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 | 789,466,303 | 21,067 | filimonov | @amosbird can you check it? Looks related to WITH changes. | 2021-03-03T06:14:34 | 2021-03-03T06:14:34 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 789,630,740 | 21,361 | cyberhuman | We did an upgrade today but AFAIK it didn't help. I will check in a while.
Is there anything else I could look at? | 2021-03-03T11:03:20 | 2021-03-03T11:03:20 | {} | NONE |
ClickHouse | ClickHouse | 789,676,579 | 21,361 | cyberhuman | I took a look and there is no improvement after the upgrade.
What I noticed by the way is that on the hosts in question and other hosts with the same issue there are too many parts (tens or hundreds of thousands). The number slowly goes down as (or if) they're merged. But the number of ongoing merges is very low (1 or... | 2021-03-03T12:22:17 | 2021-03-03T12:22:17 | {} | NONE |
ClickHouse | ClickHouse | 789,740,906 | 18,978 | tavplubix | > > Possibly we should change Changelog category to "Backward incompatible" and add a note that ATTACH PART[ITION] queries may not work during cluster upgrade.
>
> @tavplubix The proposed solution doesn't change the way the replicas process the commands if the data in detached/ is missing, I don't see the reason why... | 2021-03-03T14:10:22 | 2021-03-03T14:10:22 | {} | MEMBER |
ClickHouse | ClickHouse | 790,268,922 | 21,361 | cyberhuman | I have enabled debug logging and it looks like the replication queue entries are retried but always fail. And the reason for the failure is missing parts. What's worse, those parts have been removed from the replica that had them. Even the 15-4-0_3021_3021_0 mentioned in the beginning of the report was removed on the s... | 2021-03-04T03:51:28 | 2021-03-04T03:52:21 | {} | NONE |
ClickHouse | ClickHouse | 789,311,775 | 21,390 | den-crane | it never worked. Not a bug.
Probably related to issue that aliases are not supported in views. | 2021-03-02T23:59:38 | 2021-03-02T23:59:50 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 789,351,839 | 18,578 | alexey-milovidov | Not relevant anymore. | 2021-03-03T01:22:50 | 2021-03-03T01:22:50 | {} | MEMBER |
ClickHouse | ClickHouse | 789,362,954 | 21,388 | alexey-milovidov | Need to construct a reproducer. | 2021-03-03T01:50:53 | 2021-03-03T01:50:53 | {} | MEMBER |
ClickHouse | ClickHouse | 789,902,495 | 21,376 | excitoon | @pumelo I mean, what do you have in log before this line? What is a logging level? | 2021-03-03T17:18:34 | 2021-03-03T17:18:34 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 789,292,845 | 21,239 | alexey-milovidov | It's not trivial to support global dictionary for Replicated tables since data can appear on servers in different order - it means that dictionary ids will be also selected in different way on replicas and replicated data will be not byte-identical. | 2021-03-02T23:16:15 | 2021-03-02T23:16:15 | {} | MEMBER |
ClickHouse | ClickHouse | 789,349,028 | 21,361 | cyberhuman | Thank you for looking into it!
> So you rewrite 100 partitions fully every minute (sounds bad, maybe at least remove the partitioning, I guess the table is quite small).
We're not rewriting, we're appending 🤔
The partitioning must be there because of the way the data will be queried later.
We'll try to upgr... | 2021-03-03T01:15:22 | 2021-03-03T01:15:39 | {} | NONE |
ClickHouse | ClickHouse | 789,621,986 | 21,411 | filimonov | It was already removed. See https://github.com/ClickHouse/ClickHouse/pull/11637 :) | 2021-03-03T10:49:09 | 2021-03-03T10:49:09 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 789,694,082 | 21,381 | akuzm | The failures are from master. | 2021-03-03T12:54:17 | 2021-03-03T12:54:17 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 789,710,702 | 21,376 | excitoon | @pumelo Can you please provide more detailed log? | 2021-03-03T13:23:12 | 2021-03-03T13:23:12 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 789,932,578 | 21,087 | alexey-milovidov | Yes, the basic rule - leave only what is relevant for the tests. | 2021-03-03T17:51:02 | 2021-03-03T17:51:02 | {} | MEMBER |
ClickHouse | ClickHouse | 789,558,518 | 21,233 | alesapin | No related failures | 2021-03-03T09:08:23 | 2021-03-03T09:08:23 | {} | MEMBER |
ClickHouse | ClickHouse | 789,579,457 | 20,315 | qoega | Is it related?
