PR to promote them.\nare you still planning to send aPR to promote the DaemonRestart tests out of flaky? I think this is the right thing to do.\nAck; sorry, I lost this in the muck."}
{"_id":"doc-en-kubernetes-5bed2d3142dabe6e3fc7269ef850324d324032cbc87a456c8f2e9b79085fd590","title":"","text":"<!-- Please use this template while reporting a bug and provide as much info as possible. Not doing so may result in your bug not being addressed in a timely manner. Thanks! If the matter is security related, please disclose it privately via --What happened: flag is not supported on Azure Stack, should switch to use original way to remove a disk from data disk list, refer to for more details What you expected to happen: How to reproduce it (as minimally and precisely as possible): Anything else we need to know?: Environment: Kubernetes version (use ): Cloud provider or hardware configuration: OS (e.g: ): Kernel (e.g. ): Install tools: Network plugin and version (if this is a network-related bug): Others: /kind bug /assign /priority important-soon /sig cloud-provider /area provider/azure"}
{"_id":"doc-en-kubernetes-153297a62268f5892b05d52975dc41aef2505a641a424b2f4dec851bfaac7cd3","title":"","text":"only check that the output includes the correct keys - they don't confirm that it has appropriate values. Doing full validation on the version data, or at least enough to catch null-string values, would guard against failures like the bad functionality seen in OpenSUSE's binaries, which have been shipping with many of the fields left completely blank (as I discovered ).\n/sig testing\n/remove-sig testing /sig cli Edit: testing owns the framework, binaries, infrastructure etc. for testing. SIG CLI is nominally responsible for their test cases.\nIf SIGs are responsible for their own testcases, then this touches test cases for api-machinery as well (I'm not sure what the non-CLI version of this is, but there's definitely a schema adherence test that should fail at the API level). /sig api-machinery\n/triage accepted /assign\nGitHub didn't allow me to assign the following users: SaiHarshaK. Note that only , repo collaborators and people who have commented on this issue/PR can be assigned. Additionally, issues/PRs can only have 10 assignees at the same time. For more information please see <detailsIn response to : /triage accepted /assign Instructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\n/assign"}
{"_id":"doc-en-kubernetes-2a1f139ca790a16bd0882043a2e2916760e4c33264d1a31f1254bc354f4822db","title":"","text":"After KCM restart, any Pod that reuses a volume that failed to detach gets stuck ContainerCreating with message: User \"system:node:ip-10-0-243-\" cannot get resource \"volumeattachments\" in API group \"\" at the cluster scope: no relationship found between node 'ip-10-0-243-' and this object See steps to reproduce below, it's quite a complex condition. It happens when a pod is restarted, a CSI driver \"busy\" (i.e. times out) + KCM restarted at a wrong time. Pods start just fine. a Pod that uses an attachable CSI volume. the Pod. Make sure it unmounts successfully, but any fails with a transient error, such as timeout. The external attacher continues to detach the volume at this point. kube-controller-manager (KCM) and start a new one. Wait for any leader election etc - it's important for the new KCM to populate A/D controller with no pod using the detaching volume. Let the A/D controller to observe detach error. a new Pod on the same node that uses the same volume. the CSI driver to detach the volume (i.e. stop injecting any transient errors). At this point: The external attacher succeeds detaching the volume, VolumeAttachment is deleted. The volume is marked as attached in node status (!!!) despite no VolumeAttachment exists. A/D controller does not even try to create a new VolumeAttachment, the volume is marked as attached in its Actual State of World (ASW). Kubelet tries to mount the volume, but WaitForAttach fails with: , basically forever. I created . The script expects that the cluster is created by , because it restarts KCM in the middle. It uses my hacked hostpath CSI driver to inject detach errors. Sources are at / the last commit(s) there. There are two issues that together lead to this situation: fails, , i.e. puts it back to . With the volume in , kubelet may try to mount it, while the external attacher tries to detach it. It is not as bad as it sounds, because kubelet can't read VolumeAttachment with the deletion timestamp and it will not call / of un-attached volume."}
{"_id":"doc-en-kubernetes-86a0f830cda4a62b3e964f9aa561f07fe82e259bc1038677353226e3d67190fe","title":"","text":"KCM starts, all VolumeAttachments that don't have a Pod (that's good), but also (that's probably bad). When the newly started KCM re-tries to detach the volume and gets an error again, it marks the volume as fully attached (same as the previous point) and adds it to . Now the volume is both mounted + attached in ASW and A/D controller actually stops here and waits forever for kubelet to unmount the volume. <detailsToday's k/k master, </details<detailsNone </details<detailsN/A, it's KCM issue. </details<details, but seen in a downstream Kubernetes distro too. </details<details</details<details</details\n/sig storage\n/assign\n/triage accepted\nI have the same problem: Seems restart of Kubelet solves this problem. Kubernetes v1.25.13"}
{"_id":"doc-en-kubernetes-754a337748c77aa830912aa5c369fd083977cc37177bb1280fd450d5c939bfe3","title":"","text":"<!-- Please use this template while reporting a bug and provide as much info as possible. Not doing so may result in your bug not being addressed in a timely manner. Thanks!--What happened: In the test failed due to namespace leakage: we created e2e-tests-load-30-nodepods-1-pfn9m, but never got response from api server (got 0-length response instead) we retried and created e2e-tests-load-30-nodepods-1-m9tq5 e2e-tests-load-30-nodepods-1-m9tq5 to f.namespacesToDelete and leaked -pfn9m next test case failed because e2e-tests-load-30-nodepods-1-pfn9m wasn't deleted. What you expected to happen: While it's unclear why we got 0-length response instead ( investigates that), we always can get error while the namespace was created (e.g. connection reset, timeout, io error). We should handle this case in retrying correctly. How to reproduce it (as minimally and precisely as possible): Anything else we need to know?: Environment: Kubernetes version (use ): v1.14.0-alpha.0.793+ Cloud provider or hardware configuration: gce OS (e.g. from /etc/os-release): Kernel (e.g. ): Install tools: Others: <!-- DO NOT EDIT BELOW THIS LINE --/kind bug\nThere are no sig labels on this issue. Please add a sig label by either: a sig: e.g., to notify the contributor experience sig, OR the label manually: e.g., to apply the label Note: Method 1 will trigger an email to the group. See the . The in method 1 has to be replaced with one of these: bugs, feature-requests, pr-reviews, test-failures, proposals. <detailsInstructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\n/cc"}
{"_id":"doc-en-kubernetes-02478d007e8fb94d5569c598d90648b98c08e1979b9683524b76b29f0b58f1e9","title":"","text":"Hello, the SRC folks asked that I file this publicly: In the client-go repo, , function creates a private key and stores it into a file with mode : Typically keys are created with more restrictive modes to prevent other users or services on the system from being able to copy them before the user fixes the permissions. Of course, if a more restrictive umask is in place, this may not be a big deal, but it's an extremely fickle piece of the environment to rely upon for security. It's better to be explicit. Thanks\n/kind bug Not sure if this is /sig api-machinery or /sig auth\nis this related to us?\nHi IMO, mode make sense for the key file. I would like to work on this, waiting for Thanks!!\n/triage accepted\n/assign\nThe Kubernetes project currently lacks enough contributors to adequately respond to all issues and PRs. This bot triages issues and PRs according to the following rules: After 90d of inactivity, is applied After 30d of inactivity since was applied, is applied After 30d of inactivity since was applied, the issue is closed You can: Mark this issue or PR as fresh with Mark this issue or PR as rotten with Close this issue or PR with Offer to help out with [Issue Triage][1] Please send feedback to sig-contributor-experience at . /lifecycle stale [1]:\n/remove-lifecycle stale\n/lifecycle frozen /good-first-issue I think changing the code to use should be fine. Anyone is free to pick this up. Would be good to include a unit test.\nThis request has been marked as suitable for new contributors. Please ensure that the issue body includes answers to the following questions: Why are we solving this issue? To address this issue, are there any code changes? If there are code changes, what needs to be done in the code and what places can the assignee treat as reference points? Does this issue have zero to low barrier of entry? How can the assignee reach out to you for help? For more details on the requirements of such an issue, please see and ensure that they are met. If this request no longer meets these requirements, the label can be removed by commenting with the command. <detailsIn response to : /lifecycle frozen /good-first-issue Instructions for interacting with me using PR comments are available ."}
{"_id":"doc-en-kubernetes-97af64e63f750f1585d6dae219445a2123c113559188cb8ace853ac6ec6d5ee9","title":"","text":"If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\n/assign\nHello, this is my first ever open source contribution. I am trying to understand and get this fixed. Any suggestions would be helpful!\n/assign\n/assign"}
{"_id":"doc-en-kubernetes-6a9e001ff60e5ca04752d0e51696e4d68773c676c7cdd69e5d88aa83e0129d5a","title":"","text":"=== RUN TestStructuredAuthenticationConfigReload/oldinvalidconfigtonewvalidconfig first seen in 3/11 No response /sig api-machinery instrumentation\n/sig auth /remove-sig api-machinery instrumentation /triage accepted /assign"}
{"_id":"doc-en-kubernetes-54abd4c5b24af7b3b07bb9048a102e7524bce4cd2b3e4eeee922b6e64ea9d4ce","title":"","text":"attemting to run throws and error with ERROR: Error: - does not specify a valid ReplicationController\nThe documentation is a little ambiguous, as the second example suggests that - should work but the Synopsis section does not. Could you either fix the documentation or the command?\nCalvin, can you post the contents of , and the schema of the original rc? There are some for the schema when providing a target from file, which may not be met.\nrunning works, I was interpreting the error message that it though - was a rc\nwill work for any object schema file. I.e, that is the same command you use to create a podreplicationcontroller from file. The type being created is determined by the contents of the file. Looking back at the code, the error you saw indicates that does not contain a replicationcontroller spec. If you can create it, that must mean it's something else; can you post/link the contents?\nyes I need to clean it up first\nThat looks like a valid rc spec, but the labels may be insufficient for rollingupdate. In order to do a rollingupdate you need to have a label to control the update. Typically this would be a \"version\" label. Example Basically the selector for the old and new rc must share at least one label key with a differing value. By the way, rollingupdate is currently getting a simplified version, which generates the new rc spec for you. This is wip that will be available in next week's GKE release. See\nah I see, I actually have a dynamic version in there so that will be easy to do thanks Jeff Lowdermilk wrote:\nso with an updated config still does not work while does work with the first one still erroring out with so likely my missing selector issue was an unrelated issue\nFrom your last comment, seems this is resolved. If you are still having trouble, lmk and I will reopen.\nstill not fixed because I can't do\nI'm pretty sure that's a piping issue with the wrapper command that likely doesn't occur if you use kubectl directly. Try Note that steps 1,2 will be unnecessary with the next gcloud release; you will be able to use kubectl directly after creating a cluster.\nWith cloudsdk 0.9.60, standalone kubectl works by default with GKE clusters."}
{"_id":"doc-en-kubernetes-34e6555626ce71d3ee72ecabeeb61d74a5a04012a9d26021a4a4f8a40e19ef7f","title":"","text":"The wrapper command is no longer needed and will be removed in a future release. Let me know if you have problems running against a GKE cluster.\nI am running across this issue with a recent build of .\nis still throwing errors with version 0.17 used directly (without wrapping)"}
{"_id":"doc-en-kubernetes-6758a4fd9fc03e20cf227e8e04e978c36b819698b12d653642006e59891b2248","title":"","text":"<!-- Please use this template while reporting a bug and provide as much info as possible. Not doing so may result in your bug not being addressed in a timely manner. Thanks! If the matter is security related, please disclose it privately via --Currently when using it adds as one of the limiters. This does not work as intended because there is no immediate sleep following the call to the limiter as seen in which does work correctly. The issue with the in its current usage is there is no forget nor a sleep immediately following the call so if is called thousands of times in a row, the timeout grows very quickly with no max. For example, if a clusters has 10k secrets and those are all enqueued using the the will end up with a 16 min wait before the tail end of objects can be enqueued and processed. I would expect some max backoff to take affect. This can be seen at\n/sig api-machinery\n/cc /triage accepted\nit possible to set maxWaitingTime for BucketRateLimiter? when r.Limiter.Reserve().Delay() maxWaitingTime, the return of BucketRateLimiter When() is maxWaitingTime. defined is 1000ms / r.limit * MAXQPS(5000) this issue example (r.tokens = 100, r.limit = 10), if call When() times, the tail end of the object is enqueued waiting for 8.33 min. i'm not sure how to set the value of MAXQPS please help review it\nThe fix merged to close this issue does not fix the underlying issue with the DefaultControllerRateLimiter. The bucket is not being used correctly so this does not work as intended, please read my issue description again as to what is actually happening.\nis another fix needed to solve this issue?\nA token bucket rate limiter will only refill at the defined rate. If enough requeues are , it will take a long time for the refill rate to clear the queue. If a max wait is introduced, then requeues will be inserted into the queue faster than the refill rate of the token bucket rate limiter. A max wait will violate the token bucket rate limiter and we want token bucket rate limiting behavior."}
{"_id":"doc-en-kubernetes-9fcfa37c5fd88e21dae3cf641aefb45ac17c97dc04df9b3a60101bffeb7f8b4c","title":"","text":"Failed: https://k8s- Run so broken it didn't make JUnit output!\nFailed: https://k8s- Run so broken it didn't make JUnit output!\nFailed: https://k8s- Run so broken it didn't make JUnit output!\nFailed: https://k8s- Run so broken it didn't make JUnit output!\nFailed: https://k8s- Run so broken it didn't make JUnit output!\nFailed: https://k8s- Run so broken it didn't make JUnit output!\nFailed: https://k8s- Run so broken it didn't make JUnit output!\nFailed: https://k8s- Run so broken it didn't make JUnit output!\nFailed: https://k8s- Run so broken it didn't make JUnit output!\nFailed: https://k8s- Run so broken it didn't make JUnit output!\nFailed: https://k8s- Run so broken it didn't make JUnit output!\nFailed: https://k8s- Run so broken it didn't make JUnit output!\nFailed: https://k8s- Run so broken it didn't make JUnit output!\nFailed: https://k8s- Run so broken it didn't make JUnit output!\nFailed: https://k8s- Run so broken it didn't make JUnit output!\nFailed: https://k8s- Run so broken it didn't make JUnit output!\nFailed: https://k8s- Run so broken it didn't make JUnit output!\nFailed: https://k8s- Run so broken it didn't make JUnit output!\nFailed: https://k8s- Run so broken it didn't make JUnit output!\nFailed: https://k8s- Run so broken it didn't make JUnit output!\nFailed: https://k8s- Run so broken it didn't make JUnit output!\nFailed: https://k8s- Run so broken it didn't make JUnit output!\nFailed: https://k8s- Run so broken it didn't make JUnit output!\nthis appears to have always been broken, I'm going to disable the job for now. https://k8s-\nFailed: https://k8s- Run so broken it didn't make JUnit output!"}
{"_id":"doc-en-kubernetes-1f38d50a57cfbeef9d8517599fecefb7d45971350552d6bdbc65534ca41945d0","title":"","text":"Aggregated discovery v2beta1 fixtures should be removed. /sig api-machinery /triage accepted"}
{"_id":"doc-en-kubernetes-55c62ef7d3a20811fc11fc78d986a318ea3d711ec0d8f690353464985c4986dd","title":"","text":"What happened: After creating Kubernetes service on a Kubernetes cluster built on Azure IaaS (not AKS), nodes which are not master nodes but are also not worker nodes and which are labeled with are included in the backend pool for the load balancer. What you expected to happen: All nodes labeled with should be excluded from the load balancer's backend pool. How to reproduce it (as minimally and precisely as possible): Create a node with a type other than and label it with Anything else we need to know?: Environment: Kubernetes version (use ): Client Version: version.Info{Major:\"1\", Minor:\"16\", GitVersion:\"v1.16.10\", GitCommit:\"\", GitTreeState:\"clean\", BuildDate:\"2020-05-20T14:00:52Z\", GoVersion:\"go1.13.9\", Compiler:\"gc\", Platform:\"linux/amd64\"} Server Version: version.Info{Major:\"1\", Minor:\"16\", GitVersion:\"v1.16.10\", GitCommit:\"\", GitTreeState:\"clean\", BuildDate:\"2020-05-20T13:51:56Z\", GoVersion:\"go1.13.9\", Compiler:\"gc\", Platform:\"linux/amd64\"} Cloud provider or hardware configuration: Azure OS (e.g: ): CentOS Linux release 7.8.2003 (Core) Kernel (e.g. ): Linux master- 3.10.0-1127.8.2.el7.x8664 SMP Tue May 12 16:57:42 UTC 2020 x8664 x8664 x8664 GNU/Linux Install tools: Custom terraform stack Network plugin and version (if this is a network-related bug): Others:\n/sig cloud-provider These SIGs are my best guesses for this issue. Please comment if I am incorrect about one. I am a bot run by vllry. ‍\n/priority important-soon /sig cloud-provider /area provider/azure\n/priority critical-urgent"}
{"_id":"doc-en-kubernetes-9ba650efd322225fce569f5f04b924197d000dece17d628f51cd31eac5f4f289","title":"","text":"I followed the document to create an e2e cluster However, doesn't work after the cluster is up. I got the following error: /cc\n/cc\nwhere are you trying to ssh from? Can you ssh to a Linux instance in GCE? If you run do you see ssh keys there? Do any of them match a public key that you have under ? There could be a delay the very first time you try to ssh; we noted this in , but we could add it to the README too. If you try a second time after a couple minutes does it still fail?\nYeah I can ssh to a linux instance in the same cluster. It still fails after a long. I've removed all my ssh keys from cloud metadata, and recreating a cluster. Let me see how it goes this time.\nI still got the same error after recreating the cluster. I can ssh into linux nodes in the same cluster just fine."}
{"_id":"doc-en-kubernetes-a26579e34c4e1ee413ed781fcfba121a4c6a53c97161cd6f60e138eced3ebf5d","title":"","text":"Keep running but not crashing Have a long volume attach path This corresponds to these lines We are using k0s <details</detailsbare metal <details</details<details</details<details</details<details</details\nI turned to use the JSON logger, it is gorgeous no wonder it failed\nrelated\n/sig node storage\n/cc\nVolumeToAttach contains Pod pointers, so yes, I think we should not log the entire struct, and only log important parts to it. /triage accepted /help-wanted\n/priority important-longterm /assign"}
{"_id":"doc-en-kubernetes-efc4d6f641fd1b1bcaa275636a136a6176382b556456e87b10de85fb3c9c026d","title":"","text":"Every time watch is reset, event exporter exports an entry about it to Stackdriver. To make it less irritating, this has to be done once per container run Related image PR: /cc"}
{"_id":"doc-en-kubernetes-8cc0000ebdb21aeba4d3dda4212069010f35d863300d721e65a4eeef908ef7b3","title":"","text":"As planned for 1.12 in kubernetes/features: After , the next step is to add the logic to the reconciler to detach the volumes when the nodeShutdown taint is applied. /assign /sig storage /kind feature looping\nHow do you know that node is Shutdown? is it a planned shutdown or a unplanned shutdown?\nAFAIK regardless of whether it's planned or not, it's detected by the cloud provider. In case of bare-metal node it doesn't happen automatically, as there is no universal way to reliably determine if a node is powered off, so the node must be tainted as by some external source.\nIssues go stale after 90d of inactivity. Mark the issue as fresh with . Stale issues rot after an additional 30d of inactivity and eventually close. If this issue is safe to close now please do so with . Send feedback to sig-testing, kubernetes/test-infra and/or . /lifecycle stale\nany update on this issue?\n, I am currently working with sig-node team on handling node shut down. I will make a proposal and discuss in sig-node meeting next Tuesday. Will post the update soon.\nStale issues rot after 30d of inactivity. Mark the issue as fresh with . Rotten issues close after an additional 30d of inactivity. If this issue is safe to close now please do so with . Send feedback to sig-testing, kubernetes/test-infra and/or . /lifecycle rotten\n/remove-lifecycle rotten /kind bug /priority critical-urgent /assign /sig node\nThose labels are not set on the issue: <detailsIn response to : /remove-kind feature Instructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\nIssues go stale after 90d of inactivity. Mark the issue as fresh with . Stale issues rot after an additional 30d of inactivity and eventually close. If this issue is safe to close now please do so with . Send feedback to sig-testing, kubernetes/test-infra and/or . /lifecycle stale\nany update on this? is it going to be part of 1.15?\nStale issues rot after 30d of inactivity. Mark the issue as fresh with ."}
{"_id":"doc-en-kubernetes-8b5c9612057f9871f14258bc136d8e86f57d2327861ec200bbe8402ae4819108","title":"","text":"Rotten issues close after an additional 30d of inactivity. If this issue is safe to close now please do so with . Send feedback to sig-testing, kubernetes/test-infra and/or . /lifecycle rotten\n/remove-lifecycle rotten\nIssues go stale after 90d of inactivity. Mark the issue as fresh with . Stale issues rot after an additional 30d of inactivity and eventually close. If this issue is safe to close now please do so with . Send feedback to sig-testing, kubernetes/test-infra and/or . /lifecycle stale\n/remove-lifecycle stale\n/priority important-longterm\nIssues go stale after 90d of inactivity. Mark the issue as fresh with . Stale issues rot after an additional 30d of inactivity and eventually close. If this issue is safe to close now please do so with . Send feedback to sig-testing, kubernetes/test-infra and/or . /lifecycle stale\n/remove-lifecycle stale /lifecycle frozen\n/triage accepted /kind feature\nThis issue has not been updated in over 1 year, and should be re-triaged. You can: Confirm that this issue is still relevant with (org members only) Close this issue with For more details on the triage process, see /remove-triage accepted\nThis issue is currently awaiting triage. If a SIG or subproject determines this is a relevant issue, they will accept it by applying the label and provide further guidance. The label can be by org members by writing in a comment. <detailsInstructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\nI think this can be closed now as we have non-graceful node shutdown.\n/close\nClosing this issue. <detailsIn response to : /close Instructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details"}
{"_id":"doc-en-kubernetes-6f8ca0211eec05ccf55bb519cf8df0569345414eb205c19a4215b8fda8e40904","title":"","text":"<!-- Please only use this template for submitting reports about failing tests in Kubernetes CI jobs --Which jobs are failing: https://k8s- Which test(s) are failing: should unmount if pod is force deleted while kubelet is down [Disruptive][Slow] Since when has it been failing: 11/07/18 15:02 Testgrid link: https://k8s- Reason for failure: The test case stops and starts kubelet, and it seems like kubelet is unable to start again Anything else we need to know: <!-- DO NOT EDIT BELOW THIS LINE --/kind failing-test\ncc\nThanks for opening this Looking at , looks like could be the cause? /assign can you please look into this as priority. This is failing critical release blocking tests. /cc /sig cli /priority critical-urgent /milestone v1.13\nIt looks like kubelet keeps restarting. The error right before restarting is:\n/assign\nThis bug could have been exposed by the newly tests in\nHmm, so I see two errors: Which is a Fatal error and would prevent Kubelet from starting anyways. In which case this error is handled correctly by the pluginwatcher, but I'm not sure preventing Kubelet from starting is the desired behavior. Should we log this error and skip it? There also seem to be a cleanup issue here on that file is this expected? This one is clearly a bug and not really handled correctly. It's due to the fact that we encounter an error at start, cleanup but don't wait for the gorountines to finish before cleaning up. I can send in a fix pretty easily for the second issue, basically add a waitgroup in the . For the first error, what do we think is the expected behavior?\nThis was mentioned in as well. /cc\nIs this related to the issue where plugin watcher is recursively scanning for sockets?\nAlso, it should be noted that the path it's erroring on is not a socket type.\nThat's a bit different, as it is intended to scan recursively. Also note that the error here is produced by . What should be our behavior in such a case? Ignore that error?\nI'm not sure if we want to be recursively scanning. The volume directories are mount points, which means there could be lots of files/directories existing in those volumes."}
{"_id":"doc-en-kubernetes-5fd14dcabf74b34a3f64d7f7feeb636f4d13fa903d9084a91cc242778a1c3d1e","title":"","text":"maybe we can put a limit on depth of plugin directory for on the same PR:\nIdeally the directory that the plugin sockets are dropped in should be different from the directory where mounts happens. Currently the plugin socket directory is: ) The mounts directory is (so far so good). Problem is the \"global mount\" directory is also :( So options are: recursively scanning like suggested, and say plugins will can only be directly under This will break existing plugins, for example, the GCP plugin is at (see ). recursively scanning but change the plugin socket directory to ? This will also break existing plugins, like the GCP one pointed out above. the plugin dir the same, and leave recursion in, but put a limit on depth of plugin directory as suggested by This is ugly but may be the cleanest solution.\nThis conversation is completely different from the bug surfaced by this issue. Let's move that to the corresponding issue: I have a fix ready to go out for the panic, we should figure out what behavior we expect Kubelet to have when encountering this error from filepath.Walk: Should we ignore it?\nIgnoring the error (with a warning maybe) instead of failing all of kubelet seems reasonable to me. If the plugin did not setup their registration socket correctly, then the end result should be that the plugin fails to register, but all other plugins should continue to work correctly.\nthe latest runs of this job no longer timesout - Instead I am seeing a new CSI related failure Is this expected. Do we want to track that failure in this bug now or do we have a separate issue?\nthe latest test logs indicate some test-infra issue:\nthe same shows up as a timeout on that particular test. Lets wait for the next run. but somehow the timeout seems to have gone. Are we still planning to merge and associated with this issue?\nshould be merged as it fixes a bug. can you help with approval? Thanks!\nare you missing the right PR# in the comment above?\nthis is the correct PR -\n/reopen Keeping it open until we can see the job pass a couple of rounds\nReopened this issue. <detailsIn response to : /reopen Instructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository."}
{"_id":"doc-en-kubernetes-7a1e75b2dd238a9264dd7ee20adbbea7f5a6949137fb440b1e66e3d97a7ebee9","title":"","text":"</details\nI see just one CSI test timing out in serial job Is that something the fixes here attempted to resolve? if not I can open a new issue if it persists and close this one out. Thanks.\nThe test mentioned in my comment above was a flake.. closing this issue now. will open a new issue for the flake\nI'm getting the following related error on a busy Node: In the above case it's a full filesystem, but I imagine the same error would pop up with a RO filesystem as well. Looking at the patch, I'd expect kubelet would still be failing to start when it tries to watch it. (above was from a v1.12.3 cluster) , what do you reckon?\nplugin watcher fixes for 1.12 are expected to be released with 1.12.5:\n, I understand that I'm missing that patch. The patch code, however, would not make a difference in this case, if I read it correctly. It would not prevent the recursive dive into mounted volumes, as the errors happening at the deeper levels of that dive.\nThe patch blacklists the plugin watcher from searching the directory\nOK, I was looking at the wrong patch after all (). Thanks for the clarifications."}
{"_id":"doc-en-kubernetes-b66f5b7782acce0f091a6b4a6784e7025204ffa447d61b006db9acac747a220f","title":"","text":"master-informing: capz-windows-master ! 07-10 17:46 IST No response /sig cli cc\nThis issue is currently awaiting triage. If a SIG or subproject determines this is a relevant issue, they will accept it by applying the label and provide further guidance. The label can be by org members by writing in a comment. <detailsInstructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\n/kind failing-test\nLooking at the failure and tracing the recent code changes, the test was recently modified for the regex pattern match as part of this PR which comes with deprecation of the golang-based short version information in favor of making the --short the new default. But the output of kubectl version is still reflecting the older long format which is causing this test to fail.\n/assign"}
{"_id":"doc-en-kubernetes-c8f75f1ad6e3f2bccb3ac0da7b9de068a195504da38201c6a8a8d5b4a34c6c73","title":"","text":"Is this a BUG REPORT or FEATURE REQUEST?: BUG /kind bug What happened: If you are using a mac and GCE and you specify NODELOCALSSDS=0 when using GCE via kube-up or e2e --up, then you end up getting 2 local SSDs. What you expected to happen: I would expect to get 0 local SSDs by default and when I specify NODELOCALSSDS=0 explicitly. How to reproduce it (as minimally and precisely as possible): On mac execute: NODELOCALSSDS=0 go run -- -v --up Anything else we need to know?: I've discovered that this issue is due to If NODELOCALSSDS=0, this evaluates to $(seq 0) From man seq on mac osx This was run on mac with the seq manpage indicating it comes from BSD Feb 19 2010. Note: some unix/linux variants, e.g. a test on ubuntu with coreutils 8.25 and a test on centos-7 with coreutils 8.22, had a different behavior that does not demonstrate this bug. Environment: Cloud provider or hardware configuration**: GCE OS (e.g. from /etc/os-release): mac osx\nThere are no sig labels on this issue. Please by: a sig: e.g., to notify the contributor experience sig, OR the label manually: e.g., to apply the label Note: Method 1 will trigger an email to the group. You can find the group list and label list . The in the method 1 has to be replaced with one of these: bugs, feature-requests, pr-reviews, test-failures, proposals\n/sig storage"}
{"_id":"doc-en-kubernetes-5963c982feb2169d82fc17ad89ab9be2967cbad808254e7a712ad33fb952ec83","title":"","text":"If the namespace controller gets an error when doing a sync and the error is not a error the error will not be handled. Ran in to this when we found When deleting a namespace it received a error that wasn't reported anywhere and the namespace stuck in because a finalizer was left.\nOther things noticed in the namespace controller: need a crash handler in sweep for all error return cases and determine whether should be returned determine what to do with non- errors at ... I think that currently it just hangs termination of that namespace until a resync\nI will take a look. Thanks for the heads up."}
{"_id":"doc-en-kubernetes-618a17ee1a2bfcaa30e720bd6c59c3ac91ee900494c2c92ac1b2427757bb09d9","title":"","text":"reported a variety of failure modes in .\n+1\nI figured I'd add to this issue rather than create a new one. I attempted to create a vagrant cluster today (dev tree is up to date) with 1 Master and 1 Minion and the cluster creation is stuck in the 'Validating master' loop ( for over 10 minutes in the validating loop ). salt-master and salt-minion logs seem to be empty ( they don't appear to be running, in fact ).\nI still have the cluster running in case anybody wants additional information.\n/cc please let me know if there is any other information about my vagrant setup that I need to post here.\nI have been working to get more stability in this environment over last few days. Can you try again from HEAD after this merges:\nCan you also verify that you built a release prior to running cluster/Kube-up?\nyes, I started from scratch and built a release before running cluster/kube- - cluster/kube- seems to fail fairly early in case there is no release build available.\nIt looks like got merged. I'll try again and get back to you.\nThe vagrant cluster seems to come up ( completes master and minion validation ). However, there is new credentials related error that is showing up at the end of running cluster/kube- :\nI'll add one more problem I've encountered. Runing from head:\nNew to me. Will investigate more tomorrow. Sent from my iPhone\nAlso new to me :-) - will see if I can reproduce either. Sent from my iPhone\nYour credendials and certificates are not working with new environment. Just delete your .kubernetesvagrant* files and recreate the cluster. It would be nice to automate this task, if you tear down the cluster.\nI am able to reproduce the 502 Bad Gateway error and am working on determining what changed in order to provide a resolution over the last couple days. Please assign this issue to me.\nThe issue appears to me that the kube-apiserver is not getting provisioned by salt correctly. Parsing Salt logs to determine why.\nLooks like the file is not getting laid down by salt for some reason ;-)\nThe Jinja template was changed as part of identifying the cluster. Fixing up now, and will send a PR to close this issue.\nthat was it, thanks!"}
{"_id":"doc-en-kubernetes-b47abd0a7beb2bc1a841f65e5737673c1fa7ed1c22ab52e8ed01b146c2722dfe","title":"","text":"What happened: In the PMEM-CSI CI, we rapidly deploy and remove a custom resources which then owns all objects created for that resource. Deletion of those objects is handled through ownership references, i.e. deleting the CR is expected to cascade to all objects. This worked fine up to Kubernetes 1.17. When starting to test with 1.18.2, we pretty reliably get into a state where the CR and the service objects owned by it are marked for deletion, but don't go away. Debugging with the garbage collector's dependency graph (see ) showed that endpointslices were blocking the deletion of the services and thus the CR. In the graph dump (= a snaphot in time), they were not even marked for deletion. A closer look showed that endpointslices for the services were getting created and deleted continuously for over 15 minutes. Eventually, the garbage collector won and managed to delete endpointslices and services before new endpointslices were created. What you expected to happen: There seems to be a missing check in if the service is marked for deletion (i.e. has a deletionTimestamp), then the controller should not create new endpointslices. How to reproduce it (as minimally and precisely as possible): I tried to boil it down to a script, but it just ran fine. The \"easiest\" way to reproduce it is to run PMEM-CSI's E2E testing, which is probably not easy enough :disappointed: Anything else we need to know?: Disabling the EndpointSlice feature makes the problem go away. Environment: Kubernetes version (use ): 1.18.2 Cloud provider or hardware configuration: QEMU virtual machines OS (e.g: ): Fedora Install tools: kubeadm /sig network\n/triage unresolved Comment when the issue is assessed and confirmed. I am a bot run by vllry. ‍\ncc for endpointslice\nThanks for the detailed bug report! Will work on getting a fix in soon. /assign\n/remove-triage unresolved\nCan you describe the general process to reproduce this bug? Would it involve hundreds of EndpointSlices for a Service? I've tried to recreate this in a , but unfortunately I haven't been able to get anything to break. Any ideas for what I could try?\nThere's one endpointslice per service. Our operator creates two services."}
{"_id":"doc-en-kubernetes-884db7f4f09e6e30f5114a0e6bbb9ebc21f2d686aa6c190d6a4d9470af013127","title":"","text":"One peculiarity may be that we don't actually want the pods to start during these tests because we only want to verify object creation, not the usage of the result, and therefore all pods reference non-existent images. All objects immediately get deleted once the test has verified that they exist, so the timing is probably very different from \"normal\" usage patterns. I suspect that it is more important to create and delete services quickly than it is to have many endpointslices per service.\nI had tried to come up with a script that triggers the problem, without success. That was before I realized that endpointslices are involved. On Monday I can have another look.\nThat was because I thought that (the default) enables foreground deletion. It doesn't, and there's also no option to select it. After explicitly selecting it via curl, the following script failed for me eventually: Output: Note that for the endpointslice is greater than the of its service."}
