owner stringclasses 14
values | repo stringclasses 14
values | id int64 116k 4.61B | issue_number int32 1 181k | author stringlengths 1 39 | body stringlengths 1 262k | created_at timestamp[us]date 2000-06-06 02:40:44 2026-06-02 20:16:18 | updated_at timestamp[us]date 2000-06-06 02:40:44 2026-06-02 20:22:56 | reactions unknown | author_association stringclasses 5
values |
|---|---|---|---|---|---|---|---|---|---|
ClickHouse | ClickHouse | 651,107,111 | 11,948 | CurtizJ | Some benchmarks.
Used table `hits` with query `INSERT INTO hits_compact(CounterID) SELECT rand() from numbers(100)`
Regular `hits` table.
```
Queries executed: 1000.
localhost:9000, queries 1000, QPS: 102.008, RPS: 10200.778, MiB/s: 0.078, result RPS: 0.000, result MiB/s: 0.000.
0.000% 0.006 sec.
... | 2020-06-29T13:04:45 | 2020-06-29T13:04:45 | {} | MEMBER |
ClickHouse | ClickHouse | 651,420,212 | 4,810 | alexey-milovidov | Also see #10266 | 2020-06-29T23:26:42 | 2020-06-29T23:26:42 | {} | MEMBER |
ClickHouse | ClickHouse | 651,558,130 | 12,023 | mjodeh | Will update the version.
No we don't have alter delete/update but we have `alter add column` | 2020-06-30T05:54:30 | 2020-06-30T05:54:30 | {} | NONE |
ClickHouse | ClickHouse | 651,679,150 | 12,051 | liuzhizu | the outdate guide can result in build fail,please attention! | 2020-06-30T09:32:04 | 2020-06-30T09:32:04 | {} | NONE |
ClickHouse | ClickHouse | 651,712,354 | 4,810 | siradjev | > @siradjev It works perfectly on recent versions.
Confirmed with 20.3.11.97. Works. | 2020-06-30T10:38:04 | 2020-06-30T10:38:04 | {} | NONE |
ClickHouse | ClickHouse | 650,976,162 | 11,993 | akuzm | Some more stats made with this script: https://gist.github.com/akuzm/d1677a2810c2a6f7cbd056de1384c7ec
They show which metric from ProfileEvents changed (with statistical significance), for how many queries, by how much.
```
num_queries min med max avg metric_name top_ch... | 2020-06-29T07:11:18 | 2020-06-29T07:11:18 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 651,119,519 | 10,286 | alexey-milovidov | > We can't lose data after answering OK to user in case of sudden power failure or single disk failure?
No, you cannot.
There are multiple reasons why it's possible to lose your data if you use single replica:
- RAID write hole;
- non-atomic update of single sector on HDD;
- filesystem does not report fsync er... | 2020-06-29T13:26:19 | 2020-06-29T13:28:24 | {} | MEMBER |
ClickHouse | ClickHouse | 651,175,960 | 12,003 | alexey-milovidov | Yes, I had the same idea. We can support custom dictionary and store it directly in repository. But it will be very fragile if we don't normalize words. We can invest into text processing functions in ClickHouse but it's a big project. | 2020-06-29T14:57:41 | 2020-06-29T14:57:41 | {} | MEMBER |
ClickHouse | ClickHouse | 651,315,609 | 9,713 | kononencheg |
Query `ALTER TABLE _storage._source_update_rollup MOVE PARTITION tuple() TO TABLE _storage._source_update` leads to
```
2020.06.29 19:26:58.749833 [ 203 ] {1e98d644-7590-4e64-8e19-da4839b0e71a} <Warning> _storage._source_update: Tried to add obsolete part all_7_7_0 covered by all_2_22_2_102 (state Committed)
.... | 2020-06-29T19:31:27 | 2020-06-29T19:31:27 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 651,323,117 | 11,498 | vitlibar | > Some simple signature can be used to confirm that the data is created by proper person and was not changed, for example
> `alter_signature = sha256(initiator_username, query_id, query, signing_server, target_servers, secret_to_sign_alters, salt ), salt`
Ok.
