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 | 726,055,277 | 16,838 | akuzm | Turns out I misunderstood how merge works -- it uses a constant amount of memory, so the size of parts to merge is not relevant. | 2020-11-12T12:44:00 | 2020-11-12T12:44:00 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 726,104,942 | 16,865 | filimonov | @alexey-milovidov merge? (to backport to today's releases?) | 2020-11-12T14:16:37 | 2020-11-12T14:16:37 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 726,142,843 | 16,922 | den-crane | try to disable distributed_aggregation_memory_efficient as well. | 2020-11-12T15:18:17 | 2020-11-12T15:18:17 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 726,157,441 | 16,932 | nikitamikhaylov | > > ~3000 parts.
>
>
>
> @nikitamikhaylov do we use pieces for tiny partitions too?
Yes. But we split only partitions into pieces. How many partitions are there? | 2020-11-12T15:41:25 | 2020-11-12T15:41:25 | {} | MEMBER |
ClickHouse | ClickHouse | 725,798,753 | 16,857 | lishoulian | > If you set `max_bytes_before_external_group_by` lower, say, 50 GB -- does it still fail? 10 GB?
>
> My guess is that this setting is based on the size of input rows and not the amount of required memory, which might turn out to be larger.
I set it to a smaller value, and the same error is reported.
There's 6... | 2020-11-12T03:03:58 | 2020-11-12T17:57:29 | {} | NONE |
ClickHouse | ClickHouse | 726,333,831 | 16,954 | Avogar | I will add tests at the beginning of the next week. Let see if performance of insert will change or if some stateless tests will fail with `optimize_on_insert` enabled as default. | 2020-11-12T20:48:19 | 2020-11-12T20:55:11 | {} | MEMBER |
ClickHouse | ClickHouse | 726,416,888 | 16,957 | l1t1 | if i add having clause, it raises error quicklyer
```
localhost :) drop table sfzcm;
DROP TABLE sfzcm
Ok.
0 rows in set. Elapsed: 0.004 sec.
localhost :) create table sfzcm engine=MergeTree()order by id as select id,count(*)c from sfz15y group by id having count(*)>1;
CREATE TABLE sfzcm
ENGINE = Merg... | 2020-11-13T00:02:31 | 2020-11-13T00:02:31 | {} | NONE |
ClickHouse | ClickHouse | 726,478,745 | 16,957 | l1t1 | following sql runs more than 15 minutes.
```
create table sfzct engine=MergeTree()order by id as select id,count(*)c from sfz15y group by id;
``` | 2020-11-13T03:08:08 | 2020-11-13T03:08:08 | {} | NONE |
ClickHouse | ClickHouse | 726,493,530 | 15,087 | fastio | > @fastio
> Your suggestion about extending parameters list of ReplicatedMergeTree() with immediate declaration of ZK endpoints in SQL statement is quite interesting, although we would need to `ALTER TABLE zk_endpoint` every time any of ZK-node addresses/ports change.
>
> I would say that for now (as long as PR #14... | 2020-11-13T04:01:29 | 2020-11-13T04:01:29 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 725,957,098 | 15,713 | ghost | > What if the order is defined on the finalized columns of aggregation?
To clarify my understanding I have a question.
Is it means that a query has to contain `ORDER BY clause` to reduce high-cardinality intermediate states? If then, it's OK. The query I imagine is as below.
```SQL
SELECT page, browser, os, count... | 2020-11-12T09:28:10 | 2020-11-12T09:31:26 | {} | NONE |
ClickHouse | ClickHouse | 726,015,638 | 7,127 | filimonov | https://github.com/bitquery/activecube-graphql | 2020-11-12T11:19:22 | 2020-11-12T11:19:22 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 726,485,677 | 16,926 | Zhile | @filimonov Can you help to add a bug label or assign to someone to address this?