{"_id":"doc-en-kubernetes-4510e09c6fe68ed5e12ffcf7126f6af8b2755c8c7c2f631bec51ed41fca52919","title":"","text":"What happened: One pod stuck at CreateContainerError and can't get recovered automatically. Kubelet log shows that the requested name (let's say Container2) is already in use. shows the Container2 Running, and it's created before Container1 This should never happen. How to reproduce it (as minimally and precisely as possible): I wish i knew. Anything else we need to know?: I know this can be caused by time shifting, but I didn't find any time shifting sign in system log and chrony log, and there is no rebooting between the create timestamp of Container1 and Container2, so this should be caused by something else. This happens intermittently in our env, like 3 times in half a year, and I didn't find any ime shifting sign each time. Environment: Kubernetes version (use ): Client Version: version.Info{Major:\"1\", Minor:\"15\", GitVersion:\"v1.15.6\", GitCommit:\"\", GitTreeState:\"clean\", BuildDate:\"2019-11-13T11:20:18Z\", GoVersion:\"go1.12.12\", Compiler:\"gc\", Platform:\"linux/amd64\"} Server Version: version.Info{Major:\"1\", Minor:\"15\", GitVersion:\"v1.15.6\", GitCommit:\"\", GitTreeState:\"clean\", BuildDate:\"2020-08-28T11:18:26Z\", GoVersion:\"go1.12.12\", Compiler:\"gc\", Platform:\"linux/amd64\"} Cloud provider or hardware configuration: OS (e.g: ): CentOS Linux release 7.5.1804 (Core) Kernel (e.g. ): Linux node-1 3.10.0-957.21.3.el7.x8664 SMP Wed Jul 17 06:57:50 UTC 2019 x8664 x8664 x86_64 GNU/Linux Install tools: kubeadm Network plugin and version (if this is a network-related bug): Others:\n/sig node\nAny update on this issue, we are also facing the same issue. EKS version : 1.14.9 Docker version: 19.03.6-ce AWS CNI: 1.6.2 Intermittently we are also facing this , what might the issue, could please guide us here? Below are the kubelet logs. )\""}
{"_id":"doc-en-kubernetes-cf3a33a19ef0552fe6ccd0d8ccfc3a21fead851f6c57fd16801184b4e1e277da","title":"","text":"with CreatePodSandboxError: \"CreatePodSandbox for pod \"-f0fe-11ea-9872-)\" failed: rpc error: code = Unknown desc = failed to create a sandbox for pod \"aiqx-server-dep--b9lmv\": operation timeout: context deadline exceeded\" Sep 07 11:40:40 kubelet[2899]: W0907 11:40:40. 2899 ] No ref for container \"docker://7b9c86b529530f15d941f8e68973a202c8f348ab5db864a8371e8c82ddbf9dc4\" (r(-e845-11ea-bd96-) Sep 07 11:40:43 kubelet[2899]: W0907 11:40:43. 2899 ] Unable to create pod sandbox due to conflict. Attempting to remove sandbox \"a73bd9a03d56bc7855e4e89e163e13c5e7173397f10edcc200fb686e1b6885c8\" Sep 07 11:40:43 kubelet[2899]: E0907 11:40:43. 2899 ] Failed to remove the conflicting container \"a73bd9a03d56bc7855e4e89e163e13c5e7173397f10edcc200fb686e1b6885c8\": Error: No such container: a73bd9a03d56bc7855e4e89e163e13c5e7173397f10edcc200fb686e1b6885c8\nIssues go stale after 90d of inactivity. Mark the issue as fresh with . Stale issues rot after an additional 30d of inactivity and eventually close. If this issue is safe to close now please do so with . Send feedback to sig-testing, kubernetes/test-infra and/or . /lifecycle stale\nStale issues rot after 30d of inactivity. Mark the issue as fresh with . Rotten issues close after an additional 30d of inactivity. If this issue is safe to close now please do so with . Send feedback to sig-testing, kubernetes/test-infra and/or ."}
{"_id":"doc-en-kubernetes-af4ac4b9e185e9334a90bdc75507d53790e8bdc2018fef7b4f1e5e30a35b49e6","title":"","text":"/lifecycle rotten\nRotten issues close after 30d of inactivity. Reopen the issue with . Mark the issue as fresh with . Send feedback to sig-contributor-experience at . /close\nClosing this issue. <detailsIn response to : Rotten issues close after 30d of inactivity. Reopen the issue with . Mark the issue as fresh with . /close Instructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details"}
{"_id":"doc-en-kubernetes-81aeaa915020279b8f6f88fad69c7f8c675b68383da3729b9f05c07134ed836b","title":"","text":"This issue is here to satisfy PR requirements of issue association. A lot of times we need to do version bumps for minor bug fixes that don't deserve a separate issue in the kubernetes/kubernetes repo. /kind bug /sig node\n/lifecycle frozen\nall the linked PRs are emrged and no activity since more than 3 years /close\nClosing this issue. <detailsIn response to : all the linked PRs are emrged and no activity since more than 3 years /close Instructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details"}
{"_id":"doc-en-kubernetes-03a1b5e4c6941c765bd244d07f4ddb990e16cb6fe45dcc8cf0eeb5fc4aec846d","title":"","text":"committed a changed version of go into godeps. Was this deliberate? I don't see any discussion of it ( I've almost done this by mistake as well, so I'm inclined to think it was just an oversight. (I think this is giving me errors building, but I'm not yet 100% sure this is the cause)\nHmm, this is a good point. It does look to be accidental. I don't know. I think and would have context though I believe they are out— do you know offhand whether this is probably an accident? I guess were talking specifically about in the .\nI believe it's an accident.\nThanks, The linked PR should fix this (assuming I don't have to do anything else with Godeps).\nThis was almost certainly an accident. Whenever I've updated godeps it's changed automatically to 1.3 and I've had to adjust it back manually.\nActually, jeff lowdermilk was super helpful and showed me how to update the deps on this PR. it was his commit which I merged into my PR, so not sure it it was accidental or not. Fortunately, now that k8petstore is in the go build, we will know if it breaks anything when you revert the version back. Thanks for noticing this. , Alex Robinson wrote:"}
{"_id":"doc-en-kubernetes-1a4f2a91f513677e5b237ae427502f2b96572c4f540bf286aa5c4a8d97d94caf","title":"","text":"If I have probe and then all good. Previous version of the pod is terminated after startup probe success, but if wasn't specified the pod becomes ready immediately. It would be nice to add new state (for example starting) which would have pods until a is successful. Also, it would mean that container not ready for connection. Thank you for attention!\n/sig cluster-lifecycle\n/remove-sig cluster-lifecycle /sig apps try asking in #kubernetes-users on k8s slack.\nthey are silent in the chat. Who is involved in the development of startup probe?\ni think this topic better fits SIG Node. try asking in their channel on slack , but be patient ;) /sig node /remove-sig apps\nThose labels are not set on the issue: <detailsIn response to : i think this topic better fits SIG Node. /sig node /remove-sig apps </details\nI think this is a feature request, not support, so I changed labels. On its face it SEEMS reasonable, but I am not close enough to the details to say. This is definitely sig-node.\nAdding a new pod lifecycle state has been determined to be intractable in the past. Is the proposed change keeping the pod in !Ready until the first success from ?\nWhere do you usually discuss this? Just wondering. Yes, I want that pod considered not ready until will not be successful. If readiness probe doesn't present then a pod can become ready. How can a pod be ready if it is not running?\nhow long wait?\nLooking at this issue, I can confirm it, at least superficially - a pod with a perma-failing is \"ready\" if there is no readinessGate defined and \"not ready\" if it is defined. The question I have is whether we think that is \"correct\" or not? Initially I thought it was correct, but the more I think about it the less convinced I am. If I have defined a startupProbe and it is not passing, my app should not be ready. Does anyone disagree?\nAlso, the app should not be ready in status. It's funny, but without a readiness probe, pods in the final state have the status ready and accept requests. This is unexpected.\nWhether a pod is ready while terminating is an entirely different question, IMO."}
{"_id":"doc-en-kubernetes-3f629dda069dd58c9ce26e66996c26ea4a1a35ac59c48a136ee3c51de61527c3","title":"","text":"Apps that server network traffic probably need to \"drain\" and to do that they need to stay \"ready\" for a while as upstream LBs de-program. Let's not conflate issues, please.\nThe fix seems simple - lets see what CI says.\nSorry, I don't spend much time on slack... you could have checked the release notes in 1.18 ;-) Tim's fix seems good, thanks!\n/reopen Until we merge all related PRs...\nReopened this issue. <detailsIn response to : /reopen Until we merge all related PRs... Instructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\nAll merged. /close\nClosing this issue. <detailsIn response to : All merged. /close Instructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details"}
{"_id":"doc-en-kubernetes-94ebb6d5cc5a883444cc57b7fe675619934dadcea635fbafcba0058503bec1c7","title":"","text":"<!-- Please use this template while reporting a bug and provide as much info as possible. Not doing so may result in your bug not being addressed in a timely manner. Thanks! If the matter is security related, please disclose it privately via --What happened: When service is set with ILB annotation and empty string subnet, e.g. The service would be pending forever with following error: What you expected to happen: ILB should be provisioned successfully. How to reproduce it (as minimally and precisely as possible): Anything else we need to know?: Environment: Kubernetes version (use ): Cloud provider or hardware configuration: OS (e.g: ): Kernel (e.g. ): Install tools: Others: /sig azure /kind bug /assign\n/milestone v1.14 We're entering and this looks relevant to the 1.14 release. Please if I am incorrect\nThanks"}
{"_id":"doc-en-kubernetes-1d9713caefc449b5fa12697f2f529990c13dff1a712a56404b9120eb41d78bc3","title":"","text":"<!-- Please use this template while reporting a bug and provide as much info as possible. Not doing so may result in your bug not being addressed in a timely manner. Thanks! If the matter is security related, please disclose it privately via --When attempting to upgrade kubernetes using kubeadm upgrade apply, kubeadm fails to parse an etcd imagetag containing build metadata in the form of as a valid semver. An example upgrade produces an error similar to: Note this will occur even if the etcd version is unchanged. The +` character. This pattern is used to support custom kubernetes images, for instance kube-apiserver:v1.20.4+patch.0 will be converted to imagetag v1.20.4_patch.0 when running make quick-release-images. Add the following to kubeadm- Install kubernetes. Then upgrade kubernetes with kubeadm upgrade apply. I have tested a fix for this issue and will PR shortly. Kubernetes version (use ): N/A Cloud provider or hardware configuration: N/A OS (e.g: ): N/A Kernel (e.g. ): N/A Install tools: kubeadm Network plugin and version (if this is a network-related bug): N/A Others: N/A\n/sig cluster-lifecycle\n/area kubeadm\n/triage accepted"}
{"_id":"doc-en-kubernetes-a3e5f991085cecca4a970d807dd5157be9d11e288da6daa6aadb8b1d3ef3056a","title":"","text":"In asked the following question. While the user-facing documentation answers this question clearly enough, John was reading the HELP string in the Prometheus scrapes. That thing only says the following. The metric HELP strings should be clear enough. Scrape or read the code (). No response <detailsThis problem started when we generalized to allow width other than 1 and got worse when we introduced borrowing. </details<detailsN/A </details<detailsN/A </details<details</details<details</details<details</details\n/assign\nActually, this metric should be removed. It is a duplicate of but with an inferior name.\nLooks good (Sorry I never got around to logging this issue myself. Have been super busy. And your wording is better anyway :-) )\n/triage accepted Thank you."}
{"_id":"doc-en-kubernetes-7c0443d3746fe3757c9195c5c33416ecd39f97575569d1247bf005f30b3d366d","title":"","text":"<!-- This form is for bug reports and feature requests ONLY! If you're looking for help check and the . If the matter is security related, please disclose it privately via --Is this a BUG REPORT or FEATURE REQUEST?: What happened: I visited the Kubernetes 1.11 release blog post and clicked on the link to the article on resizing persistent volumes. The server returned a 404. What you expected to happen: I expected to navigate to the article on resizing persistent volumes in Kubernetes. How to reproduce it (as minimally and precisely as possible): Visit the release announcement blog post (), scroll down to the \"4 Day Features Blog Series\" section, and click on the link to the article on resizing persistent volumes. Anything else we need to know?: The link currently points to: The correct link is: Environment: Kubernetes version (use ): N/A Cloud provider or hardware configuration: N/A OS (e.g. from /etc/os-release): N/A Kernel (e.g. ): N/A Install tools: N/A Others: N/A\nReiterating the mentions to trigger a notification: <detailsIn response to : <!-- This form is for bug reports and feature requests ONLY! --/kind bug /kind feature What happened: I visited the Kubernetes 1.11 release blog post and clicked on the link to the article on resizing persistent volumes. The server returned a 404. I expected to navigate to the article on resizing persistent volumes in Kubernetes. Visit the release announcement blog post (), scroll down to the \"4 Day Features Blog Series\" section, and click on the link to the article on resizing persistent volumes. The link currently points to: The correct link is: Kubernetes version (use ): N/A Cloud provider or hardware configuration: N/A OS (e.g. from /etc/os-release): N/A Kernel (e.g. ): N/A Install tools: N/A Others: N/A Instructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\nIt should be reported to kubernetes/website, but doesn't matter."}
{"_id":"doc-en-kubernetes-7f95224274155bb39af2e05742451e79cefc1f176d05622797c3ba56ecf46808","title":"","text":"is failing with: Examples:\nThis is the failing line of code...\nPASS: FAIL: Seems like this disappeared: Also this is flipped: we run the following\nmy leading theory is this is something in\nThe issue is that the above PR changed a bunch of to be and thus we no longer wait for getArchive() to complete before starting tests. And unfortunately the previous PR cannot be reverted\nAfter merging, kubelet-gce-e2e-ci still failed with the same issue. reopen the issue. Currently the failure blocked the merge-bot.\ndid stop the bleeding. Now kubelet-gce-e2e-ci is green: and merge-bot is unblocked.\nI think the \"go arc\" piece was a red herring. I suspect the issue is that the same PR tries to copy the binary into the archive directory, but PR stops building ginkgo, so the copy fails. That is why this line then fails..."}
{"_id":"doc-en-kubernetes-f8d5fa52ad40b7a48721555fc6e9ed2d9d20bf180da0265ac5e3a79ed043fc75","title":"","text":"Intermittent failures on Travis.\nHrm, this passes 100/100 locally for me. I'll look at the code and see if I see anything. --brendan"}
{"_id":"doc-en-kubernetes-78c3aac202d1e5d328f496362409af5f5e81f7978b044bd0fb2f29b228ab13af","title":"","text":"We should drop the bit once in favor of a once gcp has a replacement. xref: /assign /milestone v1.24 /triage accepted /sig auth\nThe Kubernetes project currently lacks enough contributors to adequately respond to all issues and PRs. This bot triages issues and PRs according to the following rules: After 90d of inactivity, is applied After 30d of inactivity since was applied, is applied After 30d of inactivity since was applied, the issue is closed You can: Mark this issue or PR as fresh with Mark this issue or PR as rotten with Close this issue or PR with Offer to help out with [Issue Triage][1] Please send feedback to sig-contributor-experience at . /lifecycle stale [1]:\nThe Kubernetes project currently lacks enough active contributors to adequately respond to all issues and PRs. This bot triages issues and PRs according to the following rules: After 90d of inactivity, is applied After 30d of inactivity since was applied, is applied After 30d of inactivity since was applied, the issue is closed You can: Mark this issue or PR as fresh with Close this issue or PR with Offer to help out with [Issue Triage][1] Please send feedback to sig-contributor-experience at . /lifecycle rotten [1]:\n/remove-lifecycle rotten /lifecycle frozen\nI'm in the release 1.24 bug triage shadow. Does this issue still target release 1.24? If yes, would you please update the kind and priority fields for us to track it better?\nYes, definitely targeting 1.24 /priority important-soon /kind cleanup deprecation\nwould you please link the PR that solves this issue and update the status?\nThere is none yet. Waiting on GKE folks.\naddresses this issue."}
{"_id":"doc-en-kubernetes-24df73a6db05c45f0cac4e827cf56a2dc89af2bfb599ee7ed6ea71c4b1f5c613","title":"","text":"Should be able to tell from logs if a minion isn't present because cloud provider didn't tell us about it, or if it's failing health checks."}
{"_id":"doc-en-kubernetes-89b11fc4c25f0b72d5f027a149a029fe0341e3d491bf3200f9e9687686d8fec6","title":"","text":"<!-- Please use this template while reporting a bug and provide as much info as possible. Not doing so may result in your bug not being addressed in a timely manner. Thanks! If the matter is security related, please disclose it privately via --On Windows 20H2 we are observing crashloops with fluent-bit due to a malformed Windows Event Log entry in the Security log. Fluent-bit to not crash loop. Bring up a GCE Windows cluster on Windows 20H2. You may see a loss of logs. Kubernetes version (use ): Cloud provider or hardware configuration: OS (e.g: ): Kernel (e.g. ): Install tools: Network plugin and version (if this is a network-related bug): Others: /sig windows /triage accepted"}
{"_id":"doc-en-kubernetes-7164ee736ec8b7198f38602535de086e7396a736ea12447e5be8a654710fb76e","title":"","text":"Is this a BUG REPORT or FEATURE REQUEST?: /kind cleanup What happened: It seems kubelet did not update inode pressure; and only test codes are using . So prefer to deprecate or remove it directly. (seems no node-misc group :). )"}
{"_id":"doc-en-kubernetes-43bb87e0e7e2bbc9a63f125545e4751ea88971868bf29aa570529777682e8f1f","title":"","text":"Problem Running the e2e density tests on a local cluster or any other cluster where ssh to master doesn't work fails, even though enough pods are created, because of SSH attempts. Specifically \"should allow starting 30\" is the argument to reproduce. Solution Allow density tests to pass even if some meta test operations fail after the fact. cc"}
{"_id":"doc-en-kubernetes-17da90f267168da4485057cb3c70835a09c33080424540be2f99c7292698f22f","title":"","text":"In my CSI controller I have enabled the feature gate HonorPVReclaimPolicy for csi-provisioner. Based on the documentation Finalizers can be on a PersistentVolume to ensure that PersistentVolumes having Delete reclaim policy are deleted only after the backing storage are deleted. The newly introduced finalizers and external- are only to dynamically provisioned volumes. I can see the finalizer set on the PV while creating static Persistent Volume. PV deletion is stuck due to this finalizer apiVersion: v1 kind: PersistentVolume metadata: creationTimestamp: \"2023-11-08T11:39:24Z\" deletionGracePeriodSeconds: 0 deletionTimestamp: \"2023-11-08T11:54:31Z\" finalizers: external- name: csi-infoscale-pv PV should be deleted when PVC is deleted. Persistent Volume yaml apiVersion: v1 kind: PersistentVolume metadata: name: csi-infoscale-pv spec: storageClassName: csi-infoscale-sc-prod persistentVolumeReclaimPolicy: Delete capacity: storage: 2Gi accessModes: ReadWriteOnce csi: driver: org.veritas.infoscale # Please provide pre-provisioned Infoscale volume name. volumeHandle: clust21432/vrtskube_dg-/vol1 fsType: vxfs Persistent Volume Claim apiVersion: v1 kind: PersistentVolumeClaim metadata: name: csi-infoscale-pvc-static spec: accessModes: ReadWriteOnce resources: requests: storage: 2Gi storageClassName: csi-infoscale-sc-prod volumeName: csi-infoscale-pv Same thing works fine on OCP. WARNING: This version information is deprecated and will be replaced with the output from kubectl version --short. Use --output=yaml|json to get the full version."}
{"_id":"doc-en-kubernetes-0ba3f4f18a7007434ced94c47e051d55455d5fb79cd77986fc1ae4d5bc653af8","title":"","text":"Client Version: version.Info{Major:\"1\", Minor:\"27\", GitVersion:\"v1.27.5\", GitCommit:\"\", GitTreeState:\"clean\", BuildDate:\"2023-08-24T00:48:26Z\", GoVersion:\"go1.20.7\", Compiler:\"gc\", Platform:\"linux/amd64\"} Kustomize Version: v5.0.1 Server Version: version.Info{Major:\"1\", Minor:\"27\", GitVersion:\"v1.27.5\", GitCommit:\"\", GitTreeState:\"clean\", BuildDate:\"2023-08-24T00:42:11Z\", GoVersion:\"go1.20.7\", Compiler:\"gc\", Platform:\"linux/amd64\"} Vanilla K8s setup <details</details<details</details<details</details<details</details\nThis issue is currently awaiting triage. If a SIG or subproject determines this is a relevant issue, they will accept it by applying the label and provide further guidance. The label can be by org members by writing in a comment. <detailsInstructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\n/sig storage maybe similar issues should go in\nOpened similar issue in external-provisioner repo also There is no reply on that as well. Can you please help understand the behavior of the feature HonorPVReclaimPolicy.\nIf you want Persistent Volume (PV) to be deleted when the associated Persistent Volume Claim (PVC) is deleted, you donot need to use finalizers explicitly. Kubernetes handles this behavior by default. It is ok that if you use to get the desire output using Finalizer we can do it.\nAs per my understanding Please follow the steps : Add finalizer in metadata of as you did above need to make change in the controller which is handling PV deletion in .go file. See whether the controller administrator is working or not. I think it will work.\nFinalizer is because the feature gate HonorPVReclaimPolicy for csi-provisioner is enabled. We are not adding the finalizer explicitly. From the documentation, \"The newly introduced finalizers and external- are only to dynamically provisioned volumes.\" Not sure why is the finalizer getting set."}
{"_id":"doc-en-kubernetes-592cda3e1f42c2eb1f47fff2bb790922cec52a4f953f170e43c5acb028d04e55","title":"","text":"Also, for deleting PV it requires the annotation \"\" but this annotation is set in case of dynamic PVs. Should it be set explicitly for static PV?\nThis feature is alpha so the feature gate is not enabled by default. Did you enable the feature gate on kube-controller-manager?\nYes, i have enabled this feature.\nexternal- is also applied to statically provisioned volumes. FYI\nHi Which version of the external provisioner is the driver using? Does the driver meet the requirements of the CSI spec on the DeleteVolume PRC? FYI: Is this problem occasional or permanently reproducible? Can you provide the provisioner logs at that time?\n/triage needs-information\ncc\ncc\nExternal provisioner used is : \"csi-provisioner:v3.5.0\" This problem is always reproducible Provisioner logs: I1123 12:22:38. 1 ] handleProtectionFinalizer Volume : &PersistentVolume{ObjectMeta:{csi-infoscale-pv -420a-4e9e-be82- 0 2023-11-23 12:21:18 +0000 UTC <nil<nilmap[] map[{\"apiVersion\":\"v1\",\"kind\":\"PersistentVolume\",\"metadata\":{\"annotations\":{},\"name\":\"csi-infoscale-pv\"},\"spec\":{\"accessModes\":[\"ReadWriteOnce\"],\"capacity\":{\"storage\":\"2Gi\"},\"csi\":{\"driver\":\"org.veritas.infoscale\",\"fsType\":\"vxfs\",\"volumeHandle\":\"clust40559/vrtskubedg-/vol1\"},\"persistentVolumeReclaimPolicy\":\"Delete\",\"storageClassName\":\"csi-infoscale-sc-prod\"}} ] [] [ external-] [{kubectl-client-side-apply Update v1 2023-11-23 12:21:18 +0000 UTC FieldsV1 {\"f:metadata\":{\"f:annotations\":{\".\":{},\"\":{}}},\"f:spec\":{\"f:accessModes\":{},\"f:capacity\":{\".\":{},\"f:storage\":{}},\"f:csi\":{\"."}
{"_id":"doc-en-kubernetes-3ecd202a68dc777f32874cd48d4b39291a9bf67f2771bfe1a1dd4808649976b8","title":"","text":"\":{},\"f:driver\":{},\"f:fsType\":{},\"f:volumeHandle\":{}},\"f:persistentVolumeReclaimPolicy\":{},\"f:storageClassName\":{},\"f:volumeMode\":{}}} } {csi-provisioner Update v1 2023-11-23 12:21:22 +0000 UTC FieldsV1 {\"f:metadata\":{\"f:finalizers\":{\"v:\"external-\"\":{}}}} } {kube-controller-manager Update v1 2023-11-23 12:21:22 +0000 UTC FieldsV1 {\"f:metadata\":{\"f:annotations\":{\"\":{}}},\"f:spec\":{\"f:claimRef\":{\"."}
{"_id":"doc-en-kubernetes-fb611cee6f984fef598fa329474fd3d3abef5728e2a0833d8129f0a5ee9ed92e","title":"","text":"\":{},\"f:apiVersion\":{},\"f:kind\":{},\"f:name\":{},\"f:namespace\":{},\"f:resourceVersion\":{},\"f:uid\":{}}}} } {kube-controller-manager Update v1 2023-11-23 12:22:38 +0000 UTC FieldsV1 {\"f:status\":{\"f:phase\":{}}} status}]},Spec:PersistentVolumeSpec{Capacity:ResourceList{storage: {{ 0} {<nil} 2Gi BinarySI},},PersistentVolumeSource:PersistentVolumeSource{GCEPersistentDisk:nil,AWSElasticBlockStore:nil,HostPath:nil,Glusterfs:nil,NFS:nil,RBD:nil,ISCSI:nil,Cinder:nil,CephFS:nil,FC:nil,Flocker:nil,FlexVolume:nil,AzureFile:nil,VsphereVolume:nil,Quobyte:nil,AzureDisk:nil,PhotonPersistentDisk:nil,PortworxVolume:nil,ScaleIO:nil,Local:nil,StorageOS:nil,CSI:&CSIPersistentVolumeSource{Driver:org.veritas.infoscale,VolumeHandle:clust40559/vrtskubedg-/vol1,ReadOnly:false,FSType:vxfs,VolumeAttributes:map[string]string{},ControllerPublishSecretRef:nil,NodeStageSecretRef:nil,NodePublishSecretRef:nil,ControllerExpandSecretRef:nil,NodeExpandSecretRef:nil,},},AccessModes:[ReadWriteOnce],ClaimRef:&ObjectReference{Kind:PersistentVolumeClaim,Namespace:default,Name:csi-infoscale-pvc-static,UID:-59a5-4f02-83ac-,APIVersion:v1,ResourceVersion:,FieldPath:,},PersistentVolumeReclaimPolicy:Delete,StorageClassName:csi-infoscale-sc-prod,MountOptions:[],VolumeMode:Filesystem,NodeAffinity:nil,},Status:PersistentVolumeStatus{Phase:Released,Message:,Reason:,},} I1123 12:22:38. 1 ] delete \"csi-infoscale-pv\": started I1123 12:22:38. 1 ] GRPC call: /csi.v1.Controller/DeleteVolume I1123 12:22:38. 1 ] GRPC request: {\"volumeid\":\"clust40559/vrtskubedg-/vol1\"} I1123 12:22:39."}
{"_id":"doc-en-kubernetes-21563431fea9ed5a14c1fde8160d9fcab633c35237ee882cf06625b1768a9487","title":"","text":"1 ] GRPC response: {} I1123 12:22:39. 1 ] GRPC error: <nilI1123 12:22:39. 1 ] delete \"csi-infoscale-pv\": volume deleted I1123 12:22:39. 1 ] delete \"csi-infoscale-pv\": failed to remove finalizer for persistentvolume: Operation cannot be fulfilled on persistentvolumes \"csi-infoscale-pv\": the object has been modified; please apply your changes to the latest version and try again W1123 12:22:39. 1 ] Retrying syncing volume \"csi-infoscale-pv\", failure 0 E1123 12:22:39. 1 ] error syncing volume \"csi-infoscale-pv\": Operation cannot be fulfilled on persistentvolumes \"csi-infoscale-pv\": the object has been modified; please apply your changes to the latest version and try again I1123 12:22:39. 1 ] handleProtectionFinalizer Volume : &PersistentVolume{ObjectMeta:{csi-infoscale-pv -420a-4e9e-be82- 0 2023-11-23 12:21:18 +0000 UTC 2023-11-23 12:22:38 +0000 UTC 0xc0007f63d0 map[] map[{\"apiVersion\":\"v1\",\"kind\":\"PersistentVolume\",\"metadata\":{\"annotations\":{},\"name\":\"csi-infoscale-pv\"},\"spec\":{\"accessModes\":[\"ReadWriteOnce\"],\"capacity\":{\"storage\":\"2Gi\"},\"csi\":{\"driver\":\"org.veritas.infoscale\",\"fsType\":\"vxfs\",\"volumeHandle\":\"clust40559/vrtskubedg-/vol1\"},\"persistentVolumeReclaimPolicy\":\"Delete\",\"storageClassName\":\"csi-infoscale-sc-prod\"}} ] [] [external-] [{kubectl-client-side-apply Update v1 2023-11-23 12:21:18 +0000 UTC FieldsV1 {\"f:metadata\":{\"f:annotations\":{\".\":{},\"\":{}}},\"f:spec\":{\"f:accessModes\":{},\"f:capacity\":{\".\":{},\"f:storage\":{}},\"f:csi\":{\"."}
{"_id":"doc-en-kubernetes-11586616fd718035f7c04dbcead8d898c73d9e9840ec80b2b32382f64622ed92","title":"","text":"\":{},\"f:apiVersion\":{},\"f:kind\":{},\"f:name\":{},\"f:namespace\":{},\"f:resourceVersion\":{},\"f:uid\":{}}}} } {kube-controller-manager Update v1 2023-11-23 12:22:38 +0000 UTC FieldsV1 {\"f:status\":{\"f:message\":{},\"f:phase\":{}}} status}]},Spec:PersistentVolumeSpec{Capacity:ResourceList{storage: {{ 0} {<nil} 2Gi BinarySI},},PersistentVolumeSource:PersistentVolumeSource{GCEPersistentDisk:nil,AWSElasticBlockStore:nil,HostPath:nil,Glusterfs:nil,NFS:nil,RBD:nil,ISCSI:nil,Cinder:nil,CephFS:nil,FC:nil,Flocker:nil,FlexVolume:nil,AzureFile:nil,VsphereVolume:nil,Quobyte:nil,AzureDisk:nil,PhotonPersistentDisk:nil,PortworxVolume:nil,ScaleIO:nil,Local:nil,StorageOS:nil,CSI:&CSIPersistentVolumeSource{Driver:org.veritas.infoscale,VolumeHandle:clust40559/vrtskubedg-/vol1,ReadOnly:false,FSType:vxfs,VolumeAttributes:map[string]string{},ControllerPublishSecretRef:nil,NodeStageSecretRef:nil,NodePublishSecretRef:nil,ControllerExpandSecretRef:nil,NodeExpandSecretRef:nil,},},AccessModes:[ReadWriteOnce],ClaimRef:&ObjectReference{Kind:PersistentVolumeClaim,Namespace:default,Name:csi-infoscale-pvc-static,UID:-59a5-4f02-83ac-,APIVersion:v1,ResourceVersion:,FieldPath:,},PersistentVolumeReclaimPolicy:Delete,StorageClassName:csi-infoscale-sc-prod,MountOptions:[],VolumeMode:Filesystem,NodeAffinity:nil,},Status:PersistentVolumeStatus{Phase:Failed,Message:error getting deleter volume plugin for volume \"csi-infoscale-pv\": no deletable volume plugin matched,Reason:,},} I1123 12:22:40."}
{"_id":"doc-en-kubernetes-372f094f1905cb03e404d407c686a802e84a00b41a059cf5fa90382ee5601de2","title":"","text":"1 ] handleProtectionFinalizer Volume : &PersistentVolume{ObjectMeta:{csi-infoscale-pv -420a-4e9e-be82- 0 2023-11-23 12:21:18 +0000 UTC 2023-11-23 12:22:38 +0000 UTC 0xc0007f63d0 map[] map[{\"apiVersion\":\"v1\",\"kind\":\"PersistentVolume\",\"metadata\":{\"annotations\":{},\"name\":\"csi-infoscale-pv\"},\"spec\":{\"accessModes\":[\"ReadWriteOnce\"],\"capacity\":{\"storage\":\"2Gi\"},\"csi\":{\"driver\":\"org.veritas.infoscale\",\"fsType\":\"vxfs\",\"volumeHandle\":\"clust40559/vrtskubedg-/vol1\"},\"persistentVolumeReclaimPolicy\":\"Delete\",\"storageClassName\":\"csi-infoscale-sc-prod\"}} ] [] [external-] [{kubectl-client-side-apply Update v1 2023-11-23 12:21:18 +0000 UTC FieldsV1 {\"f:metadata\":{\"f:annotations\":{\".\":{},\"\":{}}},\"f:spec\":{\"f:accessModes\":{},\"f:capacity\":{\".\":{},\"f:storage\":{}},\"f:csi\":{\".\":{},\"f:driver\":{},\"f:fsType\":{},\"f:volumeHandle\":{}},\"f:persistentVolumeReclaimPolicy\":{},\"f:storageClassName\":{},\"f:volumeMode\":{}}} } {csi-provisioner Update v1 2023-11-23 12:21:22 +0000 UTC FieldsV1 {\"f:metadata\":{\"f:finalizers\":{\"v:\"external-\"\":{}}}} } {kube-controller-manager Update v1 2023-11-23 12:21:22 +0000 UTC FieldsV1 {\"f:metadata\":{\"f:annotations\":{\"\":{}}},\"f:spec\":{\"f:claimRef\":{\"."}
{"_id":"doc-en-kubernetes-8654731eb8139b66a1cecad4a317b2078534f6977aa1c7223978f7fff83e9392","title":"","text":"\":{},\"f:apiVersion\":{},\"f:kind\":{},\"f:name\":{},\"f:namespace\":{},\"f:resourceVersion\":{},\"f:uid\":{}}}} } {kube-controller-manager Update v1 2023-11-23 12:22:38 +0000 UTC FieldsV1 {\"f:status\":{\"f:message\":{},\"f:phase\":{}}} status}]},Spec:PersistentVolumeSpec{Capacity:ResourceList{storage: {{ 0} {<nil} 2Gi BinarySI},},PersistentVolumeSource:PersistentVolumeSource{GCEPersistentDisk:nil,AWSElasticBlockStore:nil,HostPath:nil,Glusterfs:nil,NFS:nil,RBD:nil,ISCSI:nil,Cinder:nil,CephFS:nil,FC:nil,Flocker:nil,FlexVolume:nil,AzureFile:nil,VsphereVolume:nil,Quobyte:nil,AzureDisk:nil,PhotonPersistentDisk:nil,PortworxVolume:nil,ScaleIO:nil,Local:nil,StorageOS:nil,CSI:&CSIPersistentVolumeSource{Driver:org.veritas.infoscale,VolumeHandle:clust40559/vrtskubedg-/vol1,ReadOnly:false,FSType:vxfs,VolumeAttributes:map[string]string{},ControllerPublishSecretRef:nil,NodeStageSecretRef:nil,NodePublishSecretRef:nil,ControllerExpandSecretRef:nil,NodeExpandSecretRef:nil,},},AccessModes:[ReadWriteOnce],ClaimRef:&ObjectReference{Kind:PersistentVolumeClaim,Namespace:default,Name:csi-infoscale-pvc-static,UID:-59a5-4f02-83ac-,APIVersion:v1,ResourceVersion:,FieldPath:,},PersistentVolumeReclaimPolicy:Delete,StorageClassName:csi-infoscale-sc-prod,MountOptions:[],VolumeMode:*Filesystem,NodeAffinity:nil,},Status:PersistentVolumeStatus{Phase:Failed,Message:error getting deleter volume plugin for volume \"csi-infoscale-pv\": no deletable volume plugin matched,Reason:,},}\nThank you I will try to fix it. /remove-triage needs-information /assign"}
{"_id":"doc-en-kubernetes-4915683bef189ef7cc4e0ae9494c2fe05f18b93b7b90a4cf05226bf1e9ad07a8","title":"","text":"<!-- Please only use this template for submitting enhancement requests --What would you like to be : and functions need unit tests Why is this needed: /kind test /sig cloud-provider /area provider/azure /assign\nThe label(s) cannot be applied, because the repository doesn't have them <detailsIn response to : What would you like to be : /sig cloud-provider /area provider/azure Instructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\nrelated to\ncc\n/good-first-issue\nThis request has been marked as suitable for new contributors. Please ensure the request meets the requirements listed . If this request no longer meets these requirements, the label can be removed by commenting with the command. <detailsIn response to : /good-first-issue Instructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\nHi I'd like to give this one a go, if no one is currently working on it.\nsure, go ahead, thanks\nI pushed some first tests for"}
{"_id":"doc-en-kubernetes-cd494847d4bddf9ea80027a9691be245174d67d67dcf4a48743332a9d199dc09","title":"","text":"This was the root cause of rancher/rancher. In , when is called, it is given the oldVersion of an object and the newVersion of the object. It then uses the indexers to calculate the old values (from the old version) and the new updated values. In cases where are size of 1, are of size greater than 1, and contain the sole value from , the key will be deleted for the old value will be deleted and won't be re-. This is due to the two for loops at the end of the method which are used for optimization. When the first for loop is processed, . Because of this, the key is deleted from the index for this value. In the second for loop, the , and for at least one of the values . The continue statement causes the key to not be to the index for this value (and since the value was deleted in the prior for loop, it no longer exists in the index for this value), despite the fact that the indicated that this is the desired state. Since specified that this value was still valid for this object, the key should not be deleted from the index for this value, or it should be re- when it is re-discovered in . Either way, the end state of the index should include this key in the set for the specified value. I've created a which demonstrates the issue (branch is ). To use it (assuming that you have a running kubernetes cluster to use with): the controller: . the foo crd: the controller (assumes that you have $KUBECONFIG set to the file path of your kubeconfig file). an example foo with only one label: . Observe, in the logs of the running sample-controller, that the indexer was able to locate this foo using the indexer. the foo to include a second label: . Observe, in the logs of the running sample-controller, that the indexer was not able to locate the foo using the first key any longer, despite the fact that the label was still on the object. The sample-controller fork adds an indexer for foo objects which uses the indexerFunction of . This function looks for labels that have a key with a prefix of and uses their values as index keys for the foo objects. For example, if I have a with a label of , the indexer will allow you to retrieve using ."}
{"_id":"doc-en-kubernetes-582c669d9e852f593828621e1df0588db01f01d305d2c20cc4978aeea726760c","title":"","text":"The controller's method has been altered to just ensure that the foos can be retrieved using the indexer for all valid labels. If the controller can't retrieve foo by using the value from a valid label, then it returns an error (it tries for all valid labels). This was originally discovered in a fork of client-go that we maintain. After we reverted the changes introduced by , our issues stopped occurring. N/A - This was introduced in and appears to be in all versions of the client v0.23+. N/A - issue is with client-go No response No response No response No response\n/sig api-machinery\nI think that you have background in this issue/area of the code. Would you mind taking a look at this issue and the PR that I have up to fix it? Please let me know if I've misunderstood anything.\n/assign Would you mind to take a look? Thank you /triage accepted\nThanks for reporting the issue. will address this issue\nReopening - since we need to cherrypick back to 1.23\nCherrypick is merged - closing."}
{"_id":"doc-en-kubernetes-598a5e6d79587d97464d296fe938aac9741dc183fde75d2da349cdf55e566259","title":"","text":"What happened: node-local-dns caches negative queries, causing things that do polling for orchestration to fail (like operators with StatefulSets). What you expected to happen: Have a minimal TTL for negative responses (alternatively just disable caching by setting the cache size to 0) that follows what was done for kube-dns: How to reproduce it (as minimally and precisely as possible): With node-local-dns cache deployed, in my cluster it was the that failed to create clusters because the cached entry didn't expire before it had exhausted all of it's retries. Environment: Kubernetes version (use ): 1.11.6 Cloud provider or hardware configuration: GKE OS (e.g. from /etc/os-release): any Kernel (e.g. ): any\n/sig network"}
