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 | 646,956,022 | 11,819 | alexey-milovidov | Does it reproduce if you update to the current master? | 2020-06-20T07:13:23 | 2020-06-20T07:13:23 | {} | MEMBER |
ClickHouse | ClickHouse | 646,958,218 | 6,549 | alexey-milovidov | I have tried to reproduce the issue in recent release and I see that everything is working well even for non-replicated tables.
I have added a test. | 2020-06-20T07:36:27 | 2020-06-20T07:36:27 | {} | MEMBER |
ClickHouse | ClickHouse | 646,989,911 | 11,822 | alexey-milovidov | @alesapin It did not help for unknown reason. | 2020-06-20T12:43:47 | 2020-06-20T12:43:47 | {} | MEMBER |
ClickHouse | ClickHouse | 647,047,504 | 11,806 | alexey-milovidov | @azat I have no idea why we have 20.5 branch - it should not exist. | 2020-06-20T21:29:47 | 2020-06-20T21:29:47 | {} | MEMBER |
ClickHouse | ClickHouse | 647,108,850 | 11,590 | azat | >There is some code from a sibling project in Yandex to cache middle-sized allocations on top of allocator. We can bring this code into ClickHouse... I did not tested it, but we can try.
Hope that you are not talking about LFAlloc (#5369)
>BTW it's possible that Linux kernel in our CI does not support "restartabl... | 2020-06-21T10:22:37 | 2020-06-21T10:22:37 | {} | MEMBER |
ClickHouse | ClickHouse | 647,113,571 | 11,715 | alexey-milovidov | >
optimize_skip_unused_shards_nesting -- accept nesting level on which the
optimization will work
force_skip_optimize_shards_nesting -- accept nesting level on which
the optimization will work
Much better than `optimize_skip_unused_shards=2` | 2020-06-21T11:10:25 | 2020-06-21T11:10:25 | {} | MEMBER |
ClickHouse | ClickHouse | 646,955,638 | 6,549 | thyn | > If it's not fixed, why close?
After starting using zookeeper I don't have that problem. | 2020-06-20T07:09:23 | 2020-06-20T07:09:23 | {} | NONE |
ClickHouse | ClickHouse | 646,963,366 | 9,577 | alexey-milovidov | @excitoon Should we close this? | 2020-06-20T08:34:26 | 2020-06-20T08:34:26 | {} | MEMBER |
ClickHouse | ClickHouse | 646,989,465 | 11,827 | alexey-milovidov | @alesapin I don't know how to test it, because testing Docker images already contain `tzdata` for the needs of other tools and programs. | 2020-06-20T12:39:14 | 2020-06-20T12:39:14 | {} | MEMBER |
ClickHouse | ClickHouse | 647,034,354 | 11,825 | alexey-milovidov | #11835 | 2020-06-20T19:06:58 | 2020-06-20T19:06:58 | {} | MEMBER |
ClickHouse | ClickHouse | 647,106,994 | 11,590 | alexey-milovidov | I name allocation "middle sized" if it is less than what should go directly to mmap (64 MiB in my experience) but greater than a few KiB. We use these allocations a lot for columns and buffers.
There is some code from a sibling project in Yandex to cache middle-sized allocations on top of allocator. We can bring thi... | 2020-06-21T10:03:24 | 2020-06-21T10:03:24 | {} | MEMBER |
ClickHouse | ClickHouse | 647,109,160 | 11,590 | alexey-milovidov | > Hope that you are not talking about LFAlloc (#5369)
That separate caching layer was implemented to mitigate the issue that LFAlloc does mmap too frequently :) | 2020-06-21T10:25:29 | 2020-06-21T10:25:29 | {} | MEMBER |
ClickHouse | ClickHouse | 647,109,736 | 11,774 | alexey-milovidov | I'm interested in this pull request. It is better to use the same allocator on all systems.
BTW, do you have setups to test it? I'm asking because we don't run tests on non-Linux.
ClickHouse is used on Mac for two use cases:
- for clickhouse-client;
- for development.
ClickHouse is also used on FreeBSD serve... | 2020-06-21T10:31:30 | 2020-06-21T10:32:06 | {} | MEMBER |
ClickHouse | ClickHouse | 647,123,715 | 11,153 | HooliBox | I am on version version 20.4.5.36 (official build).
