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 | 736,179,441 | 17,622 | TCeason | @tavplubix All of the error is about `test_grpc_protocol`. I'm not sure is I make it error？ | 2020-12-01T02:40:55 | 2020-12-01T02:40:55 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 736,728,207 | 17,646 | azat | Rebased against upstream/master (to include #17658) | 2020-12-01T18:12:57 | 2020-12-01T18:12:57 | {} | MEMBER |
ClickHouse | ClickHouse | 735,947,883 | 17,638 | den-crane | Can you show `select version() ` ? | 2020-11-30T18:03:16 | 2020-11-30T18:03:16 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 736,241,381 | 2,203 | zhoudehui | > No.
>
> The reason is that the "left" subquery is not buffered in memory. It is processed in a streaming fashion with lookups in the right table. That's why we have to first make a lookup-optimized data structure for the "right" subquery and then proceed with streaming reading and joining of the "left" subquery. I... | 2020-12-01T05:58:16 | 2020-12-01T05:58:16 | {} | NONE |
ClickHouse | ClickHouse | 736,279,252 | 17,597 | alexey-milovidov | Did not help. | 2020-12-01T07:27:08 | 2020-12-01T07:27:08 | {} | MEMBER |
ClickHouse | ClickHouse | 736,404,898 | 16,925 | BaptisteBdn | Thanks @Amberon, it seems to work aswell with `GRANULARITY = 1` on my side.
However it would be better if we could choose another value, hope you guys can find a fix.
| 2020-12-01T10:52:03 | 2020-12-01T10:52:03 | {} | NONE |
ClickHouse | ClickHouse | 736,487,861 | 17,666 | alexey-milovidov | Worth to note that ClickHouse is successfully tested on both AVX2-capable and non-capable instances. | 2020-12-01T11:44:37 | 2020-12-01T11:44:37 | {} | MEMBER |
ClickHouse | ClickHouse | 736,537,762 | 17,438 | akuzm | The test failures are not related.
This feature is very simple, but tricky to test, so I think we don't have to add a test for it. | 2020-12-01T13:02:57 | 2020-12-01T13:02:57 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 736,607,551 | 17,634 | nikitamikhaylov | These tests have to be disabled in Arcadia anyway. | 2020-12-01T14:59:46 | 2020-12-01T14:59:46 | {} | MEMBER |
ClickHouse | ClickHouse | 736,711,039 | 17,646 | azat | >What was the initial motivation to ignore memory limit in buffer flush?
@alexey-milovidov Buffer have two different modes of flushing:
- direct
- background
And in pre #7552 direct flush is accounted for max_memory_usage (and similar) per-query settings, but background does not affected by them anyhow, so it w... | 2020-12-01T17:42:44 | 2020-12-01T17:42:44 | {} | MEMBER |
ClickHouse | ClickHouse | 736,791,261 | 17,705 | MasterOdin | > > Do the other system tables expose information only available to the current user (e.g. system.tables)?
>
> yes.
Given this, would it then make sense to leave the verbiage on this page as is to resemble the other pages, and add a line to https://clickhouse.tech/docs/en/operations/system-tables/ about these tab... | 2020-12-01T20:12:58 | 2020-12-01T20:12:58 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 736,448,056 | 17,602 | vitlibar | Cherry-picked manually. | 2020-12-01T11:19:26 | 2020-12-01T11:19:26 | {} | MEMBER |
ClickHouse | ClickHouse | 736,468,570 | 17,393 | alexey-milovidov | It's more complicated.
First we calculate intHash64 of array size. Then combine it with all array elements in chain.