{"_id":"doc-en-kubernetes-7f1e6e7766f66194a4a76f7bcbc82118fdef24c779ae808892b54bf44f2870dc","title":"","text":"<!-- Please use this template while reporting a bug and provide as much info as possible. Not doing so may result in your bug not being addressed in a timely manner. Thanks! If the matter is security related, please disclose it privately via --E2E test fails. The test is implemented at: I would expect the test pass with these objects in place: AFAIK this fulfills the general requirement of the test that there's a corresponding endpointslice object with same addressing information as is on the endpoint object. IMO the test should rather look for epslice which is owned by the endpoint rather than relying on a hardcoded name. Add flag to kube-apiserver Manually create the endpoint with proper addressing for the api-server epslice \"mirroring\" create new endpointslice with name like --The test fails as it's hardcoded to look for endpointslice named I'm working on a PR to provide a fix proposal Kubernetes version (use ): 1.22.1 Cloud provider or hardware configuration: OS (e.g: ): Kernel (e.g. ): Install tools: k0s Network plugin and version (if this is a network-related bug): kube-router 1.3.1 Others:\n/sig testing\n/remove-sig testing /sig network EndpointSlices are owned by sig network.\n/kind failing-test\n/remove kind-bug /remove kind-failing-test The test is correct, apiserver must generates both endpoint and enpointslices with the special name \"kubernetes\" , that are the backends for the \"kubernetes.default\" service with the IP addresses of the apiservers. This special service \"kubernetes.default\" and its endpoints are used by clients with In-Cluster auto discovery. I think that whatever cluster that is not generating the endpoint and endpointslices for the \"kubernetes.default\" service should not meet Conformance, at least I can't see how automatic in-cluster configuration will work without that ... I only see option used for webhook tests, the other 2 reconcilers always create the endpoints and slices for \"kubernetes.default\" should we clarify the goal of this \"none\" option?\nYes, I do understand this. And in our case this is true, there is a svc called as there is also endpoints object called . This makes the in-cluster discovery work."}
{"_id":"doc-en-kubernetes-e7b624ebdf04c04b0fd86ae888848c4f91744879af316f1243574b67dd031473","title":"","text":"Otherwise we'd get much more errors in the conformance instead of this only one... :) In our case we're relying on the endpointslice mirroring to take care of creating the endpointslice object, which makes this test fail as it uses generated \"suffix\" on the object name. I have not dived too deep (yet) into the internals of endpointslices but I fail to see how the endpointslice object name really matters when both svc and endpoint object does have proper name.\noh, this seems interesting then ... who is creating the corresponding endpoint? can you paste a kubectl get endpoints -o yaml?\nthis is how it looks like in my cluster\nIn our case (k0s distro) it's our custom controller that is triggered when the controlplane is configured in HA mode. So this basically sets the api to mode . When our controller runs it has a configured LB address which it resolves to IP address(es) and creates the corresponding EP object. We do NOT create the endpointslice as we rely on the mirroring functionality to let it create and sync it. Here's the object details: Reading the code for API built-in endpoint reconcilers I see they set and create the endpointslice object by themselves. Yes, we could do that too in k0s but I fail to see why the name of the endpointslice object really matters at all.\nI agree the conformance test should not be looking up endpointslice by name. It should be using a label selector of to locate one or more endpointslices and verifying the ports and endpoints across them.\nFair enough, on the good side this also will simplify the KEP from to dual-stack the apiserver you are the expert in this area, can you PTAL at this discussion? /kind failing-test\nhmm the test description explicitly says the name of the object has to be \"kubernetes\", if the name is mandatory the test is correct, otherwise we need to rewrite the test description and talk with conformance people /area conformance\nthe description should be rewritten... the test is enforcing stricter requirements on endpointslices than required/expected\n/triage accepted"}
{"_id":"doc-en-kubernetes-91b918bf4349bc4d86d996ff7ec289a620469d416a95bd3acf270a016c4f134a","title":"","text":"Once e2e tests are running, the tests keep running even if I press and exit the process that controls the TTY. But the tests actually keep running in the background and they keep printing stuff on a shell prompt which has already returned and thereby clobbering the STDOUT\n/assign\nGitHub didn't allow me to assign the following users: oomichi. Note that only can be assigned. <detailsIn response to : /assign Instructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details"}
{"_id":"doc-en-kubernetes-937f1cb17c20e5fd2b4a1f81201690c3dbe6990ef3c1bc74ff71daaa02b6db56","title":"","text":"Solution: I will add one.\n/cc\nxref\nThanks for sending a PR so quickly!\n[MILESTONENOTIFIER] Milestone Issue Needs Attention Action required: During code slush, issues in the milestone should be in progress. If this issue is not being actively worked on, please remove it from the milestone. If it is being worked on, please add the label so it can be tracked with other in-flight issues. Note: If this issue is not resolved or labeled as by Wed, Nov 22 it will be moved out of the v1.9 milestone. <details<summaryIssue Labels</summary- : Issue will be escalated to these SIGs if needed. : Escalate to the issue owners and SIG owner; move out of milestone after several unsuccessful escalation attempts. : Fixes a bug discovered during the current release. </details<details<summaryHelp</summary<ul<li<a href=\"\"Additional instructions</a</li<li<a href=\"\"Commands for setting labels</a</li</ul</details"}
{"_id":"doc-en-kubernetes-d2ad17ee33cbb9dd0cc3cd8888bfcb21e558f50039e64e1a5ba0cdc518d0afba","title":"","text":"During OPNFV test execution we noticed the folder selinux context changed. e2e test log How the folder changed after the test executed. I expected the TC won't change the selinux context. The question here, should this sigstorageserial TC set work on an selinux enabled K8S cluster or not? This question valid for the another sigstorage TC set with several test case, where selinux block a lot of TC too. See above. No response_ <details</details<detailsOpenStack </details<details</details<details</details<details</details<details</details\n/sig testing\n/sig storage\nThe Kubernetes project currently lacks enough contributors to adequately respond to all issues. This bot triages un-triaged issues according to the following rules: After 90d of inactivity, is applied After 30d of inactivity since was applied, is applied After 30d of inactivity since was applied, the issue is closed You can: Mark this issue as fresh with Close this issue with Offer to help out with [Issue Triage][1] Please send feedback to sig-contributor-experience at . /lifecycle stale [1]:\n/remove-lifecycle stale\n/assign\nReproducer with simple kubernetes/kubernetes artifacts:\n/triage accepted"}
{"_id":"doc-en-kubernetes-a63e01ebe6efecf34b1eda7468cd7472762d7f902ee9117cf6f0f2ff92b9f93a","title":"","text":"Forking from To summarize, here's what we observed: daemonset calls coming every minute from both fluentd-scaler and addon-manager (verified this by individually turning them on/off). Things we need to understand here: Is the daemonset object continuously toggling b/w 2 states? (we know here that it's RV is increasing continuously) If yes, what field(s) in the object are changing? IIRC value of some label/annotation (I think 'UpdatedPodsScheduled') is changing (probably related to 2 below) Also, why/should the fluentd-scaler send any api request if the resources are already set to the right value? pods are getting deleted and recreated by daemonset-controller when the scaler is enabled (as was also seen in ). Why this is happening? One thing to note here - all those delete calls are preceded by calls from respective kubelets (but maybe that's expected). cc /priority critical-urgent /assign\nMarked this as release-blocker as at least some part of it seems like a correctness issue (i.e fluentd pods dying and getting recreated). Note also that this is causing increased mem-usage with daemonset-controller (depicted here - )\nThanks for filing this! The only thing that was changing (apart from ds/fluentd-gcp status) was the resource version.\nI modified fluentd scaler to use --dry-run in to check whether there would be a change, but this still keeps on updating, even though there is no change. Apparently, the logic inside kubectl is broken...\nBtw, addon manager sends tons of PATCH requests for all resources every minute anyway, so even though I will fix fluent-scaler not to send them, it won't solve everything.\nShould this issue actually be closed? It was auto-closed by the bot, but nobody has confirmed that the actual issue is fixed.\nI'll let to confirm (looks like he'll be back on Monday)\n/status in-progress\nYes, we should keep this issue open until we fully understand what's going wrong and fix the root-cause. I believe much of the discussion wrt this is happening under related issue\nACK. In progress ETA: 20/03/2018 Risks: May delay the release if not satisfactorily resolved\nNeed a number of successive green runs, then close."}
{"_id":"doc-en-kubernetes-2efbc26b22516e4405a8d5724e6f4faec26f16538d69327dd97918ad592fbe81","title":"","text":"So I took a look and seems like the fluentd-scaler version bump (which was supposed to fix) was merged b/w runs and of our . However, I don't see any visible improvement in controller-manager memory usage around that time: !\nAlso, I'm noticing such lines occasionally (once every 5-10 secs in the start and once every ~1min later on) from of latest run: And back when the fluentd-scaler wasn't doing anything useful (i.e before ), the had such lines only sparingly.\nNow it isn't even sending requests (other than GET) to the apiserver. How do you generate that graph? Is it 99th percentile of memory usage over time? Or what exactly? One difference after introducing fluentd-gcp-scaler is that now there is a change occurring on the DaemonSet (once, but still). Also, what test is failing because of this? I'd like to take a look into the logs.\nAnswering your questions: I fetched the ResourceUsageSummaries for density test from historical runs (e.g ) and then obtained the 99% mem-usage for controller-manager for each run. Yes. More precisely, we're collecting mem-usage from controller-manager once every 1 min (or sth like that) and then taking 99th %ile of the resulting sample. Density and load tests. Where we're checking that mem usage of controller-manager is less than the given threshold (i.e 500MB). Ref: The tests are occasionally failing due to just crossing the threshold. Overall, the mem-usage had seen an increase back during the scaler change (see graph above). The job is this - https://k8s- Inside , you'll find all necessary logs (also a file named which has the resource-usage data).\nThe fluentd-scaler update eliminated fluentd restarts problem. Verified from controller-manager logs that we no longer see such logs that we were : This is great, and I believe that moves this out of release-blocker as we no longer face problems I mentioned in\nThat said, and me will continue looking into increase in mem-usage (which is not that serious). We'll start by verifying that turning the scaler off improves mem-usage.\nFeel free to override me if you have a good reason :)\nI agree that this doesn't sounds to be release-blocker."}
{"_id":"doc-en-kubernetes-ad76c0c3dd106ecff60d8298c8700ef88a9f32d4985b54e29d57ef6ad7351ef8","title":"","text":"But I would like to see follow-ups at least for issues like: does this memory increase come from? what was claimed to be \"no-op\" changes for daemon set were causing problems [is that issues in kubectl, apiserver, somewhere else?] Something like that may hit us again in the (near?) future if we won't understand it now. The current solution is really volatile [if there is a bug somewhere we shouldn't workaround it on our side, but instead have this fixed] at the cluster start (maybe it's fine to leave as is, but I would like to see a discussion of that and a conscious decision being made here)\nDefinitely agree those are questions we need to answer.\nYup. 1) is tested in . 2) I believe apiserver sends watch events even when only resource version changes, which causes the daemonset controller to recreate the pods. So either: resource version shouldn't change when there is no change anywhere else in the resource apiserver shouldn't send watch events when only resource version is changing daemonset should check whether watch event it got is actually changing anything 3) Isn't that an inherent part of the way addon manager works? It creates all sorts of resources, some dependent on the other (e.g. cluster roles and cluster role bindings), all at once. If some dependency is not satisfied, it will get fixed during the next run.\nThis is the code that is supposed to protect us from situation that something must change when we write object to etcd (resource version is not stored in etcd): So there should never be a situation that only resource version is changing. If it is, there is bug and need to be fixed. - FYI Reconciliation is a different thing that creating something in one state only to change that to a different state by a different component. I'm not saying this is for sure wrong approach. But I would like to see a conscious decision process discussing pros-and-cons of that.\nThat's what I'd expect as well. This I'm not sure if should be correct behaviour as IIUC watchers need to receive every RV update for correctness reasons ( ) Maybe. But are we sure if RV is just increasing without any object change? For creation of addons, yes (that race will likely last only for starting few seconds)."}
{"_id":"doc-en-kubernetes-87ee63a141c821e91ed597807f3f6830646e2e9b40e1991334df493e9d39b5a9","title":"","text":"But this is wrt addon changing cluster resources, which can potentially create race at arbitrary time after cluster start (though your fix may have gotten rid of this).\nOk.. Seems like I missed Wojtek's comment above.\nSo I just checked the ds-controller code and it seems like we're updating the ds status only if this condition is being violated: Which means one of those fields is probably mismatching. I'll try to add some logs and run the test at --v=4 to get more info.\nanything is possible, but we do have tests ensuring this works properly (no-op updates do not actually result in an update+resourceVersion bump in etcd):\nSo I ran some 100-node experiments against with more debug logs and I think I now understand the reason for this. The controller-manager logs and mainly show the following difference: This logs show-up when the scaler is active. And these pod deletions are happening as part of rolling-upgrade of daemonset to newer version (the one having resources set). You can see they're getting deleted one-by-one, which also explains why it takes so long (and so are happening even 30 mins after the cluster creation). And the reason why they're deleted one-by-one is because ds-controller has a rolling-upgrade strategy where by default. Proof: And the increased mem-usage FMU is because of the rolling-upgrade performed by ds-controller (over the prolonged period) - as it's the only difference I observe from logs Further, the reason why the RV was increasing was because of few fields 'NumUnavailable', 'NumAvailable', etc fluctuating while the rolling-upgrade\nI spoke offline with and suggested that we should increase maxUnavailable for the fluentd daemonset to a large enough value so that we're not bottlenecked by it. My reasoning is: fluentd is not some user workload which will have some service downtime in case there are many unavailable at once fluentds on nodes are independent of each other (i.e one fluentd shouldn't need to wait for another node's fluentd to come up) we should try to make the scaling happen as fast as possible - otherwise it may be too late before they cope-up with increased log-load in the cluster I'm going to make that change and test it against my PR (thanks for pointing that we can change maxUnavailable directly in ds config)."}
{"_id":"doc-en-kubernetes-90b3139f2f371dfe70c636d4111a7217036a6164fbce02b650a75f688275a9fd","title":"","text":"So increasing maxUnavailable indeed seems to help (ref: ): All the 101 replicas were deleted at once instead of one-by-one and happened pretty much within 2 mins of the original fluentd-ds seen by the controller: We should get that PR into 1.10.\nThanks for confirming this! Did increasing maxUnavailable also reduce c-m memory usage?\nSo the run was still over the threshold, but I was expecting it as I ran c-m with increased verbosity (which ). That said, I'll observe how the usages are once that PR gets in - and FMU it's still possible to have higher values due to the rolling-upgrade potentially happening during the test.\nIs there anything else to do here or should this be closed now?\nLet's close it once we're sure there are no more such failures. We may also need to bump the threshold. On Mon, Mar 26, 2018, 9:46 AM Kubernetes Submit Queue < :\nSo bumping maxUnavailable for fluentd rolling-upgrade in is causing couple of issues: Our latest 5k-node performance failed as part of BeforeSuite where we're checking that ~all kube-system pods are running or ready. The reason is because quite some fluentd pods were killed as part of the rolling-upgrade and we timed out before all those pods could come up Our 100-node performance runs due to high CPU/mem-usage of c-m observed during density test, in cases when the fluentd upgrade happens during the test - which is kind of expected So at this point, we might want to choose among one of the 2 options: Rollback , which comes at the cost of tradeoffs I explained in Increase our CPU/mem thresholds to eliminate those flakes in 100-node job and increase the \"max-allowed-not-ready-pods\" in kube-system ns to a higher value, so that our 5k-node test doesn't fail in BeforeSuite Thoughts?\nTBH, I don't know why it was merged without waiting for opinions... I really vote for reverting that. We don't want this - this may mean making cluster \"not fully usable\" from user perspective.Which is not really what we want to test.\nSGTM. Let's revert it and increase our thresholds for the time-being. The discussion about changing fluentd rolling-upgrades can continue to happen in the background."}
{"_id":"doc-en-kubernetes-59f7438c09573642bfc1d12812df1bf9a90b62da276e49f75dd1387c9fcb0e09","title":"","text":"[MILESTONENOTIFIER] Milestone Issue Needs Approval Action required: This issue must have the label applied by a SIG maintainer. If the label is not applied within 5 days, the issue will be moved out of the v1.11 milestone. <details<summaryIssue Labels</summary- : Issue will be escalated to these SIGs if needed. : Escalate to the issue owners and SIG owner; move out of milestone after several unsuccessful escalation attempts. : Fixes a bug discovered during the current release. </details<details<summaryHelp</summary<ul<li<a href=\"\"Additional instructions</a</li<li<a href=\"\"Commands for setting labels</a</li</ul</details\nThis issue is now fixed after my above PRs. We're having continuous for the job"}
{"_id":"doc-en-kubernetes-d3815452ec6a4b65682bad0d5bef0b8e8ccaebedbf3ea598534c49251f831668","title":"","text":"In and , I started seeing these flakes. The job in question has not been very flaky before so this is odd. In and , I started seeing these flakes. The job \"Job should not create pods when created in suspend state\" is the one that is flaking. Triage says a few days ago. No response No response /sig apps\nDid some digging and I did not see any changes that could account for an increase of flakiness in the job controller code. There was a change made to the e2e tests: The logs say: The code is pretty simple so I wonder if we are hitting a case where sometimes the suspend condition is not set and we fail. Maybe we could add a consistent call here?\n/triage accepted /priority important-soon\n/cc\n/reopen /assign do you mind monitoring the CI state for 1 or 2 days to confirm your PR really fixed it ?\nReopened this issue. <detailsIn response to : /reopen Instructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\nIt makes sense to me.\ncc I think this issue has been fixed. Can we close it?\n/close Thanks for the fix. The test is not flaky now.\nClosing this issue. <detailsIn response to : Instructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details"}
{"_id":"doc-en-kubernetes-2c221590d5e1ff106fb9cb8927cfb7dd0bc0677f71da23dc8e18ac35421ec6b8","title":"","text":"We're seeing an issue with large memory growth in NoExecuteTaintManager when pods with tolerations are rapidly created/deleted. The item to the workqueue contains the full pod objects received from the informer: That means those pod objects are retained until they are processed. Additionally, it looks like the processing loop never removed the keys from the workqueue when it was done. It also means that multiple updates for the same pod result in multiple items in the workqueue, rather than enqueuing the pod to be processed once, like other controllers do. Simplifying this controller to enqueue pod updates only by name/namespace, and operate on the latest pod obtained from the store (like other controllers do) would speed up the processing and relieve the memory burden of the workqueue. /sig scheduling /cc\n/assign\nThanks for finding this Jordan - looking into your fixes.\nthe memory issue was fixed by /close"}
{"_id":"doc-en-kubernetes-91422af253157e56510335693740251e17b7c26d119e77363e85ed4155980546","title":"","text":"<!-- This form is for bug reports and feature requests ONLY! If you're looking for help check and the . If the matter is security related, please disclose it privately via --Is this a BUG REPORT or FEATURE REQUEST?: What happened: fsgroup is only intended for RWO volumes because it recursively changes the permissions of all the directories in the volume. This can cause problems for nfs-type volumes because multiple pods could access the same volume and have the permissions changed out from underneath. What you expected to happen: Determine whether or not the plugin supports fsgroup either via: access mode. This is still a bit ambiguous for nfs volume types, that can theoretically support rwo, especially since Kubernetes does not do any sort of access mode enforcement. a new csi capability\nCC\nAre you saying that CSI internal driver should check driver capability for RWO, if supported, only then apply value ? What would the be ?\nYes one possible way is to infer fsgroup is supported for all rwo volumes, which may not be true? Like a nfs volume technically supports rwo but not fsgroup. So the other option is to make it a new CSI capability. I'm not sure exactly sure how best to describe it though, maybe something about volume ownership management.\nWill it work if is not set then don't chown the ownership? I would think that for shared file system types(nfs, glusterfs) - won't be set. For raw block storage too - I think we shouldn't have to do this.\nwhat do you mean by ? Shouldnt have to solve this issue ?\nI meant - which is obvious. I just meant that the entire mechanism of changing ownership of files should apply only to volume types with block storage filesystems. I am just wondering if we can get away with detecting fsType and not have to introduce new capability in CSI.\nagreed, do not map well to current csi.Capabilities enums which makes it hard to determine RWO when mounting. Perhaps, instead of a new capability, we revisit the existing capabilities to see if there are any way RWO can be reliably inferred from returned capabilities."}
{"_id":"doc-en-kubernetes-e4259f265013a5d1d3262148ae5650fa802161db5ab8c8ccc1d210e68217a7fd","title":"","text":"Currently, this is what the code is doing which we knew needed to be revisited:\nUsers may not fully understand what fsgroup is and set it anyway, or a default psp with some fsgroup may be defined in which it will automatically be set on all pods. The first case we can attribute to user error, but the second case may not have a good workaround and could cause permission issues on rwx volumes. So I think we do need to solve this.\nYes, agreed this has to be solved. As you pointed out user error or improperly configured PSP can set value that does not match the correct attribute, causing permission issues. Here is what I think can be done without new CSI capabilities: , then do not apply fsgroup then: , apply fsgroup , do not apply fsgroup to void stated issue then: , apply fsgroup I think that may help deduce when to apply fsgroup.\nIf mount reference count <= 1, apply fsgroup If mount reference count 1, do not apply fsgroup to void stated issue A RWM volume (say NFS) can be mounted on multiple nodes. Do you have reliable count of mounts across all nodes?\nIn addition, that would be a change in semantics to how we support rwx volumes today. If we don't have a new CSI capability, then I think the best course of action is to only apply fsgroup for RWO volumes. Some questions remain: Do all RWO volumes support fsgroup? RWX is a superset of RWO. So if a user requests a RWO volume, it could still be satisfied by a RWX volume type. Can we detect this and still not apply the fsgroup even though the volume capability will say RWO?\nIf there's not a reliable way to enumerate node mounts (thinking about it, for CSI external drivers, the answer is NO). Theres is no reliable way to have k8s autocorrect a bad fsgroup/PV.AccessMode combo. So we would have to rely on the config provided by user/admin. I have come to similar conclusion as -- mainly, apply fsgroup only to volumes with access modes. The other k8s specified access modes, , cannot be reliably applied."}
{"_id":"doc-en-kubernetes-d8c35fcb948670df81783d48737c49dba78eee3bd98e80f83b95ce53a92518be","title":"","text":"Even if we were to introduce additional plugin capabilities that the CO can query (from driver) to find out what modes can be applied, there would still be the possibility of user/admin mis-configuration. Probably not, Is this where you think an additional capability would help ( ) That is a good point. Can't think of a way to detect/guess RWX when RWO is specified, this without the driver providing that info. Can we adopt the rule until we decide what capabilities would look like ?\nI will just try to summarize the conversation from Slack. We can safely assume that only applies to block storage type with file system on top of it. So for all - volumes that need to have called MUST have a file system() on them. Is that assumption correct? If it is - can we not use presence of field or we can query the local volume for fs type (like how we query selinux relabelling capability) and call only when - volume is RW and has a valid block storage file system type. Will this not solve our use case? EDIT: renamed to .\nWe'll need to make it clear that is only for block-based storage systems, and the field cannot be reused for some multi-protocol file server (ie, nfsv3, nfsv4, smb, etc). For that case, then the plugin will need to expose its own new storageclass parameter + volume attribute to specify protocol.\nCan we not use to get fsType - and use that information for calling ? It may be problematic to rely entirely on present inside csi source.\nthe only problem with is that in CSI volume info is intentionally opaque to the CSI code inside k8s. All mount logic is delegated to the driver.\nInstead of a new capability or trying to deduce how best to apply fsgroup, why not pass the to the driver and let it decide how to apply it (either as a attribute for the mount or top level request param)? Right now, all mount logic has been delegated to the external CSI driver except for applying permission. Thoughts ?\nthat is true but we can still inspect file system on the mounted path because by the time gets called volume is already mounted. This should be no different than how selinux support is determined. what you think?\nis this approved for 1.12 milestone? Thank you for working on this,\nstill discussing how best to approach this."}
{"_id":"doc-en-kubernetes-5cbdf2214fb613203f89164533ba44fdfe78b2afafa2ed81aceb81e111086f6f","title":"","text":"It would be a 1.12.\n[MILESTONENOTIFIER] Milestone Issue: Up-to-date for process <details open<summaryIssue Labels</summary- : Issue will be escalated to these SIGs if needed. : Never automatically move issue out of a release milestone; continually escalate to contributor and SIG through all available channels. : Fixes a bug discovered during the current release. </details<details<summaryHelp</summary<ul<li<a href=\"\"Additional instructions</a</li<li<a href=\"\"Commands for setting labels</a</li</ul</details\nApproved for 1.12\nBased on previous post, this is a summary: If then is not applied (because it could be an indication of a nonblock fs, or error) Done is provided if fstype is invalid Done if fstype is valid and then apply fsgroup logic else ignore fsGroup Done\n- any updates on ?\nWaiting for a review and approval.\nyou have a outstanding comment on the PR. Can you please answer that?\nHow does this approach relate to inline CSI volumes? There's no PersistentVolume involved, and hence no way to specify an access mode. , but my driver ignores it (always using tmpfs). I considered having my CSI driver read the mounting pod's spec to see if it has a security context with \"fsGroup\" set, since that detail is not supplied to the driver, as . My goal is to have the volume mounted with the files on it readable by the container's user, but there's no way for my driver to know who that user will be. Am I supposed to just use 0444 for the ownership bits, to allow everyone to read the files?"}
{"_id":"doc-en-kubernetes-c565e19bceebcf0dd9ac2effd9df046efbfbf8278a160afc85d83d70045c5f06","title":"","text":"returns It's because we recently change HorizontalPodAutoscalerSpec.MinReplicas from int to *int."}
{"_id":"doc-en-kubernetes-07ac40315577ee77e8734022e934cb4c363bc3e85cf45880c776268568485961","title":"","text":"In our Tanzu Supervisor Cluster k8s setup, we observe that the \"should verify that a failing subpath expansion can be modified during the lifecycle of a container\" conformance test fails. The root-cause seems to be that the test is updating the pod annotations, thereby resulting in removing the other annotations present in the pod object. Removing the custom annotations that we use causes the test to fail during pod deletion. Should this test perhaps use Patch instead of Update so that the annotations are merged and not replaced? Thanks ginkgo.By(\"updating the pod\") podClient.Update(pod.ObjectMeta.Name, func(pod *v1.Pod) { pod.ObjectMeta.Annotations = map[string]string{\"mysubpath\": \"mypath\"} })\nThis issue is currently awaiting triage. If a SIG or subproject determines this is a relevant issue, they will accept it by applying the label and provide further guidance. The label can be by org members by writing in a comment. <detailsInstructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\n/sig storage /sig conformance I personally agree that it should not remove existing annotations, but better ask the test owners\nThe label(s) cannot be applied, because the repository doesn't have them <detailsIn response to : /sig storage /sig conformance I personally agree that it should not remove existing annotations, but better ask the test owners Instructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\nMay I request the original author to weigh in on this? Thank you.\nI dont have any issues with the code change to a merge but what would bother me is that conformance tests are being altered by outside influences which doesnt seem particularly right. I think this would be for the sig storage owners to decide\nIssues go stale after 90d of inactivity. Mark the issue as fresh with . Stale issues rot after an additional 30d of inactivity and eventually close. If this issue is safe to close now please do so with . Send feedback to sig-contributor-experience at . /lifecycle stale"}
{"_id":"doc-en-kubernetes-164397ea6f5fa2ca928d4dd76377d3f8103ac21b9ae3014f9a95c921fb791826","title":"","text":"WRT , container registry credentials should be included in a docker file instead. We deprecated for this reason in , but is more tricky, because the file it references is not static. We need to figure out what the alternative is for dynamically providing pull credentials to nodes. /cc\nI think we need to enable Managed Service Identity (MSI) for all instances before deprecating . To do this, we also need to grant VM access to Resource Group in Azure IAM (e.g. config Azure AD in acs-engine). WDYT?\nIssues go stale after 90d of inactivity. Mark the issue as fresh with . Stale issues rot after an additional 30d of inactivity and eventually close. If this issue is safe to close now please do so with . Send feedback to sig-testing, kubernetes/test-infra and/or . /lifecycle stale\nI think this is still active. Right, /remove-lifecycle stale\nThanks, this is still active\nIssues go stale after 90d of inactivity. Mark the issue as fresh with . Stale issues rot after an additional 30d of inactivity and eventually close. If this issue is safe to close now please do so with . Send feedback to sig-testing, kubernetes/test-infra and/or . /lifecycle stale\n-- What's the status of this?\nSorry, no progress yet. Any ideas on this?\n-- pinging again on this one.\nWe're not making any progress yet in this release. Still seeking the possible solutions for non-MSI cases, secret is one option for this. /milestone v1.15 Related and\nNone one is working on this. Let's move this out of 1.15 /milestone v1.16\nthanks. This missed v1.15 feature timeline. LGTM of moving to v1.16.\nHello! I'm the bug triage lead for the 1.16 release cycle and considering this issue is tagged for 1.16, I'd like to check its status. The code freeze is starting on August 29th EOD PST (about 5 days from now) and once the code freeze starts, only release-blocking changes will be considered to be merged. Is this issue still planned to be fixed for the 1.16 release?"}
{"_id":"doc-en-kubernetes-d697b49623b35e9f2d1bfe1612db7307739262948b622701b4cd63789983d676","title":"","text":"/milestone clear is preparing a KEP for this, but it's still not ready yet.\nWill follow this KEP for implementation\nThis should be addressed by /milestone v1.18\nHello ! Bug Triage team here for the 1.18 release. This is a friendly reminder that code freeze is scheduled for 5 March, which is a few weeks from now. Is this issue still intended for 1.18? Thanks in advance!\nC/O Code freeze for release 1.18 is starting in 5 March (less than 1 week from now) and we want to ensure that each issue has a chance to be finished on time. Is this still on track for this deadline?\n/milestone v1.19 The work on the KEP hasn't started yet, so moving this to 1.19\nHello from Bug Triage team! Wanted to follow up on this issue with a friendly reminder that the code freeze for 1.19 is starting June 25th (about 4 weeks from now). As this issue is tagged for 1.19, is it still planned for this release?\n/milestone v1.20\nHey Bug Triage here. Wanted to follow up on the status of this issue as we're approaching code freeze on 12.11.2020. This issue is tagged for 1.20, is it still planned for this release?\nHi Bug Triage here, seeing that there is not much recent activity here, and we are pretty late into code freeze, I am going to remove the milestone label. /milestone clear\n/triage accepted /priority backlog I think we would be able to remove the flag once external kubelet credential provider promoted to GA\n/kind deprecation\nThis issue has not been updated in over 1 year, and should be re-triaged. You can: Confirm that this issue is still relevant with (org members only) Close this issue with For more details on the triage process, see /remove-triage accepted\nThis issue is currently awaiting triage. If a SIG or subproject determines this is a relevant issue, they will accept it by applying the label and provide further guidance. The label can be by org members by writing in a comment. <detailsInstructions for interacting with me using PR comments are available ."}
{"_id":"doc-en-kubernetes-d4a21b3be577ab214d5d61faf9bcc6243731916e2690559ffed63dd63be604bd","title":"","text":"If you have questions or suggestions related to my behavior, please file an issue against the repository. </details"}
{"_id":"doc-en-kubernetes-7b220aad8e787390cf311abd9e152a85d982510e255be9f3d76952f410cd2fe7","title":"","text":"https://k8s- It's quite flaky.\nThere are no sig labels on this issue. Please by:<br(1) mentioning a sig: <br(2) specifying the label manually: <br<brNote: method (1) will trigger a notification to the team. You can find the team list and label list\n/assign\n/sig networking\nThere are two failures in gci-gce-slow, so it's not specific to gke. https://k8s- Also in gce-slow: https://k8s-\n/sig network\nThis is strange, seems like this test is trying to reach the test-netexec pod through servicePublicIP:serivcePort, but the request hit the gce default backend instead. cc\nThere is another type of failure: Seems to be the port collision between test service and fluentd daemonset, which changed to use few days ago ().\ncc\nChecked the kube-proxy logs and found this is not the case. Port was not allocated (although that could happen in theory) to any service during test:\nThis turned out to be an interesting corner case. It happens at the beginning after we set up the GCE network LoadBalancer for OnlyLocal service --- when the LB still forwarding traffic onto nodes that don't have endpoints on. For more context, when network LB forward traffic to nodes, packet arrives node with below src:dst: If no endpoints are on the node, and there happens to have a pod using hostPort with the same port number, packet will be DNAT to that pod IP instead. These are observations and I haven't traced down to the exact iptables entries. I'm changing service port in test to avoid collision for now. cc"}
