Datasets:
open-index
/
open-github-issues

Dataset card Data Studio Files
xet
 Community
1
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
835,361,305
23,958
kssenii
@electrical I suppose it will not get into LTS, because it is not a bug fix..
2021-05-08T13:20:34
2021-05-08T13:20:34
{}
MEMBER
ClickHouse
ClickHouse
835,869,391
23,979
azat
>@azat please provide some benchmark where there will be significant performance improvements. @kitaisreal There are some numbers in the initial PR (#15454): ``` $ ninja bench-sparse_hash-run [1/1] cd /src/ch/hashtable-bench/.cmake && ...ch/hashtable-bench/.cmake/bench-sparse_hash sparse_hash/insert: 7.574 <!-...
2021-05-09T19:27:06
2021-05-09T19:27:06
{}
MEMBER
ClickHouse
ClickHouse
835,880,676
23,979
alexey-milovidov
How do you have one hour loading time? If dictionary occupies 100 GB, it means just about 30 MB/sec - very slow even for google::sparse_hash_map.
2021-05-09T20:26:05
2021-05-09T20:26:05
{}
MEMBER
ClickHouse
ClickHouse
836,053,446
8,482
alexey-milovidov
I'd like to make a test that will force filling a cache - just in case... We can add it in next PR if you like.
2021-05-10T01:43:26
2021-05-10T01:43:26
{}
MEMBER
ClickHouse
ClickHouse
836,083,832
23,987
mythmoon
异常重启后，出现, It is a bug., Stack trace (when copying this message, always include the lines below):
2021-05-10T02:34:30
2021-05-10T02:34:30
{}
NONE
ClickHouse
ClickHouse
834,920,711
22,086
yiguolei
> To be honest, I didn't think too much of transactions outside of INSERT usecase (there seem to be a lot of pitfalls), for me the main use case is the atomicity of inserts. Some additional guarantees for reading (except "do not read what is not commited") seem to be less in demand. Therefore, I can have a one-sided lo...
2021-05-08T01:02:17
2021-05-08T01:02:17
{}
CONTRIBUTOR
ClickHouse
ClickHouse
835,240,309
23,903
azat
>Integration tests (asan) — Timeout, fail: 0, passed: 1314, flaky: 1 >Integration tests (thread) — Timeout, fail: 1, passed: 1073, flaky: 0 Unrelated
2021-05-08T09:25:13
2021-05-08T09:25:13
{}
MEMBER
ClickHouse
ClickHouse
835,370,610
23,958
kssenii
@electrical hm, I agree, may be it will get into lts.
2021-05-08T13:46:22
2021-05-08T13:46:22
{}
MEMBER
ClickHouse
ClickHouse
835,405,524
23,935
den-crane
duplicate https://github.com/ClickHouse/ClickHouse/issues/21717 Why did you use 21.3.4.25 instead of 21.3.9.83 ?
2021-05-08T15:38:15
2021-05-08T15:38:22
{}
CONTRIBUTOR
ClickHouse
ClickHouse
835,486,858
21,659
vzakaznikov
@zvonand, let's experiment with either disabling host network (https://github.com/ClickHouse/ClickHouse/blob/master/tests/testflows/runner#L73) or creating a dedicated custom network between containers.
2021-05-08T19:44:15
2021-05-08T19:44:15
{}
CONTRIBUTOR
ClickHouse
ClickHouse
835,813,536
21,594
sergey-gt-ekb
On official docker images since versions ~20, got the same crushes.... :( CH totally not work in containers?! Version 18.16 work more stable, without crushes. But i have got another issue(and same on latest images): At random moment server stop working normally, and in the server logs begin writes a lot of binar...
2021-05-09T14:01:09
2021-05-09T14:01:40
{}
NONE
ClickHouse
ClickHouse
835,880,883
23,979
alexey-milovidov
For large hash table the difference in loading time will probably be even lower, because resizes will take less relative amount of time.
2021-05-09T20:27:03
2021-05-09T20:27:03
{}
MEMBER
ClickHouse
ClickHouse
834,908,587
23,947
alexey-milovidov
I've changed comment in XML to CDATA, because XML comments cannot have `--` character sequence inside.
2021-05-08T00:43:48
2021-05-08T00:43:48
{ "+1": 1 }
MEMBER
ClickHouse
ClickHouse
834,910,345
23,941
alexey-milovidov
It can be a mode of EXPLAIN query.
