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 | 822,069,645 | 23,265 | alexey-milovidov | This is more intrusive but also better way to do it: #23266 | 2021-04-18T22:06:22 | 2021-04-18T22:06:22 | {} | MEMBER |
ClickHouse | ClickHouse | 822,071,247 | 23,257 | alexey-milovidov | Also I don't understand why `max_distributed_connections` does not work with hedged requests. | 2021-04-18T22:19:08 | 2021-04-18T22:19:08 | {} | MEMBER |
ClickHouse | ClickHouse | 813,320,064 | 22,661 | CurtizJ | | test_name | c | last_time | first_seen | last_seen | fix |
|:-:|-:|:-:|:-|:-|:-|
| 01666_merge_tree_max_query_limit | 10 | 2021-04-03 20:40:03 | 19e4a33f9d2dfa5293d74c195d916ba3ddff3a0e | 89db2b1ffc49aed82efcabffed1ccb594d6d550c | |
| 01149_zookeeper_mutation_stuck_after_replace_partition | 5 | 2021-04-03 20:36:... | 2021-04-05T10:09:38 | 2021-04-19T11:18:33 | {} | MEMBER |
ClickHouse | ClickHouse | 822,496,221 | 23,020 | tavplubix | I'm asking about version because the issue looks like duplicate of #6704. It was fixed in #13450, but the fix was not backported to versions older than 20.5. We also have fixed a lot of other issues related to `ON CLUSTER` queries recently, so I recommend to upgrade to newer ClickHouse version, because 20.3 is no longe... | 2021-04-19T14:08:31 | 2021-04-19T14:08:31 | {} | MEMBER |
ClickHouse | ClickHouse | 821,995,908 | 19,320 | alexey-milovidov | This feature can be postponed until we will add heredocs.
The examples of heredocs:
Postgres:
```
$here$
doc
$here$
```
Perl, sh:
```
<<HERE
doc
HERE
```
PHP:
```
<<<HERE
doc
HERE
```
Python (incomplete):
```
'''
doc
'''
```
C++:
```
R"here(
doc
)here"
```
XML (incomplete)... | 2021-04-18T13:58:38 | 2021-04-18T13:59:46 | {} | MEMBER |
ClickHouse | ClickHouse | 822,050,244 | 23,258 | azat | >@azat We build our sources with -Werror.
> All builds in CI are using -Werror with the only exception of "unbundled" build.
Oh, you right, missed the default value for WERROR in cmake.
However -Wundef is not added for gcc, I'm fine to replace this with `add_warning(undef)` for gcc then (and a fix for __clang__ ... | 2021-04-18T19:50:49 | 2021-04-18T19:51:08 | {} | MEMBER |
ClickHouse | ClickHouse | 822,058,321 | 23,212 | azat | The motivation was to provide a knob to disable mmap for big parts, since with multiple queries that just reads those parts the queries may fail due to memory limit exceeded (for total memory tracker), after #23211 had been merged. | 2021-04-18T20:45:12 | 2021-04-18T20:45:12 | {} | MEMBER |
ClickHouse | ClickHouse | 822,071,713 | 23,261 | alexey-milovidov | Isn't the issue fixed, can we close it? | 2021-04-18T22:22:51 | 2021-04-18T22:22:51 | {} | MEMBER |
ClickHouse | ClickHouse | 822,072,787 | 23,211 | alexey-milovidov | We suspect that performance test is broken after this PR. Could you please resubmit it? | 2021-04-18T22:30:51 | 2021-04-18T22:30:51 | {} | MEMBER |
ClickHouse | ClickHouse | 822,076,634 | 17,186 | alexey-milovidov | Rule of thumb: progress is returned when there is a progress.
But some queries are processed instantly. | 2021-04-18T22:56:10 | 2021-04-18T22:56:10 | {} | MEMBER |
ClickHouse | ClickHouse | 822,081,742 | 19,816 | olgarev | > @olgarev no this feature is not needed to be documented.
OK, thanks! Can you delete the "doc-alert" label then, please? | 2021-04-18T23:18:59 | 2021-04-18T23:18:59 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 822,185,941 | 23,211 | azat | >We suspect that performance test is broken after this PR.
Yes.
> Could you please resubmit it?
Ok, will resubmit with perf tests adjusted. | 2021-04-19T05:47:17 | 2021-04-19T05:47:17 | {} | MEMBER |
ClickHouse | ClickHouse | 822,263,038 | 23,212 | azat | >Does not look as an issue. Memory is calculated in some reasonable way (even if it is calculated multiple times for shared mappings, it is still reasonable for memory accounting). The user will get an error about memory limit maybe slightly sooner than expected.