{"_id":"doc-en-kubernetes-2c743938ba30f2fba7527b26804dcfbb8cfcfa24fc2e107c2d5fed7244e0552d","title":"","text":"At least on AWS, if the clock drifts 300 seconds (which it will), then we get the rather cryptic message: This is because AWS includes the timestamp in the request. Installing ntpd would stop this drifting from happening. If the drift does happen, the likely symptom is that EC2 calls would stop working: attach/detach volumes on the minions, instance sync and routing configuration on the master. I don't believe that GCE has the same problem."}
{"_id":"doc-en-kubernetes-3129ecf685ad2fea6cb3ff97d12eed1692df1703f934c30f7ebae50375bb63d6","title":"","text":"From we can find that, hardcode \"private\" and \"public\", seems only support that network with such name. , it fetch the first ip. usually, it can include both v6 and v4 ip. what's the meaning to just fetch the first one ? is it v4 or v6 ?\nping and\nYes. Zoom out a level or two in the code, and you'll see we're trying to either a) setup an openstack loadbalancer to point to the minion node (a single address) using , or b) report to kubernetes all the addresses we know about the node and whether they are \"internal\" or \"external\" in . All we have at the starting point is the server name (from kubernetes), and we need to make up the rest of this information somehow. (I agree (a) should reuse the NodeAddresses() function from (b), but the latter is a newer function and the code hasn't yet been refactored in that way. This doesn't really change the substance of this discussion.) Re (1): The logic as currently implemented is: Assume any network named \"private\" is \"internal\" Assume any network named \"public\" is \"external\" Assume the AccessIPs (IPv4 and IPv6) are \"external\" Any other network addresses are ignored. I implemented the first two (\"private\" and \"public\") following the Rackspace public cloud precedent, and on the basis that users in other OpenStack clouds can use these names if they want to encourage an address other than the Access addresses. For the (a) use case, we just take the first address we find in order of the above; for (b) we report that full list. I preferred \"private\" to \"public\" since (again, at least on Rackspace public cloud), the traffic on \"private\" network is sometimes cheaper/free. I agree this isn't great, and please suggest something better if you have a use case in mind. Perhaps we should pull a list of network names that should be considered \"internal\" and \"external\" from the provider config? Perhaps I could assume that provider networks are public and non-provider networks are internal? I don't want to add too much configuration here if there's a good algorithm that will Just Work."}
{"_id":"doc-en-kubernetes-73739ae9bad159746241b14eae0e38aaa61fc839e6a2ae622df67e262b940fc6","title":"","text":"I think in practice it's probably ok to continue to prefer \"internal\" to \"external\" for loadbalancer setup, but I can contrive hypothetical cases that need more flexibility here too. Again, if you have a concrete need for such a thing, please file a bug so we have some idea of the flexibility required. Complicating all this is that Kubernetes never actually defined what an \"internal\" address was. just says \"Generally the IP address of the node that is routable only within the cluster\" which isn't a sufficiently strict definition. Looking at where Kubernetes actually uses it: just uses the first address reported (preferring InternalIPs to ExternalIPs) and the Kubernetes master configures an \"SSH Tunnel Check\" which just uses the first ExternalIP ... which kind of makes it pointless to ever even bother reporting anything beyond the first address. In both cases, it's assumed that this address is reachable by the Kubernetes master nodes, which makes the Internal vs External distinction a bit pointless. Re (2): is meant to return more than one IP, and the loadbalancer setup can only use a single IP for each node (or would unnecessarily track/healthceck multiple paths to the same backend for multiple addresses). If you're seeing single addresses returned by then you should file a bug. All the above treats IPv4 and IPv6 as equally important, if the OpenStack cloud has both, and the \"first\" address used for the loadbalancer will be an arbitrary choice between the two protocols (not quite true: will use AccessIPv4 before AccessIPv6). I think that's how it should be (the rest of the code should be protocol agnostic), but I wouldn't be surprised if there's some part of Kubernetes that doesn't support IPv6 properly. There certainly used to be lots of places that assumed IPv4 in Kubernetes, and GCE doesn't support IPv6 iirc so it gets less testing - if you've found something that breaks with IPv6, please file bugs. Now this bug: what would you like done here? Would you like the current situation to be documented further? Is the \"private\"/\"public\" naming thing too restrictive for your usecase? Something else?"}
{"_id":"doc-en-kubernetes-fcd218ff1291684683bf7f5abec9b421b4d5bd08568265528c48e934f7742a2f","title":"","text":"Hi, if we talked about openstack, I expect it not bind to rackspace, although, the kubernetes called RK SDK, But from my understanding, seems RK itself SDK not bind network to \"private\" and \"pubic\". User can define public network with their names, like site-network, for example. or \"testnet\", And for private, it should be seem. For nova list, the addresses, it can have \"OS-EXT-IPS:type\":\"fixed\" or \"OS-EXT-IPS:type\":\"floating\" for floating, I think what you means \"Public\", for fixed, I think its what you means \"Private\". In other words, we'd better to find related information from other fields in nova or neutron infromation. Instead of bind to \"private\" and \"public\" name.\nI this issue and wrote a patch that implements what suggests. I will send a PR soon\nGlad to see some improvements here. And wish to use it if implemented"}
{"_id":"doc-en-kubernetes-f1fd45d6e4c1b622cbcb5debe7400d2388f299758e0c5fe9f2d35eac274ddbc5","title":"","text":"I am trying to bring up a kind cluster with containerd in order to run e2e test on mac + kind. Not sure whether this is a bug or I missed some prerequisites. Any help would be greatly appreciated :) building the image dra/node:latest works as expected:bringing up the kind cluster failed (seems that kubelet is unhealthy and healthchecked failed) below the control plane logs: below the logs of one worker: successfully bring up the kind cluster with containerd Tried on M2 macbook pro I followed NONE Darwin MacBook-Pro-de- 22.1.0 Darwin Kernel Version 22.1.0: Sun Oct 9 20:19:12 PDT 2022; root:xnu-8792.41.9~3/RELEASEARM64T6020 arm64 arm Darwin No response No response\nThere are no sig labels on this issue. Please add an appropriate label by using one of the following commands: - Please see the for a listing of the SIGs, working groups, and committees available. <detailsInstructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\nThis issue is currently awaiting triage. If a SIG or subproject determines this is a relevant issue, they will accept it by applying the label and provide further guidance. The label can be by org members by writing in a comment. <detailsInstructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\nwhat type of mac you're using? for m1/m2 mac, you need to replace the linux-amd64 with linux-arm64(your log seems like using m1/m2 mac which same as the problem I've met before)\nthanks I am going to try that\nping me if you can run e2e on mac. I setup the cluster but failed to run the e2e test..\ncan you please share the exact command you tried and the failure ?\nIf you use a newer version of kind you might not need to clobber the included containerd, not sure what version is required And yes, on M1/M2 you need to use arm binaries/images.\nthis is the command that I tried !"}
{"_id":"doc-en-kubernetes-360d9c6c864de0dc81a219b1de4cd5a5370dd74d19a3f94773debf0837eaf62c","title":"","text":"It worked for me just fine on M2 machine with this obvious fix for the build script: Here is a test output:\nI'll submit a PR with the fix. Updating the documentation would be also nice to have. I hope and can do that.\nDRA tests fail on kind 0.18.0. Which Containerd version is used there?\nIt was the latest 1.6.x at release time. 1.6 is their LTS branch before 2.0 comes which has some tricky changes for us to deal with (notably the CRI hosts config deprecation which will break all our local registry users :/)\nUnfortunately 1.6 doesn't have CDI support, which DRA depends on. The version used in the build script is 1.6.0-830-g34d078e99, but it was probably built from containerd master after 1.6 was released.\nthanks for providing the details. you could run the e2e as shown by and find the result"}
{"_id":"doc-en-kubernetes-9945f53575b60dc39792ed32e02eb9c66a039731b5284891454b6dce2289c963","title":"","text":"<!-- Please use this template while reporting a bug and provide as much info as possible. Not doing so may result in your bug not being addressed in a timely manner. Thanks! If the matter is security related, please disclose it privately via --What happened: When starting pods with AFS volumes, there is an initial delay of five seconds until the pod is transitioning from the \"Scheduled\" state. The reason for this is that currently the volume mounting happens inside a wait.Poll which will initially wait a specified interval (currently 5 seconds) before execution. Event log for comparison with and without AFS volume attached (seconds since event): What you expected to happen: How to reproduce it (as minimally and precisely as possible): Anything else we need to know?: Environment: Kubernetes version (use ): Cloud provider or hardware configuration: OS (e.g: ): Kernel (e.g. ): Install tools: Network plugin and version (if this is a network-related bug): Others: /assign /priority important-soon /sig cloud-provider /area provider/azure\nsee"}
{"_id":"doc-en-kubernetes-0f269e408c0c5ca583af123d217b4d1fb7160d426de10d2601450f917f382cd5","title":"","text":"I an ELB to the Kibana cluster log viewer but it is not reported by which, aside from having a URL, should check each cluster service and if it has an external IP it should report that instead of the master proxy address.\ncc"}
{"_id":"doc-en-kubernetes-dbcf83004ea3f0459daec6a4d94a4bece404fe8f97225f1a2fba429c343bf057","title":"","text":"I've been pulling my hair out trying get kubectl talk to my local cluster. It seems that kubectl merges the local file with values pulled down from the cloud. I've done the following: The I just set is silently overridden. Questions: is this documented anywhere? should kubectl merge in my GCE cluster when I set my cloud-provider to something else? should kubectl override my previously set current-context?\nworkaround is to use flag to override the config file value.\nbut not a good UX long term\nSomeone else hit this as well: . Better fix is to have the local-up script set a context, which I will try to add today. As for the overwriting behavior, it's documented in , but based on what you describe, the intended load order is not working; local .kubeconfig should overwrite global. cc\nThat sounds infuriating. I'll take a look this morning."}
{"_id":"doc-en-kubernetes-94306baf6d6408dd320a641269af3f429c4d2a949a311ade7ed7994b23100633","title":"","text":"After this () has been merged and EnpointSlice feature is now enabled by default, we noticed Kube-proxy will not work on Windows. Only message I can see in the logs is: /sig windows\n/cc /cc /cc\nHow about separating the above check into two if statements so that the log can pinpoint which field is false ?\nIt seems long term, EndpointSlice support should be so that is kept in sync with other implementations.\ncc\nThanks for reporting this! Windows proxiers were still intended to be reading from Endpoints, but apparently that's not working as intended. EndpointSlices are moving back to disabled by default for 1.17, so that should fix the failing tests for now. I'll work on an actual fix for this as well, but it may not make it in until after KubeCon. /assign /sig network /priority important-soon\n/triage unresolved Comment when the issue is assessed and confirmed. I am a bot run by vllry. ‍\n/remove-triage unresolved\n/assign cc\nGitHub didn't allow me to assign the following users: jocelynberrendonner. Note that only , repo collaborators and people who have commented on this issue/PR can be assigned. Additionally, issues/PRs can only have 10 assignees at the same time. For more information please see <detailsIn response to : /assign cc Instructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\n: Please let me know if you need any help on the fix! I'll be more than happy to review/help test the fix on Windows."}
{"_id":"doc-en-kubernetes-8f6e6c1f685c98cc169477efadd1a12031a77645ead6fa9fa244a9ca04430443","title":"","text":"is sent as a fix. cc It's a test failure reported by the publishing robot.\nThe line it fails at:\nIt looks like a vendor copy vs gopath copy issue.\nOne possible fix is removing from the vendor/ directory. I'm testing locally.\nConfirmed removing from the vendor/ directory fixed the test in robot.\nWe need to add the \"issue filer\" feature to the publishing robot, we can't rely on to report robot issues everytime :P I'm just not sure how long the publishing robot will be needed (i.e., when we will get rid of the staging/ directory) so not sure whether we should invest more time on the robot."}
{"_id":"doc-en-kubernetes-62f595d5f46f490d9eea1d39e1ba000bf29c92ed637615b4bae439425896a1f8","title":"","text":"<!-- This form is for bug reports and feature requests ONLY! If you're looking for help check and the . If the matter is security related, please disclose it privately via --Is this a BUG REPORT or FEATURE REQUEST?: What happened: patched out the Size field (was PR ), it will cause issue again. See also and . What you expected to happen: PVC resizes of Cinder volumes successfully. How to reproduce it (as minimally and precisely as possible): Anything else we need to know?: Environment: Kubernetes version (use ): Cloud provider or hardware configuration: OS (e.g. from /etc/os-release): Kernel (e.g. ): Install tools: Others:\n/kind bug /sig storage /sig openstack\n/assign"}
{"_id":"doc-en-kubernetes-af2890d14d0991d34b559e2e2a7effc23a0b7cafb90e6ce93894f3fac456697c","title":"","text":"We have a new vanity url, , which redirects to the GCS release bucket - e.g. Right now, the https endpoint doesn't work, since isn't in the list of domains on the certificate. Once it works, we should switch all references over to the nicer vanity URL. x-ref\nSo what it required to make this issues actionable? Where to add this for the certificate?\nthe issue tracking that is\nThis needs to be triaged as a release-blocker or not for 1.5\nnot a release blocker.\nIs it appropriate to move this to the next milestone or clear the 1.5 milestone? (and remove the non-release-blocker tag as well)\ncan this be closed?\nI forgot to get to this in 1.6, but it'd still be good to do.\ndo you want this in v1.7? is there a sig to own this?\ndoesn't need to block 1.7, no.\nThere are no sig labels on this issue. Please by:<br(1) mentioning a sig: <br(2) specifying the label manually: <br<brNote: method (1) will trigger a notification to the team. You can find the team list .\n/sig release"}
{"_id":"doc-en-kubernetes-95b3abd37032c873ce8d7d9904ff018c8e3e7e0f9808f24d7c83d05cb3882c93","title":"","text":"Our GKE scalability jobs (https://k8s-) are failing in the cluster teardown step (even though the e2e tests themselves passed). E.g failures: https://k8s- https://k8s- Seems like it's caused by some leaked network-related resources: /assign Krzysiek - Could you PTAL into this when you find time? It'll be great to have our GKE suite back to green. cc\nI think this is related to an earlier bug we've seen (similar one for GCE): Basically service-controller is creating some cloud resources (like firewalls) for lb-type services which are not brought down properly when the k8s cluster is brought down.\nAlso another related issue - There should also be an internal bug for this somewhere. IMO we may be able to fix this in test-infra (without waiting for proper long-term fix on gke side).\nI am adding clean-up method. This way all of the firewall rules will be cleaned up after tests. If this method has to remove any firewall rule, the teardown phase will return exception. It's not a fix for the bug - it only allows us to easy isolate tests leaking firewall rules.\nTests: should have session affinity work for LoadBalancer service with ESIPP on should be able to switch session affinity for LoadBalancer service with ESIPP on should have session affinity work for LoadBalancer service with ESIPP off should be able to switch session affinity for LoadBalancer service with ESIPP off leak resources related to the LoadBalancer. PR will fix this."}
{"_id":"doc-en-kubernetes-68f56cf00eaca7b36f313a8fbb47f02ec3096dc918e863d7bad5040791bff0ba","title":"","text":"From After , Multi-AZ Clusters tests in 1.19 and 1.20 started timing out as createBalancedPodForNodes is extremely inefficient in scale: for each node (out of 5k), it lists all pods in a cluster serially (each list takes ~5s), which translates to ~7 hours. Optimizing this function by listing pods once and then in memory by Spec.NodeName /kind bug /priority critical-urgent /sig scheduling\n/triage accepted\n/assign\n/reopen Looks that ComputeCPUMemFraction took only 12 minutes, but it stuck later, on It is scheduling one pod per each node: Which is waiting 5s to be ready: 5k nodes * 5s = ~7h\nReopened this issue. <detailsIn response to : /reopen Instructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\n/assign Can you do the pod creations in parallel?\nYes, good idea. I'll update to do that"}
{"_id":"doc-en-kubernetes-d581fd23524a17c2e094f8c22d9237840c9557b1634d31d91db6cf64598bf32e","title":"","text":"<!-- Please only use this template for submitting reports about flaky tests or jobs (pass or fail with no underlying change in code) in Kubernetes CI --ci-kubernetes-e2e-ubuntu-gce [sig-storage] Subpath Container restart should verify that container can restart successfully after configmaps modified links to appreciated links to specific failures in spyglass appreciated Triage shows that this has been flaking quite frequently: It has flaked in 5 of the last 10 runs: /sig storage /priority important-soon <!-- Please see the deflaking doc () for more guidance! --\nThis issue is currently awaiting triage. If a SIG or subproject determines this is a relevant issue, they will accept it by applying the label and provide further guidance. The label can be by org members by writing in a comment. <detailsInstructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\nNoticed some more flakes on this one\nYep, looks like this is still flaking quite frequently: Going to re-ping SIG Storage.\n/assign\nTest is flaking maybe half of the time. ~This test is failing in gce-ubuntu-master-default - not sure if it's failing in any other dashboards, nor whether this is run anywhere else.~ edit: This is failing on Ubuntu only. Not sure if this is limited to master branch. Fix to the original issue was in since 1.19. This is a subpath test which a configmap a container that mounts the configmap a liveness probe and waits for the container to restart. liveness probe and waits for the container to stop restarting. The test failed at the last step - the container never stopped restarting. Timeout is 2min. The liveness probe continuously checks whether a \"probeFile\" exists in (an emptyDir), and if it doesn't exist, whether the configmap file contains the modified value. The probeFile essentially acts as a condition which guards the probe from checking the configmap file. The test fails the liveness probe by removing the probeFile without updating the configmap just yet. It then fixes the liveness probe by updating the configmap to the modified value."}
{"_id":"doc-en-kubernetes-fe4ad65b6af4a4370d93bc829526a20d829ebc4486fe6305d4213476a53bb63c","title":"","text":"So if the container keeps restarting, the liveness probe could've been failing continuously within 2min, due to one of the following: (? not sure about this one) Command can't be executed (for example if container can't start) configmap file was never updated. Within 2min, the pod only restarted 5 times. The liveness probe's period is 2s, initial delay is 1s, and failure threshold is 1, so a restart due to liveness probe failure is triggered every 3s. The image is busybox. The command is just , so it's unlikely for the container to have crashed. There were no liveness probe unhealthy events after the test fixes the probe, and yet it continued to restart. No more events after probe is fixed.\nCould we check if the configmap value was updated for sure to rule out that scenario ? I have not tried to run an end to end test locally :) so am not sure how to debug it\nWill do! I ran the test in a GKE Ubuntu cluster running 1.19.6 and couldn't reproduce the failure. I'm now trying to run a cluster on GCE with Ubuntu nodes running k8s master, but haven't been successful after trying a number of techniques. If anyone happens to know how to do this let me know; I'll keep asking around otherwise.\nI got a repro on a GKE cluster backed by COS. After the test failed the container ends up in Running state eventually, and the configmap file outputs the expected modified value. The fix is to increase the test timeout. PR coming up."}
{"_id":"doc-en-kubernetes-828b2f2bc9a1c73b1b130d20e1091ab566168df681bda6f542cc66f3f230828b","title":"","text":"/kind failing-test <!-- If this is a flake, please add: /kind flake --<!-- Please assign a SIG using: /sig SIG-NAME --\n/sig cli\nThere may have two issues to fix: doesn't support run with assertion must check whether schema is nil first\n/assign\n/triage accepted"}
{"_id":"doc-en-kubernetes-e65d433be70b14f59611269e2f1c0fa746f312a0f5c96a0c29a874f17f7c06ea","title":"","text":"<!-- Please use this template while reporting a bug and provide as much info as possible. Not doing so may result in your bug not being addressed in a timely manner. Thanks! If the matter is security related, please disclose it privately via --What happened: If you create a service that reference a Basic IP in a standard load balancer cluster, the service will stuck at \"Pending\" state for days without detailed error from Azure response. When you describe the service, it only has following: What you expected to happen: In this case Azure responded 400 with detailed error message that the referenced IP is a basic IP and cannot be used on standard LB. I would expect the detailed error message show up as events for easier diagnostic. How to reproduce it (as minimally and precisely as possible): Create a service that reference a Basic IP in a standard load balancer cluster Anything else we need to know?: Environment: Kubernetes version (use ): Cloud provider or hardware configuration: OS (e.g: ): Kernel (e.g. ): Install tools: Network plugin and version (if this is a network-related bug): Others:\n/assign /area provider/azure /sig cloud-provider\n/area cloudprovider"}
{"_id":"doc-en-kubernetes-7c861bbb23da2f1ab2b6ca68ce44086076d9635fab7f36a80cef91e7d1c385de","title":"","text":"It looks like there was a change that first appeared in Kubernetes 1.29, where in 1.29 the flag on kubelet is no longer being used to populate the Node's Addresses.Hostname value. Here is the relevant output from a for a 1.28 cluster with : And here's the output from a 1.29 cluster with the same configuration: Here is the full list of parameters in case that is helpful: This is from an IBM Cloud Kubernetes Service 1.28 and 1.29 cluster. We do not think there was any cloud-configuration-manager or any other related change that would cause this difference. Is it possible it is a side effect of ? I expected the behavior to be consistent between Kubernetes releases, since I didn't find any reference to this change in Create an IBM Cloud Kubernetes Service cluster at 1.28 and 1.29 and see the difference No response <details</details<details</details<details</details<details</details<details</details<details</details\nThis issue is currently awaiting triage. If a SIG or subproject determines this is a relevant issue, they will accept it by applying the label and provide further guidance. The label can be by org members by writing in a comment. <detailsInstructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\nyeah, it seems likely my change broke this, it should keep working as you say, let me take a look /assign\n/sig network /sig cloud-provider\n/assign /unassign Thanks\nFYI it was an unintended side effect of:\nDo the sig cloud provider folks need to help here? We did make external cloud providers the default as of 1.29, which may have uncovered this. Let us know!\nNo, it's just a dumb kubelet startup bug. Well, not dumb, but easy to fix.\nIt seems the problem is that once kubelet sets the hostname the cloud controller manager does not override it so maybe the right behavior is only override it if is actually overridden ?\nwhat problem?"}
{"_id":"doc-en-kubernetes-cb25003b88a3fd857f2f7edabf578ed21b587771e769d62a703fb5f227a9b05a","title":"","text":"we are discussing if the fix should be backported or not It seems you are relaying on the with type Hostname and this not clearly documented We also find out recently that the behavior of this address will be overriden if the external cloud provider configures it if the external cloud provider returns the hostname , then this problem will be solved without need to backport\n(Well, to clarify my comments there: it's not well-defined between different clouds and node linux distros. If you know what cloud and distro you are on, then the behavior might be consistent and usable, but that still makes for a lousy kubernetes API.) Anyway, the question here is: was this actually breaking anything major, or did you just happen to notice this and think \"that's weird\"? (Also, if anyone else was broken by this bug, feel free to chime in; trying to figure out what to do about backporting this, because the original fix for this bug ended up introducing another bug, and the fix for that_ bug is maybe not backport-able...)\nSo far I have only heard from a single customer (who was using IBM Cloud Kubernetes Service) who had something in their cluster relying on cluster node's Address.Hostname value, and so when they updated to 1.29 they were broken. I believe they just modified their code to get this \"Hostname\" value (which was really the node IP) from somewhere else. So, to answer the question, it is somewhere between the \"I noticed it and thought 'that's weird'\" option and the \"it broke something major\" option. We have this documented in our 1.29 release notes for IBM Cloud Kubernetes Service, so other customers will have fair warning before updating. So at this point I don't consider fixing this in 1.29 as essential."}
{"_id":"doc-en-kubernetes-6612882cbe44a332a8a18637718123dbd71ae9f680fe3a41129c9067a30b426a","title":"","text":"Kubernetes version (use ): kubernetes/master Environment: na What happened: What you expected to happen: How to reproduce it (as minimally and precisely as possible): see above Anything else we need to know: nope, have a fix.\nThere are no sig labels on this issue. Please by:<br(1) mentioning a sig: <br(2) specifying the label manually: <br<brNote: method (1) will trigger a notification to the team. You can find the team list and label list"}
{"_id":"doc-en-kubernetes-237446a2101b1d0e3896762c3997d0b07e17e89bcd32bef5edcdc8a493c92d82","title":"","text":"Kubernetes version (use ): 6 Environment: GCE What happened: I created a cluster for Heapster testing purposes to debug failing test in It turned out that Summary API does not export system/misc stats in SystemContainers section (though there are kubelet and docker stats). Everything works as expected on nodes. See also https://k8s-\nWhat node OS were you using? We don't create the system (misc) container on systemd, so it's expected that GCI nodes don't report those stats.\nThe default configuration for 1.4, which is: master: GCI nodes: Container VM\nAre the nodes running 1.4.6 too? There was a bug with this in 1.4.5, but it should have been fixed in 1.4.6.\nYes, both are running in 1.4.6. If you want to reproduce from Heapster current head run . It will take some time and as a side effect you should have a cluster with the mentioned problem. This was catch by integration tests and I temporary disabled the check here\nif you have spare cycles, can you look into this?\nfriendly ping\nLooking at it now\nI have found a couple things so far. On the containerVM nodes, cadvisor is exporting a container with \"name\": \"/system\" On the GCI Master, cadvisor is exporting a container with \"name\": \"\" The has this: \"{% set system_container = \"--system-cgroups=/system\" -%}\" (/system is the default misc container name) So the real question is: Should cadvisor export \"/system\" instead of \"\" on GCI? Or should we be configuring GCI to use \"--system-\"?\nI thought we fix this very issue for 1.4 branch by a while back. Are you running into the same issue with 1.5?\nI think I told you the wrong thing earlier. We don't support the misc (/system) cgroup on systemd systems (I can't remember why, but it was an explicit decision - or might know). The is something different, I believe a base systemd cgroup (slice). As for the misc container on ContainerVM nodes, we've been testing for the misc container in 1.5+ in the summary api test (https://k8s-) for a while, and I haven't seen this issue."}
{"_id":"doc-en-kubernetes-c190eb06e0eb058949f528f1af5886fe17cab35cd1b33a69ea2cddc6a3fc8812","title":"","text":"So if this is still reproducible, it's limited to 1.4. See this the bug was created for the post-fix.\nFound the explanation here: The should use the flag to know the location that has the processes that are associated with , but it should not modify the cgroups of existing processes on the system during bootstrapping of the node. This is because is the on the host and it has not delegated authority to the to change how it manages .\nIIUC this is intended behavior on systemd nodes. Based on the explanation provided, the kubelet doesnt have the authority to control \"misc\" processes on systemd.\n+1. \"misc\" is just a bandaid for cvm. , David Ashpole <:\nOk, so I'm closing this issue."}
{"_id":"doc-en-kubernetes-432f38af47f0e48a0119df4c6a3e98c546702de6b66b5c01890c1ec73b3675a0","title":"","text":"The last 4 runs of https://k8s- have failed on the and steps with : https://k8s- https://k8s- https://k8s- https://k8s- The GCE version of this test suite appears to be fine. This is blocking the submit queue. CC\nIs there a way to see what does /cluster/gce/list- output in the runs... exit status 1 is not very informative\nHere is a non-gke failure: https://k8s-\nGiven that the last thing output from the listing of resource is: I would bet that the defined is failing to find a match from (see )\nHummm I ran it locally against but did not get the same failure\nyou might be more familiar with list-\nThis started happening on the 16th and became more regular on the 18th-- do you think the new janitor code might be causing this? More details here:\nis giving me exit status 1, let me find which command it is.\nYeah, and have figured it out. Because of the janitor, we no longer have any default firewall rules in these projects (which might be a bug), and so now the is failing. I think in the distant past always printed out a header, even if nothing was found, though it hasn't done that in a long time. Fix for is to the on L52. We may also want to decide if we should be deleting the default firewalls.\nWe'll want to cherrypick that to the old branches, too.\nhummm I'll let it skip default-firewall\nKeeping open until we've verified fix"}
{"_id":"doc-en-kubernetes-8b28935a84f88ac18c7fc65931585a933c76320eb13cd6ef7170c583b31097e6","title":"","text":"<!-- This form is for bug reports and feature requests ONLY! If you're looking for help check and the . --Is this a BUG REPORT or FEATURE REQUEST?: /kind bug What happened: cAdvisor has a option to only collect docker related cgroups. Kubelet stats api now supports allocatable cgroup, /kubepods, for the runtime cgroup and the kubelet cgroup. But with settings, Kubelet stats api can not query pod related cgroups metrics. FYI: What you expected to happen: should be compatible with kubelet stats api.\n/cc\n/sig node /sig instrumentation\nadd to cadvisor in A comma-separated list of cgroup path prefix that needs to be collected even docker-only is specified bump cadvisor dep for Kubernetes setting to\nIssues go stale after 90d of inactivity. Mark the issue as fresh with . Stale issues rot after an additional 30d of inactivity and eventually close. If this issue is safe to close now please do so with . Send feedback to sig-testing, kubernetes/test-infra and/or . /lifecycle stale\n/remove-lifecycle stale\nIssues go stale after 90d of inactivity. Mark the issue as fresh with . Stale issues rot after an additional 30d of inactivity and eventually close. If this issue is safe to close now please do so with . Send feedback to sig-testing, kubernetes/test-infra and/or . /lifecycle stale\n/remove-lifecycle stale\nIssues go stale after 90d of inactivity. Mark the issue as fresh with . Stale issues rot after an additional 30d of inactivity and eventually close. If this issue is safe to close now please do so with . Send feedback to sig-testing, kubernetes/test-infra and/or . /lifecycle stale\nStale issues rot after 30d of inactivity. Mark the issue as fresh with . Rotten issues close after an additional 30d of inactivity. If this issue is safe to close now please do so with . Send feedback to sig-testing, kubernetes/test-infra and/or . /lifecycle rotten\nRotten issues close after 30d of inactivity. Reopen the issue with . Mark the issue as fresh with . Send feedback to sig-testing, kubernetes/test-infra and/or ."}
{"_id":"doc-en-kubernetes-c993cfc824e8b144526f19f7f90c37dca7b5b7104e06f115b259fa72f1d9ae68","title":"","text":"/close\nClosing this issue. <detailsIn response to : Rotten issues close after 30d of inactivity. Reopen the issue with . Mark the issue as fresh with . /close Instructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details"}
{"_id":"doc-en-kubernetes-124d4c7fb9ee696a83909f715e9bd8706a0f4465e08a35801f7cd2c1ed445555","title":"","text":"/kind failing-test <!-- If this is a flake, please add: /kind flake --/kind flake <!-- Please assign a SIG using: /sig SIG-NAME --/sig testing apps\nThis issue is currently awaiting triage. If a SIG or subproject determines this is a relevant issue, they will accept it by applying the label and provide further guidance. The label can be by org members by writing in a comment. <detailsInstructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\ncan you PTAL, it seems this test was in It doesn't seem to have a high rate of failure , but seems flakes are legit\n/assign\nside note: it's really hard to make it fail in a fast machine :) I had to temporarily add a sleep in the reconciler.\nhave you tried Jordan's instructions compiling and running with ? for integration you should have an external etcd instance\nI haven't. Good to know for next time :)"}
{"_id":"doc-en-kubernetes-0045c1600a60d9073df9d5127e356e4d0c19e0407dd07c166d371541fb200a6a","title":"","text":"Update the godoc on the encryption config API on how to specify group/resources to be encrypted ref: godoc has the details na No response_ na na <details</details<details</details<details</details<details</details\n/sig auth /triage accepted\n/assign Cool-Coder174\nThanks for volunteering for this LMK if you need any help with it.\nJust for clarification, I just need to add documentation on how to configure encryption on groups/resources?\nYou can update godoc to mention that custom resources can now be to the list of to be encrypted as well and how to add the of the resource to the list. e.g. from this test\n/assign /unassign Release v1.26 deadlines are here so we need to get this completed ASAP.\n/milestone v1.26\nI apologize for the slow actions. I know time and contributions are valuable. May I request a walk through. This is my first time ever contributing and I really want to learn. I have been reading some of the code base on my spare time.\nthis feature is on the harder side of things to get started with. Reading the blog post and the associated documentation is probably the best starting point:\n& what is the status on this issue? 1.26 is coming up! cc\n/milestone v1.27"}