I did the grep max_server_memory_usage /var/log/clickhouse-server/clickhouse-server.log
```
2020.06.21 11:20:40.539962 [ 506 ] {} <Information> Application: Setting max_server_memory_usage was set to 2.86 TiB
2020.06.21 11:21:59.810640 [ 679 ] {} <Information> A... | 2020-06-21T12:47:15 | 2020-06-21T12:47:15 | {} | NONE |
ClickHouse | ClickHouse | 647,126,911 | 11,765 | alexey-milovidov | `SELECT tuple(1) < '';` | 2020-06-21T13:15:15 | 2020-06-21T13:15:15 | {} | MEMBER |
ClickHouse | ClickHouse | 647,132,894 | 11,153 | HooliBox | I pasted log file [here](https://gist.github.com/HooliBox/894105b306bd90e789d4a0f21f9745e0).
I tried both tcp and http, they all failed with same error. The only difference is tcp takes couple more iterations to fail, I guess it is using less memory? Not sure. | 2020-06-21T14:06:22 | 2020-06-21T14:06:22 | {} | NONE |
ClickHouse | ClickHouse | 647,135,550 | 11,826 | alexey-milovidov | > ClickHouse build check — 15/19 builds are OK
> Build failed without build log. It's inner CI problem. Maybe something wrong with docker or host. Just wait for build restart.
It is an issue in infamous "Sandbox" infrastructure. | 2020-06-21T14:28:37 | 2020-06-21T14:28:37 | {} | MEMBER |
ClickHouse | ClickHouse | 647,143,459 | 11,841 | den-crane | hm, it's rather not a memory leak.
Can you show
```
select name,value from system.settings where name like '%mem%'
select name,value from system.settings where name like '%group%'
```
first execution :
2020.06.21 14:20:59.067416 [ 480 ] {dc6cf0de-74ff-4c1a-9e47-09a401755aff} <Debug> MemoryTracker: Peak m... | 2020-06-21T15:34:17 | 2020-06-21T15:34:27 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 646,955,509 | 6,549 | alexey-milovidov | Note that @blinkov is just a robot and its comment does not make any sense. | 2020-06-20T07:08:01 | 2020-06-20T07:08:01 | {} | MEMBER |
ClickHouse | ClickHouse | 646,969,904 | 11,773 | jangorecki | Thank you @den-crane Option does work as expected now and I am able to complete query that was failing. | 2020-06-20T09:43:27 | 2020-06-20T09:43:27 | {} | NONE |
ClickHouse | ClickHouse | 647,046,740 | 11,806 | azat | @alexey-milovidov
>We don't have 20.5 prestable release yet
Yep, but AFAIU the release will be done from the 20.5 branch not from the master. Am I understand the release cycle correctly?
>and the checklist is not done: #11288
P.S. I was monitoring it and before writing comment checked the checklist, and s... | 2020-06-20T21:21:33 | 2020-06-20T21:21:33 | {} | MEMBER |
ClickHouse | ClickHouse | 647,060,345 | 10,362 | zlobober | By reading this topic, I understood follwing; please correct me if I am wrong:
1. max_memory_usage_for_all_queries is now obsolete;
2. max_memory_usage & max_memory_usage_for_user are still working;
3. max_server_memory_usage is now an intended way to prevent server from running out of memory;
4. ClickHouseMetrics_... | 2020-06-21T00:13:42 | 2020-06-21T00:17:57 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 647,110,952 | 11,774 | azat | >You can run built binary in Vagrant/Virtualbox/Qemu and test that it does not crash on trivial queries.
I'm not the fan of freebsd, but I completely forgot that I can just download the binary from CI, I guess this I can do | 2020-06-21T10:44:45 | 2020-06-21T10:44:45 | {} | MEMBER |
ClickHouse | ClickHouse | 647,116,468 | 10,242 | alexey-milovidov | The code does not have enough comments for methods and classes.
Example: the function `changeIO` can be documented... | 2020-06-21T11:39:00 | 2020-06-21T11:39:19 | {} | MEMBER |
ClickHouse | ClickHouse | 646,944,925 | 11,718 | alexey-milovidov | > Unit tests relwithdebuginfo gcc-9 — fail: 1, passed: 5184
Should investigate that. | 2020-06-20T05:33:28 | 2020-06-20T05:33:28 | {} | MEMBER |
ClickHouse | ClickHouse | 646,955,703 | 6,549 | alexey-milovidov | Yes, the issue is related to FREEZE partitions.