Suppose you have the following pattern:
- 50-100 sma... | 2021-04-19T08:04:17 | 2021-04-19T08:05:05 | {} | MEMBER |
ClickHouse | ClickHouse | 822,356,680 | 23,300 | Sallery-X | > P.S. That is a feature for some VERY specific use cases and can make sense only if you see CPU is used heavily by merges, and network is fast / underloaded. For 'normal' usage patterns (one/few huge tables with batch inserts) we don't recommend using it.
I want to use nfs, all replicas use one file system. So i n... | 2021-04-19T10:20:14 | 2021-04-19T10:20:14 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 822,071,044 | 23,261 | alexey-milovidov | @azat Do you know what to answer here? https://github.com/ClickHouse/ClickHouse/issues/21167#issuecomment-820907631 | 2021-04-18T22:17:25 | 2021-04-18T22:17:25 | {} | MEMBER |
ClickHouse | ClickHouse | 822,077,966 | 17,186 | alexey-milovidov | But I cannot reproduce it on version 20.9.3.45 either.
`docker run --network host -it --rm yandex/clickhouse-server:20.9.3.45`
```
$ curl -v 'http://localhost:8123/?send_progress_in_http_headers=1&http_headers_progress_interval_ms=100&query=SELECT+max(hex(SHA256(toString(number))))+FROM+(SELECT+number+FROM+syste... | 2021-04-18T23:04:52 | 2021-04-18T23:04:52 | {} | MEMBER |
ClickHouse | ClickHouse | 822,252,001 | 23,038 | alesapin | No related failures. I'll merge this PR and will continue performance improvements in the next one. | 2021-04-19T07:47:25 | 2021-04-19T07:47:25 | {} | MEMBER |
ClickHouse | ClickHouse | 808,636,198 | 14,893 | pingyu | I think we can split this issue to the following subtasks:
- [x] cmake #22207
- [x] union (`uniqThetaSketch`) #22609
- [ ] intersection & difference (functions take `uniqThetaSketchState` as argument ?)
- [ ] function to convert Array(UInt8) to intermediate state (to be used for Merge/Intersetion/Difference, to ... | 2021-03-27T03:17:56 | 2021-04-19T11:30:09 | {
"+1": 2
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 822,472,736 | 23,197 | mergify[bot] | **Command `rebase`: success**
> **Branch has been successfully rebased**
>
| 2021-04-19T13:37:11 | 2021-04-19T13:37:11 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 822,559,627 | 23,029 | alexey-milovidov | ```
CREATE TEMPORARY TABLE t (c String);
SELECT isNull(c) OR COUNT(1)
FROM t
GROUP BY c
HAVING isNull(c)
UNION ALL
SELECT isNull(c) OR COUNT(1)
FROM t
GROUP BY c
HAVING isNull(isNull(c))
SETTINGS enable_optimize_predicate_expression = 0
``` | 2021-04-19T15:28:10 | 2021-04-19T15:28:10 | {} | MEMBER |
ClickHouse | ClickHouse | 822,051,418 | 23,264 | azat | trace_log size:
- snapshot: 33M
- tsv: 491K | 2021-04-18T19:58:45 | 2021-04-18T19:58:45 | {} | MEMBER |
ClickHouse | ClickHouse | 822,071,582 | 23,257 | alexey-milovidov | If I got it right, could you please add a *very* descriptive comment into a test? | 2021-04-18T22:21:49 | 2021-04-18T22:22:22 | {} | MEMBER |
ClickHouse | ClickHouse | 822,518,089 | 23,305 | den-crane | It's not about multi-column.
The same issue with
```
create table test(d Date, k Int64, s String)
Engine=MergeTree partition by toYYYYMM(d) order by tuple()
``` | 2021-04-19T14:37:31 | 2021-04-19T14:37:31 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 822,573,165 | 23,313 | genzgd | Yes it does seem related, although thought #10368 issue explicitly references the situation where the part does not exist on any replicas. | 2021-04-19T15:46:07 | 2021-04-19T15:46:07 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 822,059,940 | 23,212 | azat | Also note that after #23211 merged perf tests started to fail with memory limit exceeded error (but for query), I will fix this, but let's decide with this first.