{"_id":"doc-en-kubernetes-3e146b385ed02c4a2f1c99c93d7470431d31b35da9a5c1943b70b3f73ebe8376","title":"","text":"tests are failing atm on GCI https://k8s- because VOLUMEPLUGINDIR=/home/kubernetes/flexvolume set in takes precedence over KUBELETTESTVOLUMEPLUGINDIR=\"/etc/srv/kubernetes/kubelet-plugins/volume/exec\". As far as I can tell /home/kubernetes/flexvolume is not readable yet by containerized kubelet so the test needs to keep using \"/etc/srv/kubernetes/kubelet-plugins/volume/exec\" and the bash needs to be shuffled around to solve this. cc\n/sig storage\nThis is more an acknowledgment of the issue than anything, can't write a PR until later\ncc /kind e2e-test-failure\nthis is currently blocking alpha.3 release\nwe are just waiting for approval\nI think they’re still failing, not sure what the release process is, but these tests started running only yesterday and so failure indicates something wrong with the test themselves, they shouldn’t block release. (They succeed on gce cluster with Debian nodes but fail on gci)\nwell, let's see if we can get the kinks worked out before the beta release on 9/27 or we will need to disable it from release-blocking\n/assign /reopen\nI0824 17:49:28. 8562 ] operationExecutor.VerifyControllerAttachedVolume started for volume \"flex-volume-0\" (UniqueName: \"flexvolume-k8s/dummy-e2e-tests-flexvolume-wf48v/-88f4-11e7-a3f9--flex-volume-0\") pod \"flex-client\" (UID: \"-88f4-11e7-a3f9-\") edit: VerifyControllerAttachedVolume must succeed since the plugin is non-attachalbe and no error is reported. But the voluem is absent from node.volumesInUse I0824 17:44:06. 4 ] AttachVolume.Attach succeeded for volume \"flex-volume-0\" (UniqueName: \"flexvolume-k8s/dummy-attachable-e2e-tests-flexvolume-j8962/flex-volume-0\") from node \"bootstrap-e2e-minion-group-9913\" W0824 17:44:06."}
{"_id":"doc-en-kubernetes-2c104bfa2b3eddfc8805d23cd7bd64cf6563fee869d8467e10776a93a804d4bb","title":"","text":"4 plugin-] flexVolume driver k8s/dummy-attachable-e2e-tests-flexvolume-j8962: using default GetVolumeName for volume 0x1c5df60 I0824 17:43:59. 3728 ] operationExecutor.VerifyControllerAttachedVolume started for volume \"flex-volume-0\" (UniqueName: \"flexvolume-k8s/dummy-attachable-e2e-tests-flexvolume-j8962/flex-volume-0\") pod \"flex-client\" (UID: \"-88f3-11e7-a3f9-\") E0824 17:43:59. 3728 ] Operation for \"\"flexvolume-k8s/dummy-attachable-e2e-tests-flexvolume-j8962/flex-volume-0\"\" failed. No retries permitted until 2017-08-24 17:44:00. +0000 UTC (durationBeforeRetry 500ms). Error: Volume has not been to the list of VolumesInUse in the node's volume status for volume \"flex-volume-0\" (UniqueName: \"flexvolume-k8s/dummy-attachable-e2e-tests-flexvolume-j8962/flex-volume-0\") pod \"flex-client\" (UID: \"-88f3-11e7-a3f9-\") Seems there is something wrong with VerifyControllerAttachedVolume, still at a loss as to why it's showing up in GCI but not Debian edit: get node -o yaml shows: yet it errors because the volume hasn't been to volumesInUse??? Aug 24 20:31:42 mawong-e2e-minion-group-npjh kubelet[9431]: E0824 20:31:42. 9431 ] Operation for \"\"flexvolume-k8s/dummy-attachable-e2e-tests-flexvolume-z8xc4/flex-volume-0\"\" failed. No retries permitted until 2017-08-24 20:31:44. +0000 UTC (durationBeforeRetry 2s)."}
{"_id":"doc-en-kubernetes-522cbbb5c0d9fc5ea9cc6970c506130e31c26f402262297f89cf1b73ad15c110","title":"","text":"Error: Volume has not been to the list of VolumesInUse in the node's volume status for volume \"flex-volume-0\" (UniqueName: \"flexvolume-k8s/dummy-attachable-e2e-tests-flexvolume-z8xc4/flex-volume-0\") pod \"flex-client\" (UID: \"-890b-11e7-9392-\")\nNever mind the GCI vs Debian issue, our test fix would have worked but in between and , something else changed and is causing both GCI and Debian tests to fail.\nmight be the cause, so we might be looking at legitimate failures caught by the test at this point (yay?) and not faulty tests. investigating..."}
{"_id":"doc-en-kubernetes-0d64a06982b9b42eb12ec4911c298e718209b3fb4fbfb723e1ec7aab0e2e7e0b","title":"","text":"It seems that exec probes must emit \"ok\" to be considered success, rather than return 0. If this is what we really want, we should document it somewhere.\nThis is (was?) a limitation of docker exec, which didn't return the code. Brendan On Jan 28, 2015 8:38 AM, \"Tim Hockin\" wrote:\nThe doc says it looks at exit status, so we need to fix one or the other.\nfor kubelet\nfor exec\ncould you take a look at this? Feel free to re-assign it back to me.\nFWIW the current behavior isn't SO bad, and it makes clear a failure in the thing that probed vs a failed probe. Just saying, we don't NECESSARILY need to change it. On Jan 28, 2015 5:26 PM, \"Dawn Chen\" wrote:\nWith docker 1.5 we can go back to using the exitCode instead of \"ok\" string. OOM kills will also have special exitCodes in docker 1.5. , Tim Hockin wrote:\nok, I'm going to change the docs for now, we can change back when we standardize on 1.5..."}
{"_id":"doc-en-kubernetes-f3af4e5ee08342f3d9a5775ee2c1466702ed34eb1665ac9c4b71bf76c8f54023","title":"","text":"Nodes Network when a minion node becomes unreachable [replication controller] recreates pods scheduled on the unreachable minion node AND allows scheduling of pods on a minion after it rejoins the cluster(Failed 17 times in the last 30 runs. Stability: 43 %)Failed with 0 retries\nfor triage\nI'll look into this.\nI talked to about this and he had a pretty good guess in about two seconds about what the problem is. ReplicationController sync method does this: but DaemonSet controller doesn't. (I guess when DaemonSet was copy-pasted from ReplicationController, this was accidentally removed.) Pod reflector and syncer are running in separate goroutines, and if pod reflector doesn't run then syncer will think there are no pods on the node.\nTo be clear: this is about the test Controller Manager should not create/delete replicas across restart which is the one that is actually failing.\nI'm confused you're saying that this is about , but you're updating . I thought they are about two different tests...\nYes sorry, I put my comment in the wrong issue. Reassigning this one to you, and I'll copy my comment into the other one.\nFirst failure: We check if all namespace are deleted. If other tests failed to delete their namespace this test will fail. I'm going to delete this logic as we already check if we have successfully deleted our namespace. See also for some more details.\nSecond failure: We disable network by running the following command: I checked and it seems that it results in ing master to hang. I suspect that Kubelet hangs on updating node status and depending when it breaks the connection (by timeout) it will fit in test timeouts or not.\nWhat is the default timeout for connections using client libraries to apiserver? I'm checking the code, but can't find it.\nI'm not sure if we set one, but we should. I guess I would expect to see that in the config, w/ default set by this function: , Filip Grzadkowski wrote:\nSo: let's add a timeout, and revert .\nI think that adding a timeout and reverting are orthogonal. I we want to simulate network problems with node-master communication then using REJECT is better anyway. At the same time adding timeout makes sense for many other reasons."}
{"_id":"doc-en-kubernetes-171a96c8c8138970f18a9330482c8efd4960f1ea370d03cb243605d2473f3591","title":"","text":"OK then, let's add a timeout and test both DROP and REJECT-- no sort of network problem should hang a component.\nI think it didn't hang. It just that the kubelet was retrying too slow and sometimes it fit test timeout, sometime it didn't.\nCould we reduce the timeout to make it always fit the test timeout? , Filip Grzadkowski\nI guess we can. I'll try tomorrow.\nIt seems we have a new type of failure: It seems that kubelet on a machine where network is blocked doesn't notice new pod. After reading logs I assume that kubelet doesn't notice that the watch is broken and doesn't try to relist everything. It notices new pod only after watch timeout.\nUgh, I hope we don't have to start sending watch heartbeats.\nIt seems that all recent failures are with the following error: which means the reason is here: - can you please take a look or delegate?\nI did some more debugging of it. I was looking into this failure: There is something strange happening here: run performTemporaryNetworkFailure: succesfully block the load from kubelet: the succesfully wait for node to become not ready: some time nodecontroller successfully removes the pod: replication controller creates a new pod: here is something I don't undertand - because of some reason, the test doesn't observe that the pod actually disappeared. - any thoughts on it?\nBTW - this is exactly the same situation in:\nOK - I think this is related to default 30s deletion grace period.\nYes - it seems that this is exactly the case. I will send out PR fixing this out for review.\nI'm reopening it for tracking. If there won't be any failures within next few days - I will move the test back from flaky and close this issue.\nThanks for fixing, .\nIt seems that the main issue is fixed. However, there was one failure in the last 30 runs: in which one pod simply didn't started (and it was \"preparation\" for the test, not the test itself).\nDuring next 2 days there was exactly one more failure of this test: it is exactly the same as the previous one. However, those are duplicates of So I'm going to move this test out of flaky suite."}
{"_id":"doc-en-kubernetes-53ca4ed8e960fbba0bd8112d5cddbb22b80695a5a78dce628bb98a2132843341","title":"","text":"There were two patches merged recently in the kernel that will significantly improve the IPVS proxier: The first patch fixes an issue where a packet is dropped whenever a stale conntrack entry is reused. This caused a 1s delay whenever the stale conntrack entry was used. This was the original reason for setting . More details in The second patch fixes an issue when a source port is reused from a given client and the corresponding destination was deleted. This is easily reproducible when a single client sends a lot traffic to a small set of destinations and one of the destination servers are removed. More details in Based on these two patches, we should start setting (default) starting on Linux v5.9 when those two patches make it to the main line. The new sysctls for the IPVS proxier would be: We should probably wait for v5.9 before making any changes to kube-proxy, this is just a tracking issue so we don't forget.\n/sig network /area ipvs\nShould probably hold on the PR until v5.9 is released later this year. cc\n/triage unresolved Comment when the issue is assessed and confirmed. I am a bot run by vllry. ‍\n/remove-triage unresolved\nYes it makes sense to wait until a kernel with the patch exists ;) I think we can also avoid touching it and follow the approach \"only touch the sysctl we absolutely need, otherwise trust the admin to know what they need\"\nI install linux-5.9-rc2 in centos7. set /proc/sys/net/ipv4/vs/connreusemode = 1 i have three service: service-a ---service-b --service-c in k8s. service-c return \"helloworld\" I use ab to benchmark service-a. [root ~]# ab -n -c 1000 http://172.17.2.66:/ This is ApacheBench, Version 2.3 &lt;Copyright 1996 Adam Twiss, Zeus Technology Ltd, Licensed to The Apache Software Foundation, Benchmarking 172.17.2.66 (be patient) Completed 1000 requests Completed 2000 requests Completed 3000 requests Completed 4000 requests Completed 5000 requests Completed 6000 requests Completed 7000 requests Completed 8000 requests Completed 9000 requests Completed requests Finished requests Server Software: Server Hostname: 172."}
{"_id":"doc-en-kubernetes-2c3bb3dca2a83bb247d336067e1c335109a2eb22fe4a06bf3bed6ebfcc858325","title":"","text":"17.2.66 Server Port: Document Path: / Document Length: 11 bytes Concurrency Level: 1000 Time taken for tests: 4.128 seconds Complete requests: Failed requests: 0 Write errors: 0 Total transferred: bytes HTML transferred: bytes Requests per second: 2422.76 [#/sec] (mean) Time per request: 412.753 [ms] (mean) Time per request: 0.413 [ms] (mean, across all concurrent requests) Transfer rate: 302.84 [Kbytes/sec] received Connection Times (ms) min mean[+/-sd] median max Connect: 0 16 71.5 4 1004 Processing: 4 216 92.9 202 2340 Waiting: 4 215 92.7 202 2340 Total: 4 232 121.5 215 2341 Percentage of the requests served within a certain time (ms) 50% 215 66% 253 75% 280 80% 299 90% 348 95% 412 98% 490 99% 556 100% 2341 (longest request) [root ~]# ab -n -c 1000 http://172.17.2.66:/ This is ApacheBench, Version 2.3 &lt;Copyright 1996 Adam Twiss, Zeus Technology Ltd, Licensed to The Apache Software Foundation, Benchmarking 172.17.2.66 (be patient) Completed 1000 requests Completed 2000 requests Completed 3000 requests Completed 4000 requests Completed 5000 requests Completed 6000 requests Completed 7000 requests Completed 8000 requests Completed 9000 requests Completed requests Finished requests Server Software: Server Hostname: 172.17.2.66 Server Port: Document Path: / Document Length: 11 bytes Concurrency Level: 1000 Time taken for tests: 7.071 seconds Complete requests: Failed requests: 0 Write errors: 0 Total transferred: bytes HTML transferred: bytes Requests per second: 1414.21 [#/sec] (mean) Time per request: 707.110 [ms] (mean) Time per request: 0.707 [ms] (mean, across all concurrent requests) Transfer rate: 176."}
{"_id":"doc-en-kubernetes-c35c3d64a37eee54a055f855f349a29d12f9312945295419d8a667917ed08399","title":"","text":"78 [Kbytes/sec] received Connection Times (ms) min mean[+/-sd] median max Connect: 0 9 66.2 2 1019 Processing: 10 661 575.2 372 3603 Waiting: 1 659 573.7 372 3603 Total: 10 670 584.6 380 3607 Percentage of the requests served within a certain time (ms) 50% 380 66% 742 75% 1008 80% 1183 90% 1535 95% 1784 98% 2376 99% 2479 100% 3607 (longest request) [root ~]# ab -n -c 1000 http://172.17.2.66:/ This is ApacheBench, Version 2.3 &lt;Copyright 1996 Adam Twiss, Zeus Technology Ltd, Licensed to The Apache Software Foundation, Benchmarking 172.17.2.66 (be patient) Completed 1000 requests Completed 2000 requests Completed 3000 requests Completed 4000 requests Completed 5000 requests Completed 6000 requests Completed 7000 requests Completed 8000 requests Completed 9000 requests Completed requests Finished requests Server Software: Server Hostname: 172.17.2.66 Server Port: Document Path: / Document Length: 11 bytes Concurrency Level: 1000 Time taken for tests: 13.580 seconds Complete requests: Failed requests: 0 Write errors: 0 Total transferred: bytes HTML transferred: bytes Requests per second: 736.40 [#/sec] (mean) Time per request: 1357.958 [ms] (mean) Time per request: 1.358 [ms] (mean, across all concurrent requests) Transfer rate: 92.05 [Kbytes/sec] received Connection Times (ms) min mean[+/-sd] median max Connect: 0 4 6.6 1 49 Processing: 11 1333 491.9 1307 3185 Waiting: 1 1330 491.9 1305 3185 Total: 17 1337 491."}
{"_id":"doc-en-kubernetes-8a42de3c47fe5733a8be9b0431ca4615a2a4c3e9126af1e5f3fc1676b07bf58b","title":"","text":"9 1311 3191 Percentage of the requests served within a certain time (ms) 50% 1311 66% 1565 75% 1679 80% 1748 90% 1911 95% 2129 98% 2384 99% 2650 100% 3191 (longest request) [root ~]# ab -n -c 1000 http://172.17.2.66:/ This is ApacheBench, Version 2.3 &lt;Copyright 1996 Adam Twiss, Zeus Technology Ltd, Licensed to The Apache Software Foundation, Benchmarking 172.17.2.66 (be patient) Completed 1000 requests Completed 2000 requests Completed 3000 requests Completed 4000 requests Completed 5000 requests Completed 6000 requests Completed 7000 requests Completed 8000 requests Completed 9000 requests Completed requests Finished requests Server Software: Server Hostname: 172.17.2.66 Server Port: Document Path: / Document Length: 11 bytes Concurrency Level: 1000 Time taken for tests: 23.720 seconds Complete requests: Failed requests: 936 (Connect: 0, Receive: 0, Length: 936, Exceptions: 0) Write errors: 0 Total transferred: bytes HTML transferred: bytes Requests per second: 421.58 [#/sec] (mean) Time per request: 2372.009 [ms] (mean) Time per request: 2.372 [ms] (mean, across all concurrent requests) Transfer rate: 47.77 [Kbytes/sec] received Connection Times (ms) min mean[+/-sd] median max Connect: 0 5 7.5 1 49 Processing: 8 2345 3364.8 333 Waiting: 0 1744 3030.9 277 Total: 8 2350 3364.8 340 Percentage of the requests served within a certain time (ms) 50% 340 66% 1434 75% 3390 80% 5490 90% 8669 95% 9233 98% 99% 100% (longest request) the last time tcp connections of service-b :\n/assign Sorry, but I do not understand what are you trying to say."}
{"_id":"doc-en-kubernetes-78960ac39e24e400e47e6916fdf5044032c9ae95fde2e36f307e50e5504537d0","title":"","text":": (\n/cc\nWill this be implemented in 1.21?\n/area kube-proxy\nIssues go stale after 90d of inactivity. Mark the issue as fresh with . Stale issues rot after an additional 30d of inactivity and eventually close. If this issue is safe to close now please do so with . Send feedback to sig-contributor-experience at . /lifecycle stale"}
{"_id":"doc-en-kubernetes-3d2f3dab1a1037d1573c264392a8fb35a501948caa8df40cf12a2cd970b123e2","title":"","text":"<!-- Please use this template while reporting a bug and provide as much info as possible. Not doing so may result in your bug not being addressed in a timely manner. Thanks! If the matter is security related, please disclose it privately via --command fails because of I think the link is from k/k's Am I right?? kubetest command succeeds With installed, OR Kubernetes version (use ): Cloud provider or hardware configuration: OS (e.g: ): Kernel (e.g. ): Install tools: Network plugin and version (if this is a network-related bug): Others:\nThis issue is currently awaiting triage. If a SIG or subproject determines this is a relevant issue, they will accept it by applying the label and provide further guidance. The label can be by org members by writing in a comment. <detailsInstructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\n/sig release /sig testing\n/assign\nNeed cherry-picks for 1.21, 1.20 to address this. is updated but downloaded downloads and executes its /reopen\nReopened this issue. <detailsIn response to : Need cherry-picks for 1.21, 1.20 to address this. Instructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\ncc\nthis was cherry picked AFAICT, thank you!"}
{"_id":"doc-en-kubernetes-9f300d1173b6c69710c4c78dc86f3d1b70206f2452121d5fb8dd9b7678b638cc","title":"","text":"The function in accepts an parameter which it describes as a way to \"controller specific flags to be included in the complete set of controller manager flags\". However, when the usage string is shown (either via or when parsing options fails), these additional options are not included in the usage string. This was identified via code inspection while trying to figure out what options did in . For e.g. , the output of should include a section corresponding to . This section should in turn list all of . a cloud provider executable that includes additional options, such as the additional options are not shown in the output No response_ n/a (all cloud providers are affected) n/a n/a n/a n/a\nThis issue is currently awaiting triage. If a SIG or subproject determines this is a relevant issue, they will accept it by applying the label and provide further guidance. The label can be by org members by writing in a comment. <detailsInstructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\n/sig cloud-provider"}
{"_id":"doc-en-kubernetes-9100812604bb274f029e79ebee13bfe2ce369a45e8a3a70a5fa62bc6c27fbac5","title":"","text":"<!-- Thanks for filing an issue! Before hitting the button, please answer these questions.--Is this a request for help? No What keywords did you search in Kubernetes issues before filing this one? static manifest Is this a BUG REPORT or FEATURE REQUEST? (choose one): Bug report Kubernetes version (use ): Environment: Cloud provider or hardware configuration: Bare metal and reproduced in Vagrant OS (e.g. from /etc/os-release): Kernel (e.g. ): Install tools: custom scratch install and reproduced with kubeadm Others: Calico for networking. Used (Also able to reproduce without calico) What happened: Modified one of the static manifest files in . The kubelet killed the running pod and did NOT create a new pod. (waited 30 minutes, still no pod) What you expected to happen: Kubelet should have killed the pod and started a new one. How to reproduce it (as minimally and precisely as possible): Started with single node, vanilla Ubuntu Xenial, installed kubernetes via kubeadm, nothing special. Modify , for example change the logging level or even just whitespace at the end of the file. Observe the kubelet kill the controller manager pod, and does not restart it. Anything else we need to know: If I remove the file from the directory, and then replace the file, kubelet will start it up. Kubelet logs from the period where I updated the file:\nDid you modify the yaml file using an editor (e.g., vim)? If the editor creates a temporary file in the same directory, kubelet can get confused and thinking that's a new pod (with the same name & namespace). When the temporary file gets removed, kubelet'd inadvertently delete the pod from its cache. Since kubelet relies on inotify for changes, if it doesn't get notified for any new changes from the original yaml file, it'd forever ignore the pod. There is an issue open to add periodic listing in addition to inotify (introduced in 1.5). This would make sure kubelet eventually recover from the situation. I can't find the issue, but there is another one for inotify: . Another fix is to make sure kubelet ignore such known temporary files (it already ignores files starting with ). In general, don't edit the file directly."}
{"_id":"doc-en-kubernetes-0117d4a9a190c22a46ca0d65f470fd275b0d70394b42857314fa11d87e816ed2","title":"","text":"Copy the file to another place, edit, and copy it back to the manifests directory.\nI'm seeing a similar problem except I never edited the file, might be fit for a separate issue\nYes I was using vim, and this explanation makes sense. Would be nice if it was smart about it. Would a PR to ignore the vim temp file be acceptable\nI think it's reasonable. I doubt if anyone uses those file names on purpose.\nThe dot ignore should have covered the vim temp file too, but it doesn't work because the filename passed in is the absolute filename including the path. So the first character is not a dot. I'll try to submit a PR fix.\nPR submitted\nfixed?\nHow to work around? I coudn't start api server when using kubeadm to deploy k8s. Thanks!"}
{"_id":"doc-en-kubernetes-942aeb92abd77b2734de9fd7148076d7883d9fdc897495b27676df96d5abc03a","title":"","text":"<!-- This form is for bug reports and feature requests ONLY! If you're looking for help check and the . If the matter is security related, please disclose it privately via --Is this a BUG REPORT or FEATURE REQUEST?: What happened: Not able to change the default storage class. What you expected to happen: Default storage class to change. How to reproduce it (as minimally and precisely as possible): But after few mins. Anything else we need to know?: Tried to remove the annotation but no luck. Environment: Kubernetes version (use ):Cloud provider or hardware configuration: minikube OS (e.g. from /etc/os-release): Buildroot 2018.05 Kernel (e.g. ): 4.16.14 Install tools: minikube Others:\n/assign\nThose labels are not set on the issue: <detailsIn response to : /remove-sig api-machinery Instructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\n/sig cluster-lifecycle\nThe cause is that kube-addon-manager is reconciling the standard storage class.\nThose labels are not set on the issue: <detailsIn response to : /remove-sig api-machinery Instructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\nThanks for looking into it. Sorry for native question, I want to keep only ceph as a default storage class how can I do that?\nhi I'm not really sure about the actual command to fix this. Basically, you could log into the VM and do the same modification as the PR above does, then restart minikube.\nHas the fix made it into minikube yet? I'm still seeing the issue in 0.31.0\nLooking at PR , the fix was made in kubernetes core, not minikube. The error still exists in minikube:"}
{"_id":"doc-en-kubernetes-e039746a8b3b20db88e55e1a23e49575ef9c4052776f0c6f0c425699707a07a4","title":"","text":"<!-- This form is for bug reports and feature requests ONLY! If you're looking for help check and the . If this may be security issue, please disclose it privately via --Is this a BUG REPORT or FEATURE REQUEST?: What happened: kubectl top nodes didn't report one of my nodes. I have a cluster of 8 and only 7 printed."}
{"_id":"doc-en-kubernetes-3b3a7005744c97bc2a779a3634e9ae4963750e882f0e9fc3a43aa2969cd1ed2b","title":"","text":"<pre:::: fang 10:04:02 (cgthayer-pchat) 0 platform; kubectl top nodes NAME CPU(cores) CPU% MEMORY(bytes) MEMORY% gke-charles-test-cluster-default-pool--jkrj 311m 33% 2470Mi 93% gke-charles-test-cluster-default-pool--krsh 154m 16% 1763Mi 66% gke-charles-test-cluster-default-pool--vhgq 265m 28% 1954Mi 73% gke-charles-test-cluster-default-pool--x3bs 186m 19% 1990Mi 75% gke-charles-test-cluster-default-pool--073f 270m 28% 1729Mi 65% gke-charles-test-cluster-default-pool--09zf 403m 42% 2698Mi 102% gke-charles-test-cluster-default-pool--p7h1 191m 20% 1582Mi 59% :::: fang 10:04:07 (cgthayer-pchat) 0 platform; kubectl get nodes NAME STATUS ROLES AGE VERSION gke-charles-test-cluster-default-pool--073f Ready <none7d v1.7.12-gke.1 gke-charles-test-cluster-default-pool--09zf Ready <none7d v1.7.12-gke.1 gke-charles-test-cluster-default-pool--cl65 Ready <none6d v1.7.12-gke.1 gke-charles-test-cluster-default-pool--jkrj Ready <none8d v1.7.12-gke.1 gke-charles-test-cluster-default-pool--krsh Ready <none8d v1.7.12-gke.1 gke-charles-test-cluster-default-pool--p7h1 Ready <none7d v1.7.12-gke.1 gke-charles-test-cluster-default-pool--vhgq Ready <none8d v1.7.12-gke.1 gke-charles-test-cluster-default-pool--x3bs Ready <none8d v1.7.12-gke.1 <preWhat you expected to happen: I expected all nodes to be printed."}
{"_id":"doc-en-kubernetes-ba6ca4dcc9f7eacf0c6e395f7904a902222263b10161fd5f101ea154df88d32e","title":"","text":"Since the missing node might have bad data I might expect it to have the same number of columns with \"-\" values and a note in a new 6th column saying something like \"unavailabe\" \"unresponsive\" \"stale-data\" etc. e.g. <pregke-charles-test-cluster-default-pool--p7h1 - - - - unreachable </preHow to reproduce it (as minimally and precisely as possible): Not sure. Perhaps run a pod where you can intentionally trigger it to use too much cpu or memory. Anything else we need to know?: Environment: Kubernetes version (use ): kubectl version Client Version: version.Info{Major:\"1\", Minor:\"8\", GitVersion:\"v1.8.6\", GitCommit:\"\", GitTreeState:\"clean\", BuildDate:\"2017-12-21T06:34:11Z\", GoVersion:\"go1.8.3\", Compiler:\"gc\", Platform:\"linux/amd64\"} Server Version: version.Info{Major:\"1\", Minor:\"8+\", GitVersion:\"v1.8.8-gke.0\", GitCommit:\"\", GitTreeState:\"clean\", BuildDate:\"2018-02-16T18:26:58Z\", GoVersion:\"go1.8.3b4\", Compiler:\"gc\", Platform:\"linux/amd64\"} Cloud provider or hardware configuration: GKE OS (e.g. from /etc/os-release): Kernel (e.g. ): Install tools: Others:\n/sig cli\n/kind feature\nI raise a PR to print when the metrics is missing. As for the 6th column you mentioned, I think it could be debatable for its necessity. :)"}
{"_id":"doc-en-kubernetes-d5b156fb720dd6621e83bb68112b4d593c6df8e76f8835be3c4f5531ff7b5602","title":"","text":"See Jenkins job kubernetes-upgrade-1.0-1.1-gke-step3-e2e-old's last few runs. The relevant line is . Note that this is running e2e tests off of the branch. It's possible that this is also caused by , but it's not clear to me.\nThis is also failing on kubernetes-upgrade-1.0-1.1-gke-step5-e2e-old, where we're running tests against a cluster that's been fully upgraded, (master and nodes,) from 1.0 to 1.1.\nStatus update: this may be caused by , so that issue should be prioritized, but it's not totally clear if that will fix this problem.\nsaid he would take a look and update the issue with any findings. It may be a bit premature to look into it before is resolved, but at least we can try to figure out how to manually repro this now, so that we will know whether fixes this issue.\nI'm unable to find logs from failed runs, which makes everything slightly harder. I'll try to reproduce the problem on GCE cluster, but it may not work. FYI: most of was caused by waiting for skydns to come back up (which doesn't happen: ). I see that there's a difference in number of system pods between version 1.1 and 1.0 - there's only 5 of them in 1.1 and 8 of them in 1.0. I'm wondering if this may cause problems in 'old' version of test which we're running there. I'm going to try reproduce it on GCE (as this is the setup I understand better) and if I'll fail I'll try again on GKE.\nAFAICT there's a problem with the test, not the upgrade itself. The test from 1.1 branch fails on 1.0 cluster.\nIt's caused by . Can we please stop renaming things? AFAICT the problem is with test only, everything should be working 'on production'. In v1.0 scheduler Events reasons start with a small letter (e.g. failedScheduling) and v1.0 test is looking for those kind of events to occur. In v1.1 we changed Event reasons to start with a capital letter (e.g."}
{"_id":"doc-en-kubernetes-a9b54e968993d2debcdfb36dbc5942943d975a32ddc74ac1464196bee7cec493","title":"","text":"FailedScheduling) and modified test accordingly. Effects are obvious. We can: revert patch it to v1.0 merge and patch a test fix which adds handling both versions () I'm for the last option, but whatever we do (except reverting) will break backward compatibility for everyone depending/counting/filtering/whatnot on scheduler Event reasons\nthanks for your work on this. From on : Furthermore, in my (rookie) opinion, seems like it should be considered a bug fix rather than an API change. Thus I think makes good sense.\nThanks a lot for debugging this, and I agree we should go with your PR. I talked to Brian and he said we don't consider changes to values of fields in events to be breaking API changes. Unfortunately people do seem to depend on these in various places. One solution is what I proposed in so the compiler could prevent these incompatibilities (at least for components and tests that are compiled together).\n+1 to creating a file containing constants for the event reasons that are currently in use.\nI'll prepare a cherry pick.\nReopening until cherry pick is merged.\nCherry pick is .\nThe test mentioned in the first entry in this thread is still failing but not due to the error message related to this issue. Since the cherrypick has been merged, can we close this issue, and followup with additional debugging in ?\nWorks for me. Thanks to again for your work on this.\nI actually still see failures of the same kind in (master 1.1, nodes 1.0) and (master 1.1, nodes 1.1). I assume this is because we haven't cherrypicked test fix into . So this is now a duplicate of ."}
{"_id":"doc-en-kubernetes-c04a0817a9f8c0cfea921bee7f0710f81a15653683e102754b69999644e85b55","title":"","text":"<!-- Please use this template while reporting a bug and provide as much info as possible. Not doing so may result in your bug not being addressed in a timely manner. Thanks! If the matter is security related, please disclose it privately via --What happened: What you expected to happen: kube-controller-manager should not panic. How to reproduce it (as minimally and precisely as possible): Kubernetes 1.12.6, , and Anything else we need to know?: It's running and , and logs says: Don't see any notable errors in . Is that because it's running the old ? Basically, is expecting , but somehow it's getting which I assume Environment: Kubernetes version (use ): v1.12.6 Cloud provider or hardware configuration: AWS EC2 OS (e.g: ): Amazon Linux 2 Kernel (e.g. ): 4.14.109-99.92.amzn2.x86_64\nCould be related to , but this happens with all the patches from the issue /sig scalability /cc\n/sig autoscaling /assign can you triage please?\n/sig-remove scalabilty\n/sig remove scalability\n/sig scalability remove\n/remove-sig scalability\nAPI returned v1.Status instead of v1beta2.MetricValueList. We should check what is the type of the object\n/sig instrumention\n/assign\n/sig instrumentation\nseems promising. Is it missing anything?\nI've root-caused the issue. Explained here -\nYep, should address the issue."}
{"_id":"doc-en-kubernetes-8ad6fcc1c21931d1df01e57d73eaefbe4f4eb3b6b6c8749d3543decdbcaf5bd3","title":"","text":"<!-- Please only use this template for submitting reports about failing tests in Kubernetes CI jobs --Which jobs are failing: N/A Which test(s) are failing: should function for node-pod communication: udp Since when has it been failing: Unknown. Testgrid link: The test is currently excluded. Reason for failure: By default, cmd is used to execute commands in Windows containers, but it has some issues executing the command (no output): echo 'hostName' | nc -w 1 -u <targetIP<targetPortBut it has no issues running (output is correct): echo hostName | nc -w 1 -u <targetIP<targetPortThis causes the test \"should function for node-pod communication: udp\" to fail when using Windows Containers. Anything else we need to know: /sig testing /sig windows"}
{"_id":"doc-en-kubernetes-2eda896003caba863d0609f67aa343029dbd4049babda0317e758512fcccca95","title":"","text":"<!-- This form is for bug reports and feature requests ONLY! If you're looking for help check and the . --Is this a BUG REPORT or FEATURE REQUEST?: /kind feature What happened: In some files, i found that there are many self-contained function, whose feature is as same as some util function. Obviously, they are redundant. What you expected to happen: Use util function to replace self-contained function How to reproduce it (as minimally and precisely as possible): Anything else we need to know?: Environment: Kubernetes version (use ): Cloud provider or hardware configuration**: OS (e.g. from /etc/os-release): Kernel (e.g. ): Install tools: Others:\n/assign\nThere are no sig labels on this issue. Please by: a sig: e.g., to notify the contributor experience sig, OR the label manually: e.g., to apply the label Note: Method 1 will trigger an email to the group. You can find the group list and label list . The in the method 1 has to be replaced with one of these: bugs, feature-requests, pr-reviews, test-failures, proposals\nsimilar comment to Can you edit the description in to say instead of ? Thanks!\nThanks. Done."}
{"_id":"doc-en-kubernetes-51391b67bb5f7406f2fc1910c3ffa31576cbedf4c6c7ca16b25a67ab404b314c","title":"","text":"TrimLeft returns a slice of the string s with all leading Unicode code points contained in cutset removed. To remove a prefix, use TrimPrefix instead. /sig release /kind bug\n/assign"}
{"_id":"doc-en-kubernetes-2fbecb3a2a825518f63af71ef5aa66b86ea1fb43ba7fb1eee661c2cc82f50635","title":"","text":"<!-- Please only use this template for submitting enhancement requests --What would you like to be : Update the help text for kubelet's flag to explicitly mention that it has no effect over the endpoint. Why is this needed: The help text for kubelet's flag describes it as \"Enable cAdvisor json and endpoints.\" which would seem to include , however this is not the case as that endpoint is required by kube-metrics-server and exposes stats from cadvisor or CRI (depending on the stat provider selected internally). Related Issues: /sig node /sig instrumentation /kind documentation\n/kind documentation /triage accepted /help\nThis request has been marked as needing help from a contributor. Please ensure the request meets the requirements listed . If this request no longer meets these requirements, the label can be removed by commenting with the command. <detailsIn response to : /kind documentation /triage accepted /help Instructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\n/assign This should be updated in as well as (unless there is some automated process that updates the website from the options in code that I'm not aware of).\nTo summarize a conversation in slack, it appears there is a documented procedure for updating the website found at The procedure is long, so it is tempting to just copy in the small amount of text that was changed from into a website PR rather than pulling down so many moving parts. In any case, we likely should keep this issue open until both the change to and an associated website change are merged.\ncan you link to the Slack convo?\nCan do - The conversation ended with a question whether there is a bug that would prevent an automated update of the documentation or if we need to submit a PR now that the fix has merged. I searched for an issue but didn't find a match. So we can either try to find someone on sig-docs who might know if there is an issue if it should be updated automagically, or just fix the doc if the website didn't get updated (and I just checked the master branch and it has not been updated yet).\nJust in case that is the right answer, I created the PR.\nI think the website gets autogenerated upon point releases but that'd be a good thing to clarify with the website team. I'll leave a comment asking as much."}
{"_id":"doc-en-kubernetes-5044536d87edb25c665421bf3dc4b536a0586e8213d754dc6188da659a17c1aa","title":"","text":"There are no sig labels on this issue. Please add a sig label. <detailsA sig label can be by either: a sig: e.g., to notify the contributor experience sig, OR the label manually: e.g., to apply the label Note: Method 1 will trigger an email to the group. See the . The in method 1 has to be replaced with one of these: bugs, feature-requests, pr-reviews, test-failures, proposals Instructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\nprobably caused by\nCould be a clash between these two?\ncc\nI still see the issue with local-cluster on the latest master: F0225 17:06:49. ] error: v1alpha1.KubeProxyConfiguration: FeatureGates: ReadMapCB: expect { or n, but found \", error found in byte of ......\nlooks like that's another issue, can you please file a new bug? with more logs? this one was about none of the hyperkube based processes starting up.\nwill do as it is also broke container kube-proxy as well. Same type of error."}