After you have created a backup, you cannot execute mutations, because backup is consisted with hard links, and mutations will alter backup that you don't expect.
Probably we need to introduce UNFREEZE command? | 2020-06-20T07:10:09 | 2020-06-20T07:10:09 | {} | MEMBER |
ClickHouse | ClickHouse | 647,034,265 | 11,834 | alexey-milovidov | > Docs check — Found 1 errors
Failed on master, @blinkov | 2020-06-20T19:06:01 | 2020-06-20T19:06:12 | {} | MEMBER |
ClickHouse | ClickHouse | 647,046,331 | 11,827 | alexey-milovidov | > 01305_replica_create_drop_zookeeper
Flaky test. | 2020-06-20T21:17:00 | 2020-06-20T21:17:00 | {} | MEMBER |
ClickHouse | ClickHouse | 647,110,090 | 11,774 | azat | >BTW, do you have setups to test it?
No, I started looking into this just because noticed some issues in generated headers for jemalloc | 2020-06-21T10:35:10 | 2020-06-21T10:35:10 | {} | MEMBER |
ClickHouse | ClickHouse | 647,110,230 | 11,590 | alexey-milovidov | > That's true, but you need to pass size of the object to the allocator, which is not done right now (AFAICS)
Our new_delete.cpp does it for jemalloc. | 2020-06-21T10:36:30 | 2020-06-21T10:36:30 | {} | MEMBER |
ClickHouse | ClickHouse | 647,110,369 | 11,774 | azat | @alexey-milovidov BTW it is disabled by default for now, but can be manually enabled (while before you cannot even enable it), since I did not checked the result
Once someone will report that it works, it can be enabled by default | 2020-06-21T10:37:54 | 2020-06-21T10:37:54 | {} | MEMBER |
ClickHouse | ClickHouse | 647,114,900 | 11,753 | alexey-milovidov | Could you please merge with master, so tests will be run again? | 2020-06-21T11:23:09 | 2020-06-21T11:23:09 | {} | MEMBER |
ClickHouse | ClickHouse | 647,145,479 | 11,669 | alexey-milovidov | I'm not sure how admin is supposed to choose the value of this setting. | 2020-06-21T15:51:13 | 2020-06-21T15:51:13 | {} | MEMBER |
ClickHouse | ClickHouse | 647,047,586 | 11,645 | azat | >@nikitamikhaylov @azat It does not work: https://clickhouse-test-reports.s3.yandex.net/11829/7b2b726c177b5beff53b53bcb0a1df487a81f326/integration_tests_(asan).html
@alexey-milovidov Yeah, this is due to config reload (at start server *always* reload the configuration regardless it was changed or not, due to substit... | 2020-06-20T21:30:41 | 2020-06-20T21:30:41 | {} | MEMBER |
ClickHouse | ClickHouse | 647,107,260 | 11,590 | alexey-milovidov | > As the last steps:
We need to make a test case that will show when tcmalloc is worse than jemalloc and send it to Google as an issue. | 2020-06-21T10:06:00 | 2020-06-21T10:06:00 | {} | MEMBER |
ClickHouse | ClickHouse | 647,110,516 | 11,590 | azat | >Our new_delete.cpp does it for jemalloc.
Yep (actually `deleteSized` in `memory.h`) | 2020-06-21T10:39:38 | 2020-06-21T10:39:38 | {} | MEMBER |
ClickHouse | ClickHouse | 647,113,490 | 11,715 | alexey-milovidov | > P.S. Looks like settings can be converted between SettingUInt64 and
SettingBool without breaking binary protocol?
Yes. | 2020-06-21T11:09:36 | 2020-06-21T11:09:36 | {} | MEMBER |
ClickHouse | ClickHouse | 647,114,375 | 11,334 | alexey-milovidov | @nikitamikhaylov
> This feature has unexpected results: (with big epsilon functions with anon_ prefix must be equal to ordinary function without prefix)
Or with small epsilon? | 2020-06-21T11:18:01 | 2020-06-21T11:18:01 | {} | MEMBER |
ClickHouse | ClickHouse | 647,134,709 | 11,153 | HooliBox | I used the same query and no new data was inserted.