I think it's very common for the users of Kafka table engine. And it's really normal in current Cloud situation where the url can be an AWS Route53 and the Route53 can be bond to an ELB which the IP could possibly change. | 2020-11-13T03:29:40 | 2020-11-13T03:29:40 | {} | NONE |
ClickHouse | ClickHouse | 726,542,150 | 16,848 | trollhe | > Do you have some sample data + queries that we can use to reproduce the problem?
thank you,my data and query :
data sample
data_file.gz
```
aaa|bbbb|cccc|ddd|eeee
```
import script
```
insert into logs.user_logs_real
select * from file('user_logs_real/*2020061113220100*.gz','CSV','aa String,bb Strin... | 2020-11-13T06:04:52 | 2020-11-13T06:04:52 | {} | NONE |
ClickHouse | ClickHouse | 725,932,832 | 16,843 | azat | @alexey-milovidov can you please take a look? (it is "required" for more extensive tests of #12601) | 2020-11-12T08:44:39 | 2020-11-12T08:44:39 | {} | MEMBER |
ClickHouse | ClickHouse | 725,976,706 | 15,713 | amosbird | > So that there will be loss of accuracy.
It's more like completely random, which heavily depends on how the data is sharded.
> How about ClickHouse optionally supports Limit Pushdown as Citus does?
I'm not aware of any limit pushdown w.r.t aggregations. | 2020-11-12T10:04:10 | 2020-11-12T10:04:10 | {} | COLLABORATOR |
ClickHouse | ClickHouse | 725,806,653 | 16,857 | lishoulian | > See [#10461 (comment)](https://github.com/ClickHouse/ClickHouse/issues/10461#issuecomment-623006030)
So how am I supposed to fix it?The format of my table needs to be changed？ | 2020-11-12T03:30:53 | 2020-11-12T17:57:44 | {} | NONE |
ClickHouse | ClickHouse | 726,621,548 | 16,587 | KochetovNicolai | Well, it is just an integer overflow which happens when `sumMap` state is serialized. Cause value type is `UInt8`.
```
SELECT sumMap(sm)
FROM
(
SELECT sumMap([1], [toUInt64(1)]) AS sm
FROM my_table
GROUP BY
col1,
col2
)
SETTINGS max_bytes_before_external_group_by = 100000
Quer... | 2020-11-13T08:46:16 | 2020-11-13T08:46:16 | {} | MEMBER |
ClickHouse | ClickHouse | 726,705,347 | 16,980 | akuzm | ```
SELECT fuzzBits('stringstring', 0.5) from ( SELECT 1 AS x UNION ALL SELECT NULL ) group by x
2020.11.13 14:09:29.862241 [ 1659982 ] {} <Trace> BaseDaemon: Received signal 11
2020.11.13 14:09:29.864007 [ 1660071 ] {} <Fatal> BaseDaemon: ########################################
2020.11.13 14:09:29.865355 [ ... | 2020-11-13T11:11:08 | 2020-11-13T11:11:08 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 725,906,247 | 16,372 | lostsnow | >
>
> Try to build the newest version?
Upgrade to version 20.11.2.1 (official build) still has this error. | 2020-11-12T07:55:08 | 2020-11-12T07:55:08 | {} | NONE |
ClickHouse | ClickHouse | 725,981,565 | 16,875 | alesapin | `00804_test_delta_codec_compression` -- flaky test? | 2020-11-12T10:13:24 | 2020-11-12T10:13:24 | {} | MEMBER |
ClickHouse | ClickHouse | 726,059,783 | 16,920 | filimonov | Work is in progress. Check the PR https://github.com/ClickHouse/ClickHouse/pull/16621
You can help the community by testing [builds](https://clickhouse-builds.s3.yandex.net/16621/7f6e6296a1fdcc8e4162ab01da24dfcc9e895ae9/clickhouse_build_check/report.html) from that PR, and sending your feedback.
Or just wait till... | 2020-11-12T12:53:43 | 2020-11-12T12:53:43 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 726,222,657 | 16,925 | BaptisteBdn | From what I understood, the Index created with `ORDER BY (name, timestamp, intHash32(timestamp))` would be different from the one I created manually with `INDEX timestamp_idx timestamp TYPE minmax GRANULARITY 4`.
> Does it help at all?
Honestly I'm not sure, I'm still experimenting with clickhouse right now so I'... | 2020-11-12T17:27:16 | 2020-11-12T17:27:16 | {} | NONE |
ClickHouse | ClickHouse | 726,299,505 | 16,390 | alesapin | No relevant failures. | 2020-11-12T19:42:24 | 2020-11-12T19:42:24 | {} | MEMBER |
ClickHouse | ClickHouse | 726,399,083 | 16,725 | nikitamikhaylov | Hope it will be finally merged in #16883 | 2020-11-12T23:14:39 | 2020-11-12T23:14:39 | {} | MEMBER |
ClickHouse | ClickHouse | 726,453,833 | 16,857 | lishoulian | > > > See [#10461 (comment)](https://github.com/ClickHouse/ClickHouse/issues/10461#issuecomment-623006030)
> >
> >
> > So how am I supposed to fix it?The format of my table needs to be changed？
>
> There are no good fixed for now. Use another format, or smaller files, or install more ram.