For `secret` we can choose password's hash: if `initial... | 2020-06-29T19:48:24 | 2020-06-29T19:55:16 | {} | MEMBER |
ClickHouse | ClickHouse | 651,521,020 | 11,893 | excitoon | Sure! | 2020-06-30T04:16:16 | 2020-06-30T04:16:16 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 651,036,580 | 12,027 | KochetovNicolai | It doesn't seem to me as a bug pretty much.
You read from infinite multithreaded storage, and filter all the rows after limit. Some streams from storage are not even executing.
It may be solved in two ways:
1. Implement unlimited `numbers_mt` the same way as limited table function `numbers_mt(limit)` (add shared a... | 2020-06-29T09:49:23 | 2020-06-29T09:49:23 | {} | MEMBER |
ClickHouse | ClickHouse | 651,209,319 | 12,003 | filimonov | There is also
```
pip3 install scspell3k
find . -name '*.h' -or -name '*.cpp' | xargs scspell --report-only --use-builtin-base-dict ./
```
But it requires tuning (it stores & filling dictionaries incrementally etc. - smth like you describe above) check
https://github.com/myint/scspell | 2020-06-29T15:55:27 | 2020-06-29T16:00:15 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 651,409,060 | 12,021 | alexey-milovidov | @abyss7 The old test is perfectly correct but the code in this PR was wrong. | 2020-06-29T22:49:05 | 2020-06-29T22:49:05 | {} | MEMBER |
ClickHouse | ClickHouse | 650,792,620 | 11,662 | nikitamikhaylov | > @nikitamikhaylov can you help? I can't understand what is wrong with Functional stateless tests (unbundled) and Performance test. They don't see ORC output format.
I will help, of course. But there is another strange error with clang-tidy build... Which I don’t know how to fix. | 2020-06-28T16:50:57 | 2020-06-28T16:50:57 | {} | MEMBER |
ClickHouse | ClickHouse | 650,920,429 | 12,022 | filimonov | I seems it resolves #11202.
Does is also allow writing log to file? | 2020-06-29T05:39:14 | 2020-06-29T05:39:14 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 651,104,012 | 11,498 | vitlibar | I doubt there is a case where we should skip checking access rights for the initial user. Also it seems there is no reason to skip checking access rights for the remote users.
And I think we should prefer a solution which would be consistent with how other queries are processed.
When someone executes DDL on cluster... | 2020-06-29T13:00:00 | 2020-06-29T13:00:00 | {} | MEMBER |
ClickHouse | ClickHouse | 651,341,292 | 12,012 | filimonov | Duplicate of #7699 + #6104 or https://github.com/ClickHouse/ClickHouse/issues/8445
Use `joinGet('dbname.tablename',...) ` for `on cluster`. | 2020-06-29T20:24:25 | 2020-06-29T20:27:23 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 651,416,395 | 3,740 | alexey-milovidov | @orloffv Я не понял, почему коллеги забили на эту задачу. Исправление тривиально. Сделаю. | 2020-06-29T23:13:57 | 2020-06-29T23:13:57 | {} | MEMBER |
ClickHouse | ClickHouse | 651,561,764 | 11,887 | mjodeh | So once the shard is back online clickhouse will re-blance the data, during that the query will be distributed between the live shards, right?
Ho much resources does this re-insert requires?
Does this unbalanced shards affects the query performance? | 2020-06-30T06:04:35 | 2020-06-30T06:04:35 | {} | NONE |
ClickHouse | ClickHouse | 651,922,177 | 9,032 | yurriy | > @yurriy is not working on this issue right now, we have to reassign.
It was fixed in #9079. Probably @den-crane reopened it because it wasn't in a stable version at that moment. | 2020-06-30T17:00:52 | 2020-06-30T17:01:18 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 650,785,297 | 12,009 | alexey-milovidov | > Integration tests
Broken in #11903 | 2020-06-28T15:55:44 | 2020-06-28T15:55:44 | {} | MEMBER |
ClickHouse | ClickHouse | 651,011,121 | 12,025 | dependabot-preview[bot] | Looks like idna is up-to-date now, so this is no longer needed. | 2020-06-29T08:24:38 | 2020-06-29T08:24:38 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 651,117,389 | 12,021 | abyss7 | You forgot to fix old test: 00197_if_fixed_string | 2020-06-29T13:22:45 | 2020-06-29T13:22:45 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 651,136,509 | 11,871 | akuzm | I like the branch name, but this is not a good size for a PR -- hard to review, and, judging by the amount of debug output left, even for yourself. If it has some smaller part that makes sense by itself (e.g. a couple of new functions + infractructure for them), it would be better to split it out and commit first.