I could create a new issue. | 2020-06-21T14:21:28 | 2020-06-21T14:21:28 | {} | NONE |
ClickHouse | ClickHouse | 647,147,914 | 11,839 | alexey-milovidov | > Build failed without build log. It's inner CI problem. Maybe something wrong with docker or host. Just wait for build restart. | 2020-06-21T16:11:28 | 2020-06-21T16:11:28 | {} | MEMBER |
ClickHouse | ClickHouse | 646,955,787 | 6,549 | alexey-milovidov | It is working with replicated tables, because FREEZE and backup creation has happened only on a single replica and another replica can execute mutation freely (and then the data will be replicated to all other replicas). | 2020-06-20T07:11:00 | 2020-06-20T07:11:47 | {} | MEMBER |
ClickHouse | ClickHouse | 646,961,786 | 11,300 | alexey-milovidov | The reason why tests failed is that you have changed something that we don't expect.
Maybe there is some regression in stack trace formatting speed? | 2020-06-20T08:15:52 | 2020-06-20T08:15:52 | {} | MEMBER |
ClickHouse | ClickHouse | 647,102,290 | 11,590 | azat | >Maybe it is just slower on middle-sized allocations (from a few KB to a few MB)?
@alexey-milovidov AFAICS the problem is not in the middle-sized but in allocations >~256K (depends on the number of classes for fast path, this depends on the size of pages is used in tcmalloc, tried 8k/256k not helps a lot), and it do... | 2020-06-21T09:16:31 | 2020-06-21T09:16:31 | {} | MEMBER |
ClickHouse | ClickHouse | 647,130,429 | 11,669 | azat | >ClickHouse build check — 11/17 builds are OK
Some CI issues
The only difference between b8ee2ea (everything is green) and bd45592 (this build fail) is the integration test anyway | 2020-06-21T13:44:44 | 2020-06-21T13:44:44 | {} | MEMBER |
ClickHouse | ClickHouse | 647,143,897 | 11,841 | HooliBox | ```
SELECT
name,
value
FROM system.settings
WHERE name LIKE '%mem%'
┌─name───────────────────────────────────────┬─value───────┐
│ distributed_aggregation_memory_efficient │ 0 │
│ aggregation_memory_efficient_merge_threads │ 0 │
│ memory_tracker_fault_probability │ 0... | 2020-06-21T15:38:06 | 2020-06-21T15:38:06 | {} | NONE |
ClickHouse | ClickHouse | 647,147,798 | 11,841 | HooliBox | thanks. with this setting max_bytes_before_external_group_by, the issue seems to be gone.
Do you mean I should use daily partition? but on a daily basis, there are actually not that much of data. | 2020-06-21T16:10:42 | 2020-06-21T16:10:42 | {} | NONE |
ClickHouse | ClickHouse | 646,955,818 | 6,549 | alexey-milovidov | > After starting using zookeeper I don't have that problem.
Ok. | 2020-06-20T07:11:13 | 2020-06-20T07:11:13 | {} | MEMBER |
ClickHouse | ClickHouse | 646,956,090 | 11,819 | alexey-milovidov | Fixed in https://github.com/ClickHouse/ClickHouse/pull/11746 | 2020-06-20T07:13:57 | 2020-06-20T07:13:57 | {} | MEMBER |
ClickHouse | ClickHouse | 646,958,490 | 6,549 | thyn | The problem was appearing on second or third mutation run and with tables about 300G size. I've checked small tables there was no any problem | 2020-06-20T07:39:33 | 2020-06-20T07:39:33 | {} | NONE |
ClickHouse | ClickHouse | 646,988,505 | 11,827 | alexey-milovidov | This should revert #10929 and #11246. | 2020-06-20T12:29:44 | 2020-06-20T12:29:44 | {} | MEMBER |
ClickHouse | ClickHouse | 647,108,395 | 11,590 | alexey-milovidov | BTW it's possible that Linux kernel in our CI does not support "restartable sequences". | 2020-06-21T10:17:31 | 2020-06-21T10:17:31 | {} | MEMBER |
ClickHouse | ClickHouse | 647,110,730 | 11,774 | alexey-milovidov | You can run built binary in Vagrant/Virtualbox/Qemu and test that it does not crash on trivial queries. | 2020-06-21T10:41:58 | 2020-06-21T10:41:58 | {} | MEMBER |
ClickHouse | ClickHouse | 647,113,214 | 11,830 | alexey-milovidov | > Integration tests (asan) — Some exception occurred :(
Stress test (memory) — Some exception occurred :(
S3UploadFailedError: Failed to upload /place/sandbox-data/tasks/9/3/705650539/test_output/stderr.log to clickhouse-test-reports/11830/2ff7f3496812479c1511ae29bd0b65ec53641e93/stress_test_logs/test_output/stde... | 2020-06-21T11:06:47 | 2020-06-21T11:06:47 | {} | MEMBER |
ClickHouse | ClickHouse | 647,046,238 | 11,288 | alexey-milovidov | #11806 must be included in 20.5. | 2020-06-20T21:16:02 | 2020-06-20T21:16:02 | {} | MEMBER |
ClickHouse | ClickHouse | 647,096,760 | 11,832 | bohutang | ACK | 2020-06-21T08:25:41 | 2020-06-21T08:25:41 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 647,108,846 | 11,590 | alexey-milovidov | > tcmalloc RSEQ (restartable sequences) requires linux 4.18+
We can print Linux kernel version at server startup.