2021-05-08T00:46:33
2021-05-08T00:46:33
{}
MEMBER
ClickHouse
ClickHouse
835,283,639
23,931
UnamedRus
``` CREATE TABLE goods ENGINE = Memory AS SELECT rand() % 4 AS product_id, now() + toIntervalHour(number) AS lastupdated, number AS qty FROM numbers(100) SELECT date, ABS(sum(qty - lag_qty)) AS orders FROM ( SELECT product_id, date, qty, lagInFr...
2021-05-08T10:41:09
2021-05-08T10:41:09
{}
CONTRIBUTOR
ClickHouse
ClickHouse
835,406,119
23,955
den-crane
Check clickhouse server err.log. It does not look like CH error. This error is in java / jdbc.
2021-05-08T15:40:25
2021-05-08T15:40:25
{}
CONTRIBUTOR
ClickHouse
ClickHouse
835,420,133
21,770
alexey-milovidov
@buyology You can also create another PR with changes on top of this PR.
2021-05-08T16:26:58
2021-05-08T16:26:58
{}
MEMBER
ClickHouse
ClickHouse
835,876,666
21,594
azat
@Maximus43 do you have an ability to check some **development** version of clickhouse (I wouldn't recommend to do this on production data)? If so I would recommend you to check version with ASAN, for this you need to: - [go to the commits list](https://github.com/ClickHouse/ClickHouse/commits/master) - [press on chec...
2021-05-09T20:06:08
2021-05-09T20:06:08
{}
MEMBER
ClickHouse
ClickHouse
834,880,455
23,941
fastio
Is it ok like following? ``` show estimates database_name.table_name; ┌─parts─┬─marks─┬────rows─┐ │ 343 │ 1126 │ 1032192 │ └───────┴───────┴─────────┘ ```
2021-05-08T00:01:49
2021-05-08T00:02:48
{}
CONTRIBUTOR
ClickHouse
ClickHouse
835,268,409
23,943
kssenii
No such issue in 20.3.
2021-05-08T10:15:39
2021-05-08T10:15:39
{}
MEMBER
ClickHouse
ClickHouse
835,280,717
23,935
amosbird
Good question. I don't know how to find the exact releases. Maybe you can try checking some latest versions and see if the ChangeLog contains this fix.
2021-05-08T10:36:24
2021-05-08T10:36:24
{}
COLLABORATOR
ClickHouse
ClickHouse
835,342,989
19,969
alexey-milovidov
@ys-zhaoyh I've missed your update. But I'm still not being able to answer your question. If you are trying to query HDFS data with ClickHouse on a cluster with Kerberos authentication... \- I can only say that it is expected to work. If it does not - I need an isolated example to reproduce.
2021-05-08T12:37:25
2021-05-08T12:37:25
{}
MEMBER
ClickHouse
ClickHouse
835,353,819
23,958
electrical
@kssenii Ahh okay :-) Thank you. I made the wrong assumption it was in there. Do you have an idea if this will be in an LTS release soon or should i revert back to the ODBC driver?
2021-05-08T13:02:22
2021-05-08T13:02:22
{}
NONE
ClickHouse
ClickHouse
835,382,928
23,935
filimonov
21.3.9 should also have it backported
2021-05-08T14:23:53
2021-05-08T14:23:53
{}
CONTRIBUTOR
ClickHouse
ClickHouse
835,417,694
23,913
alexey-milovidov
I don't have any idea how to implement it properly, let's ask @CurtizJ, @KochetovNicolai.
2021-05-08T16:18:17
2021-05-08T16:18:17
{}
MEMBER
ClickHouse
ClickHouse
835,485,679
23,934
kitaisreal
@hexiaoting we recently checked and a lot of users rely on our system tables, for example there are tools that are not longer supported by maintainers that rely on some system.tables structure. As additional change we need to keep old columns as virtual storage columns. Check for example system.dictionaries. That way t...