Okay, thank you，fil... | 2020-11-13T01:45:54 | 2020-11-13T01:45:54 | {} | NONE |
ClickHouse | ClickHouse | 725,982,612 | 16,914 | dependabot-preview[bot] | OK, I won't notify you again about this release, but will get in touch when a new version is available. If you'd rather skip all updates until the next major or minor version, let me know by commenting `@dependabot ignore this major version` or `@dependabot ignore this minor version`.
If you change your mind, just re-... | 2020-11-12T10:15:17 | 2020-11-12T10:15:17 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 726,131,917 | 16,907 | kmatt | Example works as documented on 20.11.2.1 | 2020-11-12T15:01:23 | 2020-11-12T15:01:23 | {} | NONE |
ClickHouse | ClickHouse | 726,538,126 | 16,850 | trollhe | > You have to also configure a Storage Policy that determines how the data is spread across the disks.
> Please refer to this guide: https://altinity.com/blog/2019/11/29/amplifying-clickhouse-capacity-with-multi-volume-storage-part-2
thank u | 2020-11-13T05:53:40 | 2020-11-13T05:53:40 | {} | NONE |
ClickHouse | ClickHouse | 726,691,764 | 16,968 | alesapin | Nice flaky check catch. | 2020-11-13T10:42:45 | 2020-11-13T10:42:45 | {} | MEMBER |
ClickHouse | ClickHouse | 726,708,781 | 16,946 | filimonov | ```
addr2line -afiCe /usr/bin/clickhouse 0xdec92a9 0xdec7dbd 0xe568ea6 0xe575ca7 0x10d4dd6f 0x10d4f77e 0x10e80a39 0x10e7c96a 0x7f180ee80fa3 0x7f180eda24c
0x000000000dec92a9
DB::executeQueryImpl(char const*, char const*, DB::Context&, bool, DB::QueryProcessingStage::Enum, bool, DB::ReadBuffer*)
/build/obj-x86_64-li... | 2020-11-13T11:18:25 | 2020-11-13T11:23:49 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 725,996,660 | 16,813 | akuzm | I fixed the 01293_show_clusters in the branch, it's not related to the changes. | 2020-11-12T10:41:32 | 2020-11-12T10:41:32 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 726,013,787 | 16,869 | filimonov | One more test:
```
SELECT timeSeriesGroupSumMerge(x)
FROM
(
SELECT timeSeriesGroupSumState(id, ts, val) AS x
FROM values('id UInt64, ts Int64, val Float64', (1, 1, 1), (2, 50, -50), (1, 100, 100))
UNION ALL
SELECT timeSeriesGroupSumState(id, ts, val) AS x
FROM values('id UInt64, ts Int64, ... | 2020-11-12T11:15:39 | 2020-11-12T11:15:39 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 726,137,236 | 16,931 | filimonov | @alexey-milovidov i suppose that
`if (!fs::exists(main_config_file))`
doesn't work as expected becase that config file is installed by a deb package, see:
https://github.com/ClickHouse/ClickHouse/blob/master/debian/clickhouse-server.install | 2020-11-12T15:09:51 | 2020-11-12T15:09:51 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 726,142,086 | 16,922 | den-crane | @yard you are describing very weird things.
group_by_two_level_threshold_bytes=1 --- forces two-level-aggregation
group_by_two_level_threshold_bytes=0 --- disables two-level-aggregation
Please show examples with a sample data. | 2020-11-12T15:17:09 | 2020-11-12T15:17:09 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 726,600,371 | 16,922 | yard | @den-crane I am trying to prepare an isolated docker-compose stack + seed data to reliably reproduce the issue. Do you know if there is any way to force shards to return data from queries like `SELECT name, count(*) FROM table GROUP BY name` in a form of multiple blocks (since judging by the logs this seems to be the t... | 2020-11-13T08:27:07 | 2020-11-13T08:27:07 | {} | NONE |
ClickHouse | ClickHouse | 726,713,042 | 16,896 | abyss7 | Yandex check failure looks unrelated | 2020-11-13T11:28:05 | 2020-11-13T11:28:05 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 725,904,136 | 16,395 | macEar | @Jokser thanks! | 2020-11-12T07:50:26 | 2020-11-12T07:50:26 | {} | NONE |
ClickHouse | ClickHouse | 725,973,574 | 15,713 | ghost | It is hard to get the correct order without the final cnt.