> doc states that it depends from THP
THP was working extremely bad at least a few years ago, so all the reccomendations state that you should disable them (otherwise you get memory fragmentation w... | 2020-06-21T10:22:37 | 2020-06-21T10:22:56 | {} | MEMBER |
ClickHouse | ClickHouse | 647,139,013 | 11,841 | den-crane | Can you provide
1. linux kernel version `uname -a`
2. `nproc --all`
3. `show create table minute_bar`
4. `cat /proc/sys/vm/overcommit_memory`
5. A data sample of `minute_bar` would be nice to have to make reproducible example. | 2020-06-21T14:57:05 | 2020-06-21T15:00:28 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 647,145,208 | 11,841 | den-crane | as workaround try to change `max_bytes_before_external_group_by='6G'`
>I thought it was the issue that it didn't release memory to os after the query is done.
CH never releases immediately allocated memory because of jemalloc tries to save CPU on allocations/deallocations. | 2020-06-21T15:49:18 | 2020-06-21T16:09:22 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 646,956,343 | 6,549 | thyn | If so, unfreeze command would be very helpfull | 2020-06-20T07:16:35 | 2020-06-20T07:16:35 | {} | NONE |
ClickHouse | ClickHouse | 646,956,399 | 6,549 | const-shi | > It is working with replicated tables, because FREEZE and backup creation has happened only on a single replica and another replica can execute mutation freely (and then the data will be replicated to all other replicas).
data will be replicated to all other replicas and alter backup at this replicas? | 2020-06-20T07:17:14 | 2020-06-20T07:17:14 | {} | NONE |
ClickHouse | ClickHouse | 647,109,937 | 11,590 | azat | >THP was working extremely bad at least a few years ago, so all the reccomendations states that you should disable them (otherwise you get memory fragmentation when program runs for significantly long time).
I'm seeing this everywhere without any details, but not sure that it is true for recent kernels (I saw some s... | 2020-06-21T10:33:35 | 2020-06-21T10:33:35 | {} | MEMBER |
ClickHouse | ClickHouse | 647,123,177 | 11,153 | azat | >I got the same problem and setting max_server_memory_usage_to_ram_ratio does not fix the problem. Anyone have a solution yet?
