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 | 521,723,082 | 6,430 | josepowera | Sorry for misunderstanding... Yes - we used chain of MVs... That was the only was it used to work then in old versions. And that's the reason I linked to #2174 and #4587 in my initial question (without specificying good reason for links).
Now I have seen your PR for test, and we will update to use straight KAFKA->MV... | 2019-08-15T17:19:48 | 2019-08-15T17:20:13 | {} | NONE |
ClickHouse | ClickHouse | 521,957,840 | 6,515 | akuzm | Could you check whether this happens on the latest stable release `v19.13.2.19-stable`? If it does, it would be helpful to have the table definition (`show create table`) and some sample data, so that we can try to reproduce this locally. | 2019-08-16T10:02:50 | 2019-08-16T10:02:50 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 522,341,011 | 6,532 | kreuzerkrieg | Good to know and thanks for prompt reply.
Could you please elaborate about manual traffic configuration? Like, why and when?
Thanks!
| 2019-08-18T17:41:31 | 2019-08-18T17:41:31 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 521,371,926 | 6,448 | alexey-milovidov | More info: https://randomascii.wordpress.com/2012/06/26/doubles-are-not-floats-so-dont-compare-them/ | 2019-08-14T18:51:17 | 2019-08-14T18:51:17 | {} | MEMBER |
ClickHouse | ClickHouse | 521,459,401 | 3,174 | den-crane | > So, the same primary key could be in different parts forever? Because it is at least in one part that has reached its max size?
yes | 2019-08-14T23:42:24 | 2019-08-14T23:42:24 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 521,464,433 | 6,501 | den-crane | Then now the last/current stable will be 18.14.19. Very funny. | 2019-08-15T00:09:08 | 2019-08-15T00:11:22 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 521,658,616 | 6,503 | lamberken | > Try to set in JDBC!!! socket_timeout = 300000
Thanks for your solution, it works 👍
associated config `http_send_timeout` and `http_receive_timeout` | 2019-08-15T14:20:55 | 2019-08-15T14:23:17 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 521,660,919 | 6,501 | iameugenejo | Thank you for more helpful comment, @alex-zaitsev. It’s helpful that you linked the Altinity blog for the reference. I’ll keep an eye on that blog. | 2019-08-15T14:27:21 | 2019-08-15T14:27:21 | {} | NONE |
ClickHouse | ClickHouse | 521,711,961 | 6,445 | alex-zaitsev | Do you control your collector? | 2019-08-15T16:46:53 | 2019-08-15T16:46:53 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 521,824,253 | 6,479 | alexey-milovidov | Approved. You only have to change submodule address to either ClickHouse_Extras or upstream. | 2019-08-15T22:49:27 | 2019-08-15T22:49:27 | {} | MEMBER |
ClickHouse | ClickHouse | 521,943,397 | 6,349 | ramazanpolat | What is `localhost replica`? | 2019-08-16T09:10:04 | 2019-08-16T09:10:04 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 522,277,835 | 6,363 | alexey-milovidov | I forgot that ProfileEvents must be differentiated (the same as in GraphiteSender) | 2019-08-17T23:24:11 | 2019-08-17T23:24:11 | {} | MEMBER |
ClickHouse | ClickHouse | 522,278,522 | 6,363 | alexey-milovidov | BTW, your implementation is totally and absolutely wrong.
It calculates metrics on flush time instead to do it each second. | 2019-08-17T23:41:29 | 2019-08-17T23:41:29 | {} | MEMBER |
ClickHouse | ClickHouse | 522,339,950 | 6,532 | alexey-milovidov | Parts fetching done in parallel with a maximum of "background_pool_size" connections.
Usually it saturates network 100% (at least for 10GBit) and cannot be made any faster.
It may have negative consequences though - sometimes you have to shape replication traffic manually. | 2019-08-18T17:27:06 | 2019-08-18T17:27:06 | {} | MEMBER |
ClickHouse | ClickHouse | 521,505,240 | 6,503 | den-crane | So you've suspect an issue in clickhouse-jdbc then?