W... | 2020-06-29T13:52:57 | 2020-06-29T13:52:57 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 651,208,749 | 11,932 | santoshdvn | I have created table with below definition
CREATE TABLE test.testing_clickhouse (
column1 String,
column2 String
) ENGINE = Log; | 2020-06-29T15:54:30 | 2020-06-29T15:54:30 | {} | NONE |
ClickHouse | ClickHouse | 651,421,444 | 11,372 | alexey-milovidov | @KochetovNicolai Signed integer overflow will be also reported by UBSan. We should fix it. | 2020-06-29T23:30:22 | 2020-06-29T23:30:22 | {} | MEMBER |
ClickHouse | ClickHouse | 651,461,843 | 12,012 | peaksnail | Hi,
It's not work for me,
clickhouse version 19.17.4.11 and it's not cluster | 2020-06-30T01:24:33 | 2020-06-30T01:24:33 | {} | NONE |
ClickHouse | ClickHouse | 651,463,814 | 11,861 | wallace-clickhouse | @den-crane I want to pass a password by using a global variable,which can avoid writing password in the metrika.xml.But I don't know where to define the global variable is situable.Can you give me some advise? Thank you!
PS:By this way,because I can get the password from a password management service by calling the se... | 2020-06-30T01:30:24 | 2020-06-30T01:30:24 | {} | NONE |
ClickHouse | ClickHouse | 650,785,159 | 12,007 | tavplubix | Could you please provide a test? | 2020-06-28T15:54:33 | 2020-06-28T15:54:33 | {
"+1": 1
} | MEMBER |
ClickHouse | ClickHouse | 650,819,560 | 11,498 | filimonov | Proposal:
1) keep initial_user 'as is' . Don't introduce any changes in cluster config
2) introduce `effective_user` - the one who is used for checking permissions, etc.
2) if logged in user has a PROXY grant and `initial_user` passed, then use `initial_user` as the effective user.
3) Otherwise currently logged in ... | 2020-06-28T20:46:58 | 2020-06-29T07:44:01 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 651,025,521 | 1,841 | shwangdev | Really need this feature. Any update yet? | 2020-06-29T08:51:23 | 2020-06-29T08:51:23 | {} | NONE |
ClickHouse | ClickHouse | 651,176,595 | 11,360 | alexey-milovidov | It does not reproduce in release build (that I use on my workstation). But it's easy, I already have the fix. | 2020-06-29T14:58:42 | 2020-06-29T14:58:42 | {} | MEMBER |
ClickHouse | ClickHouse | 651,185,786 | 11,498 | filimonov | @vitlibar
> And I think we should prefer a solution which would be consistent with how other queries are processed.
> When someone executes DDL on cluster (for example, CREATE TABLE ON CLUSTER)
> ClickHouse checks whether the initial user has the grant CREATE TABLE on the initial node.
That is not done good & ... | 2020-06-29T15:14:22 | 2020-06-29T15:14:22 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 651,188,600 | 11,662 | nikitamikhaylov | Tests `01308_orc_output_format_arrays` and `01307_orc_output_format` are now disabled for unbundled build. Because tests for orc input format are already disabled. | 2020-06-29T15:19:11 | 2020-06-29T15:19:11 | {} | MEMBER |
ClickHouse | ClickHouse | 651,260,549 | 11,498 | vitlibar | > b) if we are talking about the situation when cluster is homogenous - we expect that all permissions are the same across the nodes (no need to pass them conditions back and forward, we can just pick them by initial_user locally).