https://clickhouse-test-reports.s3.yandex.net/0/a1424bc6750826250786071b2c80dfb44c07d164/fuzzer_debug/report.html#fail1
``` sql
SELECT 9223372036854775807, materialize(js2.k), NULL FROM (SELECT toLowCardinality(number) AS k, 1048576, number + NULL FROM numbers(1048577)) AS js1 FULL OUTER JOIN (SEL... | 2021-03-03T09:40:39 | 2021-03-03T09:40:59 | {} | MEMBER |
ClickHouse | ClickHouse | 789,601,893 | 21,406 | alex-zaitsev | cc @hexiaoting | 2021-03-03T10:16:14 | 2021-03-03T10:16:14 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 789,745,659 | 21,395 | Sallery-X | > > conflict
>
> Everything will be ok. See: #11637
>
> You can explain it like that: leader now do an attempt to put the merge task in the queue. If in meanwhile queue changed he will remove the task, will reread the queue, and will do another attempt with respect to the changed queue state.
the code follows ... | 2021-03-03T14:17:20 | 2021-03-03T14:17:20 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 789,818,717 | 20,608 | Exus | I still cannot reproduce crashes on my test environment (4 servers) - two servers 21.2.4.18+asan and two old 21.2.4.6, all replicating same tables. I've added some data (2bln rows with 2k/s ingest), and it's stable.
Only crash I found is clickhouse-client on asan servers (cannot reproduce on other versions): when I ... | 2021-03-03T15:56:16 | 2021-03-03T15:56:38 | {} | NONE |
ClickHouse | ClickHouse | 790,209,575 | 21,393 | alexey-milovidov | @kitaisreal Bad cast is possible in `dictGet` function. | 2021-03-04T01:25:22 | 2021-03-04T01:25:22 | {} | MEMBER |
ClickHouse | ClickHouse | 789,285,348 | 14,740 | alexey-milovidov | I confirm, it's fixed. | 2021-03-02T23:00:31 | 2021-03-02T23:00:31 | {} | MEMBER |
ClickHouse | ClickHouse | 789,375,266 | 21,359 | haoch | @yiguolei Glad to know that you are interested.
We may support multiple bucket distribution strategies, here explain with a very simple example.
> 1. how to rebalance data when new node is added to the cluster?
It could assign partition (say partition key is `date` and split into N buckets) to 1 of the virtua... | 2021-03-03T02:20:40 | 2021-03-03T02:49:24 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 789,495,385 | 21,239 | Teeeto | It might require shipping new dictionary entries together with insert data to replicas then?
If it is common to insert to different replicas, it might be possible to reserve some bits of 64-bit ID as a replica ID ? | 2021-03-03T07:14:40 | 2021-03-03T07:39:07 | {} | NONE |
ClickHouse | ClickHouse | 789,621,372 | 21,395 | filimonov | > conflict
Everything will be ok. See: https://github.com/ClickHouse/ClickHouse/pull/11637
You can explain it like that: leader now do an attempt to put the merge task in the queue. If in meanwhile queue changed he will remove the task, will reread the queue, and will do another attempt with respect to the change... | 2021-03-03T10:48:05 | 2021-03-03T10:48:05 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 789,704,266 | 21,404 | smagellan | max_block_size option maybe + take a look at jdbc driver docs: https://github.com/ClickHouse/clickhouse-jdbc. I know nothing about mybatis, though | 2021-03-03T13:11:56 | 2021-03-03T13:11:56 | {} | NONE |
ClickHouse | ClickHouse | 789,743,600 | 18,978 | tavplubix | Also it's not possible to rollback to older ClickHouse version after executing `ALTER ... ATTACH` query in new version. | 2021-03-03T14:14:28 | 2021-03-03T14:14:28 | {} | MEMBER |
ClickHouse | ClickHouse | 789,772,432 | 18,610 | filimonov | > Ah, right. Though something was the reason of
>
> > DB::Exception: Cannot parse input: expected 'quorum: ' before: 'merge_type: 2\n'
>
> some structure somewhere must have changed so those versions were incompatible.