`combine(prev_hash, x) = hash(prev_hash || hash(x))`
`hash(['a', 'b', 'c']) = hash(hash(hash(intHash64(3) || hash('a')) || hash('b')) || hash('c'))` | 2020-12-01T11:32:15 | 2020-12-01T11:32:15 | {
"+1": 1
} | MEMBER |
ClickHouse | ClickHouse | 736,479,885 | 17,667 | vitlibar | The conflict doesn't seem to be easy to resolve. | 2020-12-01T11:39:29 | 2020-12-01T11:39:29 | {} | MEMBER |
ClickHouse | ClickHouse | 736,531,480 | 17,665 | amosbird | P.S. It's worth mentioning that the editor feature is better suit with `-n`, that is, single line editing with multiple queries (bracketed_paste enabled) | 2020-12-01T12:50:08 | 2020-12-01T12:50:08 | {} | COLLABORATOR |
ClickHouse | ClickHouse | 736,595,139 | 17,640 | tavplubix | Integration tests - known issue
ClickHouse special build check - pull rate limit, known issue
Yandex synchronization check - broken in master | 2020-12-01T14:43:05 | 2020-12-01T14:43:05 | {} | MEMBER |
ClickHouse | ClickHouse | 736,712,900 | 17,658 | azat | >Integration tests (asan) Pending — Started
>Integration tests (release) Pending — Started
>Integration tests (thread) Pending — Started
Seems that due to test_grpc_protocol errors they will be constantly restarted on and on?
Since there was already one report https://clickhouse-test-reports.s3.yandex.net/17658... | 2020-12-01T17:45:56 | 2020-12-01T17:45:56 | {} | MEMBER |
ClickHouse | ClickHouse | 736,789,456 | 17,705 | den-crane | >Do the other system tables expose information only available to the current user (e.g. system.tables)?
yes.
>Seems though that your suggestion would be to rewrite the first line to read something like:
yes | 2020-12-01T20:09:46 | 2020-12-01T20:09:46 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 736,790,558 | 17,648 | gavinju | thank you! It's ok! | 2020-12-01T20:11:36 | 2020-12-01T20:11:36 | {} | NONE |
ClickHouse | ClickHouse | 735,989,642 | 17,510 | azat | @vitlibar 20.10 does not have original PR (#15074), so it is not affected | 2020-11-30T19:22:18 | 2020-11-30T19:22:18 | {} | MEMBER |
ClickHouse | ClickHouse | 736,346,712 | 4,717 | filimonov | Actually, the current behavior is VERY questionable.
And it's hard to call it logical.
I think the user should be able to choose (even per dictionary actually) when to load it:
* on first access
* on server start (asynchronously / non-blocking)
* on server start, before processing connections (synchronously ... | 2020-12-01T09:35:26 | 2020-12-01T09:35:26 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 736,401,193 | 17,458 | alesapin | `01415_sticking_mutations` flaky test | 2020-12-01T10:49:46 | 2020-12-01T10:49:46 | {} | MEMBER |
ClickHouse | ClickHouse | 736,481,347 | 17,666 | alexey-milovidov | It may happen if server CPU has no support for AVX2 but the `cpuid` instruction is being virtualized and returns wrong info about the support for AVX2. | 2020-12-01T11:40:25 | 2020-12-01T11:40:25 | {} | MEMBER |
ClickHouse | ClickHouse | 736,550,747 | 17,109 | akuzm | Test failures don't seem to be relevant. | 2020-12-01T13:27:28 | 2020-12-01T13:27:28 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 736,602,053 | 17,686 | alexey-milovidov | Ref: #16787. | 2020-12-01T14:51:36 | 2020-12-01T14:51:36 | {} | MEMBER |
ClickHouse | ClickHouse | 736,717,045 | 17,658 | alexey-milovidov | Yes, that's the reason.
Let's not wait. | 2020-12-01T17:53:07 | 2020-12-01T17:53:07 | {} | MEMBER |
ClickHouse | ClickHouse | 736,747,075 | 17,646 | alexey-milovidov | BTW, I don't like to introduce a new setting for every implementation detail.
But as the initial behaviour of ignoring memory limit in Buffer was introduced by you, I'm also Ok with this change.
