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 | 832,758,377 | 23,875 | ice1x | @alexey-milovidov pulled request: https://github.com/ClickHouse/ClickHouse/pull/23898/files | 2021-05-05T14:58:39 | 2021-05-05T14:58:39 | {
"+1": 1
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 831,423,092 | 23,867 | den-crane | @ntrhieu89 Have you created a table with TTL expression in 21.3.8.76 or it was created a long time ago without TTL and you added TTL after? | 2021-05-03T17:50:02 | 2021-05-03T17:50:02 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 832,743,121 | 23,897 | kssenii | @tobiaslins
> Allow to limit max connections for external connection.
By the way, there is a setting for that: `postgresql_connection_pool_size`. | 2021-05-05T14:39:40 | 2021-05-05T14:39:40 | {
"+1": 1
} | MEMBER |
ClickHouse | ClickHouse | 832,924,936 | 23,880 | filimonov | Btw: this is not a my idea (and i have no plans to work on that in near future).
That subject popped up during some discussions (some users were asking about that) and i decided it's worth to have it in the backlog (even if it will not be implemented in the close future).
Once somebody will have time - the issue ... | 2021-05-05T18:46:28 | 2021-05-05T18:52:42 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 833,117,769 | 23,892 | den-crane | You need number of simultaneous merges * 20*2 + move_factor 0.2 | 2021-05-05T23:43:25 | 2021-05-05T23:43:25 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 833,220,243 | 20,202 | amosbird | > Integration tests (asan) — Timeout, fail: 0, passed: 1314, flaky: 1
What does this mean? (stably reproduced in last three commits) | 2021-05-06T04:45:44 | 2021-05-06T04:46:12 | {} | COLLABORATOR |
ClickHouse | ClickHouse | 832,394,280 | 23,827 | gyfis | Gotcha! I think we can close this now, thanks for your time! | 2021-05-05T03:52:14 | 2021-05-05T03:52:14 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 832,898,410 | 17,847 | alexey-milovidov | You can do it similar to the `allow_suspicious_codecs` setting. | 2021-05-05T18:06:16 | 2021-05-05T18:06:16 | {} | MEMBER |
ClickHouse | ClickHouse | 832,926,041 | 23,175 | alexey-milovidov | @tavplubix
That's a good idea.
We can also make `CREATE TABLE ... AS SELECT` atomic if database is `Atomic`.
(table won't be created in case of insertion error) | 2021-05-05T18:48:08 | 2021-05-05T18:48:08 | {} | MEMBER |
ClickHouse | ClickHouse | 832,996,216 | 23,904 | alexey-milovidov | Do you use any proxy servers in front of ClickHouse?
What do you see in ClickHouse logs? | 2021-05-05T20:47:05 | 2021-05-05T20:47:05 | {} | MEMBER |
ClickHouse | ClickHouse | 833,267,659 | 23,035 | etah000 | @stone-wlg How does the parameter "database_atomic_delay_before_drop_table_sec" connect with your problem? I also encountered this problem some time ago. | 2021-05-06T06:37:52 | 2021-05-06T06:37:52 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 831,598,751 | 23,699 | alexey-milovidov | `select toInt128(170141183460469231731687303715884105727)`
This example is incorrect - you are passing floating point value to the function.
Looks like all your examples are incorrect.
This is correct:
```
SELECT toInt128('170141183460469231731687303715884105727')
Query id: 60171897-21f2-4f0d-af9d-35230cb23... | 2021-05-03T23:31:44 | 2021-05-03T23:31:44 | {} | MEMBER |
ClickHouse | ClickHouse | 832,166,292 | 23,827 | alexey-milovidov | The problem is that we plan to extend dots usage to support semistructured data.
E.g. you insert arbitrary JSON into a column, the data is transparently split to subcolumns and then you can query it like `a.b.c::String`. | 2021-05-04T18:51:08 | 2021-05-04T18:51:29 | {
"+1": 1
} | MEMBER |
ClickHouse | ClickHouse | 832,225,875 | 23,062 | tavplubix | Thank you for this attempt to improve usability of ON CLUSTER queries and sorry it took so long to answer.