Caused by https://github.com/ClickHouse/ClickHouse/pull/14490/ | 2021-03-03T14:55:49 | 2021-03-03T14:55:49 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 789,794,599 | 21,376 | pumelo | @excitoon how can I get more detailed log? | 2021-03-03T15:23:32 | 2021-03-03T15:23:32 | {} | NONE |
ClickHouse | ClickHouse | 789,953,423 | 21,403 | den-crane | ```sql
SELECT
co,
arraySlice(arrayReverseSort(i -> (i.2), groupArray((id, v))), 1, 2)
FROM
(
SELECT
id,
vv.1 AS co,
vv.2 AS v
FROM
(
SELECT
id,
arrayJoin([('v1', v1), ('v2', v2), ('v3', v3)]) AS vv
FROM
(
... | 2021-03-03T18:21:03 | 2021-03-03T18:21:03 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 789,288,285 | 21,358 | alexey-milovidov | @amosbird We already have it, the function is named `identity`. This function does nothing and returns its argument, but it inhibits index analysis:
```
SELECT ... WHERE identity(x = 1)
``` | 2021-03-02T23:07:09 | 2021-03-02T23:07:09 | {
"+1": 1
} | MEMBER |
ClickHouse | ClickHouse | 789,363,114 | 21,388 | alexey-milovidov | The fix is probably just to obtain a global context in dictionaries. | 2021-03-03T01:51:22 | 2021-03-03T01:51:22 | {} | MEMBER |
ClickHouse | ClickHouse | 789,387,357 | 21,329 | depressed-pho | > Please take a look at the reports of fuzzers.
Yes, it was an actual bug in my code. It should be fixed now. | 2021-03-03T02:52:20 | 2021-03-03T02:52:20 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 789,453,162 | 21,389 | filimonov | W/o: this way will work:
```sql
CREATE TABLE y2 AS x
ENGINE = MergeTree
ORDER BY number
SETTINGS index_granularity = 1
```
In your example, it was parsed as query level not a table-level settings.
There are also similar usability issues with
```sql
INSERT ... SELECT ... SETTINGS ... -- SETTINGS affec... | 2021-03-03T05:47:57 | 2021-03-03T05:47:57 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 789,493,990 | 21,334 | alesapin | `test_distributed_ddl` -- broken in master | 2021-03-03T07:12:19 | 2021-03-03T07:12:19 | {} | MEMBER |
ClickHouse | ClickHouse | 789,854,316 | 21,101 | asafcombo | @den-crane apologies I am trying to replicate a complex table and didn't overview the queries.
The below will work:
```
CREATE TABLE ip_test
(
`timestamp` DateTime Codec (DoubleDelta, LZ4),
`ip` String CODEC (LZ4),
`is_ipv6` ... | 2021-03-03T16:36:30 | 2021-03-03T16:36:30 | {} | NONE |
ClickHouse | ClickHouse | 790,033,211 | 21,395 | filimonov | > if (!is_leader) return;
It can happen if you use special setting which forbids for some replicas to be leaders. | 2021-03-03T20:28:00 | 2021-03-03T20:28:00 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 790,256,726 | 21,395 | Sallery-X | > > if (!is_leader) return;
>
> It can happen if you use special setting which forbids for some replicas to be leaders.