> b) if we are talking about the situation when cluster is NOT homogenous - the initato... | 2020-06-29T17:33:55 | 2020-06-29T17:34:17 | {} | MEMBER |
ClickHouse | ClickHouse | 651,742,825 | 11,710 | myrrc | I'll take that. | 2020-06-30T11:48:54 | 2020-06-30T11:48:54 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 651,940,096 | 12,048 | jinyan2049 | who can help me ? | 2020-06-30T17:35:41 | 2020-06-30T17:35:41 | {} | NONE |
ClickHouse | ClickHouse | 650,756,814 | 12,011 | alesapin | Wait for tests | 2020-06-28T13:21:44 | 2020-06-28T13:21:44 | {} | MEMBER |
ClickHouse | ClickHouse | 650,922,519 | 11,360 | akuzm | > You forgot to attach the stack trace.
I always skip them for easily reproducible issues (i.e. just copy+paste to clickhouse-client) to reduce visual clutter. Do you find it more convenient to always have the trace? Maybe I should start attaching them even in simple cases. | 2020-06-29T05:46:04 | 2020-06-29T05:46:04 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 651,135,306 | 12,023 | den-crane | You are using very early 20.1 version. Try to upgrade to 20.1.16.120.
Do you use mutations? Alter delete/update? | 2020-06-29T13:50:59 | 2020-06-29T13:50:59 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 651,153,832 | 11,432 | calebeaires | On my case I must create the table even if has corruption to get the flow done. Here is how I handle this problem. I dont know if it is the best to do.
```typescript
// drop first (if table exists)
// then
try {
// create table
return await this.db.query(
... | 2020-06-29T14:20:49 | 2020-06-29T14:20:49 | {
"+1": 3
} | NONE |
ClickHouse | ClickHouse | 651,203,701 | 12,003 | filimonov | ```
$ pip3 install codespell
$ pwd
/home/mfilimonov/workspace/ClickHouse/src
$ codespell
./Common/Volnitsky.h:347: occurence ==> occurrence
./Common/Volnitsky.h:536: ans ==> and
./Common/Volnitsky.h:540: ans ==> and
./Common/Volnitsky.h:540: ans ==> and
./Common/Volnitsky.h:556: ans ==> and
./Common/Volni... | 2020-06-29T15:45:33 | 2020-06-29T15:45:33 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 651,413,530 | 11,977 | alexey-milovidov | @KochetovNicolai Looks like already fixed issue? | 2020-06-29T23:04:14 | 2020-06-29T23:04:14 | {} | MEMBER |
ClickHouse | ClickHouse | 651,751,483 | 12,020 | alexey-milovidov | > Performance — 1 errors, 1 too long, 1 faster, 204 unstable
Ok. @akuzm | 2020-06-30T12:09:53 | 2020-06-30T12:09:53 | {} | MEMBER |
ClickHouse | ClickHouse | 650,771,468 | 12,013 | alexey-milovidov | The query
```
SELECT i
FROM generateRandom('i Array(Array(Array(Array(Array(Array(Array(Array(Array(String)))))))))', NULL, 1000000000, 1000000000)
```
Sometimes gives
```
Memory limit (for query) exceeded: would use 256.00 TiB (attempt to allocate chunk of 281474977759896 bytes), maximum: 9.31 GiB: While ... | 2020-06-28T14:40:29 | 2020-06-28T14:40:29 | {} | MEMBER |
ClickHouse | ClickHouse | 650,817,014 | 11,498 | filimonov | > real user authorization...
How? Storing user credentials in session is a very-very bad idea, not safe and is not acceptable. For authentication schemas like Kubernetes it require to ask for lot if tickets for executing single query, also not acceptable. | 2020-06-28T20:24:54 | 2020-06-28T20:24:54 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 650,981,596 | 11,993 | akuzm | Ah, forgot the most important thing. Here's a graph for MemoryResident (it's already in test results archive, 'metrics/MemoryResident.png'):
You can see that the memory usage is somewhat higher, b... | 2020-06-29T07:23:11 | 2020-06-29T07:23:11 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 651,024,423 | 11,983 | alesapin | Ok, let's merge | 2020-06-29T08:49:14 | 2020-06-29T08:49:14 | {} | MEMBER |
ClickHouse | ClickHouse | 651,067,893 | 11,401 | akuzm | > > Due to a bug in CI, a second performance test task just finished, compared the same commits, and now 15 arithmetic queries are faster! It's hilarious.