Does this change reasonably motivated? (by some behaviour observed in production?) | 2020-12-01T18:48:44 | 2020-12-01T18:48:44 | {} | MEMBER |
ClickHouse | ClickHouse | 736,793,307 | 17,705 | den-crane | it should be reiterated in both places | 2020-12-01T20:16:30 | 2020-12-01T20:16:30 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 736,351,155 | 17,578 | l1t1 | about speed, I found this
https://arxiv.org/search/cs?searchtype=author&query=Lemire%2C+D | 2020-12-01T09:38:11 | 2020-12-01T09:38:11 | {} | NONE |
ClickHouse | ClickHouse | 736,368,357 | 17,661 | alexey-milovidov | @tavplubix Looks like it's related to `renameat2` function. | 2020-12-01T10:00:14 | 2020-12-01T10:00:14 | {} | MEMBER |
ClickHouse | ClickHouse | 736,382,474 | 17,648 | gavinju | hi, den-crane
thanks for your replay! I tried it, but the same error is displayed.
But, use the following can run correctly:
coclickhouse-server --config-file=/etc/clickhouse-server/config.xml | 2020-12-01T10:24:15 | 2020-12-01T10:24:15 | {} | NONE |
ClickHouse | ClickHouse | 736,431,700 | 17,472 | vitlibar | The conflict doesn't seem to be easy to resolve. | 2020-12-01T11:09:07 | 2020-12-01T11:09:07 | {} | MEMBER |
ClickHouse | ClickHouse | 736,577,873 | 17,682 | mzealey | Hm it causes a reproducable crash on my live database, but it is a bit tricky to make it crash on test data.
I think it may be memory allocation related? I have the below crashing once on a local dev container but usually when i try to reproduce it throws:
```
Code: 241. DB::Exception: Received from localhost:90... | 2020-12-01T14:14:58 | 2020-12-01T14:14:58 | {} | NONE |
ClickHouse | ClickHouse | 736,782,529 | 17,276 | alexey-milovidov | It works totally alright. | 2020-12-01T19:56:12 | 2020-12-01T19:56:12 | {} | MEMBER |
ClickHouse | ClickHouse | 736,784,820 | 17,705 | MasterOdin | Do the other system tables expose information only available to the current user (e.g. `system.tables`)?
Seems though that your suggestion would be to rewrite the first line to read something like:
> Contains information about the databases that are available to the current user.
and the line above the select ... | 2020-12-01T20:00:54 | 2020-12-01T20:00:54 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 736,091,178 | 17,632 | filimonov | OK. I see, related to https://github.com/ClickHouse/ClickHouse/pull/12308/ | 2020-11-30T22:20:52 | 2020-11-30T22:20:52 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 736,204,927 | 5,576 | den-crane | ```
20.13.1.5273
select count() from system.query_log where event_date global in (today()) settings force_primary_key=1
┌─count()─┐
│ 288 │
└─────────┘
1 rows in set. Elapsed: 0.002 sec.
```
fixed in #17253 | 2020-12-01T04:09:35 | 2020-12-01T04:09:35 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 736,348,375 | 15,170 | filimonov | See also https://github.com/ClickHouse/ClickHouse/issues/4717#issuecomment-736346712 | 2020-12-01T09:36:28 | 2020-12-01T09:36:28 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 736,371,279 | 17,578 | alexey-milovidov | You posted the link to all the works by Daniel Lemire. I'm trying to follow his works, blog and repositories :)
For example, recentry he has a work about fast float parsing (but this issue is about fast float formatting - the inverse operation).
For float parsing there is a PR in progress: #16787 | 2020-12-01T10:04:37 | 2020-12-01T10:04:37 | {} | MEMBER |
ClickHouse | ClickHouse | 736,387,963 | 17,663 | liangli1213 | And also, I'd like to know does below feature has been realized?