But I check the zookeeper log found that almost only one replica put the merge task! my cluster is 1 shard 15 replicas
| 2021-03-04T03:19:06 | 2021-03-04T03:19:06 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 790,311,423 | 19,673 | azat | >Integration tests (thread) — fail: 2, passed: 1149, error: 1
Test failures looks unrelated (and those tests are flaky)
>Performance — 7 errors, 7 faster, 1 slower, 53 unstable
Same issue in upstream/master
> Marker check Pending — Waiting to start Functional stateful tests (release, DatabaseOrdinary), Func... | 2021-03-04T05:46:27 | 2021-03-04T05:46:27 | {} | MEMBER |
ClickHouse | ClickHouse | 789,350,287 | 19,229 | alexey-milovidov | Sometimes we see limited scaling for the following reasons:
1. The working set of data in every thread is larger than L2 cache size, and for all threads it is larger than L3 cache size, and the bottleneck is in memory bandwidth.
Possible solution: lower `max_block_size` or `preferred_block_size_bytes`, lower `max_c... | 2021-03-03T01:18:44 | 2021-03-03T01:19:58 | {} | MEMBER |
ClickHouse | ClickHouse | 789,403,988 | 20,463 | amosbird | Superceded by https://github.com/ClickHouse/ClickHouse/issues/21364 | 2021-03-03T03:40:02 | 2021-03-03T03:40:02 | {} | COLLABORATOR |
ClickHouse | ClickHouse | 789,834,164 | 21,415 | gpmcadam | I'm running macOS as the host, but the target is Ubuntu ARM64. | 2021-03-03T16:17:09 | 2021-03-03T16:17:09 | {} | NONE |
ClickHouse | ClickHouse | 790,369,424 | 21,402 | mjh568412187 | hello
yhyc-clickhouse-05 :) select count(*) from system.parts;
SELECT count(*)
FROM system.parts
┌─count()─┐
│ 136622 │
└─────────┘
1 rows in set. Elapsed: 4.690 sec. Processed 68.31 thousand rows, 35.01 MB (14.57 thousand rows/s., 7.47 MB/s.)
At that time, the log of clickhouse-server.log_bak202102... | 2021-03-04T06:55:28 | 2021-03-04T06:55:28 | {} | NONE |
ClickHouse | ClickHouse | 789,275,582 | 11,003 | sag1 | Good, but I found latest version as 20.12.8.5 on docker hub `docker pull yandex/clickhouse-server:20.12.8.5-alpine`
Where is hosted version 21.2? | 2021-03-02T22:39:30 | 2021-03-02T22:42:25 | {} | NONE |
ClickHouse | ClickHouse | 789,573,475 | 21,067 | adrian17 | For the record, after some experimenting I found out that even without `WITH`, adding `JOIN`s appears to slowly but quadratically increase query planning time - with 12 identical nontrivial subquery JOINs I got planning time ~1s, with 20 JOINs it reached >4s. But while this was an artificial edge case, apparently addin... | 2021-03-03T09:30:45 | 2021-03-03T09:30:45 | {} | NONE |
ClickHouse | ClickHouse | 789,603,474 | 21,406 | hexiaoting | I will try to fix this. | 2021-03-03T10:18:44 | 2021-03-03T10:18:44 | {
"+1": 2
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 789,705,161 | 21,415 | filimonov | Duplicate of https://github.com/ClickHouse/ClickHouse/issues/21191 | 2021-03-03T13:13:24 | 2021-03-03T13:13:24 | {
"-1": 1
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 789,707,039 | 21,415 | gpmcadam | To clarify, this is not building on macOS but on arm64 in Docker/Linux, so I don’t think its a duplicate as described | 2021-03-03T13:16:46 | 2021-03-03T13:16:46 | {} | NONE |
ClickHouse | ClickHouse | 789,822,665 | 21,101 | den-crane | ```
CREATE TABLE ip_test
(
`timestamp` DateTime CODEC(DoubleDelta, LZ4),
`ip` String CODEC(LZ4),
`is_ipv6` UInt8 MATERIALIZED isIPv6String(ip),
`as_name` LowCardinality(String) MATERIALIZED dictGet('tests.ip_prefix_dict', 'as_name', tuple(if(is_ipv6, toIPv6(ip), toIPv4(ip))))
)
ENGINE = MergeTre... | 2021-03-03T16:01:29 | 2021-03-03T16:01:29 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 789,937,757 | 21,413 | den-crane | Do you use Engine=Merge / Engine=Distributed?