{"_id":"doc-en-kubernetes-4c28255cc86038c3c93f9e461780faa3ed1b46443f53f929b0e421cdbd186a35","title":"","text":"The purpose of the file for each Kubernetes repository is to provide a list of people who can assist the Kubernetes in the event that a security issue related to the repository is discovered or disclosed. As described in the file, those on the list should agree to our . Please update the  file for the sample-controller repository. After finding people who are willing to work in this capacity, you should add them to the list, then remove PSC members (except any PSC member will be working as a security contact for this repository). The list is GitHub usernames, optionally followed by an email address. If no email address is listed, the PSC will use the email address found on git commits made by the listed user. The file may already have people listed who are secuirty contacts. In that case, simply remove any PSC members who aren't also security contacts for the repo. See for more information /area security /committee product-security /kind cleanup /lifecycle frozen /priority important-soon /sig api-machinery"}
{"_id":"doc-en-kubernetes-33911e735262f8481400ca2430d671c6536468b71d87ed87bbe4f11626c79fd6","title":"","text":"The purpose of the file for each Kubernetes repository is to provide a list of people who can assist the Kubernetes in the event that a security issue related to the repository is discovered or disclosed. As described in the file, those on the list should agree to our . Please update the  file for the sample-apiserver repository. After finding people who are willing to work in this capacity, you should add them to the list, then remove PSC members (except any PSC member will be working as a security contact for this repository). The list is GitHub usernames, optionally followed by an email address. If no email address is listed, the PSC will use the email address found on git commits made by the listed user. The file may already have people listed who are secuirty contacts. In that case, simply remove any PSC members who aren't also security contacts for the repo. See for more information /area security /committee product-security /kind cleanup /lifecycle frozen /priority important-soon /sig api-machinery"}
{"_id":"doc-en-kubernetes-575caba005fa0a194c5e9ab05a9e30ee6530f864ab126bf649eed87c7efec61c","title":"","text":"The purpose of the file for each Kubernetes repository is to provide a list of people who can assist the Kubernetes in the event that a security issue related to the repository is discovered or disclosed. As described in the file, those on the list should agree to our . Please update the  file for the kube-aggregator repository. After finding people who are willing to work in this capacity, you should add them to the list, then remove PSC members (except any PSC member will be working as a security contact for this repository). The list is GitHub usernames, optionally followed by an email address. If no email address is listed, the PSC will use the email address found on git commits made by the listed user. The file may already have people listed who are secuirty contacts. In that case, simply remove any PSC members who aren't also security contacts for the repo. See for more information /area security /committee product-security /kind cleanup /lifecycle frozen /priority important-soon /sig api-machinery"}
{"_id":"doc-en-kubernetes-f1cb8df6316e5a200598bb646e70f7d130ec0c75fc9f39b3e957189b80c10ad2","title":"","text":"The purpose of the file for each Kubernetes repository is to provide a list of people who can assist the Kubernetes in the event that a security issue related to the repository is discovered or disclosed. As described in the file, those on the list should agree to our . Please update the  file for the code-generator repository. After finding people who are willing to work in this capacity, you should add them to the list, then remove PSC members (except any PSC member will be working as a security contact for this repository). The list is GitHub usernames, optionally followed by an email address. If no email address is listed, the PSC will use the email address found on git commits made by the listed user. The file may already have people listed who are secuirty contacts. In that case, simply remove any PSC members who aren't also security contacts for the repo. See for more information /area code-generation /area security /committee product-security /kind cleanup /lifecycle frozen /priority important-soon /sig api-machinery"}
{"_id":"doc-en-kubernetes-b6c24ff2339a072ce41f583a86461ffd390f0efd073046044ccabd0e8baad713","title":"","text":"The purpose of the file for each Kubernetes repository is to provide a list of people who can assist the Kubernetes in the event that a security issue related to the repository is discovered or disclosed. As described in the file, those on the list should agree to our . Please update the  file for the client-go repository. After finding people who are willing to work in this capacity, you should add them to the list, then remove PSC members (except any PSC member will be working as a security contact for this repository). The list is GitHub usernames, optionally followed by an email address. If no email address is listed, the PSC will use the email address found on git commits made by the listed user. The file may already have people listed who are secuirty contacts. In that case, simply remove any PSC members who aren't also security contacts for the repo. See for more information /area security /committee product-security /kind cleanup /lifecycle frozen /priority important-soon /sig api-machinery"}
{"_id":"doc-en-kubernetes-773ad9c7c77ed09f247ebe6fe4a2e977a2482d0db8567e965fc0d687730a34f0","title":"","text":"The purpose of the file for each Kubernetes repository is to provide a list of people who can assist the Kubernetes in the event that a security issue related to the repository is discovered or disclosed. As described in the file, those on the list should agree to our . Please update the  file for the apiserver repository. After finding people who are willing to work in this capacity, you should add them to the list, then remove PSC members (except any PSC member will be working as a security contact for this repository). The list is GitHub usernames, optionally followed by an email address. If no email address is listed, the PSC will use the email address found on git commits made by the listed user. The file may already have people listed who are secuirty contacts. In that case, simply remove any PSC members who aren't also security contacts for the repo. See for more information /area apiserver /area security /committee product-security /kind cleanup /lifecycle frozen /priority important-soon /sig api-machinery"}
{"_id":"doc-en-kubernetes-fad8ee5585f13cbd059c3b23e9b0beba24a9e5cc8dc4cbda61f103e29d873595","title":"","text":"The purpose of the file for each Kubernetes repository is to provide a list of people who can assist the Kubernetes in the event that a security issue related to the repository is discovered or disclosed. As described in the file, those on the list should agree to our . Please update the  file for the apimachinery repository. After finding people who are willing to work in this capacity, you should add them to the list, then remove PSC members (except any PSC member will be working as a security contact for this repository). The list is GitHub usernames, optionally followed by an email address. If no email address is listed, the PSC will use the email address found on git commits made by the listed user. The file may already have people listed who are secuirty contacts. In that case, simply remove any PSC members who aren't also security contacts for the repo. See for more information /area security /committee product-security /kind cleanup /lifecycle frozen /priority important-soon /sig api-machinery"}
{"_id":"doc-en-kubernetes-df9e56efa1a5d0dfe5cf4c9ff5d0ae22d67682beafe379ced520bca8e339d7ed","title":"","text":"The purpose of the file for each Kubernetes repository is to provide a list of people who can assist the Kubernetes in the event that a security issue related to the repository is discovered or disclosed. As described in the file, those on the list should agree to our . Please update the  file for the apiextensions-apiserver repository. After finding people who are willing to work in this capacity, you should add them to the list, then remove PSC members (except any PSC member will be working as a security contact for this repository). The list is GitHub usernames, optionally followed by an email address. If no email address is listed, the PSC will use the email address found on git commits made by the listed user. The file may already have people listed who are secuirty contacts. In that case, simply remove any PSC members who aren't also security contacts for the repo. See for more information /area security /committee product-security /kind cleanup /lifecycle frozen /priority important-soon /sig api-machinery"}
{"_id":"doc-en-kubernetes-dbffed5438f566ecf1b631adacf4d68392348225a8f199c6d9953211f55b24b3","title":"","text":"This is the script that runs to capture event output during e2e runs: It's failing after the recent pflag merge:\ncc\nOr something. I might be completely off. But something broke it recently enough, and I know like his event log.\nThat command is actually invalid. is a flag on the command, not the root. Fix in"}
{"_id":"doc-en-kubernetes-26d82dfedb03b77df6b1c9ca183f41f5576785fb4175159c538818ffb957183e","title":"","text":"kubectl log stops working for me all the sudden: A little bit digging into the issue, I found the problem is only limited to /v1beta3: Query /v1beta1 Query through v1beta3\ncc/\nDoes replacing \"minions\" by \"nodes\" work?\nCool, looks like it should be a simple kubectl fix, I will do it shortly."}
{"_id":"doc-en-kubernetes-f4fbd406d4e0e170dfafc3e7da28855114900a8241a757fc2216c8af6fba3e29","title":"","text":"<!-- Please only use this template for submitting reports about flaky tests or jobs (pass or fail with no underlying change in code) in Kubernetes CI --Which jobs are flaking: pull-kubernetes-e2e-gce ci-kubernetes-e2e-gci-gce Which test(s) are flaking: [sig-apps] CronJob should delete successful/failed finished jobs with limit of one job Testgrid link: Reason for failure: Anything else we need to know: this is the top flake for pull-kubernetes-e2e-gce: triage link: specific failures /sig apps /priority important-soon\nthis is the second flakiest test (failing on ~15% of PR runs)\nI'll look into it. /assign\nif there's not an obvious immediate fix, I would recommend: the test as to skip the test in merge-blocking job runs a second PR that reverts the change and adds debugging commits and do on that PR to flush out issues this issue open and in the current milestone to track resolution of the test before release (either by fixing the test, finding/fixing an underlying bug, or proving we have test coverage of the function via another method and deleting the test)\nmarked as in , will leave this issue open to track root causing, fixing, restoring the test before release\nTo follow up here, this was confirmed as root-caused, fixed and restored by"}
{"_id":"doc-en-kubernetes-d17b1ff8ae06b13c6af5817488a3cb18d76c8f278663cf75b8a3d7d0d9abccc1","title":"","text":"Is this a BUG REPORT or FEATURE REQUEST?: /kind bug What happened: After starting controller manager, get the following panic: /---------------------------------- I0712 08:54:30. 1 ] Deleting route 10.245.1.0/24 10.245.1.0/24 I0712 08:54:30. 1 ] Creating route for node ab-6l35lzekrg-0-7igjn3vr43tw-kube-minion-abhezrwnqucq 10.245.1.0/24 with hint -655c-11e7-9a48-, throttled 1.602µs I0712 08:54:30. 1 ] Creating route for node ab-hfvphu5tyl-0-623ksgxczeoh-kube-master-oevhfzd3xo7s 10.245.0.0/24 with hint -655c-11e7-9a48-, throttled 651ns I0712 08:54:31. 1 ] Created route for node ab-6l35lzekrg-0-7igjn3vr43tw-kube-minion-abhezrwnqucq 10.245.1.0/24 with hint -655c-11e7-9a48- after 349.278413ms I0712 08:54:32."}
{"_id":"doc-en-kubernetes-def0ff2d7afb1041ed38dd536114d4f932b578dcf3a59e0d27da6b34b247c308","title":"","text":"1 ] Created route for node ab-hfvphu5tyl-0-623ksgxczeoh-kube-master-oevhfzd3xo7s 10.245.0.0/24 with hint -655c-11e7-9a48- after 1.583805322s panic: runtime error: index out of range goroutine 1250 [running]: panic(0x3a3d4e0, 0xc42001a080) +0x1a1 (Routes).DeleteRoute(0xc4212e15a0, 0x40ff60e, 0xa, 0xc421bf22a0, 0x0, 0x0) +0x7d5 (RouteController).reconcile.func2(0xc4210bed70, 0xc421bf05e0, 0xc421bf22a0, 0xed0f7dac6, 0x3264f461, 0x6f143e0) +0x15a created by (RouteController).reconcile +0x6e6 /---------------------------------- What you expected to happen: Successed to delete route. How to reproduce it (as minimally and precisely as possible): verytime Anything else we need to know?: Environment: Kubernetes version (use ): v1.6.4 Cloud provider or hardware configuration*: OS (e.g. from /etc/os-release): Kernel (e.g. ): Install tools: Others: cc /sig-openstack\nThere are no sig labels on this issue. Please by:<br(1) mentioning a sig: <bre.g., for API Machinery<br(2) specifying the label manually: <bre.g., for sig/scalability<br<brNote: method (1) will trigger a notification to the team. You can find the team list and label list\ncc\nReiterating the mentions to trigger a notification: <detailsIn response to : cc Instructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details"}
{"_id":"doc-en-kubernetes-aa5745d62aee7fc04015b35348d752ab217dfca76565e758f0ca6e53fe701514","title":"","text":"HPA tests are currently failing in gke-large cluster suite: I did some debugging and the problem seems to be related to pod-nanny. Looking into logs from the last run: was first scheduled on the large node dedicated to it pod-nany updated heapster deployment then heapster was scheduled on the normal node (when it actually shouldn't fit) that point it was never updated by the pod nanny. that point, heapster started crashlooping (OOMing) [which in fact was expected in too small machine] So it's pretty clear that that pod-nanny changed the amount of resources to something smaller, which in fact it shouldn't."}
{"_id":"doc-en-kubernetes-9ba58a38d1b26cddbe35d0c0365909f33d458f500f4464f9dc0ae31755bbc823","title":"","text":"What happened: The rbd plugin use \"rbd info\" error: The messages before and after json result is warning string and no effect to rbd info, but it is effect to parse the output. So the kubernetes will failed: What you expected to happen: Kubernete no failed and errors How to reproduce it (as minimally and precisely as possible): Remove the config file\n/sig storage\nWe are blocked by this issue, can't resize any PVCs\nIssues go stale after 90d of inactivity. Mark the issue as fresh with . Stale issues rot after an additional 30d of inactivity and eventually close. If this issue is safe to close now please do so with . Send feedback to sig-testing, kubernetes/test-infra and/or . /lifecycle stale\n/remove-lifecycle stale\nThat is another situation that cluster configuation is empty, aslo will return lots of warnning information: Also will trigger expand volume failed. But should we add -c=/dev/null in rbd info command?\nFolks, the pull request that would have fixed this just got closed as it was rotting away. This is an actual problem in production, can we please fix this?\n+1 we are still being plagued by this issue...\n+1 {\"name\":\"kubernetes-dynamic-pvc--147e-422c-9e67-\",\"size\":,\"objects\":2560,\"order\":22,\"objectsize\":,\"blocknameprefix\":\"rbddata.\",\"format\":2,\"features\":[\"layering\"],\"flags\":[]}, invalid character '-' after top-level value\nI solved by exceuting this command\nWow...I was not expecting this to work ... but it did ;) Any insights on why is it working with just the presence of the file? This is very strange... Appreciate any comments..."}
{"_id":"doc-en-kubernetes-ec905797ef340d09da7d9d2e34a2767d786c5cb37de40d4b7cae6a294742089b","title":"","text":"What happened: Some of the Azure CSI translation checks in and needs to perform NPE checks for pointer based fields as pointed out in and What you expected to happen: NPE checks should be there How to reproduce it (as minimally and precisely as possible): Anything else we need to know?: Environment: Kubernetes version (use ): v1.15 Cloud provider or hardware configuration: OS (e.g: ): Kernel (e.g. ): Install tools: Network plugin and version (if this is a network-related bug): Others:\n/assign\n/sig storage\nIssues go stale after 90d of inactivity. Mark the issue as fresh with . Stale issues rot after an additional 30d of inactivity and eventually close. If this issue is safe to close now please do so with . Send feedback to sig-testing, kubernetes/test-infra and/or . /lifecycle stale\nStale issues rot after 30d of inactivity. Mark the issue as fresh with . Rotten issues close after an additional 30d of inactivity. If this issue is safe to close now please do so with . Send feedback to sig-testing, kubernetes/test-infra and/or . /lifecycle rotten\n/remove-lifecycle rotten\n/priority critical-urgent"}
{"_id":"doc-en-kubernetes-e28bd845629ca30049478b8fcedafd43696075613f16aad0ce76758069ac8ba0","title":"","text":"For example, kubernetes-ro service appears with empty selector (this service was created by the hack/local cluster) - is that ok? What meaning such service has if it has no entity to fwd traffic to?\nkubernetes and kubernetes-ro are currently \"virtual\" - they are not backed by running pods, but by the master itself. , abonas wrote:\nif that's internal impl, why is it exposed to user? let's say a client pulls all entities and builds a UI around it. then a service with no selector arrives, and it's internal to kube - how the client should know not to present it/touch it? (similar to the \"net\" container in pods)\nIt's exposed exactly FOR users - so they can find the apiserver. It has a corresponding Endpoints object, it gets a portal IP and a kube-proxy rule, and it gets service env vars and a DNS name. The fact that it is virtual is an implementation detail. That it exists is all users should need to know. , abonas wrote:\nAnd FWIW, the master components may one day actually be pods... , Tim Hockin wrote:\nthanks for the explanation. so any general rule how to distinguish kube pods/services from user defined?\nDo we need to distinguish them? We could add labels, I guess, if it is really important to distinguish them.\nsometimes users would want to see only the things they defined and not implementation details of the system.\nOne could argue this is why the users should be using labels. I'm not against defining labels for \"special\" stuff, but it's new territory, so we should enter with caution. , abonas wrote:\nI think the right answer might be to put them in a special \"cluster\" or \"system\" namespace. , Tim Hockin wrote:\nSee If we put them in a \"kubernetes\" ns, the DNS name would be :) I'm not against segmenting them into a different namespace. We just need to decide on what that namespace is called, plumb through that configuration, decide on compatibility for a short while or not, and decide if that namespace is \"special\" or just \"different\" - i.e. should the skydns, heapster, etc pods go into this namespace? , Brendan Burns wrote:\nAs said, the services are there for applications that need to contact the apiserver."}
{"_id":"doc-en-kubernetes-d870c86aebdf975c7f742c9ae5f4b46feed435261a0bfef8e8c1547e324212d5","title":"","text":"Additionally, nil selectors can be used by users to import external services. Adding labels to the kubernetes services won't help. Users need to manage their objects using labels. /cc\nYeah, lots of consumers will expose \"special\" services. IN fact, in many clusters the \"special\" services will outweigh the unspecial ones. At this point, the namespace they are in is completely arbitrary, and I think because it is a name, it should be free for the end user to choose where they are surfaced.\nPer the proposal here: , I think the namespace should continue to hold these services, and they require no special labels. I sense comment is made without thinking through what Kubernetes operators do to bootstrap a cluster. I think it's reasonable that namespace holds everything you need to run, and operators may or may not choose to segment their data in other namespaces. If they want to not see this information, its an exact reason for why they should use their own namespace. I don't think we need anything new here.\n- I'm not clear on your meaning. Did you mean that we should make the \"cluster admin\" namespace a config param that everyone knows about and can be different cluster-to-cluster? Or did you mean that default is as arbitrary a name as any other? I'd be fine moving things out of default into \"kube\" or \"kubernetes\" or \"cluster\" or something, but I am less convinced that a flag is a good thing. , Clayton Coleman wrote:\n- wherever we stick this stuff, it should be the namespace that is initially bootstrapped, which today is . I continue to believe that is as good a name as any other. I do not see a need to do more here.\nI think as soon as you bring naming into the picture, making decisions based on names gets very tricky (what if I want my logging service to be everywhere?). If there is a magic namespace that exposes things to the rest of the cluster, we should ask whether there's only one magic namespace, or multiple? Are services supposed to cross namespaces? It's great that pods can find the master... but that's not the only thing they'll need to find in each namespace. And baking in special names (or customizable names) into your app code is fragile, because then your code only works inside Kubernetes."}
{"_id":"doc-en-kubernetes-19a1388f52e9d713eb0f4efe1574687db54358876426fa8c067da9ba3c2f2fbb","title":"","text":"I'm just trying to pump the brakes on special services prior to us designing them... :) ----- Original Message -----\noh, to be clear, I don't want \"special\" services - I am merely asking if moving non-user-defined objects (where that is defined very loosely) out of \"default\" is worthwhile. I tend to agree with Derek that it's not worth energy. , Clayton Coleman wrote:\nagree with you that lots of consumers will expose \"special\" services We should not use the (namespace + k8s-service + skydns) mechanism for publishing the location of those special services. We should decouple the choice of namespace to run the pods of the service in from the well-known dns address of special services. Otherwise, it will be hard to move the pods of the service to another namespace, which people will definitely want to do as they grow. Instead, we need a place to publish well-known services which is not tied to the namespace of the pods of that service.\nyou open a whole new topic. I've noodled with the idea of a Name resource type. We would generate a Name for each Service as \"service.namespace\", but users could also add their own Name resources. , Eric Tune wrote:\nI'd like to understand better your comment regarding nil selectors used to export external services - can you elaborate? give an example? my understanding is that service redirects traffic to pods based on labels. with nil selector service does nothing, so why is it a service in the first place? I'm pro decoupling special services/internal implementation services from user defined entities. whether by separate namespace, or by saying that the default namespace is reserved, or by any other solution. because otherwise internal implementation and user defined content are mixed and that's no good for a proper management. just tagging you so you could see this issue and the related discussions.\nServices abstract backends - they don't have to abstract pods. Some examples: You want to have an external database cluster in production, but in test you use your own databases. You would have a service for both and use the environment variables, but the production service has a nil selector with its endpoints set to the database cluster. You want to have an empty service because you're reserving a name (that isn't accessible yet) but will be in the future You want to point your service to a service in another namespace or on another cluster."}
{"_id":"doc-en-kubernetes-1ddfea6d99e7ccf8c03b659f81beb02dd771f14183ebec47ce91a7703edcbce2","title":"","text":"You set the endpoints for your service to the publicIPs of the other service. ----- Original Message -----\n, thanks. regarding the last point - I was under impression (based on some other discussion) that there would not be cross namespace interaction? that entities can can access only entities in the same namespace?\nCross-namespace is currently allowed and I think has to be allowed, though perhaps (in the fullness of time) that can be policy based rather than wide open. , abonas wrote:\nplease see this correspondence. my understanding is that there is isolation\nResponded there. I think there's some less-than-crystal-clear words in play. , abonas wrote:\n+1 to Eric's comment about ensuring there is a consistent manner for system services like kubernetes-ro, if that's not fixes for all clusters it's going to be a nightmare to write reusable containers for multiple different clusters. I don't care if we do that by using a constant namespace, or by giving system services two different names, but there needs to be at least one name that is constant. Regarding default vs. cluster namespace, I think the issue is one of clutter as a user is starting to understand k8s, you turn up a new cluster, launch a pod, or service, and it's hard to find your one object on the sea of system created pods and services. It's visual clutter that confuses people as they onboard. On Jan 23, 2015 3:01 PM, \"Tim Hockin\" wrote:\n- can you drop me a line (my github name at google) - I want to ask you something, and you have no contact info :) , abonas wrote:\nsorry for naive question but why would I care as an user that wants to run its existing workload on Kubernetes about kubernetes-ro ? If I write an application that uses kubernetes-ro then I'm locked into Kubernetes. I understand the need for kubernetes-ro being accessible by cluster level services such as monitoring, logging. I also can imagine that a power user might want to start its own cluster services that use kubernetes-ro. I don't see clearly why would majority of users care about kubernetes-ro being visible from their container."}
{"_id":"doc-en-kubernetes-f742a948ec6f92870d242caf8e3787bfcbe2f4250e3c7f0a894622bd17960df0","title":"","text":"I see a reason for the cluster services being visible and accessible for Kubernetes developer, but from an user perspective such services only make the system more complex: why do I see these services (pods,rc) ? are they related to my application ? should I care about them (especially when I use hosted Kubernetes) ? why does my container get all these environment variables for cluster services ?\nI will update file to explain why one would have service without a selector.\nRight now, every service goes everywhere for simplicity. Name spacing and filtering probably make sense in the long run, but it's not a high priority for v1.0 Brendan On Mar 4, 2015 5:02 AM, \"Rafał Sokołowski\" wrote:"}
{"_id":"doc-en-kubernetes-4cf2bcce2d9cf8bfbc9f2302c057ef4fa219c655a3340b1abe23f2bb07c57620","title":"","text":"In volume management code, VolumeSpec is firstly needed to find the volume plugin. After the volume plugin is found, VolumeSpec is again used to attach or mount or check volume is still attached or not. This information is reconstructed during controller failure. The assumptions used here are If pod still exists, VolumeSpec should be reconstructed when controller is restarted. However, as discover from this bug, it might not be true if PVC is deleted in the meantime because pod only has PVC information and requires to use PVC to get PV information. By fixing Issue , we could fix this problem.If pod is already deleted, VolumeSpec is not really needed for detaching. But before detaching finishes, syncState loop might check this volume's attachement status and will use VolumeSpec during checking. One safe guard we can put here is to fail FindAttachablePluginBySpec if spec is nil. This function is used by multiple places such as attachdetachcontroller, desiredstateofworld, and actualstateofworld, oeperation_generator etc. Instead of adding nil check many places, I feel FindAttachablePluginBySpec or even better FindPluginBySpec (is called by FindAttachablePluginBySpec and other places). Open an issue for this\nIssues go stale after 90d of inactivity. Mark the issue as fresh with . Stale issues rot after an additional 30d of inactivity and eventually close. Prevent issues from auto-closing with an comment. If this issue is safe to close now please do so with . Send feedback to sig-testing, kubernetes/test-infra and/or . /lifecycle stale\nStale issues rot after 30d of inactivity. Mark the issue as fresh with . Rotten issues close after an additional 30d of inactivity. If this issue is safe to close now please do so with . Send feedback to sig-testing, kubernetes/test-infra and/or . /lifecycle rotten /remove-lifecycle stale\nRotten issues close after 30d of inactivity. Reopen the issue with . Mark the issue as fresh with . Send feedback to sig-testing, kubernetes/test-infra and/or . /close"}
{"_id":"doc-en-kubernetes-6fd44da75083eb708bb282062daff0367437c77cffc0b305712ab45fdcf69fe6","title":"","text":"recently merged, adding nodes as something we can put in bound object refs (yay!) Today while reading noderestriction code for a different reason, I realized we never extended noderestriction to block Kubelets from binding tokens to nodes other than themselves. Today, noderestriction does block Kubelets from binding tokens to Pods on other nodes, so perhaps it should do the same for bound node references. noderestriction to block it, probably. Read the code here: And note that we didn't modify in No response Not yet in any released version of K8s. N/A N/A N/A N/A N/A\n/triage accepted\nTo summarize Slack discussion: TokenRequest can only bind to one object at a time. kubelet (node identity) is already required to bind to Pod by noderestriction. kube-apiserver automatically inserts the node information based on the Pod binding. kubelet does not set it. So this is technically still node bound. We should still add a comment in to avoid future confusion. There may be other reasons to add a check, like future proofing..."}
{"_id":"doc-en-kubernetes-5bc96dc4d12a3dcdc2937e744d86eb7f421f613963225d2302a19e261319d9fc","title":"","text":"Seen in and in No hits on stack strace nothing evident on the git history of the test itself, but it seems , so most probable something in the device manager has changed? /sig node /kind flake\nThis issue is currently awaiting triage. If a SIG or subproject determines this is a relevant issue, they will accept it by applying the label and provide further guidance. The label can be by org members by writing in a comment. <detailsInstructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\n/assign\n/cc as this flakes a lot in"}
{"_id":"doc-en-kubernetes-b371ee8508434fd0fa06120fe5733e14a37c0788b05c28b7017350eace3dad7d","title":"","text":"This was build from master earlier today.\ncc This should be fixed before v1.4\nThis is critical\nThis needs to be tracked as a 1.4 blocker\nETA for resolution: This afternoon This is in the process of being cherrypicked"}
{"_id":"doc-en-kubernetes-151038cc19e020d7d7996b734b8e9e17c5611ef500523f3ad799d063a04956b9","title":"","text":"What would you like to be : The Event API is being promoted to v1 in and right now v1beta1 has higher priority than v1 in We should bump this in 1.20. Why is this needed: The new Event API is being promoted.\n/sig instrumentation\nIssues go stale after 90d of inactivity. Mark the issue as fresh with . Stale issues rot after an additional 30d of inactivity and eventually close. If this issue is safe to close now please do so with . Send feedback to sig-testing, kubernetes/test-infra and/or . /lifecycle stale\nStale issues rot after 30d of inactivity. Mark the issue as fresh with . Rotten issues close after an additional 30d of inactivity. If this issue is safe to close now please do so with . Send feedback to sig-testing, kubernetes/test-infra and/or . /lifecycle rotten\nRotten issues close after 30d of inactivity. Reopen the issue with . Mark the issue as fresh with . Send feedback to sig-testing, kubernetes/test-infra and/or . /close\nClosing this issue. <detailsIn response to : Rotten issues close after 30d of inactivity. Reopen the issue with . Mark the issue as fresh with . /close Instructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\n/reopen\nReopened this issue. <detailsIn response to : /reopen </details\nThis issue is currently awaiting triage. If a SIG or subproject determines this is a relevant issue, they will accept it by applying the label and provide further guidance. The label can be by org members by writing in a comment. <detailsInstructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details"}
{"_id":"doc-en-kubernetes-88a6665f471219ba222a644b43576808b71f5f4fb39e5a18e0bc0e94ac9cb36e","title":"","text":"Two problems: on node do not have namepace, but kubectl requires namespace. However, if explicitly set namespace to 'default', this will cause an error because we are updating not allowed field (only label update is allowed). is usually created by node controller, update node label with a file is tedious since user needs to create a file with id, resource.capacity, etc. can we make it easier?\nIt would make more sense to have a special global namespace than no namespace. We should make it easy in kubectl to update a single field or subset of fields, generally, for all objects. What this would do is GET, merge, and PUT.\nI had a discussion with on IRC about namespace on node. We'd like to just use \"\" (empty string) as of now. As we are not supporting any namespace feature on node, it's less confusing if we can hide the concept for now. Thinking a little more, \"\" is the special global namespace. I'm going to update kubectl to not assume 'default' namespace on every resources. It works fine for pod, service, but not for node, external LB, etc.\nThere was some discussion about \"\" being (essentially) an alias to \"default\": is that dead? , Deyuan Deng wrote:\nI've seen two versions of NamespaceALL (\"\") NamespaceAll is the default argument to specify on a context when you want to list or filter resources across An empty namespace is equivalent to the , but \"default\" is the canonical representation. Either means , or the definition conflicts - all namespace should include \"default\" namespace. Also, for things like Node, I'd really like a NamespaceNone.\n- I think there is a confusion here because of a poor naming choice. is a constant intended to be used as a filter to LIST/WATCH across multiple namespaces. It is NOT intended to be the value that is set on a resource for persistence. The discussion was to update to which I think is still a good thing to do. I have held off doing that until after we embed the namespace as a path parameter in our URLs instead of a query parameter, which I plan to look at after the holidays. is a constant intended to be used as the namespace context of a request when it's not explicitly stated on a request URL."}
{"_id":"doc-en-kubernetes-5a4d1d083bc5e635544966baa95a7511fe688574aded9d8644d8c7753070f0a1","title":"","text":"Since create operations pull the namespace from the context during persistence, it is also the value intended to be persisted. This is referring to a request URL that does not specify a namespace. It is not referring to the persisted value in the database directly - although that is the value that gets pulled. I think a is a good thing to add AFTER we make the change to specify namespace as a path parameter instead of a query parameter, and we update/rename to something that clarifies that its a wild-card filter whose value is probably best represented as.\nok, sounds good. Thanks. Look forward to see api.NamespaceAll = \"*\" and api.NamespaceNone :)"}
{"_id":"doc-en-kubernetes-d8638ef20bd789f7becbb6da887bf6c67832db9cba785d76d8909f1b2b419007","title":"","text":"<!-- Please only use this template for submitting enhancement requests --What would you like to be : Sometimes reported in fluent-plugin-elasticsearch repository for exhausted Elasticsearch cluster issue due to elasticsearch ruby client sniffering issue. This should be fixed with but k/k fluentd-elasticsearch image doesn't use this. Why is this needed: This change should be also applied in the downstream docker images and using helm charts such as\n/sig instrumentation\n/sig cloud-provider\n/assign cosmo0920\nIssues go stale after 90d of inactivity. Mark the issue as fresh with . Stale issues rot after an additional 30d of inactivity and eventually close. If this issue is safe to close now please do so with . Send feedback to sig-testing, kubernetes/test-infra and/or . /lifecycle stale\n/remove-lifecycle stale\n/assign\nIssues go stale after 90d of inactivity. Mark the issue as fresh with . Stale issues rot after an additional 30d of inactivity and eventually close. If this issue is safe to close now please do so with . Send feedback to sig-contributor-experience at . /lifecycle stale\n/remove-lifecycle stale\nIssues go stale after 90d of inactivity. Mark the issue as fresh with . Stale issues rot after an additional 30d of inactivity and eventually close. If this issue is safe to close now please do so with . Send feedback to sig-contributor-experience at . /lifecycle stale\nThe Kubernetes project currently lacks enough active contributors to adequately respond to all issues and PRs. This bot triages issues and PRs according to the following rules: After 90d of inactivity, is applied After 30d of inactivity since was applied, is applied After 30d of inactivity since was applied, the issue is closed You can: Mark this issue or PR as fresh with Close this issue or PR with Offer to help out with [Issue Triage][1] Please send feedback to sig-contributor-experience at . /lifecycle rotten [1]:\nThe Kubernetes project currently lacks enough active contributors to adequately respond to all issues and PRs."}