Check CH logs for errors. | 2019-08-15T04:01:01 | 2019-08-15T04:01:01 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 521,874,997 | 6,446 | YiuRULE | Ok weird, thanks ! I will take a look | 2019-08-16T03:54:49 | 2019-08-16T03:54:49 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 522,131,577 | 6,443 | abyss7 | What is CH version? | 2019-08-16T19:53:17 | 2019-08-16T19:53:17 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 522,396,178 | 6,425 | vzakaznikov | @KochetovNicolai, here is another update as sometimes I still could see clickhouse-client processes left over. I think the fundamental issue is that clickhouse-client process
ignores SIGCHLD. Do you know why?
Also the tests do not attempt to exit clickhouse-client cleanly.
```
$ git diff
diff --git a/dbms/tests... | 2019-08-19T03:16:25 | 2019-08-19T03:16:25 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 521,507,715 | 6,501 | den-crane | >Solution 1: Don't label every builds that pass the tests to be stable release.
>Battle test it in production for at least a month and mark it stable.
Nobody will install test version in production. It's nonsense.
>Solution 2: Remove stable label when any critical issue is identified with the version.
Most ... | 2019-08-15T04:17:59 | 2019-08-15T04:17:59 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 521,638,932 | 6,503 | den-crane | Try to set in JDBC!!! socket_timeout = 300000 | 2019-08-15T13:23:29 | 2019-08-15T13:23:29 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 521,947,028 | 6,446 | YiuRULE | @alexey-milovidov How is exactly the configuration you used for your test ?
I tried both on this branch and on the latest `master` branch, with building clickhouse locally inside a `build` folder, with the default configuration we can find at `dbms/programs/server/`, and I tried the test locally.
```
for i in {1... | 2019-08-16T09:22:59 | 2019-08-16T09:22:59 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 522,159,562 | 6,519 | simPod | I use iTerm2 with zsh and ohmyzsh. `echo $COLUMNS` says 238 on my 13" mac and 272 on my 15" mac. If that's issue only on my PC just feel free to close this, I thought it might be more general issue. | 2019-08-16T21:38:02 | 2019-08-16T21:38:02 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 522,278,912 | 6,363 | nikitamikhaylov | Ok. And I think we should create Block objects every second and somehow flush them. Sorry for UInt, It is just copypaste. Сolumns for current metrics are created with the correct type | 2019-08-17T23:53:16 | 2019-08-17T23:58:55 | {} | MEMBER |
ClickHouse | ClickHouse | 522,279,172 | 6,363 | alexey-milovidov | Ok, I'm fixing it right now... | 2019-08-17T23:59:48 | 2019-08-17T23:59:48 | {} | MEMBER |
ClickHouse | ClickHouse | 522,320,645 | 6,531 | alexey-milovidov | You can specify exact version in `apt-get install`:
```
sudo apt-get install clickhouse-server=19.13.1.11 clickhouse-client=19.13.1.11 clickhouse-common-static=19.13.1.11
```
You can also download packages directly from
https://repo.yandex.ru/clickhouse/deb/stable/main/
and install them with `sudo dpkg -i`. | 2019-08-18T13:07:12 | 2019-08-18T13:07:12 | {} | MEMBER |
ClickHouse | ClickHouse | 522,348,990 | 6,425 | KochetovNicolai | Hi, @vzakaznikov
Now I'm just trying to rewrite tests because for some reasons most of them fails under tsan build (it's not tsan issue, but uexpect seem to hang sometimes). I think it wouldn't be very difficult.
After that I will try to merge pr just to avoid further conflicts with master. And after we could gradua... | 2019-08-18T19:35:24 | 2019-08-18T19:35:24 | {} | MEMBER |
ClickHouse | ClickHouse | 521,462,895 | 3,174 | den-crane | > So, we have to avoid that at any cost, right? Because the summing, aggregating, replacing trees and sparse indexes would be somehow broken?
Nothing is broken. The opposite.
Merges are eventual.
At any moment your queries should expect that data in the table is in not final state.
There is no difference that ... | 2019-08-15T00:01:29 | 2019-08-15T00:03:55 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 521,479,584 | 6,486 | Vxider | build fails when trying to use SettingsChanges::changes() due to ctor deduction failure | 2019-08-15T01:26:18 | 2019-08-15T01:26:18 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 521,610,730 | 6,489 | alesapin | Not done:
1) `ALTER TABLE ... MOVE PARTITION`
2) `ALTER ... MOVE` doesn't check any information about merges and mutations. Races can occur.