1. Materialize MySQL keep alive when MySQL master is down
2. Materialize MySQL can do house keeping for deleted rows
3. Materialize MySQL can keep replication when using VIP to make replication relationship and VIP has failover to another MySQL node | 2020-12-01T10:34:26 | 2020-12-01T10:34:26 | {} | NONE |
ClickHouse | ClickHouse | 736,409,007 | 17,552 | vitlibar | Depends on https://github.com/ClickHouse/ClickHouse/pull/15074 which hasn't merged to `20.10`. | 2020-12-01T10:54:42 | 2020-12-01T10:54:42 | {} | MEMBER |
ClickHouse | ClickHouse | 736,519,101 | 17,646 | alexey-milovidov | What was the initial motivation to ignore memory limit in buffer flush?
I expect that it was reasonable. | 2020-12-01T12:23:20 | 2020-12-01T12:23:39 | {} | MEMBER |
ClickHouse | ClickHouse | 736,607,912 | 17,648 | den-crane | try
sudo -u clickhouse clickhouse-server --config-file=/etc/clickhouse-server/config.xml
Services use clickhouse user. | 2020-12-01T15:00:21 | 2020-12-01T15:01:28 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 736,612,443 | 17,379 | traceon | @hoverzheng have you installed Command Line Tools from Xcode? | 2020-12-01T15:07:39 | 2020-12-01T15:07:39 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 736,735,871 | 17,646 | azat | >So, why make the setting 0 by default?
To avoid hiding possible problems, to me it sounds like it is better to show memory limit error instead of using lots of of RAM and eventually fail in another place.
Since it is way more easy to answer the question "why INSERT into Buffer fails with memory limit exceeded" ins... | 2020-12-01T18:26:54 | 2020-12-01T18:27:21 | {} | MEMBER |
ClickHouse | ClickHouse | 736,785,087 | 3,489 | mikekytyzov | Is this issue still relevant ?
I have absolutely the same issue on 20.3.19.4 version
Adding this setting to `config.xml` or `users.xml` do not affect on result. | 2020-12-01T20:01:26 | 2020-12-01T20:01:26 | {} | NONE |
ClickHouse | ClickHouse | 736,807,662 | 17,268 | alexey-milovidov | It's more complicated: sometimes we need to use immediate IP for access control (to restrict proxy addresses) and forwarded IP for quota accounting (to account users in front of proxy). | 2020-12-01T20:42:24 | 2020-12-01T20:42:24 | {} | MEMBER |
ClickHouse | ClickHouse | 736,401,584 | 17,458 | alesapin | ```
Query id: 8dcde8aa-4666-4c3e-887c-3d0e8d2578ac
┌────check_start_time─┬─test_name────────────────┬─test_status─┬─report_url─────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────┐
│ 2020-12-01 08:17:44 │ 01415_sticking_mut... | 2020-12-01T10:50:01 | 2020-12-01T10:50:01 | {} | MEMBER |
ClickHouse | ClickHouse | 736,470,510 | 17,513 | vitlibar | 20.9 does not have original PR https://github.com/ClickHouse/ClickHouse/pull/15074, so it is not affected | 2020-12-01T11:33:29 | 2020-12-01T11:33:29 | {} | MEMBER |
ClickHouse | ClickHouse | 736,603,674 | 17,666 | den-crane | BTW , allow_simdjson setting can be used as WA.