Do you think that there is no sense in this `max_bytes_to_use_mmap_io` knob? | 2021-04-18T20:56:09 | 2021-04-18T20:56:09 | {} | MEMBER |
ClickHouse | ClickHouse | 822,135,996 | 23,183 | meimeitou | > @roygbip Could you help me debugging this issue? It happens on specific cloud instance and on specific OS where I cannot obtain access. Maybe you can create a small temporary VM, temporary ssh keypair and send the address along with keypair to [clickhouse-feedback@yandex-team.ru](mailto:clickhouse-feedback@yandex-tea... | 2021-04-19T03:10:13 | 2021-04-19T03:10:13 | {} | NONE |
ClickHouse | ClickHouse | 822,188,194 | 21,904 | tlightsky | oh,
I'm using stable version before,
let me try with master | 2021-04-19T05:52:17 | 2021-04-19T05:52:17 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 822,500,471 | 23,302 | filimonov | Also have huge impact on `keeper-bench` build time.
```
ninja keeper-bench
before [6519/7905]...
after [255/255]
```
:facepalm: | 2021-04-19T14:14:32 | 2021-04-19T14:14:32 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 822,567,608 | 23,313 | tavplubix | Looks like duplicate of #10368 | 2021-04-19T15:38:38 | 2021-04-19T15:38:38 | {} | MEMBER |
ClickHouse | ClickHouse | 821,995,078 | 19,320 | alexey-milovidov | The drawback is that MySQL, Postgres have different lexers than ClickHouse.
And they are also different from each other.
Examples:
```
SELECT 'Hello\'
SELECT e'Hello\\'
$$ SELECT 1 $$
``` | 2021-04-18T13:52:53 | 2021-04-18T13:52:53 | {} | MEMBER |
ClickHouse | ClickHouse | 822,051,318 | 23,264 | azat | >I don't understand why do we need both text dump and filesystem snapshot.
I did it in the same way as other tables.
> Text dump should be enough.
I agree that one format is enough, but how about keeping filesystem snapshot instead of tsv?
Since the snapshot you can just copy and it will work, while tsv you n... | 2021-04-18T19:57:58 | 2021-04-18T19:57:58 | {} | MEMBER |
ClickHouse | ClickHouse | 822,073,639 | 23,255 | alexey-milovidov | Not important enough. | 2021-04-18T22:36:42 | 2021-04-18T22:36:42 | {} | MEMBER |
ClickHouse | ClickHouse | 822,122,687 | 22,242 | hexiaoting | > @hexiaoting can this pull request be reviewed ? Or it is still draft ?
This pr can be reviewed now. | 2021-04-19T02:21:02 | 2021-04-19T02:21:02 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 822,318,553 | 23,037 | mjh568412187 | > > Is it useful to integrate clickhouse with ldap in a production cluster?
>
> https://altinity.com/blog/integrating-clickhouse-with-ldap-part-one
> https://zhuanlan.zhihu.com/p/355554683
>
> I do have WeChat but did not use it for over a year - most likely it's frozen.
The version of clickhouse in the lates... | 2021-04-19T09:25:34 | 2021-04-19T09:25:34 | {} | NONE |
ClickHouse | ClickHouse | 822,462,074 | 23,264 | alexey-milovidov | +1, it's easy to add symbols in SELECT. | 2021-04-19T13:21:44 | 2021-04-19T13:21:50 | {} | MEMBER |
ClickHouse | ClickHouse | 822,479,967 | 23,159 | akuzm | Test failures are mostly unrelated. The test script changes exposed a broken perf test, I'll fix that separately. | 2021-04-19T13:47:09 | 2021-04-19T13:47:09 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 822,379,859 | 23,040 | sundy-li | Sorry to Interrupt this conversation, maybe we can have a sign in query_id for suggestion queries...
`clickhouse-client` is so powerful, if we want to make `clickhouse-client` work with other DBMS, we can know it's a suggestion query. | 2021-04-19T11:00:52 | 2021-04-19T11:00:52 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 822,389,703 | 23,305 | den-crane | the same with
```
create table test(d Date, k Int64, s String)
Engine=MergeTree partition by toYYYYMM(d) order by tuple()
``` | 2021-04-19T11:18:38 | 2021-04-19T11:21:44 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 822,388,591 | 19,649 | vdimir | I tested downloading different number of files with different settings.
Tests performed on localhost with minio, so throughput is high.
Also some cases compared with `s5cmd`.