So that there will be loss of accuracy.
How about ClickHouse optionally supports `Limit Pushdown` as [Citus](https://docs.citusdata.com/en/v4.x/user_guide/limit_pushdown.html) does? | 2020-11-12T09:58:40 | 2020-11-12T09:58:40 | {} | NONE |
ClickHouse | ClickHouse | 725,982,601 | 16,801 | dependabot-preview[bot] | OK, I won't notify you again about this release, but will get in touch when a new version is available. If you'd rather skip all updates until the next major or minor version, let me know by commenting `@dependabot ignore this major version` or `@dependabot ignore this minor version`.
If you change your mind, just re-... | 2020-11-12T10:15:16 | 2020-11-12T10:15:16 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 726,132,841 | 16,931 | filimonov | Actually it was created automatically before during deb installation... | 2020-11-12T15:02:54 | 2020-11-12T15:02:54 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 726,180,824 | 16,587 | KochetovNicolai | ```
CREATE TABLE my_table3
(
`col1` DateTime,
`col2` Int64
)
ENGINE = MergeTree()
ORDER BY col1
SETTINGS index_granularity = 8192
Query id: 837aa6a7-b03f-4060-a565-af2d53ab4f46
Ok.
0 rows in set. Elapsed: 0.024 sec.
:) insert into my_table3 select '2020-10-26 00:00:00', 70724110 from num... | 2020-11-12T16:18:33 | 2020-11-12T16:18:33 | {} | MEMBER |
ClickHouse | ClickHouse | 725,908,693 | 16,621 | ilejn | > BTW - i think that libhdfs can use both gsasl and cyrus sasl, may be we should switch to cyrus everywhere?
>
> ( we use cyrus for Kafka and for openldap and there is no choice there)
I kind of agree with you, it is strange to have two libraries with same functionality, although
1. I do not see any evidence tha... | 2020-11-12T08:00:15 | 2020-11-12T08:00:15 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 725,999,172 | 16,923 | zhang2014 | The reason for this problem may be binlog_checksum = NONE, you can check it with SHOW VARIABLES WHERE Variable_name ='binlog_checksum'; | 2020-11-12T10:46:36 | 2020-11-12T10:46:36 | {
"+1": 1
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 726,179,198 | 16,932 | filimonov | But probably that's not the reason: "ClickHouse server version - 20.3.12.112" (i.e. still uses old implementation of copier). | 2020-11-12T16:15:51 | 2020-11-12T16:15:51 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 726,399,643 | 16,593 | nikitamikhaylov | Hope it will be finally merged in #16883 | 2020-11-12T23:16:06 | 2020-11-12T23:16:06 | {} | MEMBER |
ClickHouse | ClickHouse | 726,528,766 | 15,598 | akuzm | The `ReadonlyReplica` metric was fixed in https://github.com/ClickHouse/ClickHouse/pull/15592 | 2020-11-13T05:28:51 | 2020-11-13T05:28:51 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 725,959,770 | 15,713 | amosbird | > a query has to contain ORDER BY clause to reduce high-cardinality intermediate states
But how do you get the correct order when you don't have the final cnt? | 2020-11-12T09:33:06 | 2020-11-12T09:33:06 | {} | COLLABORATOR |
ClickHouse | ClickHouse | 725,982,599 | 16,802 | dependabot-preview[bot] | OK, I won't notify you again about this release, but will get in touch when a new version is available. If you'd rather skip all updates until the next major or minor version, let me know by commenting `@dependabot ignore this major version` or `@dependabot ignore this minor version`.