2021-05-08T19:40:02
2021-05-08T19:40:02
{}
CONTRIBUTOR
ClickHouse
ClickHouse
834,989,021
13,159
amosbird
Close because it's a misusage of WITH statement in ClickHouse. It doesn't work in postgres either. ``` with x as (select 10) select * from generate_series(1, x); ```
2021-05-08T02:41:07
2021-05-08T02:41:07
{}
COLLABORATOR
ClickHouse
ClickHouse
835,687,788
6,757
arjenzhou
Just remind: this feature need linux kernel support (renameat2(), which was introduced after 3.15)
2021-05-09T05:13:25
2021-05-09T05:13:25
{}
NONE
ClickHouse
ClickHouse
835,347,679
23,958
kssenii
@electrical, it does not work because ability to specify postgres `schema` of table was added only for table engine / table function, not for dictionary source. I'll add it for dictionary also..
2021-05-08T12:48:16
2021-05-08T12:48:16
{}
MEMBER
ClickHouse
ClickHouse
835,358,047
23,941
alexey-milovidov
@filimonov Two options exist: - sum the values across all tables; - display the values corresponding to every table in separate rows.
2021-05-08T13:12:26
2021-05-08T13:12:26
{}
MEMBER
ClickHouse
ClickHouse
835,877,228
21,594
azat
BTW can you also provide: - linux kernel version - linux kernel config - `sysctl -a` - `/proc/meminfo`
2021-05-09T20:08:47
2021-05-09T20:08:47
{}
MEMBER
ClickHouse
ClickHouse
836,435,094
23,987
filimonov
https://github.com/ClickHouse/ClickHouse/issues/23296
2021-05-10T09:04:30
2021-05-10T09:04:30
{}
CONTRIBUTOR
ClickHouse
ClickHouse
835,136,337
23,941
amosbird
> Ok, I'll try it. Please also take distributed queries into account.
2021-05-08T06:19:55
2021-05-08T06:19:55
{}
COLLABORATOR
ClickHouse
ClickHouse
835,369,248
23,958
electrical
@kssenii I see. I would partially disagree with since it's a missing function for the postgresql source in dictionaries while, as you say, it's there for the table engine / table function.
2021-05-08T13:42:06
2021-05-08T13:42:06
{}
NONE
ClickHouse
ClickHouse
835,865,527
21,364
silashansen
This feature is very interesting <3 Is it still being tracked here, or has progress moved somewhere else?
2021-05-09T19:06:55
2021-05-09T19:06:55
{}
NONE
ClickHouse
ClickHouse
835,876,484
21,594
azat
>@azat Could it be related to jemalloc? @alexey-milovidov issue with 18.16 may be related to jemalloc, but NULL pointer in - I doubt.
2021-05-09T20:05:03
2021-05-09T20:05:03
{}
MEMBER
ClickHouse
ClickHouse
836,340,990
8,482
kitaisreal
@alexey-milovidov will add test for cache fill in scope of next pull request, tested it locally old function are removed.
2021-05-10T08:10:15
2021-05-10T08:10:15
{}
CONTRIBUTOR
ClickHouse
ClickHouse
836,364,747
23,461
excitoon
Failed test is unrelated.
2021-05-10T08:31:43
2021-05-10T08:31:43
{}
CONTRIBUTOR
ClickHouse
ClickHouse
835,265,370
23,941
filimonov
What about queries affecting non-merge tree? several merge tree tables (i.e. joins) ?
2021-05-08T10:10:18
2021-05-08T10:10:18
{}
CONTRIBUTOR
ClickHouse
ClickHouse
835,327,149
21,858
Felixoid
Looks like it will fix #3607
2021-05-08T11:58:53
2021-05-08T11:58:53
{}
MEMBER
ClickHouse
ClickHouse
835,785,943
20,539
myrrc
Last commit https://github.com/ClickHouse/ClickHouse/pull/20539/commits/2a08dd3f671b154268a892045a290e426baeaddc introduced a working version with dumping genhtml reports, but this format still isn't appropriate for accumulating large coverage reports while preserving per-test data, so I'll create my own format and wri...
2021-05-09T11:23:58
2021-05-09T11:23:58
{}
CONTRIBUTOR
ClickHouse
ClickHouse
835,791,020
23,979
kitaisreal
@azat please provide some benchmark where there will be significant performance improvements.
2021-05-09T11:53:49
2021-05-09T11:53:49
{}
CONTRIBUTOR
ClickHouse
ClickHouse
835,868,487
23,911
excitoon
@ucasFL do you use bundled Poco library or some other version?
2021-05-09T19:22:45
2021-05-09T19:22:45
{}
CONTRIBUTOR
ClickHouse
ClickHouse
836,050,690
23,942
alexey-milovidov
Let's enable everything except LDAP.