>
> @akuzm which machines are used for performance tests? Virtualization?
No virtualization.
> Maybe there is steal? (BTW may worth adding it into `Asynchr... | 2020-06-29T11:55:46 | 2020-06-29T11:58:55 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 651,329,693 | 11,995 | azat | Will rebase against uptream/master, since I moved one patch(e02cf7a652eba0e6e72e4dcd92d381727ab9bd12) into #12040 (did not want to create separate PR just for this, but two unrelated changes is worse) | 2020-06-29T20:03:12 | 2020-06-29T20:03:12 | {} | MEMBER |
ClickHouse | ClickHouse | 651,505,189 | 11,861 | den-crane | >password using from_env
BTW, I never checked is it possible or not. So first thing - check this.
Obviously you need own start script which will prompt for a password then set an environment and then run CH.
Probably the password will be available for root user `ps ewww` , I don't remember the way how to solve ... | 2020-06-30T03:28:00 | 2020-06-30T03:28:45 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 651,663,769 | 11,498 | azat | >Storing user credentials in session is a very-very bad idea, not safe and is not acceptable.
password/user already stored for the connection.
And we can use password hash for authorization (and it is already supported by the [server](https://github.com/ClickHouse/ClickHouse/blob/dd9b02c2fb2fb8e19de489da60c8c409b... | 2020-06-30T09:05:40 | 2020-06-30T09:05:40 | {} | MEMBER |
ClickHouse | ClickHouse | 650,792,229 | 11,662 | Avogar | @nikitamikhaylov can you help? I can't understand what is wrong with Functional stateless tests (unbundled) and Performance test. They don't see ORC output format. | 2020-06-28T16:47:56 | 2020-06-28T16:47:56 | {} | MEMBER |
ClickHouse | ClickHouse | 651,326,321 | 11,479 | 4ertus2 | Still cannot reproduce. | 2020-06-29T19:55:38 | 2020-06-29T19:55:38 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 651,419,650 | 4,810 | alexey-milovidov | @siradjev It works perfectly on recent versions. | 2020-06-29T23:24:56 | 2020-06-29T23:24:56 | {} | MEMBER |
ClickHouse | ClickHouse | 651,755,709 | 10,574 | nvartolomei | Caused by https://github.com/ClickHouse/ClickHouse/commit/1168ba53157ab59ffc9c38f1d7c962c6c5d7a1f9 from https://github.com/ClickHouse/ClickHouse/pull/4099 | 2020-06-30T12:19:41 | 2020-06-30T12:19:41 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 651,902,095 | 11,910 | CurtizJ | > How I understand when I add redis dictonary to CH tn run KEYS * then try do type in all keys.
Yes, now `KEYS *` is used.
> Second can I use patterns for keys? Scanning all keys for type is too long.
Do you want to set this pattern in dictionary config? Now it's impossible, but it's easy to implement.
We w... | 2020-06-30T16:23:28 | 2020-06-30T16:23:40 | {
"+1": 2,
"eyes": 1
} | MEMBER |
ClickHouse | ClickHouse | 650,813,122 | 11,498 | azat | So to summarize I'm still not sure about PROXYing over user authorization (get user credential and authorize user with them on the remote side, w/o any reconnect)...
>> provide access if both initial user and cluster's user on that node have corresponding grants.
>
>This is still not enough, since most of the time... | 2020-06-28T19:49:43 | 2020-06-28T19:49:43 | {} | MEMBER |
ClickHouse | ClickHouse | 650,814,066 | 12,013 | alexey-milovidov | > We should investigate.
It's another issue. When multiple threads are trying to allocate more memory and one of them throws an exception, the memory tracker is reset and another thread may overallocate memory. | 2020-06-28T19:58:33 | 2020-06-28T19:58:33 | {} | MEMBER |
ClickHouse | ClickHouse | 650,945,908 | 12,022 | oandrew | > I seems it resolves #11202.
> Does is also allow writing log to file?