--allow_simdjson arg Allow using simdjson library in 'JSON*' functions if AVX2 instructions are available. If disabled rapidjson will be used. | 2020-12-01T14:54:00 | 2020-12-01T14:54:00 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 736,729,015 | 17,646 | alexey-milovidov | Maybe you want something more sophisticated:
- ignore per-query and per-user memory limits;
- still respect server total memory limit? | 2020-12-01T18:14:31 | 2020-12-01T18:14:31 | {} | MEMBER |
ClickHouse | ClickHouse | 735,982,413 | 17,403 | den-crane | Can you show an output of `curl -vvv 'http://user:password@srvA:8123/?query=SELECT%201'` | 2020-11-30T19:08:10 | 2020-11-30T19:08:19 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 736,469,594 | 17,393 | alexey-milovidov | The code is here: https://clickhouse.tech/codebrowser/html_report/ClickHouse/src/Functions/FunctionsHashing.h.html | 2020-12-01T11:32:53 | 2020-12-01T11:32:53 | {
"+1": 2
} | MEMBER |
ClickHouse | ClickHouse | 736,484,687 | 17,666 | alexey-milovidov | What kind of infrastructure do you use for your server? Is it a well known cloud provider or some well known tool in private cloud or something else? | 2020-12-01T11:42:31 | 2020-12-01T11:42:31 | {} | MEMBER |
ClickHouse | ClickHouse | 736,782,625 | 17,276 | alexey-milovidov | Test needed. | 2020-12-01T19:56:23 | 2020-12-01T19:56:23 | {} | MEMBER |
ClickHouse | ClickHouse | 736,189,217 | 8,337 | LB-J | hi，do you have reslove this problems ？ | 2020-12-01T03:13:25 | 2020-12-01T03:13:37 | {} | NONE |
ClickHouse | ClickHouse | 736,234,859 | 17,631 | 7omsxx | Try to set max_server_memory_usage_to_ram_ratio=2
it works | 2020-12-01T05:42:11 | 2020-12-01T05:42:11 | {} | NONE |
ClickHouse | ClickHouse | 736,312,013 | 17,393 | ahmczsy | > For tuples, it is calculated iteratively in the following fashion:
>
> `hash(x, y) = hash(hash(x) || hash(y))`
> and
> `hash(..., x_n) = hash(hash(...) || hash(x_n))`
>
> where || is concatenation of two 64bit values to single 128bit chunk in memory.
How does the array as shards implement?
`select sipHa... | 2020-12-01T08:35:42 | 2020-12-01T08:35:42 | {} | NONE |
ClickHouse | ClickHouse | 736,427,394 | 17,156 | vitlibar | The conflict doesn't seem to be easy to resolve. | 2020-12-01T11:06:24 | 2020-12-01T11:06:24 | {} | MEMBER |
ClickHouse | ClickHouse | 736,506,792 | 15,806 | CurtizJ | Thanks. I will review and merge soon. | 2020-12-01T11:58:05 | 2020-12-01T11:58:05 | {} | MEMBER |
ClickHouse | ClickHouse | 736,554,544 | 16,947 | akuzm | > * GROUP BY on the shard is rewritten by the `QueryNormalizer`, and in case of `_shard_num` was requested with an alias (`_shard_num AS num`), on the shard `_shard_num` will be alias-to-alias, while on the initiator it will be still parent alias, so the shard, column with name `_shard_num AS foo` will be rewritten to ... | 2020-12-01T13:34:35 | 2020-12-01T13:34:35 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 736,612,576 | 17,688 | m7shapan | @den-crane thank you, you really made my day | 2020-12-01T15:07:49 | 2020-12-01T15:07:49 | {} | NONE |
ClickHouse | ClickHouse | 736,729,828 | 17,569 | alexey-milovidov | Yes, perf test shows -2.388x!
https://clickhouse-test-reports.s3.yandex.net/17569/cd8e7981e08dace480fd340646e2504301d5d96c/performance_comparison/report.html | 2020-12-01T18:16:03 | 2020-12-01T18:16:03 | {} | MEMBER |
ClickHouse | ClickHouse | 735,968,228 | 17,632 | den-crane | I think it's nothing to fix here.
`LowCardinality(String) CODEC (Gorilla, ...)` probably can work because LowCardinality dictionary key has the same size over part. Though I am not sure about it because I think a part can have several LC dictionaries? | 2020-11-30T18:41:03 | 2020-11-30T18:42:26 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 736,272,783 | 17,628 | alesapin | `Functional stateless tests (thread) — fail: 1, passed: 2424, skipped: 14`
`00840_long_concurrent_select_and_drop_deadlock.sh` -- test fixed in master. | 2020-12-01T07:14:07 | 2020-12-01T07:14:07 | {} | MEMBER |
ClickHouse | ClickHouse | 736,273,565 | 17,625 | alesapin | `test_storage_rabbitmq/test.py::test_rabbitmq_csv_with_delimiter` -- test fixed in master
stress test -- RocksDB. | 2020-12-01T07:15:51 | 2020-12-01T07:15:51 | {} | MEMBER |
ClickHouse | ClickHouse | 736,367,411 | 17,646 | azat | The test fails now because of the text_log, each text_log/query_log/... entry blocks the memory tracking in the thread, however it is not nested and hence after message in the log (for text_log) the memory tracker is not blocked anymore.