Try to replace WHERE with PREWHERE in your test-case.
https://github.com/ClickHouse/ClickHouse/issues/6143
20.3 -- support ends by the end of March. | 2021-03-03T17:58:14 | 2021-03-03T17:58:34 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 790,230,012 | 21,362 | hssrs | I solved the problem and came back here to share my solution. @nikitamikhaylov, Thank you for your help :)
The problem is caused by network. If you have a proxy configured for Docker as I do, It may help to change the proxy or simply pull the images manually (You can find the required image from `$ClickHouse/docker/t... | 2021-03-04T02:13:07 | 2021-03-04T02:13:07 | {} | NONE |
ClickHouse | ClickHouse | 789,404,165 | 21,246 | ucasfl | I think this pr should be backport, it fixed some union related problems. @KochetovNicolai @alexey-milovidov | 2021-03-03T03:40:41 | 2021-03-03T03:40:41 | {} | COLLABORATOR |
ClickHouse | ClickHouse | 789,572,802 | 11,003 | sag1 | Great ! Thanks @alexey-milovidov | 2021-03-03T09:29:41 | 2021-03-03T09:29:41 | {} | NONE |
ClickHouse | ClickHouse | 789,589,861 | 13,757 | qoega | Still flaky https://clickhouse-test-reports.s3.yandex.net/0/bd7b540b8255475ad584e814a8eeeb1386645fda/functional_stateless_tests_(release).html#fail1 | 2021-03-03T09:57:16 | 2021-03-03T09:57:16 | {} | MEMBER |
ClickHouse | ClickHouse | 789,770,887 | 21,101 | asafcombo | @filimonov can you reopen?
reproduced in `21.2.4.6`
trace
```
. DB::FunctionDictHelper::getDictionaryStructure(std::__1::basic_string<char, std::__1::char_traits<char>, std::__1::allocator<char> > const&) const @ 0xa4030b3 in /usr/bin/clickhouse1. DB::FunctionDictGetNoType<(DB::DictionaryGetFunctionType)0>::ge... | 2021-03-03T14:53:51 | 2021-03-03T14:54:37 | {} | NONE |
ClickHouse | ClickHouse | 789,771,933 | 18,609 | filimonov | Caused by https://github.com/ClickHouse/ClickHouse/pull/14490/ | 2021-03-03T14:55:10 | 2021-03-03T14:55:10 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 790,052,006 | 21,433 | azat | >But why it will print DB::Exception instead of DB::ParsingException ?
I guess because it catches only Exception, not ParsingException, for example here - https://github.com/ClickHouse/ClickHouse/blob/e8df9971f14880729fe20336388ea09e4dc44712/src/Server/TCPHandler.cpp#L327
First it prints ParsingException and afte... | 2021-03-03T21:01:30 | 2021-03-03T21:01:30 | {} | MEMBER |
ClickHouse | ClickHouse | 789,495,410 | 21,334 | alesapin | Performance -- no visible changes. | 2021-03-03T07:14:43 | 2021-03-03T07:14:43 | {} | MEMBER |
ClickHouse | ClickHouse | 790,035,711 | 19,673 | azat | Rebased again (to fix constant conflicts in ErrorCodes) | 2021-03-03T20:32:17 | 2021-03-03T20:32:17 | {} | MEMBER |
ClickHouse | ClickHouse | 790,046,221 | 21,433 | nikitamikhaylov | But why it will print DB::Exception instead of DB::ParsingException ? | 2021-03-03T20:51:08 | 2021-03-03T20:51:08 | {} | MEMBER |
ClickHouse | ClickHouse | 789,266,904 | 11,003 | alexey-milovidov | It should be already available since version 21.2. | 2021-03-02T22:29:43 | 2021-03-02T22:29:43 | {} | MEMBER |
ClickHouse | ClickHouse | 789,284,044 | 21,329 | alexey-milovidov | ```
WITH
CAST('127.0.0.1', 'Nullable(String)') AS addr,
arrayJoin([NULL, NULL, NULL, NULL]) AS prefix
SELECT
'1\07.0.0.0/8127.0.0.0/8127.0.0.0/8127.0.0.0/8127.0.0.0/8127.0.0.0/8127.0.0.0/8127.0.0.0/81\07.0.0.0/8127.0.0.0/8127.0.0.0/8127.0.0.0/8127.0.0.0/8127.0.0.0/8127.0.0.0/8127.0.0.0/8',
isIPAdd... | 2021-03-02T22:57:32 | 2021-03-02T22:57:32 | {} | MEMBER |
ClickHouse | ClickHouse | 789,288,946 | 21,103 | alexey-milovidov | Q: How to remove TTL in ClickHouse version 20.8?