Yep. `clickhouse-local --logger.log file --logger.console=false` write to file and disable console (otherwise it will write to both file and console) | 2020-06-29T06:18:45 | 2020-06-29T06:19:18 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 651,107,270 | 11,948 | CurtizJ | I've run benchmark on my cloud VM with HDD and it's strange that fsyncs took so long. On my laptop with SSD queries with fsync were executed about 10 times faster, while other queries took the same time. BTW, i didn't expect that difference will be so significant. | 2020-06-29T13:05:04 | 2020-06-29T13:33:52 | {} | MEMBER |
ClickHouse | ClickHouse | 651,751,804 | 12,021 | alexey-milovidov | > Performance
Not Ok. | 2020-06-30T12:10:41 | 2020-06-30T12:10:41 | {} | MEMBER |
ClickHouse | ClickHouse | 651,952,712 | 12,051 | filimonov | Can you send a PR with fix? There is an edit button on a page (looks like pencil) | 2020-06-30T18:00:44 | 2020-06-30T18:00:44 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 650,924,641 | 12,023 | filimonov | The described usecase is ok.
The 'instability' is most probably caused by oomkiller when clickhouse tries to allocate lot of memory. Check system logs (for example with `dmesg` command) and ClickHouse logs (/var/log/clickhouse-server/) after crash.
Adjust memory settings of ClickHouse (most probably you need to... | 2020-06-29T05:52:42 | 2020-06-29T05:52:42 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 651,070,818 | 11,401 | azat | >>Maybe there is steal? (BTW may worth adding it into AsynchronousMetrics::update too?).
>
>Sure, what exactly to add?
steal form `/proc/stat`, from `proc 5` it is 8 column and linux 2.6.11+
>By the way, we now have graphs for CPU frequency in perf test output archive
Yeah, I saw this code in `AsynchronousMe... | 2020-06-29T12:00:56 | 2020-06-29T12:00:56 | {} | MEMBER |
ClickHouse | ClickHouse | 651,590,741 | 12,033 | akuzm | Well, it does show something (average relative data tlb misses for perf tests):
```
$ clickhouse-local --query "select floor((sumMap([metric_name], [right]).2 as a)[1]/a[2], 5) tlb_misses, test from file('report/query-metric-stats.tsv', TSVWithNamesAndTypes, 'metric_name text, left float, right float, diff float, sta... | 2020-06-30T07:04:30 | 2020-06-30T07:06:51 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 650,817,895 | 11,498 | filimonov | > initial_user (it is the initiator of the Distributed queyr) is not the same as proxied_user (user who executes local query) and should not be.
It's mostly terminology.
1)We have logged in user.
2)That user wants to talk in the name of other user who was logged in remotely and initialized the query (you can... | 2020-06-28T20:32:52 | 2020-06-28T20:35:54 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 650,822,828 | 11,360 | alexey-milovidov | You forgot to attach the stack trace. | 2020-06-28T21:16:11 | 2020-06-28T21:16:11 | {} | MEMBER |
ClickHouse | ClickHouse | 651,036,520 | 11,103 | KochetovNicolai | Yes. It will probably be first optimisation for QueryPlan | 2020-06-29T09:49:16 | 2020-06-29T09:49:16 | {
"+1": 1
} | MEMBER |
ClickHouse | ClickHouse | 651,146,156 | 7,774 | abyss7 | According to our new rules all WIP PRs should be converted to the draft state. | 2020-06-29T14:07:49 | 2020-06-29T14:07:49 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 651,259,379 | 11,498 | vitlibar | @filimonov
> I.e. If users Alice and Bob both work in the same department and have the same rights Alice could pretend that she is Bob and drop some tables, and it will be impossible (or hard) to track such situation in the logs (Bob could be fired by angry boss before Bob will able to prove that it was Alice).