If I understand correctly, the main purpose of this PR is to avoid `Watching task ... is executing longer than distributed_ddl_task_timeout ...` error when some node in cluster is not available. But I'm not sur... | 2021-05-04T20:34:32 | 2021-05-04T20:34:32 | {
"+1": 1
} | MEMBER |
ClickHouse | ClickHouse | 832,639,088 | 23,880 | alexey-milovidov | I thought `ionice` does not help. Do you have experiments proving the opposite? | 2021-05-05T12:13:53 | 2021-05-05T12:13:53 | {} | MEMBER |
ClickHouse | ClickHouse | 832,640,171 | 23,875 | alexey-milovidov | You can edit the docs - press on the pencil icon at the top right corner. | 2021-05-05T12:15:48 | 2021-05-05T12:15:48 | {
"+1": 1
} | MEMBER |
ClickHouse | ClickHouse | 832,701,450 | 21,839 | tobiaslins | @kssenii @akuzm 🚨
Since the last Clickhouse update my Postgres Connections are growing over time until no connections are available anymore.
I see a ton of commits and open connections (30+) after several minutes. After i restart the CH server everything recovers and start growing again!
On 11:30 I upgraded to t... | 2021-05-05T13:47:25 | 2021-05-05T13:57:58 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 832,759,530 | 17,847 | fibersel | Yes, but how can I access global settings? | 2021-05-05T15:00:05 | 2021-05-05T15:00:05 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 833,267,015 | 23,891 | miladstar77 | no ,
| 2021-05-06T06:36:30 | 2021-05-06T06:36:30 | {} | NONE |
ClickHouse | ClickHouse | 831,523,815 | 23,864 | azat | >Is this feature really needed?
I don't have strong opinion on this, but this sounds like a reasonable functionality to me.
>why don't the user just fix the issues in the infrastructure?
User may fix it, but this can happen from time to time (for different users) due to too may rows was batched into one, and t... | 2021-05-03T20:47:45 | 2021-05-03T20:47:45 | {} | MEMBER |
ClickHouse | ClickHouse | 831,993,588 | 23,778 | jhettler | I am sorry, a bit late, but here it is ...
It starts with network usage - I expect that this is data loading from Snowflake to Clikhouse server
but there is nearly no usage of memory (this is last 15 ... | 2021-05-04T14:39:04 | 2021-05-04T14:39:04 | {} | NONE |
ClickHouse | ClickHouse | 832,054,692 | 4,419 | arctica | > It's currently possible by using CollapsingMergeTree with default `sign = 1`. We wil add ProjectionMergeTree if necessary.
Yes, as I outlined in the beginning of my proposal, there are many specialized table engines which all solve one usecase but none can do it all and you can't really combine them. Table engine ... | 2021-05-04T15:59:45 | 2021-05-04T15:59:45 | {} | NONE |
ClickHouse | ClickHouse | 832,211,084 | 23,783 | azat | @KochetovNicolai maybe you can take a look at this? This is the first PR in the patch series that will replace #16947 eventually. | 2021-05-04T20:09:53 | 2021-05-04T20:09:53 | {} | MEMBER |
ClickHouse | ClickHouse | 832,592,672 | 23,751 | alexey-milovidov | It did not help, see
https://clickhouse-test-reports.s3.yandex.net/23874/80065b08f8182936ac57408bd84492d95f2322c1/integration_tests_(asan).html#fail1
https://clickhouse-test-reports.s3.yandex.net/23874/80065b08f8182936ac57408bd84492d95f2322c1/integration_tests_(thread).html#fail1 | 2021-05-05T10:47:08 | 2021-05-05T10:47:08 | {} | MEMBER |
ClickHouse | ClickHouse | 832,614,977 | 17,847 | fibersel | @alexey-milovidov how can I get value of experimental flag from inside compression codec?
Should I derive codec form `WithContext` class? | 2021-05-05T11:29:33 | 2021-05-05T11:29:33 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 832,628,111 | 23,889 | OmarBazaraa | I think I figured out the problem. The quota is not enforced on the distributed queries. If I have a cluster of 2 shards and keep issuing a distributed query on shard 1, the quota usage will not increase on shard 2. Is there a way to enforce the quota on the `initial_user` instead of the `user`? | 2021-05-05T11:55:13 | 2021-05-05T11:55:13 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 832,763,391 | 18,695 | filimonov | One more - a bit opposite, but very popular usecase.