A: It takes two steps. Step 1: Upgrade to ClickHouse version 21.2 or newer. Step 2: Remove TTL with ALTER TABLE ... REMOVE TTL query. | 2021-03-02T23:08:33 | 2021-03-02T23:09:27 | {} | MEMBER |
ClickHouse | ClickHouse | 789,339,540 | 11,003 | alexey-milovidov | I checked, it's alright:
`docker run --rm --network=host yandex/clickhouse-server:latest`
downloads version 21.2.5 - the latest stable release with all patches. | 2021-03-03T00:54:58 | 2021-03-03T00:55:05 | {
"+1": 1
} | MEMBER |
ClickHouse | ClickHouse | 789,376,951 | 21,358 | amosbird | Then I guess it's better to just use `identity` to ignore certain skip indices manually @mzealey | 2021-03-03T02:25:08 | 2021-03-03T02:25:08 | {} | COLLABORATOR |
ClickHouse | ClickHouse | 789,515,199 | 21,395 | filimonov | Why do you think you need it? Since 20.5 all replicas are leaders (unless you explicitly forbid them to be a leader).
If you're using an older version there was a code for that (but it was quite fragile) | 2021-03-03T07:52:06 | 2021-03-03T07:53:45 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 789,273,636 | 9,596 | alexey-milovidov | Also it's a good idea to use `ColumnUnique` to implement generic version of function `transform`.
We can convert source array elements to ColumnUnique to represent them as a dictionary to offsets in the destination array.
Then we will use this dictionary for lookups and copy elements with `IColumn::insertFrom`.
... | 2021-03-02T22:35:56 | 2021-03-02T22:35:56 | {} | MEMBER |
ClickHouse | ClickHouse | 789,225,453 | 21,359 | haoch | @alexey-milovidov @amosbird Could you please help have a look? We would like to contribute the change but want to align if it's the right direction with community. Basically the code change should not be much but it will help to make `DistributedTable` sharding key feture more generic.
Our idea is similar with http... | 2021-03-02T21:22:13 | 2021-03-02T23:00:54 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 789,304,740 | 21,359 | haoch | @amosbird I see your proposal https://github.com/ClickHouse/ClickHouse/issues/21364, our approach maybe able to support similar effect but a lightweight solution which we name as `virtual_cluster` bucket distribution mode. | 2021-03-02T23:43:18 | 2021-03-02T23:44:14 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 789,896,986 | 21,385 | alexey-milovidov | I still have to find how to justify these changes... | 2021-03-03T17:10:56 | 2021-03-03T17:10:56 | {} | MEMBER |
ClickHouse | ClickHouse | 789,944,673 | 12,561 | alexey-milovidov | Or `evaluateConstantExpressionAsLiteral`. | 2021-03-03T18:08:17 | 2021-03-03T18:08:17 | {} | MEMBER |
ClickHouse | ClickHouse | 789,318,832 | 21,390 | knoguchi | I figured `WITH <id> AS <subquery expr>` is a different beast than `WITH <expression> AS <identifier>` altho the two share same syntax.