O... | 2020-06-29T17:31:38 | 2020-06-29T17:55:02 | {} | MEMBER |
ClickHouse | ClickHouse | 651,273,359 | 11,498 | filimonov | > Yes, but it's unclear what to do about that. Don't give access to your zookeeper to bad people)
Some simple signature can be used to confirm that the data is created by proper person and was not changed, for example
`alter_signature = sha256(initiator_username, query_id, query, signing_server, target_servers, sec... | 2020-06-29T18:00:23 | 2020-06-29T18:00:23 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 651,337,722 | 12,032 | filimonov | It is regreassion, so major. | 2020-06-29T20:18:12 | 2020-06-29T20:18:12 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 651,619,170 | 12,050 | dependabot-preview[bot] | Looks like python-slugify is up-to-date now, so this is no longer needed. | 2020-06-30T07:54:42 | 2020-06-30T07:54:42 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 651,722,498 | 12,045 | tavplubix | `00738_lock_for_inner_table` - seems not related, added to CI Fest | 2020-06-30T11:02:31 | 2020-06-30T11:02:31 | {} | MEMBER |
ClickHouse | ClickHouse | 650,755,819 | 12,005 | alexey-milovidov | > Integration tests
Broken in #11903 | 2020-06-28T13:16:22 | 2020-06-28T13:16:22 | {} | MEMBER |
ClickHouse | ClickHouse | 650,773,320 | 12,008 | den-crane | That's how insert deduplication works.
https://clickhouse.tech/docs/en/engines/table-engines/mergetree-family/replication/
_Data blocks are deduplicated. For multiple writes of the same data block (data blocks of the same size containing the same rows in the same order), **the block is only written once**. The r... | 2020-06-28T14:49:46 | 2020-06-28T14:50:14 | {
"+1": 1,
"hooray": 2
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 650,784,750 | 11,590 | alexey-milovidov | Thank you!
We can accept tcmalloc version with even more significant patches...
PS. Even if you write your own allocator on top of tcmalloc, it is acceptable :) | 2020-06-28T15:51:57 | 2020-06-28T15:52:33 | {} | MEMBER |
ClickHouse | ClickHouse | 650,837,685 | 12,007 | oandrew | Failed test does not seem to be related (mysql) | 2020-06-28T23:34:53 | 2020-06-28T23:34:53 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 650,937,155 | 12,003 | akuzm | > @akuzm Any ideas on how to automate this?
> I have tried `ispell` or `aspell` but it has too many false positives.
Not sure. The problem is that we don't want to support custom dictionaries? Maybe we can pipe all sources to clickhouse-local, build a list of words that are used in sources, and warn if we add a wor... | 2020-06-29T06:06:05 | 2020-06-29T06:06:05 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 651,048,806 | 11,401 | azat | >Due to a bug in CI, a second performance test task just finished, compared the same commits, and now 15 arithmetic queries are faster! It's hilarious.
@akuzm which machines are used for performance tests? Virtualization? Maybe there is steal? (BTW may worth adding it into `AsynchronousMetrics::update` too?).
I d... | 2020-06-29T11:19:57 | 2020-06-29T11:19:57 | {} | MEMBER |
ClickHouse | ClickHouse | 651,078,999 | 12,027 | filimonov | Discussed in Telegram. It's was a bit incorrect attempt to reproduce the performance issue, the real issue is here: https://github.com/ClickHouse/ClickHouse/issues/12030
The issue with infinite table and unfortunate streams / threads scheduling is quite interesting, but out of the scope for now.
Maybe we can in... | 2020-06-29T12:16:16 | 2020-06-29T12:16:42 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 651,202,637 | 11,873 | alexey-milovidov | > Integration tests
Broken in #11903 | 2020-06-29T15:43:42 | 2020-06-29T15:43:42 | {} | MEMBER |
ClickHouse | ClickHouse | 651,203,333 | 11,893 | Akazz | Do you think there can be a simple test that changes policy? | 2020-06-29T15:44:56 | 2020-06-29T15:44:56 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 651,263,759 | 11,498 | vitlibar | > I can drop all your databases just having zookeeper access. You will not even know in that case who dropped your databases.
Yes, but it's unclear what to do about that. Don't give access to your zookeeper to bad people) | 2020-06-29T17:40:39 | 2020-06-29T17:40:39 | {} | MEMBER |
ClickHouse | ClickHouse | 651,356,666 | 11,498 | filimonov | That secret should be inaccessible for users, should be configured between the servers (otherwise the user still can add bad queries into zookeeper w/o help of server).