3) **Background move for ReplicatedMergeTree**. Most important part of this feature.
4) It seems like there is no information about movements in `part_log`
5) Interesting mes... | 2019-08-15T11:32:02 | 2019-08-15T11:39:08 | {} | MEMBER |
ClickHouse | ClickHouse | 522,112,513 | 6,525 | PlatonB | Спасибо, работает. Но только `sudo apt-key adv --keyserver hkp://keyserver.ubuntu.com:80 --recv E0C56BD4 # optional` - не "optional". | 2019-08-16T18:45:27 | 2019-08-16T18:45:27 | {} | NONE |
ClickHouse | ClickHouse | 522,278,802 | 6,363 | alexey-milovidov | And you use UInt64 for CurrentMetrics - this is also wrong. | 2019-08-17T23:49:58 | 2019-08-17T23:49:58 | {} | MEMBER |
ClickHouse | ClickHouse | 522,324,501 | 6,528 | alexey-milovidov | > I'd imagine in the neighbour function the behavior would be that if the key for the neighbouring value was not the same as the current row key then it would return NULL.
Yes, great idea! | 2019-08-18T13:59:10 | 2019-08-18T13:59:10 | {} | MEMBER |
ClickHouse | ClickHouse | 522,327,118 | 6,299 | MahmoudGoda0 | Thank you very much!
2 questions please:
1- Now version 19.13.1.1088-testing including both requests "5042" & "6299" ? I'm asking this question as i see the enhancement done in "5042" effected negative when i used 1088-testing.
2- How I know that a specific version has been released to the stable version ? | 2019-08-18T14:33:40 | 2019-08-18T14:33:40 | {} | NONE |
ClickHouse | ClickHouse | 522,331,108 | 6,515 | fancyqlx | The memory usage became normal after I used FreeThreadPool in AsynchronousBlockInputStream instead of GlobalThreadPool. I guess this kind of thread pool may hold memory in an unexpected way. | 2019-08-18T15:26:38 | 2019-08-18T15:26:38 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 522,128,973 | 5,230 | 4ertus2 | Does the server crashes or this errors just exceptions and it's still running?
There was a crach in Decimal128 arrgeration fixed in recent versions #6378 It could be related if it craches. | 2019-08-16T19:43:40 | 2019-08-16T19:48:11 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 522,266,034 | 6,012 | abyss7 | > Probably, it is worth to add general settings class `RowOutputFormatParams` the same as `RowInputFormatParams`.
Don't see any reason to do it now - there is a single param. | 2019-08-17T19:55:28 | 2019-08-17T19:55:28 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 522,280,544 | 6,530 | alexey-milovidov | Now it looks Ok:
```
SELECT
event_date,
event_time,
milliseconds,
ProfileEvent_UserTimeMicroseconds,
ProfileEvent_SoftPageFaults
FROM system.metric_log
ORDER BY event_time ASC
┌─event_date─┬──────────event_time─┬─milliseconds─┬─ProfileEvent_UserTimeMicroseconds─┬─ProfileEvent_SoftPa... | 2019-08-18T00:32:50 | 2019-08-18T00:32:50 | {} | MEMBER |
ClickHouse | ClickHouse | 522,293,583 | 6,528 | zombiemonkey | This is a great idea. Can it also be added to runningDifference* functions and the new neighbour function? https://github.com/yandex/ClickHouse/pull/5925.
I'd imagine in the neighbour function the behavior would be that if the key for the neighbouring value was not the same as the current row key then it would retur... | 2019-08-18T05:53:57 | 2019-08-18T05:53:57 | {} | NONE |
ClickHouse | ClickHouse | 522,306,516 | 6,299 | MahmoudGoda0 | Hi Guys,
I think this is related to "Optimization of ORDER BY with respect to the ORDER key in MergeTree tables", I installed the new stable version (19.13.2.19) today, and i see very bad results comparing with version "19.13.1.961" for the below query:
select TOP 1000 ext_rec_num,xdr_id,xdr_grp,xdr_type,xdr_subtyp... | 2019-08-18T09:38:33 | 2019-08-18T09:38:33 | {} | NONE |
ClickHouse | ClickHouse | 522,430,711 | 5,568 | filimonov | Please check/share client configuration - it's in section:
```
<openSSL>
<client>
....