2021-05-10T01:38:38
2021-05-10T01:38:38
{}
MEMBER
ClickHouse
ClickHouse
836,332,193
23,843
kitaisreal
@MaxWk please add documentation, fix style check, and add function test for this new function.
2021-05-10T08:02:36
2021-05-10T08:02:36
{}
CONTRIBUTOR
ClickHouse
ClickHouse
835,371,425
23,916
azat
>[Functional stateless tests flaky check (address) — Timeout, fail: 0, passed: 61](https://clickhouse-test-reports.s3.yandex.net/23916/1e131f03af1fe17e2abff6212b3d9df37fa734c4/functional_stateless_tests_flaky_check_(address).html#fail1) Should this timeout be addressed somehow? Or is it fine?
2021-05-08T13:49:02
2021-05-08T13:49:02
{}
MEMBER
ClickHouse
ClickHouse
835,423,503
23,913
CurtizJ
However, it shoud work with `Distributed`. I guess (can't check now), that it fails because of conflicting ast optimization (`optimize_monotonous_functions_in_order_by`), which eliminates monotonic function and order key doesn't match after it (https://github.com/ClickHouse/ClickHouse/issues/16107). Probably we should ...
2021-05-08T16:38:52
2021-05-08T16:40:14
{}
MEMBER
ClickHouse
ClickHouse
835,497,901
23,911
excitoon
@ucasFL we need to reproduce it in test. Actually I wrote a test server which sometimes drops connections but my test is passing. Probably there are different HTTP header set with which Poco behaves differently :man_shrugging:
2021-05-08T20:20:41
2021-05-08T20:20:41
{}
CONTRIBUTOR
ClickHouse
ClickHouse
834,960,592
23,035
etah000
In that case, it should report some errors like "node already existed".
2021-05-08T02:04:07
2021-05-08T02:04:07
{}
CONTRIBUTOR
ClickHouse
ClickHouse
834,965,103
23,941
fastio
Ok, I'll try it.
2021-05-08T02:10:01
2021-05-08T02:10:01
{}
CONTRIBUTOR
ClickHouse
ClickHouse
835,485,956
22,729
vzakaznikov
@alexey-milovidov, @zvonand, let's disable these tests in regression.py to merge this PR. We will enable them in another PR that will run them in parallel.
2021-05-08T19:41:02
2021-05-08T19:41:02
{}
CONTRIBUTOR
ClickHouse
ClickHouse
835,692,605
23,926
filimonov
BTW opposite rename (from atomic to ordinary DB) of MV with implicitly created storage table also do a very weird thing - it renames the table but leaves it in atomic DB.
2021-05-09T05:28:51
2021-05-09T05:28:51
{}
CONTRIBUTOR
ClickHouse
ClickHouse
835,869,844
23,979
azat
@alexey-milovidov >Ok, but... is it worth adding a separate parameter (that almost no one will use - dead code, difficult to test and support)? Making it by default increases memory usage (that's why it was reverted in #21948) >Also, it speeds up only loading of dictionaries. But it is what is mostly irreleva...
2021-05-09T19:29:33
2021-05-09T19:29:33
{}
MEMBER
ClickHouse
ClickHouse
835,881,202
23,979
alexey-milovidov
What numbers do you have in your production?
2021-05-09T20:28:26
2021-05-09T20:28:26
{}
MEMBER
ClickHouse
ClickHouse
835,954,781
23,942
vzakaznikov
So far I can't get LDAP module to be stable under heavy system load.
2021-05-10T00:00:54
2021-05-10T00:00:54
{}
CONTRIBUTOR
ClickHouse
ClickHouse
836,086,640
23,986
alexey-milovidov
Does the issue reproduce on recent ClickHouse release?
2021-05-10T02:39:04
2021-05-10T02:39:04
{}
MEMBER
ClickHouse
ClickHouse
836,228,689
23,954
wallace-clickhouse
Can you give me some general optimization instructions?For the situation of returning many columns(about 100) in one query,under which I find the query is slower than less columns. Thanks a lot！
2021-05-10T06:11:20
2021-05-10T06:11:20
{}
NONE
ClickHouse
ClickHouse
836,330,148
23,934
kitaisreal
@hexiaoting yes, it will be best solution for us, because if we merge it as is, there will be issues that some tool that cannot be updated is commited to our system tables column names.