We want to have realtime inserts (so we need to insert into a single partition w/o fan out to hundreds of partitions) but when parts get bigger we want to partition it by something like (tenant_id, month) | 2021-05-05T15:05:05 | 2021-05-05T15:05:05 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 832,875,537 | 23,901 | alexey-milovidov | If we simply remove these two lines it will work as expected:
https://github.com/ClickHouse/ClickHouse/blob/master/src/Server/HTTP/ReadHeaders.cpp#L71 | 2021-05-05T17:30:06 | 2021-05-05T17:30:06 | {} | MEMBER |
ClickHouse | ClickHouse | 832,906,662 | 23,902 | alexey-milovidov | As WINDOW functions still require experimental flag to enable, I change the category from "bug" to "bug experimental". | 2021-05-05T18:18:45 | 2021-05-05T18:18:45 | {} | MEMBER |
ClickHouse | ClickHouse | 832,943,065 | 23,902 | vzakaznikov | Another example. With distributed we get
```
May 05,2021 15:11:53 ⟥ Step executing command
echo -e "SELECT depname, empno, salary, sum(salary) OVER (PARTITION BY depname) AS sum FROM empsalary ORDER BY depname, salary, empno FORMAT TabSeparatedWithNames" | click... | 2021-05-05T19:14:46 | 2021-05-05T19:14:46 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 832,997,053 | 23,904 | alexey-milovidov | ClickHouse never responds with HTTP 502. | 2021-05-05T20:48:26 | 2021-05-05T20:48:26 | {} | MEMBER |
ClickHouse | ClickHouse | 833,118,905 | 23,907 | den-crane | You should never use `PARTITION BY a ORDER BY a` SimpleAggregate in PARTITION/ORDER By | 2021-05-05T23:45:54 | 2021-05-05T23:45:54 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 833,268,945 | 23,907 | nkolotov | > Try Tuple
>
> ```
> CREATE TABLE test
> (
> `a` SimpleAggregateFunction(argMax, Tuple(UInt8, UInt64))
> )
> ENGINE = AggregatingMergeTree
> PARTITION BY a
> ORDER BY a
> ```
This returns `DB::Exception: Aggregate function argMax requires two arguments.` | 2021-05-06T06:40:25 | 2021-05-06T06:40:25 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 831,750,370 | 23,866 | filimonov | It looks like build issue relates to https://github.com/edenhill/librdkafka/pull/3315
It requires a deeper look, we use boringssl and some [hack](https://github.com/edenhill/librdkafka/pull/3315) in librdkafka code to use it. I don't know that part of code good so i will leave it for now.
I guess i want do update... | 2021-05-04T07:52:14 | 2021-05-04T07:53:27 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 831,885,417 | 23,875 | ice1x | Submitted this works fine.
May be it should be also inside the guide?
https://clickhouse.tech/docs/ru/development/build-osx/ | 2021-05-04T11:57:12 | 2021-05-04T11:57:19 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 832,191,088 | 6,107 | tavplubix | Fixed in #21535, set `distributed_ddl_entry_format_version` to 2 and it will pass settings. | 2021-05-04T19:33:36 | 2021-05-04T19:33:36 | {} | MEMBER |
ClickHouse | ClickHouse | 832,757,009 | 23,880 | filimonov | > Do you have experiments proving the opposite?
I think you meaned "compared to just setting niceness, on CFQ sheduler".
It's quite hard to model that scenario, and the priority of that is low. So i never finished that check.
But here the situation is a bit different - we don't support setting niceness for se... | 2021-05-05T14:56:59 | 2021-05-05T15:07:31 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 832,766,980 | 23,892 | mindrb | Thank you for the clarification.
No such messages in CH log. But that is maybe because I reduced logging to `<level>information</level>`.
Where could I read about this MERGE POOL and it's settings?