If you change your mind, just re-... | 2020-11-12T10:15:16 | 2020-11-12T10:15:16 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 725,985,928 | 16,372 | lostsnow | ```
+------------------------+------------+
| COLUMN_TYPE | COLUMN_KEY |
+------------------------+------------+
| bigint(20) unsigned | PRI |
| bigint(20) unsigned | MUL |
| tinyint(4) | |
| decimal(10,2) unsigned | |
| datetime | ... | 2020-11-12T10:21:30 | 2020-11-12T10:21:30 | {} | NONE |
ClickHouse | ClickHouse | 725,988,051 | 16,372 | lostsnow | DISTINCT records
```
mysql> select DISTINCT COLUMN_TYPE, COLUMN_KEY from COLUMNS where TABLE_SCHEMA='xxx';
+------------------------+------------+
| COLUMN_TYPE | COLUMN_KEY |
+------------------------+------------+
| bigint(20) unsigned | PRI |
| bigint(20) unsigned | MUL |
| t... | 2020-11-12T10:25:21 | 2020-11-12T10:25:21 | {} | NONE |
ClickHouse | ClickHouse | 726,135,709 | 16,931 | filimonov | And I guess it still expected to be created, but the folder is absent...
https://github.com/ClickHouse/ClickHouse/blob/e0dc004b95b1c89b252d6e0a7ec9b3342bb7b548/programs/install/Install.cpp#L324-L339
Relates to #13528 | 2020-11-12T15:07:32 | 2020-11-12T15:07:32 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 726,135,800 | 16,922 | yard | Hi Denis! You are saving the day, as always 👍
Setting _even just_ `group_by_two_level_threshold_bytes` to 1 seems to be making the problem go away.
Extra observation: setting `group_by_two_level_threshold_bytes` to 1 makes the resultset be composed of multiple blocks (i.e. `clickhouse-client` prints the result... | 2020-11-12T15:07:39 | 2020-11-12T15:07:51 | {} | NONE |
ClickHouse | ClickHouse | 726,184,650 | 16,890 | filimonov | I guess it fixed by https://github.com/ClickHouse/ClickHouse/pull/15536
Please update your ClickHouse version. | 2020-11-12T16:25:09 | 2020-11-12T16:25:09 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 726,699,580 | 16,944 | l1t1 | CREATE TABLE table2 AS select number from numbers(5)
CREATE TABLE table3 AS select number from table2 | 2020-11-13T10:59:11 | 2020-11-13T10:59:11 | {} | NONE |
ClickHouse | ClickHouse | 725,957,568 | 16,372 | TCeason | > > Could you provide you table structure under the mysql database？
>
> Over 70 tables, and I don't know which one's causing the problems.
I think you can use information_schema to verify which type of cols under the table.
e.g.
```
root@mysqldb 17:27:36 [information_schema]> select COLUMN_TYPE, COLUMN_KEY... | 2020-11-12T09:29:00 | 2020-11-12T09:29:00 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 725,972,591 | 16,877 | alesapin | Seems like something wrong with the server restart. ZooKeeper protocol works surprisingly well. | 2020-11-12T09:56:47 | 2020-11-12T09:56:47 | {} | MEMBER |
ClickHouse | ClickHouse | 726,029,183 | 15,977 | TCeason | @tavplubix Hi dear, CI report:
Fast test — Cannot finish cmake
Why does it happen? I have already run `git submodule sync --recursive && git submodule update --init --recursive;` on my instance. And It also builds successed. | 2020-11-12T11:47:17 | 2020-11-12T11:47:17 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 726,052,142 | 16,926 | filimonov | Yes, your finding look correct.
https://github.com/ClickHouse/ClickHouse/blob/fd84d16387e507763a9dfa661397c385d7ab6931/src/Processors/Formats/Impl/AvroRowInputFormat.cpp#L646-L647
@oandrew do you want to take a look at that?
| 2020-11-12T12:37:19 | 2020-11-12T12:37:19 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 726,057,401 | 16,926 | Zhile | If the URL changed the IP, then the calls will always fail and we don't have anything to do rather than restart the server. It's very risky and very heavy. | 2020-11-12T12:48:27 | 2020-11-12T12:48:27 | {} | NONE |
ClickHouse | ClickHouse | 726,121,324 | 16,895 | alesapin | Discussed in telegram -- need tiny fixes. | 2020-11-12T14:44:08 | 2020-11-12T14:44:08 | {} | MEMBER |
ClickHouse | ClickHouse | 726,154,241 | 16,932 | filimonov | > ~3000 parts.