Current implementation performs good for downloading one big file. Case that suffers from downloading each file is when we have many fil... | 2021-04-19T11:16:32 | 2021-04-19T11:56:38 | {
"rocket": 1
} | MEMBER |
ClickHouse | ClickHouse | 821,983,120 | 23,256 | alexey-milovidov | Keep in mind that `pthread_getname_np` is not signal safe. Instead a thread-local `const char *` variable can be used to stash and get the name. | 2021-04-18T12:20:51 | 2021-04-18T12:20:51 | {} | MEMBER |
ClickHouse | ClickHouse | 822,070,022 | 23,264 | alexey-milovidov | > I did it in the same way as other tables.
I remind it was for only one very special case when something was specific to query_log table. @alesapin maybe you know about it?
> trace_log size:
Let's leave tsv only. | 2021-04-18T22:09:18 | 2021-04-18T22:09:49 | {} | MEMBER |
ClickHouse | ClickHouse | 822,071,433 | 23,257 | alexey-milovidov | So, if we have 1000 shards and `max_distributed_connections` is 100 what exactly will happen with hedged requests enabled? What will be the number of threads created? 1000 or 100 or max_threads or 1?
I expect that `max_threads` will be created if `hedged_requests` are enabled and 100 threads will be created if `hedg... | 2021-04-18T22:20:41 | 2021-04-18T22:21:10 | {} | MEMBER |
ClickHouse | ClickHouse | 822,073,664 | 23,249 | alexey-milovidov | Not important enough. | 2021-04-18T22:36:54 | 2021-04-18T22:36:54 | {} | MEMBER |
ClickHouse | ClickHouse | 822,074,205 | 23,212 | alexey-milovidov | Either I don't understand the motivation or there is very little sense to add this feature.
> with multiple queries that just reads those parts the queries may fail due to memory limit exceeded (for total memory tracker)
Does not look as an issue. Memory is calculated in some reasonable way (even if it is calcula... | 2021-04-18T22:40:58 | 2021-04-18T22:40:58 | {} | MEMBER |
ClickHouse | ClickHouse | 822,173,887 | 23,264 | azat | >Let's leave tsv only.
I did it the same as it done for other tables:
- stateless: tsv only
- stress: snapshot only
I can convert everything to tsv only if that make sense. | 2021-04-19T05:19:46 | 2021-04-19T05:19:46 | {} | MEMBER |
ClickHouse | ClickHouse | 822,246,868 | 23,257 | azat | >This statement contradicts the changes in this PR or did I miss something?
The test was for max_distributed_connections does not work (#17848) even before async_socket_for_remote had been added.
So the test works incorrectly after, since with async_socket_for_remote max_distributed_connections is "ignored".
> S... | 2021-04-19T07:39:17 | 2021-04-19T07:39:43 | {} | MEMBER |
ClickHouse | ClickHouse | 822,386,789 | 23,305 | den-crane | @amosbird | 2021-04-19T11:13:04 | 2021-04-19T11:13:04 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 822,026,719 | 23,206 | den-crane | USER1 should be connected to the fast replica and use this fast replica as an initiator | 2021-04-18T17:18:32 | 2021-04-18T17:18:32 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 822,393,413 | 21,167 | KochetovNicolai | @chyy can you check if there are exceptions `Unknown packet` in logs? | 2021-04-19T11:25:58 | 2021-04-19T11:25:58 | {} | MEMBER |
ClickHouse | ClickHouse | 822,411,741 | 23,300 | filimonov | It will not work. Other replicas do fetch instead of merge
| 2021-04-19T12:00:13 | 2021-04-19T12:00:13 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 822,414,581 | 23,264 | akuzm | It's not very useful if you don't resolve and save the adresses at the same time. Something like this: https://github.com/ClickHouse/ClickHouse/blob/master/docker/test/performance-comparison/compare.sh#L321 | 2021-04-19T12:05:11 | 2021-04-19T12:05:11 | {
"+1": 1
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 822,441,073 | 23,293 | UnamedRus | BTW there is
Lazy database engine, which can be used for storing a lot of Log tables.