</client>
</openSSL>
```
in `/etc/clickhouse-server/config.xml` - for interserver connections (if used) and in `/etc/clickhouse-client/config.xml` - for connections from clickhouse-client. ... | 2019-08-19T06:26:23 | 2019-08-19T06:26:45 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 521,504,115 | 6,503 | lamberken | > Check errors in CH server log. It could a network issue.
Hi, the network should not have problems, because I found a regular that when I reduce the parallelism of writing threads, this problem does not occur.
| 2019-08-15T03:53:37 | 2019-08-15T03:53:37 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 521,832,064 | 6,467 | alexey-milovidov | Let's merge and wait for docker image to rebuild. | 2019-08-15T23:28:50 | 2019-08-15T23:28:50 | {} | MEMBER |
ClickHouse | ClickHouse | 521,953,584 | 6,446 | alexey-milovidov | clang, debug build.
It also reproduces in our CI checks in various build configurations, example: https://clickhouse-test-reports.s3.yandex.net/0/b1d1a23318758c74bb7a883bf39022db584474b5/functional_stateless_tests_(ubsan).html
| 2019-08-16T09:47:02 | 2019-08-16T09:47:02 | {} | MEMBER |
ClickHouse | ClickHouse | 522,322,098 | 6,299 | alexey-milovidov | Your testing version `19.13.1.961` had the optimization for ORDER BY, but it wasn't released in any stable version. After you installed 19.13.2.19-stable, you get lost this optimization.
Resolution: install the latest testing from https://repo.yandex.ru/clickhouse/deb/testing/main/
It is v19.13.1.1088-testing.
P... | 2019-08-18T13:27:33 | 2019-08-18T13:27:33 | {} | MEMBER |
ClickHouse | ClickHouse | 522,341,188 | 6,472 | alexey-milovidov | `Checksums of parts don't match ... We will download merged part from replica to force byte-identical result.`
No actions required. | 2019-08-18T17:44:15 | 2019-08-18T17:44:15 | {} | MEMBER |
ClickHouse | ClickHouse | 521,516,595 | 6,448 | kreuzerkrieg | > > If you are sure you are ok with == and != let it be
>
> Yes. We usually compare floats that was not affected by any floating point math. For example:
>
> ```
> float x = 0;
> ... maybe set x to some non-zero value
> if (x == 0)
> do something
> ```
>
> There is not much floating point math in Clic... | 2019-08-15T05:15:41 | 2019-08-15T05:15:41 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 521,591,624 | 6,361 | blinkov | @BayoNet ping + there are some failed doc checks | 2019-08-15T10:12:26 | 2019-08-15T10:12:26 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 522,132,314 | 5,230 | doom369 | Server up and running. Just random connection is dropped. | 2019-08-16T19:55:46 | 2019-08-16T19:55:46 | {} | NONE |
ClickHouse | ClickHouse | 522,134,753 | 5,230 | 4ertus2 | Does it occurs only with ```DB::DataTypeDecimal<DB::Decimal<__int128> >::serializeText``` or there're other data types that cannot be serialized?
Columns of what types are in group by, columns of what types are under other the aggregations? Do inserts wethout aggregations works? What mimimal query (with what aggregati... | 2019-08-16T20:04:32 | 2019-08-16T20:04:51 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 522,236,820 | 6,445 | gusevartemasd | > Do you control your collector?