2021-05-10T08:00:45
2021-05-10T08:00:45
{}
CONTRIBUTOR
ClickHouse
ClickHouse
835,047,309
23,934
hexiaoting
Functional stateless tests (address) failed query: `01455_shard_leaf_max_rows_bytes_to_read.sql`. But I run this test case on my server, it's OK. And I don't change any code related to it. Any advice?
2021-05-08T03:31:48
2021-05-08T03:31:48
{}
CONTRIBUTOR
ClickHouse
ClickHouse
835,053,777
18,762
sundy-li
continue in #23934
2021-05-08T03:41:25
2021-05-08T03:41:25
{}
CONTRIBUTOR
ClickHouse
ClickHouse
835,137,524
23,935
amosbird
It's most likely related to https://github.com/ClickHouse/ClickHouse/pull/21766
2021-05-08T06:22:16
2021-05-08T06:22:16
{}
COLLABORATOR
ClickHouse
ClickHouse
835,217,935
23,929
azat
>ClickHouse special build check — 5/6 builds are OK Looks like CI issue >Integration tests (asan) — Timeout, fail: 0, passed: 1306, flaky: 2 >Testflows check — failed: 8, passed: 833, other: 100 Unrelated
2021-05-08T08:47:27
2021-05-08T08:47:27
{}
MEMBER
ClickHouse
ClickHouse
835,266,194
23,730
vsop-479
I think we had restarted clients before restarted servers. I will check it again. The driver is ClickHouse official jdbc.
2021-05-08T10:11:46
2021-05-08T10:11:46
{}
NONE
ClickHouse
ClickHouse
835,419,675
23,963
alexey-milovidov
For what purpose?
2021-05-08T16:25:20
2021-05-08T16:25:20
{}
MEMBER
ClickHouse
ClickHouse
835,631,977
23,911
ucasfl
> @ucasFL Do you build ClickHouse yourself or use one of builds from CI? Which one? Yes, I build ClickHouse myself.
2021-05-09T02:15:50
2021-05-09T02:15:50
{}
COLLABORATOR
ClickHouse
ClickHouse
835,763,729
6,084
astudnev
Why it takes local update time instead of the last update time of the dictionary data? That would remove the nessesity of any lag. Lag is hard to control, especially when dictionary read from db slave
2021-05-09T09:30:58
2021-05-09T09:30:58
{}
CONTRIBUTOR
ClickHouse
ClickHouse
835,832,451
23,942
alexey-milovidov
Ok, it took 42 minutes.
2021-05-09T15:49:51
2021-05-09T15:49:51
{}
MEMBER
ClickHouse
ClickHouse
835,878,055
21,594
azat
>It shows amount of RAM on the host machine, while container is limited to 128GB of RAM. Yes, because clickhouse does not cgroup aware right now. >Yes, It seems that segfault is related to the wrong definition of available memory. "Bare-metal" and KVM instances have no issues at all. It should not be related t...
2021-05-09T20:13:17
2021-05-09T20:13:17
{}
MEMBER
ClickHouse
ClickHouse
835,878,244
23,631
alexey-milovidov
> Functional stateless tests flaky check (address) — Timeout, fail: 0, passed: 845 Ok (timeout due to too many changed tests).
2021-05-09T20:14:15
2021-05-09T20:14:15
{}
MEMBER
ClickHouse
ClickHouse
836,189,700
21,770
taylor12805
add array nullable datatype support in ArrowColumnToCHColumn, gonna add array(decimal) support this week
2021-05-10T05:21:34
2021-05-10T05:21:34
{}
CONTRIBUTOR
ClickHouse
ClickHouse
834,832,550
20,539
myrrc
After introducing own memory pool I managed to lower execution time to 14642.36 s ~ 4.02h
2021-05-07T22:43:33
2021-05-08T10:13:43
{}
CONTRIBUTOR
ClickHouse
ClickHouse
835,356,647
23,954
alexey-milovidov
There are multiple ways how queries can be optimized. Give me your query, the definition of your tables, your cluster configuration and the settings, the configuration of your hardware, the version of ClickHouse you are using; your domain area and load scenario, the name of your company... and maybe we will find a w...