This should be fixed in #17658.
Will rebase this PR once it will be merged. | 2020-12-01T09:58:40 | 2020-12-01T09:59:07 | {} | MEMBER |
ClickHouse | ClickHouse | 736,564,428 | 17,682 | amosbird | Please provide a minimal example to reproduce the crash | 2020-12-01T13:51:33 | 2020-12-01T13:51:33 | {} | COLLABORATOR |
ClickHouse | ClickHouse | 736,712,187 | 17,700 | alexey-milovidov | CC @alesapin | 2020-12-01T17:44:43 | 2020-12-01T17:44:43 | {} | MEMBER |
ClickHouse | ClickHouse | 736,276,321 | 17,453 | alexey-milovidov | Not necessarily to rebase.
Integration test is still executing.
Functional - known failure.
Stress MSan - bug in RocksDB, fixed in another PR.
Yandex Synchronization - don't care. | 2020-12-01T07:21:24 | 2020-12-01T07:21:24 | {} | MEMBER |
ClickHouse | ClickHouse | 736,725,414 | 17,646 | azat | >I remember: we disable memory tracking for direct flush, because the data volume is unrelated to the INSERT query:
Yes, this is also a good example | 2020-12-01T18:08:23 | 2020-12-01T18:08:23 | {} | MEMBER |
ClickHouse | ClickHouse | 736,757,839 | 17,637 | alexey-milovidov | Special build check has failed because SymbolIndex is disabled on some platforms. | 2020-12-01T19:08:25 | 2020-12-01T19:08:25 | {} | MEMBER |
ClickHouse | ClickHouse | 736,777,906 | 17,507 | alexey-milovidov | #17437 | 2020-12-01T19:47:17 | 2020-12-01T19:47:17 | {} | MEMBER |
ClickHouse | ClickHouse | 736,026,978 | 17,632 | filimonov | > > sequence of alters and version upgrade.
>
> * upgrade, fail to start
>
> * rollback
>
> * trying to alter, failing.
Why did you rollback instead of setting `allow_suspicious_codecs`? | 2020-11-30T20:34:17 | 2020-11-30T20:34:17 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 736,106,905 | 17,109 | azat | >Slightly off-topic: Why/when would you use a SimpleAggregateFunction as part of the primary key?
Good question! And the answer is pretty simple - old poor design
Actually it worth forbid such use cases (as an addition for #16601)
By the way optimize_aggregation_in_order also affected with too frequent Arena all... | 2020-11-30T23:00:53 | 2020-11-30T23:02:54 | {} | MEMBER |
ClickHouse | ClickHouse | 736,165,961 | 17,379 | hoverzheng | Hi @traceon, @alexey-milovidov
I try your method, but the error is the same, Mybe I should Install Clang not AppleClang?
| 2020-12-01T01:57:45 | 2020-12-01T01:57:45 | {} | NONE |
ClickHouse | ClickHouse | 736,250,834 | 17,638 | DmitriZhukov | 20.11.2.1
| 2020-12-01T06:22:06 | 2020-12-01T06:22:06 | {} | NONE |
ClickHouse | ClickHouse | 736,317,727 | 15,806 | hexiaoting | @alexey-milovidov @CurtizJ
hi, right now this PR is ready to review.