https://clickhouse.tech/docs/en/engines/table-engines/log-family/
https://clickhouse.tech/docs/en/engines/database-engines/lazy/ | 2021-04-19T12:50:32 | 2021-04-19T12:50:32 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 822,464,989 | 23,306 | cadet354 | в LTS версии 21.3.6.55 уже невалидные результаты, т.е. эти изменения были между 21.3.5.42 и 21.3.6.55 | 2021-04-19T13:26:01 | 2021-04-19T13:26:01 | {} | NONE |
ClickHouse | ClickHouse | 822,566,826 | 23,029 | alexey-milovidov | ```
SELECT isNull(c) OR COUNT(1) AS x FROM t GROUP BY c HAVING isNull(c)
UNION ALL SELECT 0 SETTINGS enable_optimize_predicate_expression = 0
``` | 2021-04-19T15:37:36 | 2021-04-19T15:37:48 | {} | MEMBER |
ClickHouse | ClickHouse | 822,233,775 | 19,816 | kitaisreal | @olgarev sure, done. | 2021-04-19T07:18:27 | 2021-04-19T07:18:27 | {
"+1": 1
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 822,389,493 | 23,299 | ruanwenjun | ```sql
CREATE TABLE test_map
(
`itemId` Int64,
`userId` Map[Int64,Int32],
`ts` DateTime('America/Phoenix')
)
ENGINE = SummingMergeTree
PARTITION BY toYYYYMMDD(ts)
ORDER BY (itemId, userId, ts)
SETTINGS index_granularity = 8192;
CREATE TABLE test_array
(
`itemId` Int64,
`... | 2021-04-19T11:18:16 | 2021-04-19T11:18:16 | {} | NONE |
ClickHouse | ClickHouse | 822,483,943 | 23,292 | alexey-milovidov | We've got confirmation from our developer that the issue exists and will be not hard to fix. | 2021-04-19T13:52:37 | 2021-04-19T13:52:37 | {
"+1": 1
} | MEMBER |
ClickHouse | ClickHouse | 822,524,854 | 23,199 | Jokser | @taiwhite
Thank you. I think such a result for 10k rows may be caused by in-memory parts optimization. Did you check it?
Could you please also attach CH logs of this benchmark execution for 10k and 20k rows? | 2021-04-19T14:45:53 | 2021-04-19T14:47:54 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 821,982,843 | 23,212 | alexey-milovidov | I cannot imagine what max limit should be taking into account the fact that min threshold is 1 GB and 150 GB is the max part size. | 2021-04-18T12:18:45 | 2021-04-18T12:18:45 | {} | MEMBER |
ClickHouse | ClickHouse | 822,073,648 | 23,254 | alexey-milovidov | Not important enough. | 2021-04-18T22:36:45 | 2021-04-18T22:36:45 | {} | MEMBER |
ClickHouse | ClickHouse | 822,073,654 | 23,253 | alexey-milovidov | Not important enough. | 2021-04-18T22:36:49 | 2021-04-18T22:36:49 | {} | MEMBER |
ClickHouse | ClickHouse | 822,073,854 | 22,774 | alexey-milovidov | Does not look like it needs to be backported. | 2021-04-18T22:38:25 | 2021-04-18T22:38:32 | {} | MEMBER |
ClickHouse | ClickHouse | 822,110,744 | 23,183 | meimeitou | > The issue does not reproduce on Amazon Linux (AArch64, Graviton2):
>
> ```
>
> __| __|_ )
> _| ( / Amazon Linux 2 AMI
> ___|\___|___|
>
> https://aws.amazon.com/amazon-linux-2/
> No packages needed for security; 2 packages available
> Run "sudo yum update" to apply all update... | 2021-04-19T01:34:56 | 2021-04-19T01:34:56 | {} | NONE |
ClickHouse | ClickHouse | 822,111,362 | 23,102 | nikitamikhaylov | @Mergifyio rebase
| 2021-04-19T01:37:24 | 2021-04-19T01:37:24 | {
"+1": 1
} | MEMBER |
ClickHouse | ClickHouse | 822,136,733 | 23,183 | meimeitou | > @roygbip Could you help me debugging this issue? It happens on specific cloud instance and on specific OS where I cannot obtain access. Maybe you can create a small temporary VM, temporary ssh keypair and send the address along with keypair to [clickhouse-feedback@yandex-team.ru](mailto:clickhouse-feedback@yandex-tea... | 2021-04-19T03:12:50 | 2021-04-19T03:12:50 | {} | NONE |
ClickHouse | ClickHouse | 822,229,896 | 23,261 | azat | >Isn't the issue fixed, can we close it?
Let's wait for the answer in that issue. | 2021-04-19T07:11:57 | 2021-04-19T07:11:57 | {} | MEMBER |
ClickHouse | ClickHouse | 822,337,768 | 23,300 | filimonov | with that feature enabled clickhouse decides which replica SHOULD do the merge (and which one we should wait for) based on the number of currently active replicas.