2021-05-08T13:09:01
2021-05-08T13:09:01
{}
MEMBER
ClickHouse
ClickHouse
835,381,620
23,941
filimonov
> @filimonov > > Two options exist: > > * sum the values across all tables; > * display the values corresponding to every table in separate rows. I'm voting for the second one
2021-05-08T14:19:18
2021-05-08T14:19:18
{ "+1": 1 }
CONTRIBUTOR
ClickHouse
ClickHouse
835,419,174
23,962
alexey-milovidov
Is total row calculated correctly if data was flushed to disk?
2021-05-08T16:23:34
2021-05-08T16:23:34
{}
MEMBER
ClickHouse
ClickHouse
835,444,025
23,962
azat
>Is total row calculated correctly if data was flushed to disk? Yes (I've changed the aggregate function in test to uniqCombined)
2021-05-08T17:44:27
2021-05-08T17:44:50
{}
MEMBER
ClickHouse
ClickHouse
835,804,953
23,977
vzakaznikov
I don't remember in which version it was added. But the latest stable version of ClickHouse supports it. ``` user@user-host:~$ clickhouse-client ClickHouse client version 21.6.1.6689 (official build). Connecting to localhost:9000 as user default. Connected to ClickHouse server version 21.6.1 revision 54448. ...
2021-05-09T13:08:53
2021-05-09T13:10:34
{}
CONTRIBUTOR
ClickHouse
ClickHouse
835,809,655
6,084
den-crane
For example it's the only solution for us. 1. T0: Transaction1 makes an insert (updated_at=T0) T1: Transaction2 makes an insert (updated_at=T1) T2: Transaction2 commits T3: Dictionary gets updates and sees max updated_at = T1. T4: Transaction1 commits -- changes lost forever. 2. We have a dictionary which is...
2021-05-09T13:37:21
2021-05-09T13:37:21
{ "+1": 1 }
CONTRIBUTOR
ClickHouse
ClickHouse
836,053,130
8,482
alexey-milovidov
Let's add asynchronous metrics about the size of compiled expression cache in bytes, near ``` new_values["CompiledExpressionCacheCount"] = compiled_expression_cache->count(); ```
2021-05-10T01:42:53
2021-05-10T01:42:53
{}
MEMBER
ClickHouse
ClickHouse
836,217,442
23,758
excitoon
Hi @kitaisreal . Can you look at my changes?
2021-05-10T05:57:48
2021-05-10T05:57:48
{}
CONTRIBUTOR
ClickHouse
ClickHouse
834,881,787
23,631
alexey-milovidov
Conversion from Float to Decimal is imprecise: ``` SELECT toDecimal256(0.1, 76) ``` is: ``` 0.1000000000000000127244406382982157680441515146893321226564617167921968152000 ``` must be: ``` 0.1000000000000000055511151231257827021181583404541015625000000000000000000000 ``` But I will not fix it in this...
2021-05-08T00:03:40
2021-05-08T00:04:00
{}
MEMBER
ClickHouse
ClickHouse
835,421,205
23,963
kitaisreal
@alexey-milovidov some statistics counters can be based on queries like SELECT sum(hit_rate) FROM system.dictionaries, result will be nan if such issue exists.
2021-05-08T16:30:53
2021-05-08T16:30:53
{}
CONTRIBUTOR
ClickHouse
ClickHouse
835,831,199
23,979
alexey-milovidov
Ok, but... is it worth adding a separate parameter (that almost no one will use - dead code, difficult to test and support)? Also, it speeds up only loading of dictionaries. But it is what is mostly irrelevant for performance. Lookup performance is the most important and I don't care if loading is 1.5 times faster o...
2021-05-09T15:43:08
2021-05-09T15:45:07
{}
MEMBER
ClickHouse
ClickHouse
836,059,930
23,911
ucasfl
> @ucasFL do you use bundled Poco library or some other version? I use bundled Poco library.
2021-05-10T01:54:20
2021-05-10T01:54:20
{}
COLLABORATOR
ClickHouse
ClickHouse
836,083,688
23,934
hexiaoting
@kitaisreal So for system tables, I should keep old columns like `attributes.names` and `attributes.values` as virtual columns, And also add a corresponding new Column `attributes Map(String, String)`, right?
2021-05-10T02:34:17
2021-05-10T02:34:17
{}
CONTRIBUTOR
ClickHouse
ClickHouse
834,780,419
23,909
kssenii
`PVS check — Have 1 new errors` - false positive. Other failures not related.