In this version, data in Map DataType is stored using two columns, one for storing keys and the other is for storing values. When using map['key'] to get the value I use sequential iterate method, Of course this is the slowest way。
After this P... | 2020-12-01T08:46:17 | 2020-12-01T08:46:17 | {
"+1": 2
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 736,363,472 | 17,622 | tavplubix | No, `test_grpc_protocol` tests are broken in master branch.
Functional stateless tests (address) - 01149_zookeeper_mutation_stuck_after_replace_partition - unrelated
Functional stateless tests (release, wide parts enabled) - 01584_distributed_buffer_cannot_find_column - unrelated
Stress test (address) - SIGABRT in... | 2020-12-01T09:51:56 | 2020-12-01T09:51:56 | {} | MEMBER |
ClickHouse | ClickHouse | 736,480,478 | 17,669 | vitlibar | The conflict doesn't seem to be easy to resolve. | 2020-12-01T11:39:52 | 2020-12-01T11:39:52 | {} | MEMBER |
ClickHouse | ClickHouse | 736,596,959 | 17,688 | den-crane | Your aggregation_ads_date_analytics_local is SummingMergeTree. Not ReplicatedSummingMergeTree.
Do you have Zookeeper? Do you use Replicated tables?
https://den-crane.github.io/Everything_you_should_know_about_materialized_views_commented.pdf
page 15
https://youtu.be/ckChUkC3Pns?t=10715 | 2020-12-01T14:46:01 | 2020-12-01T14:46:01 | {
"heart": 1
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 736,766,880 | 17,459 | azat | >Style check — Found 2 errors
rebase with preserving merges is tricky - fixed | 2020-12-01T19:26:24 | 2020-12-01T19:26:24 | {} | MEMBER |
ClickHouse | ClickHouse | 736,273,076 | 17,625 | alesapin | `test_storage_rabbitmq/test.py::test_rabbitmq_csv_with_delimiter` -- test fixed in master | 2020-12-01T07:14:47 | 2020-12-01T07:14:47 | {} | MEMBER |
ClickHouse | ClickHouse | 736,613,485 | 17,379 | traceon | @hoverzheng if you want to use vanilla Clang from brew:
```sh
$ brew install llvm
$ CC=$(brew --prefix llvm)/bin/clang CXX=$(brew --prefix llvm)/bin/clang++ cmake ..
```
Also, make sure you pull the latest `master`. | 2020-12-01T15:09:15 | 2020-12-01T15:11:45 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 736,683,437 | 17,657 | altinity-robot | @vitlibar, this PR is ready for your review and merge. Thanks! | 2020-12-01T16:59:29 | 2020-12-01T16:59:29 | {} | NONE |
ClickHouse | ClickHouse | 736,720,826 | 17,646 | alexey-milovidov | I remember: we disable memory tracking for direct flush, because the data volume is unrelated to the INSERT query:
e.g. you have INSERTed one record, but buffer with million of records will be flushed and it will be over memory limit.
| 2020-12-01T17:59:57 | 2020-12-01T17:59:57 | {} | MEMBER |
ClickHouse | ClickHouse | 736,823,184 | 17,707 | alexey-milovidov | It is similar to `keyed` quotas, but for key, the last element of `X-Forwarded-For` is used instead of `X-ClickHouse-Quota`. | 2020-12-01T21:11:56 | 2020-12-01T21:11:56 | {} | MEMBER |
ClickHouse | ClickHouse | 736,193,657 | 17,648 | den-crane | Do not use `/etc/init.d/clickhouse-server start`
USE:
`sudo systemctl start clickhouse-server`
`sudo systemctl status clickhouse-server`
`sudo systemctl stop clickhouse-server`
`sudo journalctl -u clickhouse-server.service` | 2020-12-01T03:28:35 | 2020-12-01T03:28:35 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 736,205,798 | 6,601 | den-crane | still relevant 20.13.1.5273 | 2020-12-01T04:13:01 | 2020-12-01T04:13:01 | {
"+1": 1
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 736,611,372 | 17,690 | alesapin | Just trying to debug the source of timeouts
| 2020-12-01T15:05:51 | 2020-12-01T15:05:51 | {} | MEMBER |
ClickHouse | ClickHouse | 736,728,391 | 17,646 | alexey-milovidov | So, why make the setting 0 by default? | 2020-12-01T18:13:18 | 2020-12-01T18:13:18 | {} | MEMBER |
ClickHouse | ClickHouse | 736,039,703 | 17,632 | den-crane | allow_suspicious_codecs -- for another cases like CODEC(ZSTD, ZSTD))
```
20.13.1
set allow_suspicious_codecs=1;
create table test(A LowCardinality(String) CODEC(Gorilla, ZSTD)) Engine=MergeTree order by tuple()
DB::Exception: Codec Gorilla is not applicable for String because the data type is not of fixed s... | 2020-11-30T20:57:29 | 2020-11-30T20:57:29 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 736,764,291 | 17,646 | azat | >BTW, I don't like to introduce a new setting for every implementation detail.