{"_id":"doc-en-kubernetes-cfbf31e1193e1637a12fc4a38528199469147183babcacd44c068bee92a1aeb9","title":"","text":"This bot triages issues and PRs according to the following rules: After 90d of inactivity, is applied After 30d of inactivity since was applied, is applied After 30d of inactivity since was applied, the issue is closed You can: Reopen this issue or PR with Mark this issue or PR as fresh with Offer to help out with [Issue Triage][1] Please send feedback to sig-contributor-experience at . /close [1]:\nClosing this issue. <detailsIn response to : The Kubernetes project currently lacks enough active contributors to adequately respond to all issues and PRs. After 90d of inactivity, is applied After 30d of inactivity since was applied, is applied After 30d of inactivity since was applied, the issue is closedReopen this issue or PR with Mark this issue or PR as fresh with Offer to help out with [Issue Triage][1] Instructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details"}
{"_id":"doc-en-kubernetes-dec167f232fc78679b57ee7f7f6400be17dd5e43f5f10aa50473841ecfd59319","title":"","text":"Is this a BUG REPORT or FEATURE REQUEST?: /kind bug What happened: During startup shows the following warning: However is set and refers to a valid X.509 private key file. All service accounts seem to be created despite the warning. How to reproduce it (as minimally and precisely as possible): start controller-manager with Anything else we need to know?: Environment: Kubernetes version (use ): v1.8.0 OS (e.g. from /etc/os-release): Debian GNU/Linux 8 (jessie)\nThere are no sig labels on this issue. Please by: a sig: e.g., to notify the contributor experience sig, OR the label manually: e.g., to apply the label Note: Method 1 will trigger an email to the group. You can find the group list and label list . The in the method 1 has to be replaced with one of these: bugs, feature-requests, pr-reviews, test-failures, proposals\nThanks for the report. The condition for printing the warning is backwards, but has no effect on behavior. Fixed in\nStill present in 1.8.3 , can it be ~backported~ cherry-picked to 1.8 branch please?\nSee for proposed cherry-pick"}
{"_id":"doc-en-kubernetes-a416b49d9932f64618fc25d3906fc3f29787f4b2fe9d0ffb0b804ce600654c09","title":"","text":"Following up on , evaluate if the test for local SSD on GKE clusters is still necessary (it is also tested internally in Google). /assign\n/sig storage\n/triage accepted\nIssues go stale after 90d of inactivity. Mark the issue as fresh with . Stale issues rot after an additional 30d of inactivity and eventually close. If this issue is safe to close now please do so with . Send feedback to sig-contributor-experience at . /lifecycle stale\n/lifecycle frozen\nThis issue has not been updated in over 1 year, and should be re-triaged. You can: Confirm that this issue is still relevant with (org members only) Close this issue with For more details on the triage process, see /remove-triage accepted\nThis issue is currently awaiting triage. If a SIG or subproject determines this is a relevant issue, they will accept it by applying the label and provide further guidance. The label can be by org members by writing in a comment. <detailsInstructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\nThis test should be removed as it's using a GKE feature that, while not deprecated, is discouraged in favor of newer APIs. Those APIs are well-tested internally in google, and I don't see value in additional toil upstream to maintain them."}
{"_id":"doc-en-kubernetes-f3221c09c6d59949b8ff055be19bd249df646b41606c69fa43fd98ef3f599715","title":"","text":"On a 1.3.6 master: But on a 1.4 master: Obviously this will cause upgrade issues. ideas? If not I will debug tomorrow.\ndefaults to using the resource name as the etcd prefix, but ingresses was not using the resource name as its prefix. fix in\nreopening until cherry-pick merges\nYikes, good catch. This would have been a disaster. , Jordan Liggitt wrote:\nmerged to release-1.4"}
{"_id":"doc-en-kubernetes-87812c2d6cc8421563555995ddc62216ea46df17dcd4476f816c4a4bdf2ab614","title":"","text":"Our 5k-node cluster jobs recently started failing even before the e2e tests start. The reason is the step which checks that all kube-system pods are running is timing out: cc\nFrom the scheduler logs I'm seeing that the pods are being scheduled quite slowly, getting worse with time: 2-3 pods/min looks quite bad. cc\nAlso finding such traces: Seems like most of the time is being spent in priority calculation.\nFrom the pod spec, I'm seeing soft inter-pod anti-affinity at node-level topology: I'm almost sure this is the one causing the mischief. We've been having performance issues with anti-affinity traditionally :p\nOk... PR is the cause. It introduced anti-affinity to kube-dns. I think we should revert this change given known issues with anti-affinity. - Opinions?\nFYI - it is taking 80m until the last pod is scheduled when I'm running 5k-node cluster locally.\nThat's quite unfortunate. It looks like we should revert the change until the anti-affinity change is scalable enough to be used more broadly, unless the scheduler folks have a target fix for this use case.\ncc\nWe are very well aware of the performance issues in our affinity/anti-affinity functions, but I am not aware of a concrete solution at this point. This is partly due to the flexible design of affinity/anti-affinity concept. My humble opinion at this point is to drop some of the flexibility as they are rarely used, in order to streamline the algorithm and improve performance. Anyway, I think the proper solution to this problem needs more time than your requirements to address the issue in hand. So, you may want to change your configuration and drop the anti-affinity rule.\nwould it make sense to add a note to the anti-affinity user documentation to hint at performance problems at large scale, until this is fixed?\nRe-opening this issue as this problem has struck again."}
{"_id":"doc-en-kubernetes-e9d9c7cf78723b98b17d4b4c5d875d12d6958caf3d4df35a73aa6ed31f64de32","title":"","text":"The following runs of our 5k-node scalability jobs failed to bring up the cluster on time: https://k8s- https://k8s- https://k8s- https://k8s- And the reason is kube-dns pods pending to be scheduled for a long time:\nThese failures started happening about 4 days ago and I'm pretty sure it's this one: Add self anti-affinity to kube-dns pods Sadly, we'll have to revert it again for now :( cc - Seems like the improvements to anti-affinity performance made under weren't enough?\nWe need to find other improvements for the predicate. If we cannot find major improvement opportunities, we may have to change its design to be more scalable at the cost of the reducing its flexibility.\n- my personal opinion is that we need to change the design anyway. Having it work both ways (i.e. existing pods everywhere impacting placing of our pod) will always be painful (especially for things like cluster autoscaler)\nI agree. The symetrical application of anti-affinity is problematic. I also think we need to change it.\nGreat. - do you agree with that too?\n, I'm wondering after some affinity/anti-affinity optimization in scheduler, do you think if it's fine to test coreDNS with anti-affinity policy in large cluster with 5000 nodes?\nWe probably should try to see where we are - but this requires testing before enabling that.\nI agree with - Could we first try checking with a benchmark integration test () as a proxy for real cluster test?\nI'm wondering how many real nodes are required for testing?\nSince the integration tests fake out the nodes, I think it's safe to crank it up to 5000 (that's the current largest integration test benchmark size) I'm also hoping to refactor our benchmarks soon, so let me know if you have any thoughts about them."}
{"_id":"doc-en-kubernetes-efbe4c5d01aeff5665bc84249f5b94afd1b0f5aa50522bcb417b43505717780e","title":"","text":"! ! Which jobs are failing: Which test(s) are failing: Since when has it been failing: failures started between 3-4PM (EST), on 1/11/2019 the failures appear limited to containerd jobs (here's a triage search for note that only containerd jobs appear as failing): /sig node\nthe relevant commit range on master appears to be ... the relevent commit range in test-infra appears to be\ncc\nfor test-infra: aws-janitor: clean up ELBs - definitely unrelated, only affects some of the AWS testing cleanup Update kubekins tag for eks jobs - definitely unrelated, affects the docker image in which some EKS testing runs only Update ubuntu image used for e2e node test - I think unrelated, appears to only affect the node e2e tests rest are merge commits\nthanks for checking. it's also possible the commit ranges I linked to need to be expanded slightly... that was my first attempt at correlating merge timestamps with the triage board failures... it's possible I got timezone correlation wrong a few hours in one direction or the other.\nah, they probably do then. otherwise I'd say something else must have changed out of band (perhaps there's a pointer to a containerd version to use somewhere or something).\nflakes or was something fixed offline (gce / gci / gke)? https://k8s- https://k8s- these turned (mostly) green without k/k and k/test-infra changes.\nHuh. Disconcerting, but ok. /close\nClosing this issue. <detailsIn response to : Huh. Disconcerting, but ok. Instructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\nAll test failed because of not being able to pull image from : I guess this is a bug in , and they fixed it now.\nnice catch. thanks\nshould the CSI images be in our regular repo?"}
{"_id":"doc-en-kubernetes-d34e56b4503184aafc2f81081d2e7d6ebfab0ded5a2f7a2bfd1277aaa2a8d441","title":"","text":"<!-- This form is for bug reports and feature requests ONLY! If you're looking for help check and the . If this may be security issue, please disclose it privately via --Is this a BUG REPORT or FEATURE REQUEST?: /kind feature What happened: I have a daemonsets with 250 replicas and maxUnavailable is 20%. When I try to update its image, the ds will scale down 50 old replicas and wait for SatisfiedExpectations, after that it will scale up 50 new replicas. But if a pod delete failed (node is notready, and pod stuck at terminating status), ds controller will never reach the expectation and just stuck here. What you expected to happen: When a pod stuck at terminating, ds controller should start new pods like deployment controller, instead of waiting for the delete batch finish and ready for next poll. How to reproduce it (as minimally and precisely as possible): Anything else we need to know?: Environment: Kubernetes version (use ): Client Version: version.Info{Major:\"1\", Minor:\"8+\", GitVersion:\"v1.8.1-8+\", GitCommit:\"\", GitTreeState:\"clean\", BuildDate:\"2017-10-25T13:04:04Z\", GoVersion:\"go1.8.3\", Compiler:\"gc\", Platform:\"linux/amd64\"} Cloud provider or hardware configuration: OS (e.g. from /etc/os-release): Kernel (e.g. ): Install tools: Others:\n/sig apps /cc\nIssues go stale after 90d of inactivity. Mark the issue as fresh with . Stale issues rot after an additional 30d of inactivity and eventually close. If this issue is safe to close now please do so with . Send feedback to sig-testing, kubernetes/test-infra and/or . /lifecycle stale\n/remove-lifecycle stale\nIssues go stale after 90d of inactivity. Mark the issue as fresh with . Stale issues rot after an additional 30d of inactivity and eventually close. If this issue is safe to close now please do so with . Send feedback to sig-testing, kubernetes/test-infra and/or . /lifecycle stale\n/remove-lifecycle stale\nDaemonSet controller waits for old pods to terminate before creating new ones for best-effort at-most-one pod per node. This is a deliberate design decision."}
{"_id":"doc-en-kubernetes-3717eeb03a72e85e977fbd3018fc04b90903ed01af39ae113b30935dae199f29","title":"","text":"In most cases, DaemonSets won't tolerate two pods scheduled on the same node at the same time; for example, it's common to communicate with DaemonSet pods via . I'm going to close this ticket as this works as designed. Please feel free to reopen the ticket if you think there are still issues to be resolved.\n/reopen\nReopened this issue. <detailsIn response to : /reopen Instructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\nwhy does the DaemonSet controller wait for all pods 50 to be done, before moving to the NEXT 50 ? I completely understand that we want only at most 1 pod per node. But if a pod deletion fails on the node, move on to the next pod as long as the total unavailable pods are less than maxUnavailable. This will allow us to make progress in the face of bad nodes or failed pod deletions due to failing tear down of bad volumes or other reasons.\n/assign\nthoughts on this one ?\nAny progress on this issue? We also encountered this problem. There are thousands of nodes in our prod cluster and nodes often go down, which cause daemonset rolling update stucked, this problem really bother us.\nPTAL\n/reopen\nYou can't reopen an issue/PR unless you authored it or you are a collaborator. <detailsIn response to : /reopen Instructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\n/reopen\nReopened this issue. <detailsIn response to : /reopen Instructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\n/reopen\nYou can't reopen an issue/PR unless you authored it or you are a collaborator. <detailsIn response to : /reopen Instructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details"}
{"_id":"doc-en-kubernetes-b91639c622bb64c70a52cad670c2f62bee84a1358c20408d54f6229d98acd356","title":"","text":"<!-- This form is for bug reports and feature requests ONLY! If you're looking for help check and the . If the matter is security related, please disclose it privately via --Is this a BUG REPORT or FEATURE REQUEST?: /kind bug What happened: Default azure file permission is not suitable for most users, dozens of users complain about this, set as is the best choice here, if users don't want , then they could use mountOptions to override and . What you expected to happen: How to reproduce it (as minimally and precisely as possible): Anything else we need to know?: Environment: Kubernetes version (use ): v1.12 Cloud provider or hardware configuration: OS (e.g. from /etc/os-release): Kernel (e.g. ): Install tools: Others: /sig azure /assign"}
{"_id":"doc-en-kubernetes-e7fba9d64f1431a5cdd542cd6f9cce6e6a943bdeb7b1334304ccc5bcf3efb9f5","title":"","text":"I ran a deployment with replicas 1 and use to set container's subpath dir, which uses NFS as the storage backend, and create about dirs in container's subPath dir,when delete the pod, it spent much more time to end status, and when the replicas increase to 10 and do the rolling update, at the same time dir nums in containers increase to , it spent about 30 minutes to totally remove pod, which is intolerant. The pod is deleted fast. a pod using subPathExpr as the volume mount. dirs in the backend storage corresponds to the container volume mount the pod kubelet log We can find that it spent about 17s to clean the mount dir in kuelet. No response <details</details<details</details<details</details<details</details<details</details<details</details\nThis issue is currently awaiting triage. If a SIG or subproject determines this is a relevant issue, they will accept it by applying the label and provide further guidance. The label can be by org members by writing in a comment. <detailsInstructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\nAfter some debugging, I inspected the pprof with flame graph when delete a pod, it shows that and cost much more time, here is the graph: ! Apparently the Walk func spent much time when doCleanPath, what we care is the top level dir,and this pr has some fast path to avoid much recursive time, but before gol1.16, the filepath.Walk is still less efficient, because when used in kubelet, it reads all the dir in the mount point before call the callback func in Walk even we don't use it after that, so it spends much time. And after go 1.16, filepath.WalkDir is introduced, which is more efficient than Walk, as the comment said in golang WalkDirFunc is called before reading a directory, so it can save much time when there are lots of dirs in the first level of mount points. So it's time wasting and meaningless when we use Walk so far because we traverse the dir but we didn't use it and return SkipDir err, the code is After I replaced it with , the took only about 41ms, which is a huge boost than 17s before."}
{"_id":"doc-en-kubernetes-2270f3670748d4dd417a0804239a121b559b1166e20ffa4186cfbb1c59e670cb","title":"","text":"When master opens up for , we need to add this back in: /sig node\nThis issue is currently awaiting triage. If a SIG or subproject determines this is a relevant issue, they will accept it by applying the label and provide further guidance. The label can be by org members by writing in a comment. <detailsInstructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\nnote to self / 1.21 and 1.22 release notes team -- let's make sure the PR body on the original PR is copied over to the new PR. will think that the old release note and documentation has been covered in 1.21, but we will (I will update this comment with link) manually override it to not show up in 1.21 release notes."}
{"_id":"doc-en-kubernetes-a9efceceb9be4ec0ae924972e4a207288eddcc2b046c2722d13a8ac3250be7d4","title":"","text":"Deployment is defined in a json file with several environment varibles. Deployment is deployed using \"kubectl apply -f \". Pods get most of the environment variables defined, but not all. Specifically, the variables defined in the are named: { \"name\": \"DBHOSTS\", \"value\": \"10.0.0.1\" }, { \"name\": \"SLACKENDPOINT\", \"value\": \"ERRORNOT\" }, { \"name\": \"LACKENDPOINT\", \"value\": \"ERRORNOT\" }, { \"name\": \"SLACK\", \"value\": \"ERRORNOT\" }, { \"name\": \"CKENDPOINT\", \"value\": \"ERRORNOT\" }, { \"name\": \"ACKENDPOINT\", \"value\": \"ERRORNOT\" }, { \"name\": \"KENDPOINT\", \"value\": \"ERRORNOT\" }, { \"name\": \"CASSANDRALOGGINGLEVEL\", \"value\": \"ERROR\" }, { \"name\": \"ENABLEGETIP\", \"value\": \"\" } the environment variables that appear in a pod are: ENABLEGETIP CASSANDRALOGGINGLEVEL KENDPOINT ACKENDPOINT CKENDPOINT DBHOSTS That is, ENV variables named \"SLACKENDPOINT\", \"LACKENDPOINT\" and \"SLACK\" never get passed on to the pods.\nIt also won't accept \"randomshit\" as an environment varible. When I add that as an environment variable to an existing deployment and run \"kubctl apply -f \" nothing happens. This is on version v1.2.4 /\nEhh? Can you post a full but minimized yaml that I can run to reproduce it? , drashkov wrote:\nWe use JSON files, here is the full one. \"SLACK_ENDPOINT\" does not appear as an environment variable in the pods. Doing \"kubectl apply -f \" does not create new pods, and does not pass these variables to the pods. If I change the number of replicas and add a new pod, that pod will be created, but won't have that varible.\nI am not able to reproduce on GKE at 1.2.4 when exec'ing into one of the Pods and printing the environment. This is the yaml I used - note the image is different: What are the symptoms you are seeing?"}
{"_id":"doc-en-kubernetes-88cd42fc07ef7063636f75ce8fb36b9d8d96cd9dd28f579d99991871766445e3","title":"","text":"Can you provide yaml using standard public image along with the specific symptoms you are seeing?\ndid you create the deployment first with out the env vars, then add them via an apply?\nThanks for looking into it I managed to reproduce it with the yaml you posted. Here's what I did: the yaml you provided as '' the \"SLACKENDPOINT\" env var from that file and saved that as \"\" (no other changes) This creates the deployment with all env vars as expected This re-creates a new set of pods and they don't have the SLACKENDPOINT var, as expected. This does not re-create the pods, so the env var is not passed along. kubectl does not return errors.\ncc cc\nThanks, I was able to reproduce the issue. It looks like the config with the ENV does get stored in the last-applied-configurations annotation: But does not get put into the pods:\nAfter a few tries I found that if you add the env as the last one you'll be able to add it. Definitely a bug.\nThe generated three way merge patch wasn't correct (the annotation is good, but look at the spec part): \"spec\":{\"template\":{\"spec\":{\"containers\":[{\"env\":[{\"name\":\"CASSANDRALOGGING_LEVEL\",\"value\":\"ERROR\"}],\"name\":\"test-api\"}]}}}}\nIt only recognizes the last env as diff\nOpened a PR ptal.\nThanks everyone."}
{"_id":"doc-en-kubernetes-2fe80e49aa5ae53d41f8add8e37445a19adaf3497eeef8427dfc711a0635077f","title":"","text":"When a Pod admission fails due to node-pressure condition then the Pod's looks like this: \"Pod The node had condition: [DiskPressure].\" The \"Pod The node...\" reads awkward. A message like: \"The node had condition: [DiskPressure].\". a job with a pod on a single node cluster, when the node is under disk pressure (then it is tainted with ). The Pod gets stuck in the phase. the node by kubectl command line: Pod is scheduled but admission by Kubelet fails as the taint is re- shortly after its manual removal. Then, the status of the Pod contains the awkward message. No response <details</details<detailsReproducible on kind </details<details</details<details</details<details</details<details</details\n/sig node /triage accepted /priority backlog\nThe first part () is returned here: . I wonder where the is being appended to the beginning. It's my first try to contribute, so I am a little bit lost.\nI think this is a candidate place which might be responsible for the prefix (but haven't checked): I would suggest you to try to reproduce the issue locally first then experiment with modifying the code.\n/assign\nthanks, I've managed to reproduce and used Delve to make sure who's responsible. I opened the changing the message. If it's accepted it will change to ."}
{"_id":"doc-en-kubernetes-067de3b7865dcc26d5939903adfdfd6f426066e830ffff2d268cf1b31c889111","title":"","text":"ref: /sig scheduling /kind cleanup Part of: The unit test in the scheduling queue only assumes the case where the queueing hint is enabled. We should make another unit test to confirm works without QHint. should have been existing since before we introduced QHint. So, just bring the old one back should be fine.\nThis issue is currently awaiting triage. If a SIG or subproject determines this is a relevant issue, they will accept it by applying the label and provide further guidance. The label can be by org members by writing in a comment. <detailsInstructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\n/priority important-soon\n/assign"}
{"_id":"doc-en-kubernetes-72455550466a39f28970619bf50fef21b9396df700b03a9fcf4c2735a10457eb","title":"","text":"Kubernetes 1.3.0. During attach disk to VM, call to cleanUpController dereferenced a null pointer. Stack trace:\nThe seems points to a comment instead of a cleanUpCotroller function in v1.3.0. There might be some modifications later which make the line number inaccurate. Would you please double check which line it really should point to? Thanks.\nthat is correct, this was a build from v1.3.0 with changes related to this: therefore the line numbers do no line up with v1.3.0. In either case, the panic happened inside\nHi Davide, I saw your PR for this issue and I have another question. Why cleanUpController could be called with an empty newSCSIController? According to the context, an empty newSCSIController should not even be returned by CreateSCSIController rt? Thanks.\nit doesn't look like it should happen, but it did and there wasn't any checks for nil pointers in our code or the Govmomi library."}
{"_id":"doc-en-kubernetes-a2e77bc9d5795a4a17af5bcc5860b35587b39d7e25d603a866fcb410798b82da","title":"","text":"seen in this inside a like loop we execute a function that return a bool and an error. If we want to fail fast we return the error directly and assert on the output of the wait function. If we want to retry we log the error and return false nil /kind flake /kind bug /sig testing /sig network\nThis issue is currently awaiting triage. If a SIG or subproject determines this is a relevant issue, they will accept it by applying the label and provide further guidance. The label can be by org members by writing in a comment. <detailsInstructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\n/cc since I think you this test but most probable the code is inherited and is a copy paste thing, but besides that, the change is to remove those there\n/assign"}
{"_id":"doc-en-kubernetes-3186f04d894d2e98a9b770821be5e33542dd055d445ea6f78bf8e42e85bb59f7","title":"","text":"<!-- Please use this template while reporting a bug and provide as much info as possible. Not doing so may result in your bug not being addressed in a timely manner. Thanks! If the matter is security related, please disclose it privately via --What happened: there is a disk resize scenario: In the beginning, it’s 3TiB with cachingMode=ReadOnly, and then disk detached, resized to 4TiB, then user find they could not attach that disk back to any agent node, I think that’s a bug we need to fix. What you expected to happen: How to reproduce it (as minimally and precisely as possible): Anything else we need to know?: Environment: Kubernetes version (use ): Cloud provider or hardware configuration: OS (e.g: ): Kernel (e.g. ): Install tools: Network plugin and version (if this is a network-related bug): Others: /kind bug /assign /priority important-soon /sig cloud-provider /area provider/azure\nThis issue is currently awaiting triage. If a SIG or subproject determines this is a relevant issue, they will accept it by applying the label and provide further guidance. The label can be by org members by writing in a comment. <detailsInstructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details"}
{"_id":"doc-en-kubernetes-8769160e6f0dabeddd11cead4810305b5f582b45774110e5e46e9ec50492bd79","title":"","text":"Testgrid: https://k8s- https://k8s- https://k8s- https://k8s- Sample failures: This started failing across multiple test jobs last night. The culprit appears to be PR over to you. /kind failing-test /priority critical-urgent /sig storage\n/milestone v1.13\n/assign\nGitHub didn't allow me to assign the following users: mkimuram. Note that only and repo collaborators can be assigned. For more information please see <detailsIn response to : /assign Instructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\nI think should delete the PVC and wait for it to be deleted.\nThank you for your comment. I haven't yet catching up, but as for the PVC deletion code in , which is changed in the PR, PVC deletion is done inside for delete policy case or for retain policy case. Isn't it enough? In addition, could you let me know how I can check the matrix for the failure that we are focusing? I couldn't find the failure for provisioning tests themselves, there.\nIs my understanding right that the purpose of is to check whether resource is leaking by checking the diff of resource between the test? If so, is it correct that the sample output above means that one pd is leaking, not the leak of pv or pvc itself? I found that in the test for retain policy is removed in this PR when it is moved to , because it is gce specific logic and looks unnecessary. Do you think that it could be a cause of this issue? If so, should deleting the backend storage be generally required before removing the dynamic provisioned PV with retain policy?"}
{"_id":"doc-en-kubernetes-182fb1698032797795262f9c7145a24707ed3bbf5827100ad6f2695fe6c12a83","title":"","text":"PriorityPidEvictionOrdering [Slow] [Serial] [Disruptive][NodeFeature:Eviction] when we run containers that should cause PIDPressure should eventually evict all of the correct pods /kind failing-test <!-- If this is a flake, please add: /kind flake --/sig node /assign /priority important-soon\nfork-bomb-container-container exited with code 2. I tried to create a pod in my env of containerd: The pod can be running and it works as designed. I need to test it in cri-o later.\n/cc\n/triage accepted"}
{"_id":"doc-en-kubernetes-0c9ef3005ed9520787f5bf63165fe010227c899755479cb10a464fa14ceede75","title":"","text":"ref:\nidentified one problem: firewall is not open for icmp in GCI setup. I opened it manually. Still does not work.\nHere are the symptoms: node can ping each other one node, it can ping all pods on it, and vice versa. one node or pod, it can ping all cbr0 gateways(10.X.X.1). pod, it cannot ping but able to curl. It cannot ping any other pod, even on the same bridge, but able to curl.\nMy first thought would be somethings blockign ping. your options are iptables, firewall, icmp is disabled alltogether (sysctl or somewhere in /proc/sys/net/ there should be a setting). Try traceroute and see where it gets to.\nI have a working version now. GCI setup icmp ACCEPT rules in INPUT chain of filter table. But not in OUTPUT and FORWARD chain.\nIn summary, 2 problems: is not opened for icmp is filtering icmp on FORWARD chain. ACCEPT on INPUT chain is not enough Will submit a fix\nGreat detective work, Is this a ping-specific problem, or are there going to be problems with other types of outgoing traffic from pods on GCI as well?\nI do not see any problem for udp/tcp traffic so far. e2e tests already covers lots of them. Unfortunately, no e2e covers ping tests and GCI happens to filter it.\nWe should add a test in networking suite for ping.\nIs this missing a milestone 1.3?"}
{"_id":"doc-en-kubernetes-673017a5e6e7ba59bdbbe8f10c13b378c1b56adec73e9d5f13b02c9f080d522d","title":"","text":"<!-- Please use this template while reporting a bug and provide as much info as possible. Not doing so may result in your bug not being addressed in a timely manner. Thanks! If the matter is security related, please disclose it privately via --What happened: ! What you expected to happen: How to reproduce it (as minimally and precisely as possible): Anything else we need to know?: Environment: Kubernetes version (use ): v1.20.1 Cloud provider or hardware configuration: OS (e.g: ): 7 Kernel (e.g. ): Linux zsm-test-2 3.10.0-1160.11.1.el7.x8664 SMP Fri Dec 18 16:34:56 UTC 2020 x8664 x8664 x8664 GNU/Linux Install tools: Network plugin and version (if this is a network-related bug): Others:\n/sig node\nDo you know how to reproduce this problem?\nYes, can reproduce the problem stably, but the steps are a little complicated.\n/triage accepted"}
{"_id":"doc-en-kubernetes-d5a9e8b7bbc510dc155bb81dd6aed03e6a233b66a01ab8a13237a98090288309","title":"","text":"When using a AWS EBS PV or on-line volume with a partition on a Nitro machine, the volume is available as NVMe device. Kubelet is not able to format & mount the partition: It seems to me that AWS EBS volume plugin does not support partitions at all. The partition is mounted. Using an in-line volume with this pod:\n/sig storage /sig cloud-provider /area provider/aws\n/triage accepted"}
{"_id":"doc-en-kubernetes-917c5aacdf98ee16adcfe96e5370eb2e1e342a70bc45d825e7dc562ccfb7cd09","title":"","text":"Since 2 days ago, the test is becoming more and more flaky, and one of the examples is: https://k8s-\nThere are no sig labels on this issue. Please by:<br(1) mentioning a sig: <bre.g., for API Machinery<br(2) specifying the label manually: <bre.g., for sig/scalability<br<brNote: method (1) will trigger a notification to the team. You can find the team list and label list\nplease triage this one. Currently I put it as the release blocker, and please feel free to remove that if you think it is not. Thanks!\nThe test deletes the node, and at the end, waits 5 minutes for the node to become ready, but it timed out. cc\nThe test logic hasn't changed after I last modified it, so the test deletes the node through gcloud, and the instance should be restarted automatically by GCE because it's part of an instance template. The likely possibility, as pointed out, is that another change made node discovery take longer, so extending the timeout would be enough. The very unlikely possibility is there's a problem discovering brand new instances."}
{"_id":"doc-en-kubernetes-7b2399b1ff0857c826d80a73801e04bffa6df7e676dd58a7e2d56a664bd364fc","title":"","text":"The test case \"executing a command with run and attach without stdin\" is inherently racy. The test runs, and expected to see in the output. The case when the test passes: case 1: Kubernetes creates and starts the container without stdin; The container exits before attach runs; kubectl falls back to get container log, which contains . case 2: Kubernetes creates and starts the container without stdin; kubectl attach before \"stdin closed\" is printed; kubectl attach contains . The case when the test fails: Kubernetes creates and starts the container without stdin; The container prints \"stdin closed\", but hasn't exited yet; kubectl attach contains nothing. One example failure: Related issue: Can we modify this test not to check the output? Or use to check output instead? /cc\nUploaded to try to address\n/reopen This test is still inherently racy, even after\nFix in (I hope!)\nMoving into the current milestone as it was not fixed in 1.14\nHello I'm a bug triage shadow for the 1.16 release cycle and considering this issue is tagged for 1.16, I'd like to check its status. Code freeze is starting on August 29th (about 3 days from now). Do we still target this issue to be fixed for 1.16?\nThanks . PR is available and awaiting review!\nWhat is the PR in question? Is it targeting 1.16 or 1.17?\n/reopen This test is having a lot of failures these days\nReopened this issue. <detailsIn response to : /reopen Instructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\nis right, it seems to be present in kind jobs now only\n/assign let's see if we start to recover a bit of stability :smile:\nThe rate of failures decreased a lot since last changes Checking the logs of the failures it seems that the problem is that the pod has not finished before checking the output, as Lantao described The test fails Collecting events from namespace The framework logs the pods the kubelet thinks is on the node\nnote that it's still racy, but it's hopefully somewhat mitigated.\n/reopen\nReopened this issue."}
{"_id":"doc-en-kubernetes-6fb855e6221755f4f8c679054412ceafcb7919572a50b214c71af07d9654c4d2","title":"","text":"<detailsIn response to : /reopen Instructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\nany updates here?\nnope, totally agree with the description, is racy by design, couldn't find any way or trick to avoid the randomness\nEDIT: It's late and I'm actually not sure I'm remembering this correctly, take this comment with a heap of salt: Specific sub-tests were fixed previously I think, but this one sub-case: That one we can't use the \"Wait for stdin input to signal we're attached\" hack, so it's just racy and broken. We could remove that case maybe. I'm not sure who I'd want to weigh in on that, probably someone more active in node / CLI.\nnot super common atm but still happening. discussing\n/unassign has a fix for this"}
{"_id":"doc-en-kubernetes-8218004df321e49189cb8c22922209b0129dd124df9f46401769a8d8c1aa8714","title":"","text":"[x] test approval issue : [x] test pr : 7 July 2022 [X] two weeks soak end date : 21 July 2022 [X] test promotion pr : According to this APIsnoop query, there is still some remaining ResourceQuota endpoints which are untested. The Status endpoints will be addressed in another test. listCoreV1ResourceQuotaForAllNamespacespatchCoreV1NamespacedResourceQuotadeleteCoreV1CollectionNamespacedResourceQuota Using a number of existing e2e test practices a new has been created for three ResourceQuota endpoints. The e2e logs for this test are listed below. This query shows the endpoints hit within a short period of running the e2e test If a test with these calls gets merged, test coverage will go up by 3 points This test is also created with the goal of conformance promotion. /sig testing /sig architecture /area conformance\n/triage accepted\n/reopen\nReopened this issue. <detailsIn response to : /reopen Instructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details"}
{"_id":"doc-en-kubernetes-dea116f1992e38bb66fa8cbf634ccd764aeb1776d3000f923339739e72460ab8","title":"","text":"What happened: I have a Kubernetes Cluster with 2 Windows Server 2019 Core VMs as worker nodes. I upgraded first the cluster, and then the first Windows node following the docs here: Every step ran without issue, but the kubelet service was not started. This was not immediately obvious, since \"restart-service kubelet\" does not propagate the error. I think \"start-service kubelet\" would be better advice. The logs said: After changing the value of cgroupDriver manually in from \"systemd\" to \"cgroupfs\", the kubelet started successfully. What you expected to happen: One of 2 things: Either the Windows upgrade docs telling me what I need to do to fix this, or the kubeadm not breaking the configuration for windows in the first place. How to reproduce it (as minimally and precisely as possible): a windows node to a Kubernetes v1.19 cluster via kubeadm the cluster via kubeadm the Windows node following the docs Anything else we need to know?: Environment: Kubernetes version (use ): 1.20.1 Cloud provider or hardware configuration: OS (e.g: ): Windows Server 2019 Core Kernel (e.g. ): Install tools: kubeadm Network plugin and version (if this is a network-related bug): Others: /sig windows\nthis may cause by the default value of defaultCgroupDriver cgroupfs Would you provide the from Windows node?\n/assign\nkubeadm doesn't do anything about cgroup drivers on Windows because cgroups don't exist on Windows. ideally the kubelet checks about this should be skipped entirely in dockershim for non-Linux.\n! However, windows docker return as cgroupfs.\nprobably 's output is the same structure on all OSes with being a default value. this field should not be checked on Windows at all.\n/retitle kubelet: don't check for the Docker cgroup driver on Windows\nadding since the PR is favorable. /triage accepted"}