@HooliBox which version do you have? and please provide startup logs (at least `Setting max_server_memory_usage ...` messages) | 2020-06-21T12:42:41 | 2020-06-21T12:42:41 | {} | MEMBER |
ClickHouse | ClickHouse | 647,140,241 | 11,841 | HooliBox | 1. linux kernel
`Linux clickhouse 5.4.34-1-pve #1 SMP PVE 5.4.34-2 (Thu, 07 May 2020 10:02:02 +0200) x86_64 x86_64 x86_64 GNU/Linux`
2. nproc --all `64`
3. show create table minute_bar
```
CREATE TABLE quant.minute_bar
(
`trade_datetime` DateTime,
`s_info_windcode` String,
`open` Float64,
`hig... | 2020-06-21T15:06:37 | 2020-06-21T15:06:37 | {} | NONE |
ClickHouse | ClickHouse | 646,983,083 | 11,826 | alexey-milovidov | No tests will be provided for this feature. | 2020-06-20T11:30:27 | 2020-06-20T11:30:27 | {} | MEMBER |
ClickHouse | ClickHouse | 647,046,695 | 11,828 | alexey-milovidov | > test_version_update_after_mutation/test.py::test_mutate_and_upgrade
- flaky test. | 2020-06-20T21:21:05 | 2020-06-20T21:21:05 | {} | MEMBER |
ClickHouse | ClickHouse | 647,114,680 | 11,753 | alexey-milovidov | The fuzz test just found a new issue that is unrelated to your changes. Don't worry. | 2020-06-21T11:21:03 | 2020-06-21T11:21:03 | {} | MEMBER |
ClickHouse | ClickHouse | 647,001,592 | 11,810 | zlobober | Closing this in favor of https://github.com/ClickHouse/ClickHouse/pull/11830 | 2020-06-20T14:17:50 | 2020-06-20T14:17:50 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 647,027,781 | 11,806 | azat | @alexey-milovidov @4ertus2 I just noticed 20.5 branch, looks like 20.5 should contains this fix?
P.S. maybe I wrote bad description, but the following query fails without this fix:
```sql
SELECT foo FROM (SELECT anyLast(number) AS foo FROM numbers(1) GROUP BY number);
Code: 47. DB::Exception: Received from loca... | 2020-06-20T18:03:07 | 2020-06-20T18:03:07 | {} | MEMBER |
ClickHouse | ClickHouse | 647,046,153 | 11,806 | alexey-milovidov | @azat Thank you for the fix!
We don't have 20.5 prestable release yet and the checklist is not done: https://github.com/ClickHouse/ClickHouse/issues/11288
Yes, this bugfix must be included in 20.5. | 2020-06-20T21:15:03 | 2020-06-20T21:15:13 | {} | MEMBER |
ClickHouse | ClickHouse | 647,046,566 | 11,827 | alexey-milovidov | > Performance — 1 errors
I have no idea what is the error. It is not highlighted in test result. | 2020-06-20T21:19:38 | 2020-06-20T21:19:38 | {} | MEMBER |
ClickHouse | ClickHouse | 647,118,378 | 10,242 | alexey-milovidov | No meaningful message is produced if there is no user.
https://youtu.be/Y7yhgjESF8s?t=8312 | 2020-06-21T11:58:14 | 2020-06-21T11:58:23 | {} | MEMBER |
ClickHouse | ClickHouse | 647,122,864 | 11,715 | azat | >BTW, what is the setup when nested Distributed tables are required?
I was going to add some block about use case into the documentation (since there can be some interesting use cases and also caveats), but did not manage to find time for this, egh.
Anyway let me try to explain it briefly here:
Suppose you hav... | 2020-06-21T12:40:04 | 2020-06-21T12:40:04 | {} | MEMBER |
ClickHouse | ClickHouse | 647,125,058 | 11,153 | HooliBox | Here is what I did, I ran the same query 10 times. The first couple time, the query works fine. Then, it stopped working and complaining about the memory.
```python
from sqlalchemy import create_engine, Column, MetaData, literal
from clickhouse_sqlalchemy import Table, make_session, get_declarative_base, types, engi... | 2020-06-21T12:59:12 | 2020-06-21T13:13:18 | {} | NONE |
ClickHouse | ClickHouse | 647,046,887 | 11,645 | alexey-milovidov | @nikitamikhaylov @azat It does not work: https://clickhouse-test-reports.s3.yandex.net/11829/7b2b726c177b5beff53b53bcb0a1df487a81f326/integration_tests_(asan).html | 2020-06-20T21:23:11 | 2020-06-20T21:23:11 | {} | MEMBER |
ClickHouse | ClickHouse | 647,119,090 | 11,153 | HooliBox | Hi,
I got the same problem and setting max_server_memory_usage_to_ram_ratio does not fix the problem. Anyone have a solution yet?
this is my error
```
ServerException: Code: 241.