| 2021-03-03T00:16:40 | 2021-03-03T00:17:00 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 789,629,481 | 21,361 | filimonov | Please post an update after version upgrade | 2021-03-03T11:01:17 | 2021-03-03T11:01:17 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 789,853,135 | 18,978 | myrrc | So we decided to leave this ticket (and the corresponding RESTORE REPLICA) as a backward incompatible change, I'll update the description and the docs shortly. | 2021-03-03T16:34:49 | 2021-03-03T16:34:49 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 789,352,915 | 14,700 | alexey-milovidov | Still relevant. | 2021-03-03T01:25:23 | 2021-03-03T01:25:23 | {} | MEMBER |
ClickHouse | ClickHouse | 789,353,984 | 15,469 | alexey-milovidov | Yes, it works perfectly in all recent releases. | 2021-03-03T01:27:06 | 2021-03-03T01:27:06 | {} | MEMBER |
ClickHouse | ClickHouse | 789,457,532 | 21,364 | filimonov | Can you extend it a bit / explain usecases better? What tasks will be better solved using that approach (comparing to existing engine=Distributed)?
In that approach, user should know names of the servers/endpoints when creating tables. And schema is tightly coupled with the current cluster. For example that will le... | 2021-03-03T05:57:10 | 2021-03-03T05:57:10 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 789,524,519 | 21,348 | kitaisreal | Hello, @MaxJPowers, could you please provide some example. Hit rate is calculated for rows in cache dictionaries. | 2021-03-03T08:10:01 | 2021-03-03T08:10:01 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 789,771,843 | 17,183 | filimonov | Caused by https://github.com/ClickHouse/ClickHouse/pull/14490/ | 2021-03-03T14:55:04 | 2021-03-03T14:55:04 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 789,961,723 | 21,402 | den-crane | >local_ods.ods_order_db__t_trade_order_item: Renaming unexpected part .... to ignored_....
This is normal behavior.
>The set of parts restored in place of ... looks incomplete. There might or might not be a data loss. Suspicious parts
This is data loss because of power outage (linux cache was not written to ... | 2021-03-03T18:33:08 | 2021-03-03T18:33:08 | {
"+1": 1
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 790,191,763 | 15,486 | alexey-milovidov | We decided to limit the range with minimum year 1925 for now. | 2021-03-04T00:40:41 | 2021-03-04T00:40:41 | {} | MEMBER |
ClickHouse | ClickHouse | 789,274,463 | 14,740 | alexey-milovidov | I will add your test just in case. | 2021-03-02T22:37:19 | 2021-03-02T22:37:19 | {
"+1": 1
} | MEMBER |
ClickHouse | ClickHouse | 789,340,336 | 21,359 | yiguolei | @haoch I am very interested into the bucket mode. But I have some questions:
1. how to rebalance data when new node is added to the cluster?
2. Is user's data in different bucket on the same shard stored in the same local table?
3. Maybe there will be many entries since the bucket is at user level, am I right? | 2021-03-03T00:56:58 | 2021-03-03T00:56:58 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 789,551,429 | 21,395 | Sallery-X |
> Why do you think you need it? Since 20.5 all replicas are leaders (unless you explicitly forbid them to be a leader).
>
> If you're using an older version there was a code for that (but it was quite fragile)
If all replicas are leaders，every replica could excute merge plan, the parts that will be merged may... | 2021-03-03T08:57:09 | 2021-03-03T08:57:09 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 789,733,801 | 21,385 | alexey-milovidov | There is no visible difference in performance test: https://clickhouse-test-reports.s3.yandex.net/21385/aecdadd02e7921ce5cd06614cd94ca5798b92d60/performance_comparison/report.html#fail1 | 2021-03-03T14:00:02 | 2021-03-03T14:00:02 | {} | MEMBER |
ClickHouse | ClickHouse | 789,830,340 | 21,415 | den-crane | @gpmcadam So you cross-compiled on arm64 in Docker/Linux for arm64 macOS as the target? | 2021-03-03T16:11:55 | 2021-03-03T16:11:55 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 789,956,003 | 21,404 | den-crane | OOM in JAVA on in Clickhouse-server? | 2021-03-03T18:24:51 | 2021-03-03T18:24:51 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 789,988,418 | 21,402 | alexey-milovidov | > The set of parts restored in place of ... looks incomplete. There might or might not be a data loss. Suspicious parts
This is data loss because of power outage (linux cache was not written to disk).