```
replica_which_should_do_the_merges = active_replicas[ hash(part_name) % number_of_active_replicas ]
```
If the number of replicas changes differ... | 2021-04-19T09:52:57 | 2021-04-19T09:58:24 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 822,380,927 | 23,299 | l1t1 | please post your test | 2021-04-19T11:02:39 | 2021-04-19T11:02:39 | {} | NONE |
ClickHouse | ClickHouse | 822,467,767 | 23,040 | alexey-milovidov | @sundy-li Yes, it's easy. See the `log_comment` setting. We can set it for suggest queries to the value like `Query for suggestions`.
This will be backward compatible, because old servers (for about a year) will ignore unknown setting. | 2021-04-19T13:29:59 | 2021-04-19T13:29:59 | {} | MEMBER |
ClickHouse | ClickHouse | 822,480,018 | 23,020 | mjh568412187 | > What version of ClickHouse do you use?
I also encountered the same problem
I am executing optimize table xxx on cluster yyyy and truncate table xxx on cluster yyyy error: Caused by: java.lang.Throwable: Code: 48, e.displayText() = DB::Exception: There was an error on [yh- clickhouse-host06:9000]: Cannot execute r... | 2021-04-19T13:47:13 | 2021-04-19T13:47:13 | {} | NONE |
ClickHouse | ClickHouse | 822,372,350 | 9,195 | JackFielding | If inserting as one the TSV, TKSV, CSV, and JSONEachRow formats. We found that setting `input_format_parallel_parsing=0` was needed, otherwise `SELECT count() FROM myTable` did not return 0 if the insert was killed part way through.
https://clickhouse.tech/docs/en/operations/settings/settings/#input-format-parallel-... | 2021-04-19T10:47:19 | 2021-04-19T10:47:19 | {} | NONE |
ClickHouse | ClickHouse | 821,982,897 | 23,212 | alexey-milovidov | Do you have any motivating use cases? | 2021-04-18T12:19:07 | 2021-04-18T12:19:07 | {} | MEMBER |
ClickHouse | ClickHouse | 822,058,925 | 23,261 | azat | >and the potential issue found here #23257 will be investigated.
Note, that the test fails not because of async_socket_for_remote or use_hedged_requests.
And the only issue that I'm aware of for async_socket_for_remote right now is that it does not respect max_distributed_connection setting (because of the nature... | 2021-04-18T20:49:12 | 2021-04-18T20:49:12 | {} | MEMBER |
ClickHouse | ClickHouse | 822,112,501 | 23,102 | mergify[bot] | **Command `rebase`: success**
> **Branch has been successfully rebased**
>
| 2021-04-19T01:42:11 | 2021-04-19T01:42:11 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 822,229,466 | 21,167 | azat | >Let's redirect that question to @chyy - can you please try the 21.3.7?
@chyy But you definitely should try 21.3.7, since it contains #21643 (and a follow up #22183) | 2021-04-19T07:11:12 | 2021-04-19T07:11:12 | {} | MEMBER |
ClickHouse | ClickHouse | 822,391,054 | 21,167 | KochetovNicolai | addr2line
```
4ea5 0x7f454dc1d8dd
0x000000000f7ea942
DB::BufferBase::Buffer::end() const
./obj-x86_64-linux-gnu/../src/IO/BufferBase.h:40
DB::readVarUInt(unsigned long&, DB::ReadBuffer&)
./obj-x86_64-linux-gnu/../src/IO/VarInt.h:146
DB::Connection::receivePacket()
./obj-x86_64-linux-gnu/../src/Client/Connectio... | 2021-04-19T11:21:14 | 2021-04-19T11:21:14 | {} | MEMBER |
ClickHouse | ClickHouse | 822,403,975 | 23,305 | den-crane | ```
SELECT * FROM test WHERE d != materialize(toDate('2020-01-01')) ;
┌──────────d─┬─k─┬─s─┐
│ 2020-01-02 │ 1 │ │
└────────────┴───┴───┘