DB::Exception: Memory limit (for query) exceeded: would use 18.63 GiB (attempt to allocate chunk of 4210950 bytes), maximum: 18.63 GiB: ... | 2020-06-21T12:04:46 | 2020-06-21T12:05:35 | {} | NONE |
ClickHouse | ClickHouse | 647,129,970 | 11,153 | azat | Ok, here are some questions to you:
- can you post into gist.github.com log for one query when it does not fails
- can you try using http interface instead of tcp? (there was issues that was in tcp interface, that has been fixed in #11022, http does not have it). AFAICS clickhouse_sqlalchemy allows HTTP interface too... | 2020-06-21T13:40:36 | 2020-06-21T13:46:29 | {} | MEMBER |
ClickHouse | ClickHouse | 647,146,185 | 11,841 | den-crane | >PARTITION BY toYYYYMMDD(trade_datetime)
>ORDER BY (trade_datetime, s_info_windcode)
>WHERE toInt64(toYYYYMMDDhhmmss(trade_datetime)) - toInt64(toYYYYMMDD(trade_datetime)) * 1000000 > 94500
> AND toInt64(toYYYYMMDDhhmmss(trade_datetime)) - toInt64(toYYYYMMDD(trade_datetime)) * 1000000 < 141500
BTW, Table/query mi... | 2020-06-21T15:57:00 | 2020-06-21T16:01:34 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 646,956,583 | 6,549 | alexey-milovidov | @const-shi
No, it will be replicated as new files and backup will not be modified.
And it cannot - because we have made FREEZE :) | 2020-06-20T07:19:19 | 2020-06-20T07:19:19 | {
"+1": 1
} | MEMBER |
ClickHouse | ClickHouse | 646,997,779 | 11,816 | alexey-milovidov | > 00652_replicated_mutations_zookeeper
Reported flaky test. | 2020-06-20T13:51:40 | 2020-06-20T13:51:40 | {} | MEMBER |
ClickHouse | ClickHouse | 647,023,951 | 11,829 | alexey-milovidov | > ClickHouse build check — 20/21 builds are OK
An error of infamous "Sandbox" infrastructure. | 2020-06-20T17:26:13 | 2020-06-20T17:26:13 | {} | MEMBER |
ClickHouse | ClickHouse | 647,029,708 | 11,826 | alexey-milovidov | #11834 | 2020-06-20T18:22:27 | 2020-06-20T18:22:27 | {} | MEMBER |
ClickHouse | ClickHouse | 647,033,151 | 11,825 | alexey-milovidov | > Unit tests MSAN clang-10
Flaky test due to ZooKeeper connection. | 2020-06-20T18:55:03 | 2020-06-20T18:55:03 | {} | MEMBER |
ClickHouse | ClickHouse | 647,046,961 | 11,829 | alexey-milovidov | > Integration tests (asan) — failed:0, passed:676, error:1
The failure was introduced in #11645 | 2020-06-20T21:23:58 | 2020-06-20T21:23:58 | {} | MEMBER |
ClickHouse | ClickHouse | 647,123,960 | 11,153 | azat | @HooliBox but in your case the query limit exceeded, not total (like in issue subject):
>DB::Exception: Memory limit (for query) exceeded: would use 18.63 GiB (attempt to allocate chunk of 4210950 bytes), maximum: 18.63 GiB: While executing MergeTreeThread. Stack trace:
So take a look at `max_memory_usage*` setti... | 2020-06-21T12:49:16 | 2020-06-21T12:49:16 | {} | MEMBER |
ClickHouse | ClickHouse | 647,139,561 | 10,520 | alexey-milovidov | @qoega has failed integrating this PR.
I will do. | 2020-06-21T15:01:13 | 2020-06-21T15:01:13 | {} | MEMBER |
ClickHouse | ClickHouse | 647,024,055 | 11,827 | alexey-milovidov | > Docs check — Found 1 errors
Broken in master. | 2020-06-20T17:27:04 | 2020-06-20T17:27:04 | {} | MEMBER |
ClickHouse | ClickHouse | 647,029,330 | 11,582 | alexey-milovidov | It did not help. | 2020-06-20T18:18:26 | 2020-06-20T18:18:26 | {} | MEMBER |
ClickHouse | ClickHouse | 647,113,765 | 11,715 | alexey-milovidov | BTW, what is the setup when nested Distributed tables are required?