I think it should be either smth completely isolated, or derived from interserver credentials (those listed in remote_servers) | 2020-06-29T20:49:33 | 2020-06-29T20:49:57 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 651,558,825 | 12,023 | mjodeh | I see from your releases that `v20.4.6.53-stable` is it production ready? | 2020-06-30T05:56:20 | 2020-06-30T05:56:20 | {} | NONE |
ClickHouse | ClickHouse | 651,673,366 | 11,498 | filimonov | > password/user already stored for the connection.
That is a security hole/weakness, categorized as [CVE-312](https://cwe.mitre.org/data/definitions/312.html) or more specific [CVE-316](https://cwe.mitre.org/data/definitions/316.html)
>> If some complicated authentication protocols will be used (like Kerberos) it... | 2020-06-30T09:21:33 | 2020-06-30T09:27:20 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 651,727,078 | 11,986 | SergeyIBM47 | A few days later my old stream has begun work without any change, but on other environments, I saw the same behaviour although I re-setup stream with new consumer id. After investigate an clickhouse server config file I found misconfiguration of my clickhouse Kafka consumer config which is followed to read old messages... | 2020-06-30T11:12:33 | 2020-06-30T11:12:33 | {
"+1": 2
} | NONE |
ClickHouse | ClickHouse | 651,951,245 | 12,048 | filimonov | Please use search first
Duplicate of #4954
Duplicate of #7690 | 2020-06-30T17:57:44 | 2020-06-30T17:57:44 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 650,782,923 | 5,157 | alexey-milovidov | Let's prefer "US" order for all variants if `parseDateTimeBestEffortUS` is called...
(But if YYYY is the first, always treat the following as month and day.)
I'm not 100% sure what is better. In most cases, people write US format using slash as a separator.
From Wikipedia:
> The most commonly used separator in ... | 2020-06-28T15:41:43 | 2020-06-28T15:41:43 | {} | MEMBER |
ClickHouse | ClickHouse | 650,811,538 | 12,011 | alexey-milovidov | It helps! But in the meantime another integration test was broken. | 2020-06-28T19:35:24 | 2020-06-28T19:36:05 | {} | MEMBER |
ClickHouse | ClickHouse | 650,846,756 | 12,020 | alexey-milovidov | > Yandex synchronization check (only for Yandex employees)
"Arcadia" build is wrong:
> Collations support is disabled, because ClickHouse was built without ICU library (version 20.6.1.1-arcadia) | 2020-06-29T00:40:31 | 2020-06-29T00:40:31 | {} | MEMBER |
ClickHouse | ClickHouse | 651,210,364 | 7,081 | tobia | Why does `set(0)` lead to trouble? | 2020-06-29T15:57:22 | 2020-06-29T15:57:22 | {} | NONE |
ClickHouse | ClickHouse | 651,349,640 | 11,479 | millin | @4ertus2 I noticed that segmentation fault occurs only at high load on the server.
When the load is light, the request simply fails with error like:
```
Poco::Exception. Code: 1000, e.code() = 224265568, e.displayText() = mysqlxx::Exception: (H��H[]A\A]A^A_�f.�:224316616) (version 20.3.8.53 (official build))
``` | 2020-06-29T20:37:49 | 2020-06-29T20:37:49 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 651,672,233 | 11,476 | adubovikov | any updates on this ? | 2020-06-30T09:19:35 | 2020-06-30T09:19:35 | {} | NONE |
ClickHouse | ClickHouse | 651,710,791 | 10,574 | nvartolomei | Simple test case:
```sql
drop database if exists test;
create database test;
use test;
create table z(v UInt64) engine = MergeTree() order by (v) settings index_granularity = 1;
insert into z select number from numbers(10000000);
select count() from z where (v, 0) IN ((1, 0));
```
Last good version availab... | 2020-06-30T10:34:57 | 2020-06-30T11:55:01 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 656,851,774 | 12,372 | qoega | It is just a label for further work with docs. Translate it, etc. | 2020-07-10T19:27:26 | 2020-07-10T19:27:26 | {} | MEMBER |
ClickHouse | ClickHouse | 656,853,371 | 235 | alexey-milovidov | Pre-built binaries are published in docs: https://clickhouse.tech/docs/en/getting-started/install/#from-binaries-non-linux | 2020-07-10T19:31:29 | 2020-07-10T19:31:29 | {} | MEMBER |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.