{"_id":"doc-en-kubernetes-4d460c4f5c8fb1c08c701d631c43381b00a0fa5b123446c3865b8df9432b458e","title":"","text":"<!-- This form is for bug reports and feature requests ONLY! If you're looking for help check and the . --Is this a BUG REPORT or FEATURE REQUEST?: /kind bug What happened: call delete endpoint when service is not found. What you expected to happen: It is no need to do delete in controller. Because in service restStorage, endpoint is deleted when delete service. See How to reproduce it (as minimally and precisely as possible): Anything else we need to know?: Environment: Kubernetes version (use ): Cloud provider or hardware configuration: OS (e.g. from /etc/os-release): Kernel (e.g. ): Install tools: Others:\n/area controller-manager /sig apps\n/kind bug"}
{"_id":"doc-en-kubernetes-9bc3939632f91b367e9735795407824b8635b986a1aa4d3e42f19abbece1efa5","title":"","text":"Is this a BUG REPORT or FEATURE REQUEST?: /kind bug What happened: kube-apiserver --help still shows this: It looks like the related feature can be turned off. However, dynamic volume provisioning Alpha has retried in 1.6. What you expected to happen: Remove from feature gate to avoid confusion. How to reproduce it (as minimally and precisely as possible): Run Anything else we need to know?: No Environment: Kubernetes version (use ): 1.7+ Cloud provider or hardware configuration**: Any OS (e.g. from /etc/os-release): Any Kernel (e.g. ): Any Install tools: Others:\nThere are no sig labels on this issue. Please by: a sig: e.g., to notify the contributor experience sig, OR the label manually: e.g., to apply the label Note: Method 1 will trigger an email to the group. You can find the group list and label list . The in the method 1 has to be replaced with one of these: bugs, feature-requests, pr-reviews, test-failures, proposals\n/sig api-machinery\n/sig api-machinery\nSIG API Machinery doesn't own the contents of the API. /sig storage"}
{"_id":"doc-en-kubernetes-9fe53756e4f224dbc05ba9a01e42da1fd9faed4696cdb60aebb7f8b8a8125e7b","title":"","text":"What would you like to be : Currently, does not have unit test coverage. Unless there's a strong reason not to, we'd like public facing functions to have unit test coverage. Why is this needed: There's a fairly small chance of GetAddressAndDialer being modified, and thus breaking, in the future without this test coverage. However, it should be relatively straightforward to add the test coverage, and could make a good first issue for a new contributor. There's also the benefit that it helps keep the test coverage in pkg/kubelet high :) And we never know, it might help catch a bug some day :) Extra note I think this ticket could be a great candidate for someone who is new to contributing to sig-node/k8s. I'd prefer to please give first dibs for picking this up to any folks who are currently underrepresented in the k8s community. Please reach out to on k8s Slack (or post directly in #sig-node and tag me) if you're interested, and I'll assign you the issue. I'm happy to mentor and be helpful in whatever way I can :)\n/sig node\n/good-first-issue\nThis request has been marked as suitable for new contributors. Please ensure the request meets the requirements listed . If this request no longer meets these requirements, the label can be removed by commenting with the command. <detailsIn response to : /good-first-issue Instructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\n/assign\nHi let's me solve it\nSounds like a plan! Can you please tag and myself for the review?\nHi, I've already created a pr which , please help me complete the review, thanks"}
{"_id":"doc-en-kubernetes-1b2bcc3aa22e5d5f53df9233e5f80a56f1f4bdaeb462b3a5016092de1417360c","title":"","text":"hack/build-cross sets it without looking at the current value. And the value has to be an array, which is a bit awkward to use.\nThis could be improved indeed. Maybe something like and Haven't tested it though... But why would one want to change the value?\nI just wanted to build one, so I did the \"obvious\" thing: hack/dev-build-and-` It failed for multiple reasons :) , Lucas Käldström wrote:\nThis issue has been a major downer for me. It's much harder right now than it ought to be to articulate the few platforms for which you'd like to build each of server binaries, client binaries, test binaries, etc. Every time I've tried to do this \"the right way,\" I've spent hours looking at scripts and feeling frustrated that build options aren't documented better. Every time, I end up cutting my losses and resorting to temporary build script hacks instead, just to get the job done.\nDoesn't fix this issue exactly, but I've found that you can easily build only if you set .\nTaking a look at this today, I also thought that setting seems to be the most reasonable.\n/sig release"}
{"_id":"doc-en-kubernetes-9ac079d1fd510b2062bc937a63b320f5cb9c70bc1e9e6c0c16b4b25caa3b72aa","title":"","text":"Encountered this error while running on master: Should we set to by default? cc\nWhy is it empty? Would be better?\nseems to expect $(whoami) based on the value of that variable for my machine. I can make that change if you would like.\nCan you keep as an option if it is specified? returns a 6 digit numeric value on my machine (corporate provisioned) and that is incompatible with GCI images, as well as error-ing out during e2e provisioning.\nthat is a great idea, so should be defaulting to whoami in the case where it has not been defined.\nOK, will file an PR to fix this.\nFiled PR at . PTAL"}
{"_id":"doc-en-kubernetes-36c3b1963e29994fcef0db8f2a8b4a629c6bfc24057a3ec2a6314c1367d30587","title":"","text":"Is this a BUG REPORT or FEATURE REQUEST? : BUG: With CRI enabled by default, kubelet reports the version of CRI instead of container runtime version: Kubernetes version (use ): Master and 1.6. What you expected to happen: Kubelet should report the version of container runtime. Will file a PR to fix this. cc"}
{"_id":"doc-en-kubernetes-ead4a113e3280b31adabcef2a84a488ae269cf78485f762926d30cc661385643","title":"","text":"Is this a BUG REPORT or FEATURE REQUEST?: /kind bug What happened: Testing bringing up k8s cluster using kubeadm from latest k8s code from master branch. kubelet is not coming up and below is the errors from syslog: upon looking at the code from where the error is happening, it is evident that the err checking is missing. What you expected to happen: deployment to succeed. How to reproduce it (as minimally and precisely as possible): Use the latest code from k8s and build a k8s release and bring up cluster using kubeadm Anything else we need to know?: Environment: Kubernetes version (use ): latest Cloud provider or hardware configuration**: GCE OS (e.g. from /etc/os-release): 16.04 LTS (Xenial Xerus) Kernel (e.g. ): Linux e2e--master 4.4.0-21-generic -Ubuntu SMP Mon Apr 18 18:33:37 UTC 2016 x8664 x8664 x86_64 GNU/Linux Install tools: kubeadm Others: /sig node /cc\nPTAL"}
{"_id":"doc-en-kubernetes-a8f763c740d3fb11e6b67f36722750f0a39e674fe923e77cebdc2dd4df02a0e7","title":"","text":"<!-- Please only use this template for submitting reports about failing tests in Kubernetes CI jobs --Which jobs are failing: ci-kubernetes-e2e-gci-gce-alpha-enabled-default Which test(s) are failing: CSI hostpath-v0 tests Since when has it been failing: 8/8 Testgrid link: https://k8s- Reason for failure: Anything else we need to know: We need to update client-go to get the fix for selfLink, build new nfs provisioner images, and update them in the e2e test. The fix for client-go is being cherry-picked to 1.15 if you want to use a tagged release:\n/assign\n/assign\nThis implies rebase to 1.15.x in external-storage, which seem to be challenging (at least to me :-) )\n14 cherry pick also in progress:\nOK, I guess this line in the lib is the culprit: So I need to update lib then external-storage to pull it in\nI don't think the lib itself needs to be updated. Only updating go-client to get the fix to is needed.\nRight, lib doesn't dictate what version of client-go to use. Will wait for client-go release/commit\n- thanks a lot for bumping the dependency. Did you have a chance to build the new container with it so that we could update it here:\nyeah, opened . (Not sure how to trigger test, or we can just merge and see.)\nYou can manually create a PVC on an alpha cluster to test."}
{"_id":"doc-en-kubernetes-8b50a6e6165c8a4dec36611b88a25cd5e6225d1dd2a0483fe2c5b587bd4e7bf4","title":"","text":"reported hitting this error: The instance had 4 security groups, and only one of them was tagged. However, the logic around picking the security group is incorrect.. if we get two non-tagged SGs before we see the first tagged SG, it throws an error. We'll want to cherry-pick this to 1.2\nI think this one was a victim of flakiness in the auto-assigner when the author is the first-listed maintainer. Assigning to you, :-)"}
{"_id":"doc-en-kubernetes-17cba5fefe0eb7064620c55e7004cb3247c052d63b398a81f9e1e6b57e9ab2c7","title":"","text":"<!-- Please only use this template for submitting enhancement requests --What would you like to be : Report request throttling by the kubelet by default (i. e. V=3) Why is this needed: Request throttling can have serious consequences, but it is reported (in ) at V=4, so not by default. While attempting to probe limits of how many pods per node we could run, we hit a situation where the test pods would go Pending and eventually fail. After investigation, it turned out that the pod was being started by the kubelet, but that status update was not provided to the apiserver in a timely manner, resulting in the pod-to-pod network not coming up, so a pod attempting to connect would time out and fail. The delay was root caused to throttling by the kubelet; raising the KubeAPIBurst and KubeAPIQPS limits resolved the issue. The throttling was caused by fetching secrets for each pod, which was in its own namespace. The subjects of pods/node, namespaces/node, and throttling parameters are complex and I do not propose to solve them here (those are scalability issues that have to be approached carefully). However, being able to easily identify that throttling is taking place without having to restart the kubelet at higher verbosity would help administrators identify the possibility of this issue. My test attempted to create 500 pods across 3 nodes (OpenShift allows 250 pods/node). Running at V=5, I captured a histogram of the throttling delay in seconds. There were 7881 throttling events reported total over perhaps a 45 minute period, with a somewhat bimodal (in this case) distribution:\n/sig node /sig scalability\nThanks for filing the pr to fix :)\nIssues go stale after 90d of inactivity. Mark the issue as fresh with . Stale issues rot after an additional 30d of inactivity and eventually close. If this issue is safe to close now please do so with . Send feedback to sig-testing, kubernetes/test-infra and/or . /lifecycle stale\nStale issues rot after 30d of inactivity. Mark the issue as fresh with . Rotten issues close after an additional 30d of inactivity. If this issue is safe to close now please do so with . Send feedback to sig-testing, kubernetes/test-infra and/or . /lifecycle rotten\n/remove-lifecycle rotten"}
{"_id":"doc-en-kubernetes-f2596026c617784a7bc3035384709bb19d64afff46adfc25b28b10c412ad8eb8","title":"","text":"The link in the following sentence is incorrect macOS Install Docker for Mac. See installation instructions . It redirects to instead of The link should get redirected to Navigate to Click on the hyperlink in the below section macOS Install Docker for Mac. See installation instructions . N/A N/A N/A N/A N/A N/A N/A\nThis issue is currently awaiting triage. If a SIG or subproject determines this is a relevant issue, they will accept it by applying the label and provide further guidance. The label can be by org members by writing in a comment. <detailsInstructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\n/assign\n/sig contributor-experience"}
{"_id":"doc-en-kubernetes-f8d2143077108332ac1fc56464c0f656ab79e989e202f2c3813e992956fd19da","title":"","text":"Hi, I have been trying to run api-server as an standalone service along with etcd. I am passing etcd-servers and cluster-ip-range as arguments, here is the error I am getting: etcd servers are up and running on that address. Any leads as to how to resolve this? TIA.\nThis is only happening with go1.8 version, not really sure what is causing issue while parsing the URI.\nCaused by this change i believe - CC as 1.8beta1 is being worked on here -\nThanks\nI'm getting this exact issue using Kubernetes 1.5.7 and Go 1.8.3. Reverting to Go 1.7 seems to resolve it. This happened after updating a Go 1.7-compiled 1.4.8 Kubernetes cluster to 1.5.7, not sure if that's a corner case. For reference, my stack trace:"}
{"_id":"doc-en-kubernetes-7b2f86b1021161afba33b63af83ddb3cce3267e73d9102e5d7e2a19b1a8a0e6b","title":"","text":"cc\nYet another failure\nThere is already fix in flight\nOk. I see\nThis was closed 5 days ago but appears to not be fixed. Many PRs have referenced it as a flake since the fix went in.\n- I referenced to it 2 after it was closed, but all of those failures were from runs before this was merged. So it's possible that it's already fixed.\nWhen was the fix merged? I saw this a couple hours ago:\nOops, same test but the error is slightly different... Maybe this is a seperate flake\nYeah - it seems to be something different.\nAlso - I'm not claiming it's definitely fixed - I just said that we need to double check, because the two that I linked was from before merge.\nI'll close this one for now. Please reopen the issue and cc us if the flake occurs again."}
{"_id":"doc-en-kubernetes-c72a20983cc57589ac47c7e12e6139151b8b668ad770346d9c321da0f1463da8","title":"","text":"When working on I a strange issue. It seems that while auto-generating conversions it works fine, but during things like or in the presubmits what is generated is different. Looking into test results: it seems that the order in which the files are processed is different then. /kind bug /sig api-machinery\nFTR: The log from above has some debug-logging and panics and the conversion-generator logs are those with\n/assign\nmodifies conversion-gen, but doesn't change the generated files. Did you intentionally exclude those changes? I wonder if it's similar to I checked out , ran I was able to reproduce the error in If that's the case, a workaround to fix the auto-generating is to first remove the affected files and re-generate them locally.\nThis is supposed to be no-op :) no auto-generated files should change That's not the fix. The problem is with the test jobs themselves: for generating swagger doc, they actually generate different things than they generate when you call \"make generated-files\" My PR is only a trigger for the problem, it's not a root cause. The auto-generated files are correct, but as part of running swagger docs, we're in-process generating a different stuff.\nIt looks that packages were missing. has a shorter / list compared to the other checks (e.g. ) when I ran . In the case of , the component-base config package was missing, thus the FIXME code path got hit. The list of packages was derived from . It seems that the dir was dirty during . I'm exploring if we can clean up the dir like does.\nIt seems that this one isn't yet fully fixed. I have updated the based on the comments on it and a similar issue on that (which I think has a very similar root cause). However, now \"govet\" and \"staticcheck\" are failing: [Ironically all other tests are passing, so I think this is again connected to some garbage somewhere] - can you please help with this?\nAs previously, I can reproduce it by:\nI'm fairly sure at this point that the inputdirs param here is too restrictive:\nIt seems that this commit might be addressing this:\nHeh - I think it helped me locally, but it doesn't help for - I probably did something wrong. But I'm fairly sure it's something along those lines that we need to do."}
{"_id":"doc-en-kubernetes-990f60044784b00bffd5b4d8f47018806db1b06c6e833c64dee7a7e61af70ed4","title":"","text":"- will you be able to help with it?\nSo yeah - I was completely wrong. Basically, running with with DBG it clearly is a problem with gen-conversion. And the problem is that we pass only limited input to it: I think we should always pass all inputs - but we should look a bit deeper into it.\nOK - so this is the problematic part: We basically process only packages that are explicitly requested by the user and ignore ones that are imported. So in order to have the conversion func imported by the generator, the package where it lives has to be explicitly imported. So we should basically always pass all input packages to conversion generator.\nI actually created a new one, as it seems to be a different thing. Closing in favor of\n/triage accepted"}
{"_id":"doc-en-kubernetes-1813ae1c1e9259cc81e5e43ebbd125b04ddddae9d4c14bbc382305b454beda00","title":"","text":"/priority failing-test /priority important-soon /kind flake /sig-api-machinery this failure is currently one of the top flake and repros in both Presubmits and CI jobs. This test is also part of the Conformance suite, which makes the flake far more critical. Can one of you please help triage this issue.\n/cc\n/cc\n/assign\nIssues go stale after 90d of inactivity. Mark the issue as fresh with . Stale issues rot after an additional 30d of inactivity and eventually close. If this issue is safe to close now please do so with . Send feedback to sig-testing, kubernetes/test-infra and/or . /lifecycle stale\n/remove-lifecycle stale\nflakes are continuing, hopefully logging gives us more info to work with\nthis test seems to flake quite a bit in the presubmits and its a conformance test on top of this. Can we please look into the new logs to see if we have more info to deflake this? Triage: Logs: https://k8s-\n/cc\nFWIW this test seems to pass consistently in our conformance dashboards - https://k8s-,%20master%20(dev). But seems to flake only in CI and Presubmits\nDo those run at different parallelism levels? It could be load/timeout sensitive\nI have also seen a general uptick in flakes in other e2e tests that could be sensitive to watch event latency, not just gc tests - for example, see the failures in https://k8s-\nIssues go stale after 90d of inactivity. Mark the issue as fresh with . Stale issues rot after an additional 30d of inactivity and eventually close. If this issue is safe to close now please do so with . Send feedback to sig-testing, kubernetes/test-infra and/or . /lifecycle stale\nHey, this test does not appear to be flaking anymore can we close this issue?\nClosing based on non-response. /close\nClosing this issue. <detailsIn response to : Closing based on non-response. Instructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\nThis was never actually fixed, looks like."}
{"_id":"doc-en-kubernetes-e8ea4b3822c8b6f9f15485ce88ff27518f982ee5d47e8079beb9d976426a7864","title":"","text":"If you spec a /8 for your service Cluster IPs, the bitmask becomes 2 MB, which is too big for etcd. We should enforce that the prefix be /12 or smaller or something like that.\nYes"}
{"_id":"doc-en-kubernetes-532d95a36e3e64a064736a1778f1c9d5e68ebcc137b6e1127e9746ac22dbf538","title":"","text":"<!-- Thanks for filing an issue! Before hitting the button, please answer these questions.--Is this a request for help? (If yes, you should use our troubleshooting guide and community support channels, see ): No What keywords did you search in Kubernetes issues before filing this one? (If you have found any duplicates, you should instead reply there.): statusFailure Is this a BUG REPORT or FEATURE REQUEST? (choose one): BUG REPORT <!-- If this is a BUG REPORT, please: Fill in as much of the template below as you can. If you leave out information, we can't help you as well. If this is a FEATURE REQUEST, please: Describe in detail the feature/behavior/change you'd like to see. In both cases, be ready for followup questions, and please respond in a timely manner. If we can't reproduce a bug or think a feature already exists, we might close your issue. If we're wrong, PLEASE feel free to reopen it and explain why. --Kubernetes version (use ): 4 Environment: Cloud provider or hardware configuration: OS (e.g. from /etc/os-release): Kernel (e.g. ): Install tools: Others: Others What happened: In FlexVolume code: The statusFailure constant should have string literal \"Failure\", not \"Failed\". Because in the test code or documentation everywhere, only \"Failure\" string is used, not \"Failed\" for the flexvolume cmd output status string. So this is a typo bug. Here is the test: here is the documentation: here is the example: What you expected to happen: The statusFailure constant should have string literal \"Failure\", not \"Failed\" at this line: How to reproduce it (as minimally and precisely as possible): code inspect: Anything else do we need to know: nothing else"}
{"_id":"doc-en-kubernetes-af5dea5ad6c50703324bee116a030e07ea1fcd8083467f9a4affa3fed2c086d0","title":"","text":"<!-- This form is for bug reports and feature requests ONLY! If you're looking for help check and the . --Is this a BUG REPORT or FEATURE REQUEST?: /kind bug What happened: There will be error when kubelet restarts on a agent if pod in that agent has an azure disk mounted, the error would be like following: azureDisk - No t a mounting point for disk andykubewin175-dynamic-pvc--92c7-11e7-b86b- on varlibkubeletpods-92c7-11e7-b86b-volumespvc--92c7-11 e7-b86b- What you expected to happen: pod could still works well even if kubelet restarts on the agent How to reproduce it (as minimally and precisely as possible): consider following scenario: A pod mounting an azure disk in a k8s agent The kubelet is restarted in that k8s agent The pod could not start up, it always reports error as following: 4d 1m 3065 kubelet, 14777acs9000 Warning FailedMount MountVolume.SetUp failed for volume \"pvc--92c7-11e7-b86b-\" : azureDisk - No t a mounting point for disk andykubewin175-dynamic-pvc--92c7-11e7-b86b- on varlibkubeletpods-92c7-11e7-b86b-volumespvc--92c7-11 e7-b86b- 4d 1m 3157 kubelet, 14777acs9000 Warning FailedMount Error syncing pod Anything else we need to know?: Environment: Kubernetes version (use ): 1.7, 1.8 Cloud provider or hardware configuration**: azure OS (e.g. from /etc/os-release): Kernel (e.g. ): Install tools: Others:\nThere are no sig labels on this issue. Please by: a sig: e.g., to notify the contributor experience sig, OR the label manually: e.g., to apply the label Note: Method 1 will trigger an email to the group. You can find the group list and label list ."}
{"_id":"doc-en-kubernetes-3017dae552427d7697f9eef98172533a65ddf628962bc7801cfca27e66cdff46","title":"","text":"The in the method 1 has to be replaced with one of these: bugs, feature-requests, pr-reviews, test-failures, proposals\n/sig azure\n/assign andyzhangx"}
{"_id":"doc-en-kubernetes-c443ff5d18f1d6c2d29d44384ec9cc1635dcc5ac37c91445fd7d9e71b7cfce65","title":"","text":"<!-- Please only use this template for submitting reports about failing tests in Kubernetes CI jobs --Which jobs are failing: Which test(s) are failing: Since when has it been failing: It has been consistently failing since 3/6, flaking a lot before that. Failure coincides with . Testgrid link: Reason for failure: cc /kind flake /kind failing-test /priority important-soon /milestone v1.14 /sig cluster-lifecycle\nsig gcp may be of help here /sig gcp\n/assign for routing from SCL side. These are google owned tests.\nLooking into this now. Any other googlers also doing so feel free to ping me :-)\nLooks like this bug is known with a fix in flight: cc\n/assign\nThanks ! Will keep an eye on this for 5 consistent passes, will reopen if there's a problem.\n/reopen Most gke jobs are able to create clusters now, however the serial jobs are still seeing failures\nReopened this issue. <detailsIn response to : /reopen Most gke jobs are able to create clusters now, however the serial jobs are still seeing failures Instructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\nNevermind, it's because the serial job with the fix is still running. I've verified that the cluster creation succeeds in the latest job. /close\nClosing this issue. <detailsIn response to : Nevermind, it's because the serial job with the fix is still running. I've verified that the cluster creation succeeds in the latest job. /close Instructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details\n/area platform/gke <img width=\"1106\" alt=\"Screen Shot 2019-03-16 at 12 00 52 PM\" src=\"https://user-\"Think we're done here"}
{"_id":"doc-en-kubernetes-73f410f7efe3cc7bf518718d7991551770a7ae4fe70afae3d631ca538f35fb20","title":"","text":"I have observed the DNS pod sometimes getting stuck. Need a health-check that tries to do a lookup or something like that.\nalso DNS should probably come up with N=2 by default\nWhy is this attached to 1.0 if it's P3? One of those is wrong.\nHealth checks I can believe. We're in the process of doing that for everything else -- Nothing else is HA, so I don't buy N=2 for DNS but not for other components.\nOther components are somewhat harder to make HA. DNS is trivially HA-able. But not too important, it's available after the fact, too. , Brian Grant wrote:\nso you mean a generic solution for health checking, or just you are adding health check for other service. In other words should I start adding health check for DNS?\nI'd still like to see a health-check for DNS, but I am not sure what is appropriate, though. Maybe have kube2sky open an HTTP server and answer /healthz iff it can complete a non-existing DNS query? Better ideas? It seems the collected other fixes add up to a more stable DNS server? , Piotr Szczesniak wrote:\nI assume this is about adding liveness probes to one or more of the containers in the dns pod: Is dig in the skydns image? We could use an exec probe to perform a DNS query.\nduh, exec probe is obviously better :) But it doesn't need to get a name, it just needs to complete and not hang. so something like sh -c \"alarm 5 nslookup foobar\" maybe? , Brian Grant wrote:\nleaving unassigned since I'm not working on it right now\nare you currently working on this? If not, I'll grab it. Thanks! --brendan\nI think that was looking into this today.\nYeah, I sat on it a little today due to not being able to get it to actually work, but I'm pretty sure it wasn't working because exec liveness probes as a whole are broken (). Sent PR to resolve this once exec probes are functional."}
{"_id":"doc-en-kubernetes-d011033a8c2d848ce3561906ba09865c53f3ece940f2eb809bc8d8929df344ea","title":"","text":"The needs updating to be in line with community agreed trajectory for the movement of cloud provider code out of k/k. For example the current statement: This statement is approximately 16 months old and implies there is still ambiguity, where there is now a much more clear plan which should be clearly communicated to those who might waste their time investing work in modifying something in this directory. The situation is one increasingly past maintenance mode, with out-of-tree implementations the goal in the 1.13 cycle followed by removal of in-tree provider code a few cycles later. See also discussion on slack: /kind cleanup /priority important-soon /sig cloud-provider /assign\nThanks for the reminder Tim, sorry this fell off my radar, will put together a PR today :)\nIssues go stale after 90d of inactivity. Mark the issue as fresh with . Stale issues rot after an additional 30d of inactivity and eventually close. If this issue is safe to close now please do so with . Send feedback to sig-testing, kubernetes/test-infra and/or . /lifecycle stale\nStale issues rot after 30d of inactivity. Mark the issue as fresh with . Rotten issues close after an additional 30d of inactivity. If this issue is safe to close now please do so with . Send feedback to sig-testing, kubernetes/test-infra and/or . /lifecycle rotten\n/remove-lifecycle rotten Need to update the README again to link out to the enhancement issues per cloud provider.\nPTAL :)\nawesome\nIssues go stale after 90d of inactivity. Mark the issue as fresh with . Stale issues rot after an additional 30d of inactivity and eventually close. If this issue is safe to close now please do so with . Send feedback to sig-testing, kubernetes/test-infra and/or . /lifecycle stale\n/close\nClosing this issue. <detailsIn response to : /close Instructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details"}
{"_id":"doc-en-kubernetes-251eef963dff19e63098abde61cb32ad4061447eb4b1f67f70c9d3499d1dc77d","title":"","text":"<!-- This form is for bug reports and feature requests ONLY! If you're looking for help check and the . --Is this a BUG REPORT or FEATURE REQUEST?: /kind bug What happened: Scheduler performance test does not work: What you expected to happen: How to reproduce it (as minimally and precisely as possible): Run schedulerperf by following: Anything else we need to know?: Environment: Kubernetes version (use ): Cloud provider or hardware configuration**: OS (e.g. from /etc/os-release): Kernel (e.g. ): Install tools: Others:\n/assign\n/sig scheduling"}
{"_id":"doc-en-kubernetes-6e8eb8f0999228ea44b81934b9bc720e607abe0b26d40545836abe46d9ceca36","title":"","text":"/kind failing-test Not sure if flake. Gubernator is showing a big spike in failures that started 06/14 around 10:00 ~UTC~ PT. This seems to be when merged. Failed on my PR: /sig testing\n/sig api-machinery\n/cc\nfiled I'm hesistant to bump it all the way back to 240, if you follow back the original PR / issue trail I ran a number of iterations of testing and it really shouldn't take 240s. four minutes is ludicrous for unit testing one package.\nthere are 39 failures in the last day in from timeout :/ /priority critical-urgent\nDid any timed-out run produce any logs?\nwe capture the test logs but itself doesn't appear to have any.\ne.g. at random (from triage above) =\"raw build-\" I'm not sure this code emits any when it fails.\nLikely it's timing out on some step, but I would have expected some logging from something related -- I was just going to look at the last thing logged and figure that something after that must be the culprit. Does that log capture stderr and stdout of the test invocations?\nit should, e.g. see a different kind of failure with logs: we're running with json output formatting then through to junit, and prow will capture stdout/stderr of whatever is running within the job as build-\nreopening to keep an eye on this.\nsigh: 180s\nit's doing better (all recent runs passed, looks like 3 flakes due to this since the PR merged this afternoon) but we may still want to bump it again\nWe're at 100% pass rate the past 3 hours but CI is not very active right now. still just 3 fails. If we have a similar rate tomorrow morning I'd opt for another small bump (maybe 180s -190s) to cover remaining limited flakes still over going all the way back to 4 minutes.\nIf we knew which sub test was slow when it is slow, that would possibly narrow it down to a haystack small enough to look for needles in.\ndug into it locally, follow up in:\nFairly confident this is now fixed after\nclosing as fixed by\nconfirmed\nexample:\nthat run happened at the fix merged at"}
{"_id":"doc-en-kubernetes-e910e751895c9a4adcc4aa9b295e34f4b420f777f06ea81f9820be7ed4c99f5c","title":"","text":"The script encounters an error when executed on Mac OS X because the option is not supported. The error message appears as follows: I expected the script to execute successfully. Execute the command on macOS. No response <details</details<details</details<details</details<details</details<details</details<details</details\n/sig testing\n/triage accepted\n/priority backlog"}
{"_id":"doc-en-kubernetes-c802a37aad85dfdce2d6232f30f1e039273d58e19f8aa63b8acb725eeb2b6040","title":"","text":"it's impossible to GET replication controllers on v3. on v1 on the same server it works fine. example request: GET on http://localhost:8080/api/v1beta3/replicationControllers returns 404 http://localhost:8080/api/v1beta1/replicationControllers returns 200\nv1beta3 moved to all lower case URL Would give that a try first Sent from my iPhone\nlooks like it's working. but it breaks backwards for clients who worked with v1. For instance, in my gem I can't now smoothly work with both versions because I need to treat the name of the resource differently. why don't at least redirect those who call replicationControllers to the lowercase too? Also, there's no way to know which resources are available and their names. A lot of REST api on their most top level at least publish the links to main resources, but if I call: http://localhost:8080/api/v1beta3/ it returns 404\nBecause we are actively trying to drop support for mixed case prior to 1.0, I did not continue supporting the old url paths. If you look at the go client, we have a \"legacy\" flag we use to control this and a few other behaviors.\nwell yes, but breaking backwards creates more work for the users... more \"if this version ...\" kind of stuff.\nWe got v1bet1/2 wrong and don't want to carry it forever. Stopping it early before v1 is our priority.\nok :( are the REST apis going through any kind of design review or code review?\nv1beta1 is really an alpha quality API. It grew organically and did not go through any kind of design review, as evidenced by the many issues you've filed. We want to get rid of it ASAP. We intend v1beta3 to be the \"release candidate\" for the v1 API, which will be the stable API. As for knowing what exists, you can now browse or GET . Additionally, pull is in progress, and will return a list of valid paths to GET .\nthanks. will check swagger for now"}
{"_id":"doc-en-kubernetes-22f9249afaa8c5861790bc63eab4a28e4fa1ff4ac55253dc70acafd4d0034e9c","title":"","text":"Already we're getting bugs from people who are missing some of the more subtle conversions.\nPoint them at the conversion tool: cmd/kube-version-\nPriority? Team? Milestone?\nDone:"}
{"_id":"doc-en-kubernetes-0635574d21f9ddc47e43e2c6f74baefafb7ed59d6808398eff3b67c17fc82a45","title":"","text":"build only 2 runs are available. ERROR: () build -dd5a-468e-9a7a- completed with status \"FAILURE\" \u001b[33m1 warning found (use --debug to expand): \u001b[0m - InvalidDefaultArgInFrom: Default value for ARG ${RUNNERIMAGE} results in empty or invalid base image name (line 16) Dockerfile.windows:22 20 | COPY etcd etcdctl /usr/local/bin/ 21 | COPY migrate-if- /usr/local/bin/ 22 | COPY migrate 23 | 24 | # NOTE(claudiub): docker buildx sets the PATH env variable to a Linux-like PATH, ERROR: failed to solve: mkdir /var/lib/buildkit/runc-overlayfs/cachemounts/buildkit3627953617/usr/local/bin: not a directory make[1]: Leaving directory '/workspace/cluster/images/etcd' make[1]: [Makefile:132: build] Error 1 make: [Makefile:217: sub-push-image-windows-amd64-1809] Error 2 ERROR ERROR: build step 0 \"\" failed: step exited with non-zero status: 2\n2024/07/25 02:39:53 Failed to run some build jobs: [error running [gcloud builds submit --verbosity info --config The etcd 3.5.15 post CI failed to build the image. needs to promote the image: This is post submit CI for /sig release etcd\n/cc\ncan you please try this?\ncc Folks, We could use your help fixing the etcd image for windows!\n/triage accepted /priority important-soon\nactually passed in"}
{"_id":"doc-en-kubernetes-84aabe4fdfca214c63d40e53a4b999fdb647fe5392c3ce898f956f6c961b8068","title":"","text":"POD is OOM. k8s v1.19.4+docker is OK. but v1.21+cri-o+buildah does not work properly. Why the limits are not effective.Not the container limit memory, It's physical memory centos7 kernel use limits memory use -XX:+UseContainerSupport Cannot read memory.limitinbytes k8s version cri-o version buildah version oracleJdk or openjdk have same problem It is normal to use xms and Xmx\n/sig node\nSupplement v19+docker v21+cri-o+buildah\n/assign taking a look, thanks for opening the issue\nseems relevant, have you checked your kernel supports the memory cgroup (I know docker does work, but let's sanity check first) I checked on my machine (fedora 34) and successfully can see usageinbytes in a container on our latest 1.22 branch. also, attaching the crio logs will likely help me debug\nalso curious about the runc version\ncri-o logs runc Looks like a problem with cgroup v1\nhm the memory cgroup seems to be enabled. what is the image you're using?\ncentos7.9.2009\n/triage needs-information\nwhat cgroup manager are you using?\nkubeadm kubectl config crio ps aux|grep kubelet\nI seem to have found this bug Delete the following two items to return to normal. deployment\nthat is unexpected... Great find!\nDoes the problem manifest if either are different? (hostNetwork: false, but dnsPolicy ClusterFirstWithHostNet)\nhostNetwork: true hostNetwork: false delete hostNetwork and dnsPolicy\nRocky 8.4, kernel-5.4, kubernetes v1.21.4 has the same problem\nah! I found the issue\nthanks for your patience\nIs crio 1.21.3 fixed? This problem still occurs after the upgrade. Branch v1.21.3 did not see the commit\nno the initial PR hasn't merged yet, I will try to revive it\nsorry this fell through the cracks, I'll try to get this through ASAP\nwill get it in 1.22.1. is that sufficient or would you like it backported to 1.21?"}
{"_id":"doc-en-kubernetes-5b183b7c89c3db39645de6f41384495ddabd41009f9474756ae64b18b81d7478","title":"","text":"Thank you Both are OK, I currently use java -xmx instead.I will update in the next version. Preferably v1.21, v1.22 version I will upgrade later.\nI've opened as well :)\n/remove-triage needs-information /triage accepted /close as this is being tracked as a CRI-O bug\nClosing this issue. <detailsIn response to : /remove-triage needs-information /triage accepted Instructions for interacting with me using PR comments are available . If you have questions or suggestions related to my behavior, please file an issue against the repository. </details"}