So if my partition is ~20GB in total I need additional ~40GB on the fast disk for the OPTIMIZE to succeed without... | 2021-05-05T15:09:35 | 2021-05-05T15:09:35 | {} | NONE |
ClickHouse | ClickHouse | 832,837,279 | 23,697 | alexey-milovidov | Version 21.1 is no longer supported. | 2021-05-05T16:34:38 | 2021-05-05T16:34:38 | {} | MEMBER |
ClickHouse | ClickHouse | 832,933,036 | 23,704 | vzakaznikov | The issue https://github.com/ClickHouse/ClickHouse/issues/23902 shows that window functions still have problems with distributed tables. | 2021-05-05T18:59:02 | 2021-05-05T18:59:02 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 833,001,748 | 23,904 | electrical | Hi, The livecheck connects directly to the docker container. I haven't seen anything in the clickhouse server logs so far.
I had it running on debug level but had to lower it to informational as the logs were getting to big. | 2021-05-05T20:56:29 | 2021-05-05T20:56:29 | {} | NONE |
ClickHouse | ClickHouse | 833,174,076 | 20,202 | amosbird | ```
2021-05-05 23:51:05 00992_system_parts_race_condition_zookeeper_long: Received exception from server (version 21.6.1):
2021-05-05 23:51:05 Code: 999. DB::Exception: Received from localhost:9000. DB::Exception: Run time inconsistency, path: /clickhouse/task_queue/ddl/query-0000003330.
2021-... | 2021-05-06T02:15:31 | 2021-05-06T02:15:31 | {} | COLLABORATOR |
ClickHouse | ClickHouse | 833,251,738 | 23,844 | excitoon | Failed tests are unrelated. | 2021-05-06T06:04:39 | 2021-05-06T06:04:39 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 833,278,567 | 23,895 | kullanici0606 | Thanks a lot for both answer and the document. | 2021-05-06T06:58:33 | 2021-05-06T06:58:33 | {} | NONE |
ClickHouse | ClickHouse | 831,446,604 | 23,864 | alexey-milovidov | Is this feature really needed?
I mean
> Add ability to split distributed batch on failures (i.e. due to memory limits, corruptions)
why don't the user just fix the issues in the infrastructure?
Do you need this feature for your scenarios?
How it is motivated? | 2021-05-03T18:29:19 | 2021-05-03T18:29:19 | {} | MEMBER |
ClickHouse | ClickHouse | 831,600,175 | 16,581 | alexey-milovidov | Conversion between DateTime64 and UInt128, Int128, UInt256, Int256 is not going to be implemented.
We are actually very happy of not having this feature. | 2021-05-03T23:36:21 | 2021-05-03T23:37:46 | {} | MEMBER |
ClickHouse | ClickHouse | 831,603,677 | 23,793 | fuwhu | > isExpiredReplicated
@alexey-milovidov
sorry, i don't get the point here.
The 'isExpiredReplicated' function here is for avoiding the usage of expired zookeeper object in ReplicatedMergeTree table, which will result in an exception "Session expired". | 2021-05-03T23:47:38 | 2021-05-03T23:47:38 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 831,905,539 | 23,703 | vzakaznikov | @alexey-milovidov, are we going to document this? | 2021-05-04T12:31:55 | 2021-05-04T12:31:55 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 832,645,659 | 23,793 | alexey-milovidov | But now we don't use ZooKeeper inside calls to getTotalRows/Bytes. | 2021-05-05T12:25:35 | 2021-05-05T12:25:35 | {} | MEMBER |
ClickHouse | ClickHouse | 832,930,787 | 23,902 | vzakaznikov | Another issue where with a distributed table the result changes all the time and does not make any sense.
```
May 05,2021 14:50:53 ⟥ Scenario empty named window, flags:TE
Check using over clause with empty window.