``` | 2021-04-19T11:45:58 | 2021-04-19T11:45:58 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 822,413,324 | 23,212 | akuzm | Why are mmapped files counted towards memory usage at all? This looks like the core problem. mmapping a file doesn't use up the available memory (RAM+swap). | 2021-04-19T12:03:02 | 2021-04-19T12:03:02 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 822,477,245 | 23,306 | cadet354 | проблема оказалась не в arrayFilter, а в функции Date, т.е. замена на step1_time > '1970-01-01 00:00:00' помогла. | 2021-04-19T13:43:29 | 2021-04-19T13:43:29 | {} | NONE |
ClickHouse | ClickHouse | 822,500,900 | 23,309 | KochetovNicolai | Test give similar crash:
```
2021.04.19 17:11:01.053397 [ 30338 ] {} <Trace> BaseDaemon: Received signal 11
2021.04.19 17:11:01.053697 [ 30640 ] {} <Fatal> BaseDaemon: ########################################
2021.04.19 17:11:01.053842 [ 30640 ] {} <Fatal> BaseDaemon: (version 21.5.1.1, build id: 105A4D93D779F2EFD6... | 2021-04-19T14:15:09 | 2021-04-19T14:15:09 | {} | MEMBER |
ClickHouse | ClickHouse | 822,584,504 | 23,261 | alexey-milovidov | @azat But also you have noticed this:
```
too huge delay, look like another issue with async_socket_for_remote/use_hedged_requests
```
So, we should suspect another issue? | 2021-04-19T16:00:49 | 2021-04-19T16:00:49 | {} | MEMBER |
ClickHouse | ClickHouse | 822,015,648 | 23,206 | UnamedRus | > Ability to connect to replica-sets depending on user profile. (Imagine we have 20 shards * 3 replica. 10 Super fast replica & 10+10 slow. And we want to route real-time queries from USER1 to the fast replicas and slow analytical queries from USER2, USER3 route to the slow replica-sets).
Probably it can be done on... | 2021-04-18T16:04:37 | 2021-04-18T16:06:28 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 822,072,616 | 23,266 | alexey-milovidov | https://github.com/ClickHouse/ClickHouse/pull/23275 | 2021-04-18T22:29:44 | 2021-04-18T22:29:44 | {} | MEMBER |
ClickHouse | ClickHouse | 822,076,538 | 17,186 | alexey-milovidov | @MikageSawatari If query is fast enough, `X-ClickHouse-Progress` is not returned, because there was no time to return progress.
Progress is returned not more frequently than `http_headers_progress_interval_ms`, default is 100ms and not more frequently than for every processed block of data.
If query is very quick... | 2021-04-18T22:55:22 | 2021-04-18T22:55:22 | {} | MEMBER |
ClickHouse | ClickHouse | 822,077,364 | 17,186 | alexey-milovidov | I see that this example shows the issue:
```
$ curl -v 'http://ymirlink:xxxx@localhost:8123/?send_progress_in_http_headers=1&http_headers_progress_interval_ms=100&query=SELECT+max(hex(SHA256(toString(number))))+FROM+(SELECT+number+FROM+system.numbers+LIMIT+10000000)+FORMAT+CSV'
* Trying 127.0.0.1:8123...
* TCP_NO... | 2021-04-18T23:01:00 | 2021-04-18T23:01:00 | {} | MEMBER |
ClickHouse | ClickHouse | 822,077,417 | 17,186 | alexey-milovidov | On current master version there is no such issue:
```
$ curl -v 'http://localhost:8123/?send_progress_in_http_headers=1&http_headers_progress_interval_ms=100&query=SELECT+max(hex(SHA256(toString(number))))+FROM+(SELECT+number+FROM+system.numbers+LIMIT+10000000)+FORMAT+CSV'
* Trying 127.0.0.1:8123...
* TCP_NODELAY... | 2021-04-18T23:01:21 | 2021-04-18T23:01:21 | {} | MEMBER |
ClickHouse | ClickHouse | 822,106,129 | 23,102 | godliness | > I think we should understand why tests with Kafka are broken and fix them.
@nikitamikhaylov
Kafka storage related test failed has been fixed, it's due to i create the `file_reader` before ClickHouse read the kafka messages.
I think the rest of test cases failed are not related to this PR, PTAL.