ofc
and i'm sure that it send just one query to ODBC clickhouse driver | 2019-08-17T13:31:04 | 2019-08-17T13:31:04 | {} | NONE |
ClickHouse | ClickHouse | 522,341,810 | 6,532 | kreuzerkrieg | Oh, I see. Thanks again | 2019-08-18T17:53:48 | 2019-08-18T17:53:48 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 521,537,903 | 6,488 | astudnev | https://github.com/yandex/ClickHouse/issues/6506 | 2019-08-15T06:58:46 | 2019-08-15T06:58:46 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 521,950,713 | 6,489 | alesapin | No, better to place it in BackgroundProcessingPool as one more task. | 2019-08-16T09:36:19 | 2019-08-16T09:36:19 | {} | MEMBER |
ClickHouse | ClickHouse | 522,070,672 | 6,486 | alexey-milovidov | Thank you! | 2019-08-16T16:30:57 | 2019-08-16T16:30:57 | {} | MEMBER |
ClickHouse | ClickHouse | 522,227,475 | 5,230 | doom369 | If this is question to me - table is next:
```
CREATE TABLE reporting_raw_data (
device_id UInt32,
stream_id UInt32 DEFAULT 0,
pin UInt8,
pin_type UInt8,
ts DateTime,
value Float64
)
ENGINE = MergeTree()
PARTITION BY toYYYYMM(ts)
ORDER BY (device_id, stream_id, pin, pin_type, ts);
```
With... | 2019-08-17T11:01:13 | 2019-08-17T11:01:13 | {} | NONE |
ClickHouse | ClickHouse | 522,378,130 | 6,425 | vzakaznikov | @KochetovNicolai, I've reproduced the hang issue in the uexpect.py. I will try to fix it. | 2019-08-19T01:15:49 | 2019-08-19T01:15:49 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 521,454,732 | 3,174 | simper66 | So, the same primary key could be in different parts forever? Because it is at least in one part that has reached its max size? | 2019-08-14T23:19:02 | 2019-08-14T23:19:02 | {} | NONE |
ClickHouse | ClickHouse | 521,505,717 | 6,501 | iameugenejo | Can you elaborate on that? Why do you think everything will be worse with any of the suggestions? And I would love to know if you have any better solution because I’m having quite a trouble with current situation. | 2019-08-15T04:04:14 | 2019-08-15T04:04:14 | {} | NONE |
ClickHouse | ClickHouse | 521,982,622 | 6,479 | vitlibar | > Approved. You only have to change submodule address to either ClickHouse_Extras or upstream.
Yes, I know, I'm going to do it. | 2019-08-16T11:49:04 | 2019-08-16T11:49:14 | {} | MEMBER |
ClickHouse | ClickHouse | 522,065,245 | 6,519 | alexey-milovidov | What terminal emulator do you use?
Possible reason: the terminal emulator reports wrong window size. | 2019-08-16T16:13:51 | 2019-08-16T16:13:51 | {} | MEMBER |
ClickHouse | ClickHouse | 522,276,261 | 6,500 | alexey-milovidov | FYI. When I tried to use `link_libraries`, CMake placed these libraries in two places: at the front of all other libraries and at the end. Example:
`glibc_compatibility.a libc.so other libraries glibc_compatibility.a libc.so`
And it doesn't work properly. | 2019-08-17T22:50:45 | 2019-08-17T22:50:45 | {} | MEMBER |
ClickHouse | ClickHouse | 522,321,797 | 6,299 | alexey-milovidov | No, it should not be included in any stable version yet. | 2019-08-18T13:23:44 | 2019-08-18T13:23:44 | {} | MEMBER |
ClickHouse | ClickHouse | 521,495,888 | 6,487 | zhang2014 | I made some changes and put it in my repo(https://github.com/zhang2014/ClickHouse/tree/zh-toc), maybe you can sync it. | 2019-08-15T02:59:42 | 2019-08-15T02:59:42 | {
"+1": 1
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 521,551,505 | 6,501 | alex-zaitsev | Originally, Yandex marked release 'stable' as soon as it is installed for Yandex.Metrica. Now they wait a bit for community feedback first I guess. In general, it makes sense to use the latest minor version from the previous major one. E.g., if latest release is 19.13.2 now, the most tested and bugfixed is 19.11.7 (19.... | 2019-08-15T07:54:59 | 2019-08-15T07:54:59 | {
"+1": 1
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 521,662,751 | 5,075 | MeteHanC | It's still the same error message in 19.13.2.19 Official Release. | 2019-08-15T14:32:29 | 2019-08-15T14:32:29 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 522,065,562 | 6,519 | alexey-milovidov | Also please run
`echo $COLUMNS`
in bash and verify that the value is correct. | 2019-08-16T16:14:51 | 2019-08-16T16:14:51 | {} | MEMBER |
ClickHouse | ClickHouse | 522,264,455 | 6,513 | alexey-milovidov | Test is pending in https://github.com/yandex/ClickHouse/pull/6514
but it only revealed more issues. | 2019-08-17T19:30:46 | 2019-08-17T19:30:46 | {} | MEMBER |
ClickHouse | ClickHouse | 522,322,939 | 6,531 | MahmoudGoda0 | Thanks alot! | 2019-08-18T13:38:20 | 2019-08-18T13:38:20 | {} | NONE |
ClickHouse | ClickHouse | 522,352,614 | 6,425 | vzakaznikov | Hi @KochetovNicolai. I will try building with clang and thread sanitizer enabled to see if I can reproduce the issue. Hopefully, I can fix it so that we don't have to rewrite all the tests. | 2019-08-18T20:26:23 | 2019-08-18T20:26:23 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 522,392,589 | 6,425 | vzakaznikov | @KochetovNicolai, here is an update to my diff to resolve an issue with clickhouse-client processes left over after test execution.