Requirements
... | 2021-05-05T18:55:30 | 2021-05-05T18:55:30 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 832,964,104 | 23,880 | alexey-milovidov | Another idea is to add IO throttling for merges and moves. E.g. merges and moves occupy not more than 50% IO bandwidth. | 2021-05-05T19:52:02 | 2021-05-05T19:52:02 | {} | MEMBER |
ClickHouse | ClickHouse | 831,392,390 | 23,793 | alexey-milovidov | `isExpiredReplicated`
Now this function is unneeded. | 2021-05-03T16:54:57 | 2021-05-03T16:54:57 | {} | MEMBER |
ClickHouse | ClickHouse | 831,885,590 | 23,868 | alexey-milovidov | Ok but the code of YT has diverged:
```
$(SOURCE_ROOT)/yt/chyt/server/config_repository.cpp:32:33: error: virtual function 'getName' has a different return type ('const std::string &' (aka 'const basic_string<char, char_traits<char>, allocator<char>> &')) than the function it overrides (which has return type 'std::st... | 2021-05-04T11:57:30 | 2021-05-04T11:57:30 | {} | MEMBER |
ClickHouse | ClickHouse | 831,902,796 | 23,430 | filimonov | 1. Can you clarify what means "submit the offset according to the original construction model" ?
2. Can you share the fragment of `/var/log/clickhouse-server/clickhouse-server.log` after table was recreated? I'm interested in log lines with trace log level containing Kafka. Do the data from Kafka topic is coming to M... | 2021-05-04T12:27:01 | 2021-05-04T12:27:01 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 833,199,924 | 23,911 | ucasfl | This should have been fixed in #22988, but it didn't.
The reason is because that `count()` always return 0 for `ReadBufferFromS3`, so when reinitialize `ReadBufferFromIStream`, it can not get correct file offset with `getPosition()`, which lead to new read error: checksum doesn't match. | 2021-05-06T03:36:26 | 2021-05-06T03:38:08 | {} | COLLABORATOR |
ClickHouse | ClickHouse | 833,247,945 | 23,822 | excitoon | > ```
> Неправильная строка для changelog.
>
> Если попадает в категорию bugfix - необходимо написать, какое поведение ClickHouse с точки зрения пользователя было ошибочным (какая ошибка исправлена).
>
> Внутренности кода никого не волнуют, их не надо писать в changelog.
>
> Ты написал:
>
> > Better handl... | 2021-05-06T05:56:12 | 2021-05-06T05:56:12 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 833,251,142 | 23,911 | excitoon | @ucasFL Can you reproduce it on some test? | 2021-05-06T06:03:16 | 2021-05-06T06:03:16 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 831,425,647 | 23,867 | ntrhieu89 | I created the table and add the TTL expression in `21.3.8.76` via Docker.
Start the server:
```
docker run -d --name some-clickhouse-server --ulimit nofile=262144:262144 yandex/clickhouse-server:21.3.8.76-alpine
```
Start the client:
```
docker run -it --rm --link some-clickhouse-server:clickhouse-server yande... | 2021-05-03T17:54:17 | 2021-05-03T17:56:46 | {} | NONE |
ClickHouse | ClickHouse | 831,543,501 | 23,864 | alexey-milovidov | It reminds me about the fact (slightly related) that we still have to fix POPULATE query for materialized views:
- make it aggregate data on block by block basis;
- make it consistent with respect to concurrent INSERTs. | 2021-05-03T21:20:41 | 2021-05-03T21:20:41 | {
"+1": 1
} | MEMBER |
ClickHouse | ClickHouse | 831,866,683 | 23,872 | den-crane | Works as expected.
MATERIALIZED VIEW is an insert trigger. It does not read the source table.
https://den-crane.github.io/Everything_you_should_know_about_materialized_views_commented.pdf | 2021-05-04T11:20:27 | 2021-05-04T11:20:27 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 832,033,876 | 4,419 | amosbird | > those projections will actually increase storage costs because they store the projected data in addition to the original data. Also it seems like if the original part is deleted, the data will be removed from the projection as well?