Me neither
>Does this change reasonably motivated? (by some behaviour observed in production?)
Yes, one Buffer table was pushed to the table, that has MV attached, with pretty heavy aggregation (but it wasn't look heavy at first) th... | 2020-12-01T19:21:11 | 2020-12-01T19:21:11 | {} | MEMBER |
ClickHouse | ClickHouse | 736,068,906 | 17,632 | siradjev | > Why did you rollback instead of setting allow_suspicious_codecs?
it only solves below error.
> SQL Error [36]: ClickHouse exception, code: 36, host: 127.0.0.1, port: 54669; Code: 36, e.displayText() = DB::Exception: It does not make sense to have codec NONE along with other compression codecs: NONE, ZSTD(1). (... | 2020-11-30T21:31:30 | 2020-11-30T21:31:30 | {} | NONE |
ClickHouse | ClickHouse | 736,319,988 | 6,462 | turningme | > > I also encountered the same problem. I modified the ttl of the original table to shorten it, and then used OPTIMIZE FINAL, but the expired data was not deleted
>
> So do I , how about make friend In weixin or QQ, then we can talk about this problem
Have you solved it ? | 2020-12-01T08:50:28 | 2020-12-01T08:50:28 | {} | NONE |
ClickHouse | ClickHouse | 736,575,894 | 17,578 | l1t1 | Daniel also contributed codes for clickhouse. great | 2020-12-01T14:11:28 | 2020-12-01T14:11:28 | {} | NONE |
ClickHouse | ClickHouse | 736,778,366 | 17,705 | den-crane | It exposes only databases available for the current user.
```
select * from system.databases;
┌─name─┬─engine───┬─data_path──────────────────────┬─metadata_path──────────────────────┬─────────────────────────────────uuid─┐
│ test │ Ordinary │ /var/lib/clickhouse/data/test/ │ /var/lib/clickhouse/metadata/test/ │... | 2020-12-01T19:48:11 | 2020-12-01T19:50:09 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 763,981,215 | 19,229 | filimonov | > > but the RAM still have it's own maximum throughput, which is reachable and can be limitation in some scenarios
>
> But I think it will be reflected as 100% CPU utilization.
>
> Though may be I am mistaken in the HT case (56 physical CPU cores (112 logical cores with HT enabled))
Well, I'm also not sure if ... | 2021-01-20T22:07:53 | 2021-01-20T22:07:53 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 764,397,003 | 19,204 | keen-wolf | > How do we control that this function will not allow to read data from other databases and private configs? It is very powerful function as it runs with the same user as clickhouse-server.
yes，as @alex-zaitsev has pointed, now the file function in this PR does the same path checking as the file() in Table Functions... | 2021-01-21T05:49:11 | 2021-01-21T05:49:11 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 764,449,301 | 3,141 | elloooooo | @alexey-milovidov Would you please paste the link of the PR which fix the issue? | 2021-01-21T07:48:02 | 2021-01-21T07:48:02 | {} | NONE |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.