2021-05-07T21:07:13
2021-05-07T21:07:13
{}
MEMBER
ClickHouse
ClickHouse
835,391,318
23,062
tavplubix
> The first step is to set the state manually, and the next step can be an automated method. I don't think that manual maintenance of host status in a distributed system is a good idea. Because unexpected sudden failures are inevitable, and such failures should be handled automatically. And it can be easily done wit...
2021-05-08T14:50:58
2021-05-08T14:50:58
{ "+1": 1 }
MEMBER
ClickHouse
ClickHouse
835,402,899
21,770
buyology
@taylor12805 — any update on this? Would you accept PRs into your branch to help out? (Not saying I'm the best suited to make such a patch, someone more versed in CH could finish it off more quickly.) We would really love for this to go in soon.
2021-05-08T15:29:07
2021-05-08T15:29:31
{ "+1": 1 }
NONE
ClickHouse
ClickHouse
835,500,412
23,911
excitoon
@ucasFL Do you build ClickHouse yourself or use one of builds from CI? Which one?
2021-05-08T20:28:51
2021-05-08T20:28:51
{}
CONTRIBUTOR
ClickHouse
ClickHouse
835,828,444
21,594
alexey-milovidov
@azat Could it be related to jemalloc?
2021-05-09T15:26:40
2021-05-09T15:26:40
{}
MEMBER
ClickHouse
ClickHouse
835,151,934
23,062
marising
> Creation of an ephemeral node on `DDLWorker` initialization will solve the problem. Existence of such node will indicate that host is available and `DDLWorker` is active. `DDLQueryStatusInputStream` will check if such node exists and ignore unavailable nodes if required. It will be much simpler than `CLUSTER PAUSE/ST...
2021-05-08T06:50:08
2021-05-08T06:50:08
{}
CONTRIBUTOR
ClickHouse
ClickHouse
835,269,372
23,935
AlexMAS
@amosbird Thank you for replying. Could you tell which version will include the fix?
2021-05-08T10:17:16
2021-05-08T10:17:16
{}
CONTRIBUTOR
ClickHouse
ClickHouse
835,303,452
23,955
a394023466
![image](https://user-images.githubusercontent.com/39642052/117537149-c4687c00-b031-11eb-8942-28e19d00bd52.png)
2021-05-08T11:15:40
2021-05-08T11:15:40
{ "+1": 1 }
NONE
ClickHouse
ClickHouse
835,314,002
23,935
AlexMAS
Seems [v21.4](https://clickhouse.tech/docs/en/whats-new/changelog/#clickhouse-release-21-4) should include the fix. I'll check it, thanks.
2021-05-08T11:34:51
2021-05-08T11:34:51
{}
CONTRIBUTOR
ClickHouse
ClickHouse
835,828,274
21,594
alexey-milovidov
@sergey-gt-ekb You should never use old versions of ClickHouse. The only supported versions are 21.4, 21.3 and 21.2.
2021-05-09T15:25:37
2021-05-09T15:25:37
{}
MEMBER
ClickHouse
ClickHouse
836,045,455
23,631
alexey-milovidov
Removing the experimental flag and adding more tests in subsequent PR.
2021-05-10T01:29:34
2021-05-10T02:05:37
{}
MEMBER
ClickHouse
ClickHouse
836,218,043
23,846
excitoon
Is anything else needed from my side?
2021-05-10T05:58:22
2021-05-10T05:58:22
{}
CONTRIBUTOR
ClickHouse
ClickHouse
831,941,256
23,703
alexey-milovidov
Better skip it.
2021-05-04T13:28:07
2021-05-04T13:28:07
{}
MEMBER
ClickHouse
ClickHouse
832,592,912
23,874
alexey-milovidov
Everything else is alright.
2021-05-05T10:47:36
2021-05-05T10:47:36
{}
MEMBER
ClickHouse
ClickHouse
832,635,242
17,847
alexey-milovidov
You can check the flag in factory - where the codecs are created.
2021-05-05T12:07:10
2021-05-05T12:07:10
{}
MEMBER
ClickHouse
ClickHouse
832,754,518
21,839
tobiaslins
@kssenii I only have one dictionary in place! after the rollback the issue is resolved again!
2021-05-05T14:53:51
2021-05-05T14:53:51
{}
CONTRIBUTOR