It's currently possible by using CollapsingMergeTree with default `sign = 1`. We w... | 2021-05-04T15:31:43 | 2021-05-04T15:31:43 | {} | COLLABORATOR |
ClickHouse | ClickHouse | 832,120,598 | 23,877 | alexey-milovidov | 0890279 | 2021-05-04T17:38:33 | 2021-05-04T17:38:33 | {} | MEMBER |
ClickHouse | ClickHouse | 832,715,294 | 21,839 | kssenii | @tobiaslins
> open connections (30+) after several minutes
these are from several different dictionaries? or just one? | 2021-05-05T14:04:50 | 2021-05-05T14:06:46 | {} | MEMBER |
ClickHouse | ClickHouse | 832,815,185 | 20,539 | myrrc | Measures with full build:
- Testing takes about 4.88h (328 errors! 2358 tests passed. 300 tests skipped. 17555.63 s elapsed), but
- CI has more powerful hardware (so we could increase the thread pool size from 16 (current) to 24 for initial symbolization and from 10 to 15 for parallel test dumping).
- Som... | 2021-05-05T16:03:57 | 2021-05-05T16:03:57 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 833,018,032 | 23,901 | abyss7 | Yes, it looks like the code is non-conformant without a reason. | 2021-05-05T21:26:04 | 2021-05-05T21:26:04 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 833,286,522 | 17,956 | zba | experiencing same issue using "WITH clause" as alias - query can execute as usual, or give this error, usually it stuck for some time and start to work after sometime (like something cached). It is something similar or need further investigation ? | 2021-05-06T07:09:41 | 2021-05-06T07:09:41 | {} | NONE |
ClickHouse | ClickHouse | 832,149,652 | 23,656 | alexey-milovidov | @kssenii Also a thought: as we output progress to stderr, we can check if stderr is a terminal and obtain it's width. | 2021-05-04T18:24:28 | 2021-05-04T18:24:28 | {} | MEMBER |
ClickHouse | ClickHouse | 832,287,651 | 23,889 | den-crane | ```
create user foo identified by '123' ;
CREATE QUOTA OR REPLACE my_quota FOR INTERVAL 4 minute MAX queries = 5 TO foo;
create table a(A Int64) Engine=Log;
grant select on a to foo;
| => clickhouse-client -u foo --password=123 -q 'select * from a'
| => clickhouse-client -u foo --password=123 -q 'select ... | 2021-05-04T22:22:26 | 2021-05-04T22:22:26 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 832,592,829 | 23,874 | alexey-milovidov | Broken test: #23751 | 2021-05-05T10:47:28 | 2021-05-05T10:47:28 | {} | MEMBER |
ClickHouse | ClickHouse | 832,132,319 | 22,398 | kimmolinna | Clickhouse vs. InfluxDB vs. Timescale vs. OpenTSDB
https://www.sciencedirect.com/science/article/pii/S1877050919310439/pdf?md5=ecc390cfc9d0be432ca0c218985c94d5&pid=1-s2.0-S1877050919310439-main.pdf | 2021-05-04T17:57:16 | 2021-05-04T17:57:16 | {
"+1": 1
} | NONE |
ClickHouse | ClickHouse | 832,137,927 | 23,827 | alexey-milovidov | Most likely this feature will not be implemented (due to increase of internal complexity).
It's expected that you should be able to refer to columns with dots if the name is wrapped in quotes:
```
"abc.def"
or
`abc.def`
```
But it is likely to break in some places in code.
PS. The proposed feature is going ... | 2021-05-04T18:06:00 | 2021-05-04T18:07:29 | {} | MEMBER |
ClickHouse | ClickHouse | 832,170,244 | 23,730 | alexey-milovidov | @vsop-479
I checked the source code... the settings should apply but only when client reconnects.
Please check if it is the case.
BTW, what driver and interface the clients use? | 2021-05-04T18:57:42 | 2021-05-04T18:57:42 | {} | MEMBER |
ClickHouse | ClickHouse | 832,729,522 | 23,895 | den-crane | it's expected behaviour. Mat.View watches inserts only into the left table (readings in your case).
https://den-crane.github.io/Everything_you_should_know_about_materialized_views_commented.pdf
| 2021-05-05T14:22:59 | 2021-05-05T14:22:59 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 832,733,241 | 23,892 | den-crane | Check CH log for "having unreserved " messages around that merges.
This is expected behaviour. Because merges book twice space on a disk, probably it was 0 free space available in MERGE POOL at hot disk.
>It is less than 70% used.
probably it was 0% in the MERGE POOL | 2021-05-05T14:27:30 | 2021-05-05T14:27:30 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 831,338,952 | 22,988 | ucasfl | This doesn't fixed my problem #21957.