```
$ git diff
diff --git a/dbms/tests/queries/0_stateless/helpers/client.py b/dbms/tests/queries/0_stateless/helpers/client.py
index de4da79480..9700c6a374 100644
--- a/dbms/tests/... | 2019-08-19T02:52:09 | 2019-08-19T02:52:09 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 521,503,360 | 6,503 | den-crane | Check errors in CH server log. It could a network issue. | 2019-08-15T03:48:26 | 2019-08-15T03:48:26 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 521,535,552 | 6,472 | 360developer | We also tried to upgrade from 19.4.1.3 to 19.13.1.11 and found the same error in the logs.
2019.08.11 12:56:23.395080 [ 14 ] {} <Error> events.custom_properties_local: Code: 40, e.displayText() = DB::Exception: Checksums of parts don't match: hash of uncompressed files doesn't match (version 19.13.1.11 (official build... | 2019-08-15T06:49:08 | 2019-08-15T06:49:08 | {} | NONE |
ClickHouse | ClickHouse | 521,751,320 | 6,472 | den-crane | @360developer
>We also tried to upgrade from 19.4.1.3 to 19.13.1.11 and found the same error in the logs.
LZ4 compression library was updated in 19.7.
19.6- & 19.7+ are incompatible (not fully compatible) because they create different LZ4 binary output with not matching checksums | 2019-08-15T18:43:08 | 2019-08-15T18:49:44 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 521,951,290 | 3,981 | doom369 | Same issue here:
```
2019.08.16 09:27:41.097746 [ 45 ] {} <Trace> HTTPHandler-factory: HTTP Request for HTTPHandler-factory. Method: POST, Address: [::ffff:xxx.xxx.xxx.xxx]:50230, User-Agent: Apache-HttpClient/4.5.2 (Java/11.0.4), Content Type: , Transfer Encoding: chunked
2019.08.16 09:27:41.097865 [ 45 ] {} <Tra... | 2019-08-16T09:38:22 | 2019-08-16T09:38:22 | {} | NONE |
ClickHouse | ClickHouse | 521,956,381 | 5,618 | jangorecki | > > should not use the disk? Why still reporting memory limits?
>
> Because final aggregation result should fit into memory.
>
> So for group by SOME_HIGH_CARDINALITY_SET_OF_COLUMNS, ch will do intermediate partial aggregations on a disk, but in the end CH will gather the result into one part in memory.
@den-c... | 2019-08-16T09:57:43 | 2019-08-16T09:59:15 | {} | NONE |
ClickHouse | ClickHouse | 522,382,666 | 6,425 | vzakaznikov | @KochetovNicolai, could you try the patch below and see if it works for you
```
$ git diff
diff --git a/dbms/tests/queries/0_stateless/helpers/uexpect.py b/dbms/tests/queries/0_stateless/helpers/uexpect.py
index d65190e2b2..6f149992fc 100644
--- a/dbms/tests/queries/0_stateless/helpers/uexpect.py
+++ b/dbms/tes... | 2019-08-19T01:48:07 | 2019-08-19T01:49:44 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 521,504,806 | 6,501 | den-crane | It's my own opinion. I am just CH user the same as you. Funny because you proposed a solution which will make everything even worse. | 2019-08-15T03:58:23 | 2019-08-15T03:58:33 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 521,507,082 | 6,503 | lamberken | > So you've suspect an issue in clickhouse-jdbc then?