If you have replicas unaffected by power outage, the missing data parts will be restored from them. | 2021-03-03T19:15:20 | 2021-03-03T19:15:20 | {} | MEMBER |
ClickHouse | ClickHouse | 790,269,147 | 20,182 | alexey-milovidov | #9404 is still in early stage. | 2021-03-04T03:52:10 | 2021-03-04T03:52:10 | {} | MEMBER |
ClickHouse | ClickHouse | 789,283,689 | 21,329 | alexey-milovidov | Please take a look at the reports of fuzzers. | 2021-03-02T22:56:46 | 2021-03-02T22:56:46 | {
"eyes": 1
} | MEMBER |
ClickHouse | ClickHouse | 789,338,980 | 11,003 | alexey-milovidov | 21.2 is in the list: https://hub.docker.com/r/yandex/clickhouse-server/tags?page=1&ordering=last_updated
I have no idea why latest points to some other version. | 2021-03-03T00:53:24 | 2021-03-03T00:53:24 | {} | MEMBER |
ClickHouse | ClickHouse | 789,626,240 | 20,608 | filimonov | @Exus any news here? | 2021-03-03T10:56:01 | 2021-03-03T10:56:01 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 789,702,600 | 9,900 | Avogar | Fixed in https://github.com/ClickHouse/ClickHouse/pull/19291. | 2021-03-03T13:09:06 | 2021-03-03T13:09:06 | {} | MEMBER |
ClickHouse | ClickHouse | 789,721,695 | 18,978 | myrrc | > Possibly we should change Changelog category to "Backward incompatible" and add a note that ATTACH PART[ITION] queries may not work during cluster upgrade.
@tavplubix The proposed solution doesn't change the way the replicas process the commands if the data in detached/ is missing, I don't see the reason why it ma... | 2021-03-03T13:41:34 | 2021-03-03T13:42:39 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 789,852,251 | 21,415 | den-crane | Ah, OK. It's not related to MacOS at all but CPU is arm64 (Apple M1). | 2021-03-03T16:33:44 | 2021-03-03T16:33:44 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 789,868,725 | 21,101 | den-crane | yes, I reproduced on 21.3
```
<Error> Application: DB::Exception: Dictionary `tests.ip_prefix_dict` not found: While processing cast(if(NOT (cast(if(ip LIKE '%:%', 1, 0)
```
BTW you can use `dictGetString` instead of `dictGet` -- works for me | 2021-03-03T16:46:30 | 2021-03-03T16:46:30 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 790,226,659 | 21,331 | godliness | @nikitamikhaylov
Seems like **Marker check** is pending. | 2021-03-04T02:04:56 | 2021-03-04T02:04:56 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 780,413,963 | 16,193 | tavplubix | Fuzzers - unrelated
Integration tests (thread) - test_storage_kafka - flacky test, unrelated
Functional stateless tests flaky check (address) - 01541_max_memory_usage_for_user - `Long test not marked as 'long'` - Looks like an issue with flacky check. Test time in flacky check is about 190 seconds, so it's considered... | 2021-02-17T09:10:51 | 2021-02-17T09:10:51 | {} | MEMBER |
ClickHouse | ClickHouse | 780,454,297 | 19,766 | vdimir | @achimbab
Sorry for long delay. Thank you for your PR!
First of all, I think that we need more details in description, about terminology and what do we call `event` and `chain`, how do we consider data in table (some special form of graph I suppose), maybe add some links to external resources about this concepts.
... | 2021-02-17T10:20:09 | 2021-02-17T10:20:09 | {} | MEMBER |
ClickHouse | ClickHouse | 780,575,861 | 20,585 | alesapin | No reconnects, no errors with ZK... why fast test fails https://gist.github.com/alesapin/6293fdb422961968c530cc34e8a9d926? https://clickhouse-test-reports.s3.yandex.net/20585/9396bae2e2051e2d50faa0d8c1005465171db481/fast_test.html#fail1 | 2021-02-17T14:01:02 | 2021-02-17T14:01:26 | {} | MEMBER |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.