I'm a little confused about the `ReadBuffer` implementation.
I use another naive way to solve this problem, which works well: I use one variable named `already_read` both in `ReadBufferFromIStream` and `ReadBufferFromS3` to record already read bytes, when exception happene... | 2021-05-03T15:26:15 | 2021-05-03T15:26:15 | {} | COLLABORATOR |
ClickHouse | ClickHouse | 832,025,784 | 4,419 | arctica | @amosbird projections seem like a very useful and interesting feature because they solve the (big) limitations of materialized views but I'm not sure if they could be used to serve the same purpose as my suggestion.
The feature I describe can reduce storage costs a lot because it aggregates the data of the table dur... | 2021-05-04T15:20:25 | 2021-05-04T15:20:25 | {} | NONE |
ClickHouse | ClickHouse | 833,305,253 | 23,911 | ucasfl | > @ucasFL Can you reproduce it on some test?
I don't know how to write test to reproduce it. Could you give me some help? | 2021-05-06T07:41:28 | 2021-05-06T07:41:28 | {} | COLLABORATOR |
ClickHouse | ClickHouse | 831,865,483 | 23,876 | den-crane | Just remove brackets `GROUP BY timestamp_h, item_id`
```
CREATE TABLE delme
(
timestamp DateTime DEFAULT now(),
timestamp_h DateTime materialized toStartOfHour(timestamp),
item_id UInt64,
bytes UInt64
)
ENGINE=MergeTree()
PARTITION BY toYYYYMMDD(timestamp)
ORDER BY (timestamp_h, item_id)
TTL time... | 2021-05-04T11:17:59 | 2021-05-04T11:17:59 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 832,140,774 | 23,827 | gyfis | @alexey-milovidov Thank you for the swift response! I thought this could get implemented, because the dots are a specific-enough feature (only used by Nested and Array types), but I can see the argument against adding this flexibility to ClickHouse. | 2021-05-04T18:10:22 | 2021-05-04T18:10:22 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 832,284,073 | 23,889 | den-crane | against a real table ?
```
clickhouse-client -u foo --password=123 -q 'select * from system.numbers limit 1'
Received exception from server (version 21.5.1):
Code: 201. DB::Exception: Received from localhost:9000. DB::Exception: Quota for user `foo` for 240s has been exceeded: queries = 8/5. Interval will end at ... | 2021-05-04T22:13:03 | 2021-05-04T22:13:13 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 832,286,223 | 23,889 | OmarBazaraa | > against a real table ?
Yes, I was querying against a real table. Something like this:
```sql
SELECT count(), currentUser() from my_table
``` | 2021-05-04T22:18:35 | 2021-05-04T22:18:35 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 832,286,664 | 23,889 | den-crane | `count(), currentUser() ` don't read the table. Only metadata from memory from system.parts | 2021-05-04T22:19:50 | 2021-05-04T22:20:00 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 832,734,237 | 23,891 | den-crane | Do you use WindowsForLinux? | 2021-05-05T14:28:39 | 2021-05-05T14:28:39 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 832,901,844 | 22,121 | vzakaznikov | There is still a bug https://github.com/ClickHouse/ClickHouse/issues/23902
| 2021-05-05T18:11:29 | 2021-05-05T18:11:29 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 832,903,129 | 23,880 | alexey-milovidov | Ok.
I'm afraid of adding more settings without good reasons.
Many clients tend to adjust some settings for the reasons of fear, uncertainty and doubt. | 2021-05-05T18:13:37 | 2021-05-05T18:13:37 | {
"+1": 1
} | MEMBER |
ClickHouse | ClickHouse | 833,118,536 | 23,907 | den-crane | Try Tuple
```
CREATE TABLE test
(
`a` SimpleAggregateFunction(argMax, Tuple(UInt8, UInt64))
)
ENGINE = AggregatingMergeTree
PARTITION BY a
ORDER BY a
``` | 2021-05-05T23:44:53 | 2021-05-05T23:44:53 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 830,580,005 | 19,627 | filimonov | > Only this will give significant boost for SELECT (i.e. avoid reading that part of data at all), I don't know does it will be often or not, but if someone care about this, this should be taken into consideration.