> Check CH logs for errors.
I think may modify some configuration item will solve this problem.
btw, I checked the `clickhouse-server.err.log`
```
2019.08.15 12:11:45.976564 [ 1751 ] {} <Error> ServerErrorHandler: Poco::Exception. Code: 1000, e.code() = 104... | 2019-08-15T04:13:46 | 2019-08-15T04:14:42 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 521,510,890 | 6,501 | iameugenejo | > Nobody will install test version in production. It's nonsense.
I didn't mention all releases should be labeled test version. Just leave it as a version without stable tag?
> Most of stable versions have critical issues. The could be critical for you but not for me.
Maybe enough critical issues warrant a remova... | 2019-08-15T04:39:09 | 2019-08-15T04:42:38 | {} | NONE |
ClickHouse | ClickHouse | 521,536,892 | 6,488 | astudnev | I need to configure replica, that is used only for reading. It is not expected that anybody will write in it. This limitation will remove the risk of corrupting the main database.
Some analogue of master-slave in MySQL | 2019-08-15T06:54:42 | 2019-08-15T06:54:42 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 521,578,779 | 5,273 | NanoBjorn | Have no idea why 00395_nullable fails, locally (under ub sanitizer) works fine
| 2019-08-15T09:27:59 | 2019-08-15T09:27:59 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 521,737,481 | 6,448 | kreuzerkrieg | @alexey-milovidov couple of notes
Some files were reformatted, of course, according to your clang format settings, if you strongly oppose the noise it creates let me know, I will revert.
Second, there were some warning suppression, which I strongly oppose, except cases when the noise originated from third party, whic... | 2019-08-15T18:01:42 | 2019-08-15T18:01:42 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 521,822,876 | 6,486 | alexey-milovidov | What build (what OS, compiler, version, additional options) and what message do you get?
We have build check in CI and all supported builds have passed: https://clickhouse-builds.s3.yandex.net/6509/c52a34fd15bc834ab40c8ef91ee281a89d58e2b3/report.html | 2019-08-15T22:43:13 | 2019-08-15T22:44:17 | {} | MEMBER |
ClickHouse | ClickHouse | 521,831,439 | 6,448 | alexey-milovidov | That's great. I quickly looked through your modifications and they look very valuable!
> Some files were reformatted, of course, according to your clang format settings, if you strongly oppose the noise it creates let me know, I will revert.
It's Ok unless it changes whole file (namespace indentation or something... | 2019-08-15T23:25:44 | 2019-08-15T23:25:44 | {} | MEMBER |
ClickHouse | ClickHouse | 522,109,756 | 6,525 | den-crane | ```
Затем для самой установки пакетов выполните:
sudo apt-get install dirmngr # optional
sudo apt-key adv --keyserver hkp://keyserver.ubuntu.com:80 --recv E0C56BD4 # optional
sudo apt-get update
sudo apt-get install clickhouse-client clickhouse-server
``` | 2019-08-16T18:36:17 | 2019-08-16T18:36:56 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 522,136,103 | 6,425 | vzakaznikov | Hi @KochetovNicolai, thanks for trying to merge the PR. I apologize for not getting back earlier. I was just distracted by other work.
What was wrong with the 00963_temporary_live_view_watch_live_timeout test? Maybe I can improve it. I also see that you have rewritten one of the tests, I guess the uexpect.py module ... | 2019-08-16T20:09:27 | 2019-08-16T20:09:27 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 522,165,160 | 6,429 | risicle | To fix the macos build I suspect you'll need to also look at `libs/libcommon/src/sleep.cpp`, which uses `clock_nanosleep` and `TIMER_ABSTIME`, neither of which are supported on macos from what I can tell. | 2019-08-16T22:02:41 | 2019-08-16T22:02:41 | {} | NONE |
ClickHouse | ClickHouse | 521,461,206 | 3,174 | simper66 | So, we have to avoid that at any cost, right? Because the summing, aggregating, replacing trees and sparse indexes would be somehow broken?