Thanks.
| 2021-04-19T01:15:43 | 2021-04-19T01:18:32 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 822,406,519 | 23,200 | tavplubix | AST fuzzer (MSan) - `==43==ERROR: MemorySanitizer: allocator is out of memory trying to allocate 0x2000 bytes`
Functional stateless tests (thread) - race in ZlibDeflatingWriteBuffer, already fixed in master
Integration tests - test_storage_hdfs - unrelated
SQLancer test - looks like #23029
Yandex synchronization ch... | 2021-04-19T11:50:40 | 2021-04-19T11:50:40 | {} | MEMBER |
ClickHouse | ClickHouse | 822,472,063 | 23,197 | nikitamikhaylov | @Mergifyio rebase
| 2021-04-19T13:36:11 | 2021-04-19T13:36:11 | {
"+1": 1
} | MEMBER |
ClickHouse | ClickHouse | 822,481,009 | 23,201 | akuzm | Test failures are unrelated. | 2021-04-19T13:48:39 | 2021-04-19T13:48:39 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 822,071,175 | 23,257 | alexey-milovidov | @azat
> Note, that the test fails not because of async_socket_for_remote or use_hedged_requests.
This statement contradicts the changes in this PR or did I miss something? | 2021-04-18T22:18:29 | 2021-04-18T22:18:29 | {} | MEMBER |
ClickHouse | ClickHouse | 822,109,982 | 23,183 | meimeitou | > What servers do you use to test these versions?
Huawei server (aarch64)
```shell
[root@Malluma ~]# cat /proc/cpuinfo | grep name | cut -f2 -d: | uniq -c
128 HUAWEI,Kunpeng 920
[root@Malluma ~]# dmidecode | grep "Product Name"
Product Name: TaiShan 200 (Model 2280)
Product Name: BC82AMDD
[root@M... | 2021-04-19T01:31:50 | 2021-04-19T01:31:50 | {} | NONE |
ClickHouse | ClickHouse | 822,224,991 | 21,167 | azat | >it seems the bug is still exists after upgrade
@chyy can you please provide the following?
- your linux kernel version (although unlikely this is related, since the only problem I'm aware of is query hangs not failures - #22109)
- and full query log for this query (using `send_logs_level=trace` or if it will no... | 2021-04-19T07:03:28 | 2021-04-19T07:03:28 | {} | MEMBER |
ClickHouse | ClickHouse | 822,335,955 | 23,076 | kozlek | Hello ! 🙂
I'm also experiencing some issues with Decimal and Nullable(Decimal).
`SELECT toDecimal64OrNull('TEST', 4)` -> NULL
which is the expected behaviour
`SELECT toDecimal64OrNull('TEST', 4), 42 / toDecimal64OrNull('TEST', 4);` -> Division By Zero
which is not the expected behaviour, as `42 / NULL -> NUL... | 2021-04-19T09:50:34 | 2021-04-19T09:50:34 | {
"+1": 1
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 822,430,120 | 23,300 | Sallery-X | > It will not work. Other replicas do fetch instead of merge
I known that. And I think It works. Other replicas will fetch instead of merge, and ensure that the cache is consistent among the replicas. | 2021-04-19T12:32:26 | 2021-04-19T12:32:26 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 822,472,839 | 23,037 | alexey-milovidov | > https://packagecloud.io/Altinity/clickhouse
packagecloud/Altinity is unofficial ClickHouse project, it may lag behind on the latest patches and stable releases.
I recommend to install as on the website https://clickhouse.tech/#quick-start (CentOS or RedHat tab) | 2021-04-19T13:37:21 | 2021-04-19T13:37:21 | {
"+1": 1
} | MEMBER |
ClickHouse | ClickHouse | 810,783,908 | 21,956 | Zhile | @alexey-milovidov Can you help to build another release for 20.12? I tried 21.1, 21.2, 21.3 and there're still critical bugs which block our upgrade.
https://github.com/ClickHouse/ClickHouse/issues/22399
https://github.com/ClickHouse/ClickHouse/issues/20193 | 2021-03-31T05:46:33 | 2021-03-31T05:46:33 | {} | NONE |
ClickHouse | ClickHouse | 810,809,470 | 16,108 | amosbird | > I thought it's already fixed.
It should. | 2021-03-31T06:31:10 | 2021-03-31T06:31:10 | {} | COLLABORATOR |
ClickHouse | ClickHouse | 810,920,960 | 22,399 | Zhile | @vdimir Can we have a quick fix for this? 21.1 21.2 21.3 can't be used at current time. | 2021-03-31T09:29:01 | 2021-03-31T09:33:45 | {
"eyes": 1
} | NONE |
ClickHouse | ClickHouse | 810,929,590 | 21,770 | taylor12805 | > @taylor12805 do you need help?
I think i can handle it. Have no time last week, will continue this weekend | 2021-03-31T09:42:18 | 2021-03-31T09:42:18 | {
"+1": 2
} | CONTRIBUTOR |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.