(the question from ClickHouse Meetup) | 2020-06-21T11:12:19 | 2020-06-21T11:12:29 | {} | MEMBER |
ClickHouse | ClickHouse | 647,115,715 | 10,242 | alexey-milovidov | Notes about what to do after merge:
StartUp -> Startup
MySQLWrite -> MySQLWire
PostgreSQLWrite -> PostgreSQLWire
| 2020-06-21T11:31:17 | 2020-06-21T11:31:17 | {} | MEMBER |
ClickHouse | ClickHouse | 646,959,546 | 11,300 | alexey-milovidov | `00816_long_concurrent_alter_column`
has failed in this PR but it does not fail in master.
| 2020-06-20T07:51:51 | 2020-06-20T07:51:51 | {} | MEMBER |
ClickHouse | ClickHouse | 647,009,742 | 11,214 | alexey-milovidov | @tavplubix It is incomplete:
```
CREATE TEMPORARY TABLE test (x INT UNSIGNED NOT NULL)
```
does not work.
| 2020-06-20T15:27:53 | 2020-06-20T15:27:53 | {} | MEMBER |
ClickHouse | ClickHouse | 647,025,503 | 10,047 | alexey-milovidov | It does not work at all:
```
SELECT max(a + b)
FROM
(
SELECT
1 AS a,
2 AS b
)
Received exception from server (version 20.5.1):
Code: 42. DB::Exception: Received from localhost:9000. DB::Exception: Number of arguments for function plus doesn't match: passed 0, should be 2.
``` | 2020-06-20T17:41:06 | 2020-06-20T17:41:06 | {} | MEMBER |
ClickHouse | ClickHouse | 647,101,557 | 11,774 | azat | @alexey-milovidov any interest in this?
I'm expecting that after the last patch the build will succeed, I can rebase (+ maybe add more comments) | 2020-06-21T09:09:03 | 2020-06-21T09:09:03 | {} | MEMBER |
ClickHouse | ClickHouse | 647,110,179 | 11,774 | azat | >ClickHouse build check — 10/18 builds are OK
Some CI internal issues, build on freebsd passed, so everything should work, but requires last rebase | 2020-06-21T10:36:06 | 2020-06-21T10:36:06 | {} | MEMBER |
ClickHouse | ClickHouse | 647,133,978 | 11,153 | azat | From your logs:
>2020.06.21 14:00:53.762580 [ 859 ] {} <Debug> MemoryTracker: Peak memory usage (for query): 6.15 GiB.
And the error when the query fails:
>DB::Exception: Memory limit (for query) exceeded: would use 18.63 GiB (attempt to allocate chunk of 4210950 bytes), maximum: 18.63 GiB: While executing MergeTr... | 2020-06-21T14:15:00 | 2020-06-21T14:15:00 | {} | MEMBER |
ClickHouse | ClickHouse | 647,143,676 | 11,840 | alexey-milovidov | > Docs check — Found 1 errors
Broken in master @blinkov | 2020-06-21T15:36:06 | 2020-06-21T15:36:06 | {} | MEMBER |
ClickHouse | ClickHouse | 646,945,173 | 11,736 | alexey-milovidov | It failed under UBSan:
```
[ RUN ] Common.PODPushBackRawMany
../src/Common/PODArray.h:241:13: runtime error: applying non-zero offset 5 to null pointer
SUMMARY: UndefinedBehaviorSanitizer: undefined-behavior ../src/Common/PODArray.h:241:13
``` | 2020-06-20T05:35:30 | 2020-06-20T05:35:30 | {} | MEMBER |
ClickHouse | ClickHouse | 647,113,224 | 11,806 | azat | >@azat I have no idea why we have 20.5 branch - it should not exist.
@alexey-milovidov will it be reseted to upstream/master? | 2020-06-21T11:06:53 | 2020-06-21T11:06:53 | {} | MEMBER |
ClickHouse | ClickHouse | 634,634,660 | 11,088 | akuzm | Oops, I accidentally added the comments the second time. I'll fix it. | 2020-05-27T12:43:42 | 2020-05-27T12:43:42 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 634,696,248 | 11,217 | WiPiton | This query fails:
```
select `tbd_date`.`quarter` as `c0`, `tbd_ak`.`code` as `c1`, `tbd_date`.`decade` as `c2`, `tbd_route_d`.`id` as `c3`
from
`default`.`tbd_date` as `tbd_date`,
`default`.`tbf_dwh_tickets01` as `tbf_dwh_tickets01`,
`default`.`tbd_ak` as `tbd_ak`,
`default`.`tbd_route_d` as `tbd_route_d`
wh... | 2020-05-27T14:25:22 | 2020-05-27T14:25:55 | {} | NONE |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.