Perhaps making the partitioning key more granular, adding shards, or whatever? | 2019-08-14T23:52:16 | 2019-08-14T23:52:16 | {} | NONE |
ClickHouse | ClickHouse | 521,715,112 | 6,489 | alesapin | In `ReplicatedMergeTree` background thread for merges and mutations replaced with thread which pulls replicated queue. This thread doesn't check disks overflow according to disks policy and doesn't move parts to other disks. I think we need to add one more thread to the replicated merge tree https://github.com/yandex/C... | 2019-08-15T16:56:03 | 2019-08-15T16:56:03 | {} | MEMBER |
ClickHouse | ClickHouse | 521,867,036 | 6,486 | Vxider | The build check is passed because this function is not called in ClickHouse. Build the following example with g++-8 will encounter the error.
```
#include <iostream>
#include <vector>
using namespace std;
using String = std::string;
struct SettingChange {
String name;
int value;
};
using Set... | 2019-08-16T03:00:40 | 2019-08-16T03:00:40 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 521,964,576 | 5,230 | doom369 | @filimonov have the same issue. Scenario:
I have 2 servers that use the same clickhouse instance, but different DBs inside this instance. Every server have minute cron job that does batch inserts (simple inserts, no aggregation). Data size is small - few hundred of records.
Once a day I see this:
![screenshot-gi... | 2019-08-16T10:29:04 | 2019-08-16T10:29:04 | {} | NONE |
ClickHouse | ClickHouse | 521,377,270 | 6,448 | kreuzerkrieg | > More info: https://randomascii.wordpress.com/2012/06/26/doubles-are-not-floats-so-dont-compare-them/
Yep, read this one the day before yesterday. If you are sure you are ok with `==` and `!=` let it be, I guess you know better what are you doing. On the other hand, why do you have special functions to compare floa... | 2019-08-14T19:05:36 | 2019-08-14T19:05:36 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 521,383,137 | 6,448 | alexey-milovidov | > If you are sure you are ok with == and != let it be
Yes. We usually compare floats that was not affected by any floating point math. For example:
```
float x = 0;
... maybe set x to some non-zero value
if (x == 0)
do something
```
There is not much floating point math in ClickHouse. And if any, we usual... | 2019-08-14T19:22:57 | 2019-08-14T19:22:57 | {} | MEMBER |
ClickHouse | ClickHouse | 522,066,298 | 6,518 | alexey-milovidov | Sorry, but this doesn't look like the fix for the root cause of the issue. Why the function should not use default implementation for constants? | 2019-08-16T16:17:05 | 2019-08-16T16:17:05 | {} | MEMBER |
ClickHouse | ClickHouse | 522,246,060 | 5,230 | 4ertus2 | Why do you think it’s the same issue? The source issue is about crash when calling insert select with aggregation. Is it your case? | 2019-08-17T15:14:48 | 2019-08-17T15:14:48 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 522,341,423 | 6,532 | alexey-milovidov | If your network is overloaded and you need to save some bandwidth for queries. | 2019-08-18T17:47:44 | 2019-08-18T17:47:44 | {} | MEMBER |
ClickHouse | ClickHouse | 521,500,839 | 6,501 | iameugenejo | Is that tagged or documented as such if that’s an official response? And what is very funny about? | 2019-08-15T03:31:46 | 2019-08-15T03:31:46 | {} | NONE |
ClickHouse | ClickHouse | 521,664,371 | 6,430 | abyss7 | Actually the title is misleading - you have a chain of MVs, which didn't ever worked until recently. But really no one tested it yet, AFAIK. The multiple MVs from a single Kafka table do work. | 2019-08-15T14:37:17 | 2019-08-15T14:37:17 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 522,451,965 | 6,472 | 360developer | @alexey-milovidov
> No actions required.
Does this mean that we can update each instance of our cluster separately? Once all nodes have been updated the `Checksum of parts` message should be gone, right?
Thanks a lot. And also many thanks for creating such a wonderful database! | 2019-08-19T07:41:50 | 2019-08-19T07:42:05 | {} | NONE |
ClickHouse | ClickHouse | 519,562,145 | 6,366 | alesapin | Hooray, I've reproduced race condition! | 2019-08-08T15:14:19 | 2019-08-08T16:21:44 | {} | MEMBER |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.