Merge/optimize final will resolve that problem in a natural way.
I agree that this may be an improv... | 2021-05-01T08:26:47 | 2021-05-03T16:00:46 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 831,877,948 | 23,875 | alexey-milovidov | You need to provide the config file in arguments:
`~/ClickHouse/build/programs/clickhouse-server --config-file ~/ClickHouse/programs/server/config.xml`
| 2021-05-04T11:42:33 | 2021-05-04T11:42:33 | {
"+1": 1
} | MEMBER |
ClickHouse | ClickHouse | 831,889,436 | 23,822 | alexey-milovidov | ```
Неправильная строка для changelog.
Если попадает в категорию bugfix - необходимо написать, какое поведение ClickHouse с точки зрения пользователя было ошибочным (какая ошибка исправлена).
Внутренности кода никого не волнуют, их не надо писать в changelog.
Ты написал:
> Better handling of URI's in Poco... | 2021-05-04T12:04:10 | 2021-05-04T12:04:10 | {} | MEMBER |
ClickHouse | ClickHouse | 831,302,187 | 21,414 | amosbird | Intergration tests look weird. | 2021-05-03T14:37:04 | 2021-05-03T14:37:04 | {
"eyes": 1
} | COLLABORATOR |
ClickHouse | ClickHouse | 831,599,322 | 23,703 | alexey-milovidov | @MyroTk These functions are not supposed to be implemented at all. | 2021-05-03T23:33:36 | 2021-05-03T23:33:51 | {} | MEMBER |
ClickHouse | ClickHouse | 832,410,601 | 23,872 | sanjaypocket52 | Den, Any solution in clickhouse to perform aggregation operations for tables who got update / insertion both. | 2021-05-05T04:57:05 | 2021-05-05T05:31:43 | {} | NONE |
ClickHouse | ClickHouse | 832,835,463 | 23,896 | alexey-milovidov | > SQLancer test — SQLancer test run. See report
SQLancer generated invalid query - @qoega. | 2021-05-05T16:31:46 | 2021-05-05T16:31:46 | {} | MEMBER |
ClickHouse | ClickHouse | 832,873,802 | 23,901 | alexey-milovidov | Yes, I think we should accept it. | 2021-05-05T17:27:13 | 2021-05-05T17:27:13 | {} | MEMBER |
ClickHouse | ClickHouse | 832,904,631 | 23,880 | alexey-milovidov | I also don't like when users have to tune ClickHouse instead of using default settings. | 2021-05-05T18:15:48 | 2021-05-05T18:15:48 | {
"+1": 1
} | MEMBER |
ClickHouse | ClickHouse | 833,120,494 | 23,904 | alexey-milovidov | If you don't see these requests in ClickHouse logs, it must be not ClickHouse, who responses with 502 code. | 2021-05-05T23:49:56 | 2021-05-05T23:49:56 | {} | MEMBER |
ClickHouse | ClickHouse | 811,963,037 | 22,398 | alexey-milovidov | ClickHouse vs. kdb+ at Deutsche Bank:
https://youtu.be/0lE-a3TIC5M?t=1036 | 2021-04-01T14:52:47 | 2021-04-01T14:52:47 | {} | MEMBER |
ClickHouse | ClickHouse | 811,992,724 | 22,398 | alex-zaitsev | > ClickHouse vs. Redshift on NYC taxi dataset:
> https://altinitydb.medium.com/clickhouse-vs-amazon-redshift-benchmark-e223429f4f95
Superceeded by https://altinity.com/blog/clickhouse-and-redshift-face-off-again-in-nyc-taxi-rides-benchmark | 2021-04-01T15:37:45 | 2021-04-01T15:37:45 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 812,018,891 | 22,431 | mdianjun | Got it, Thank you! | 2021-04-01T16:19:48 | 2021-04-01T16:19:48 | {
"eyes": 1
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 812,147,591 | 22,470 | alesapin | Waiting for tests but everything should become much more stable. | 2021-04-01T20:16:42 | 2021-04-01T20:16:42 | {} | MEMBER |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.