@nikitamikhaylov do we use pieces for tiny partitions too? | 2020-11-12T15:36:05 | 2020-11-12T15:36:27 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 726,192,847 | 16,865 | alexey-milovidov | Testflows has failed. | 2020-11-12T16:38:44 | 2020-11-12T16:38:44 | {} | MEMBER |
ClickHouse | ClickHouse | 726,262,484 | 16,933 | shawel | @filimonov . yes most definitely. insert selects! we have batch we run that stores results in S3. | 2020-11-12T18:37:24 | 2020-11-12T18:37:24 | {} | NONE |
ClickHouse | ClickHouse | 726,540,385 | 9,071 | moon03432 | I also encountered this problem, the version is 20.4.8.99 | 2020-11-13T06:00:10 | 2020-11-13T06:00:10 | {} | NONE |
ClickHouse | ClickHouse | 726,558,936 | 16,968 | amosbird | Hmm,
```
docker: Error response from daemon: toomanyrequests: You have reached your pull rate limit. You may increase the limit by authenticating and upgrading: https://www.docker.com/increase-rate-limit.
```
| 2020-11-13T06:49:22 | 2020-11-13T06:49:22 | {} | COLLABORATOR |
ClickHouse | ClickHouse | 726,136,795 | 16,916 | den-crane | No difference.
Nested it's a syntax sugar for a `create table` command and it's stored as arrays.
Nested creates columns with a dot in names (xxx.key , xxx.value). Such columns are treated in a special way during insert. CH validates that a number of elements is the same in sub-columns of xxx.
https://github.com... | 2020-11-12T15:09:13 | 2020-11-12T15:09:56 | {
"+1": 4,
"heart": 2
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 726,416,499 | 16,956 | CurtizJ | Also, there was a small bug, that optimization removed monotonous functions, but left duplicates of columns. E.g. `ORDER BY toDate(timestamp), timestamp` was transformed to `ORDER BY timestamp, timestamp` | 2020-11-13T00:01:15 | 2020-11-13T00:02:54 | {} | MEMBER |
ClickHouse | ClickHouse | 726,475,931 | 16,957 | l1t1 | it works, thanks, but i have some questions,
1. when i run the sql, it processed to 98 % very quickly, and then a very long time to process the last 2 %. why?
2. i only has 64GB system memory, why i can set it to 200GB, and why group by needs so large size.
```
localhost :) set max_bytes_before_external_group_by = ... | 2020-11-13T02:58:53 | 2020-11-13T02:58:53 | {} | NONE |
ClickHouse | ClickHouse | 726,493,255 | 15,087 | fastio | > > > In production environment, hundreds of machines hold thousands of ReplicatedMergeTree tables.
> > > The only zookeeper cluster is seriously overloaded even good performance machine used.
> > > In some scenarios, we want to change the current zookeeper cluster to another one smoothly.
> >
> >
> > I guess it... | 2020-11-13T04:00:20 | 2020-11-13T04:00:20 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 725,982,604 | 16,913 | dependabot-preview[bot] | OK, I won't notify you again about this release, but will get in touch when a new version is available. If you'd rather skip all updates until the next major or minor version, let me know by commenting `@dependabot ignore this major version` or `@dependabot ignore this minor version`.
If you change your mind, just re-... | 2020-11-12T10:15:16 | 2020-11-12T10:15:16 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 726,129,771 | 16,922 | den-crane | try to reproduce with test data and `SETTINGS group_by_two_level_threshold_bytes = 1, group_by_two_level_threshold = 1`
https://github.com/ClickHouse/ClickHouse/issues/13331
https://github.com/ClickHouse/ClickHouse/issues/9134
| 2020-11-12T14:58:01 | 2020-11-12T14:58:01 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 726,159,577 | 16,922 | yard | I think you confused `group_by_two_level_threshold_bytes` and `group_by_two_level_threshold`. The initial value of `group_by_two_level_threshold_bytes` I have in `system.settings` is 100000000.
`distributed_aggregation_memory_efficient` is set to 0 already in my setup.
Unfortunately, I am not able to reliably rep... | 2020-11-12T15:44:47 | 2020-11-12T15:48:07 | {} | NONE |
ClickHouse | ClickHouse | 726,240,788 | 16,857 | filimonov | > > See [#10461 (comment)](https://github.com/ClickHouse/ClickHouse/issues/10461#issuecomment-623006030)
>
> So how am I supposed to fix it?The format of my table needs to be changed？
There are no good fixed for now. Use another format, or smaller files, or install more ram. | 2020-11-12T17:58:40 | 2020-11-12T17:58:40 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 726,298,821 | 16,934 | alesapin | Wow, it's not flaky at all! | 2020-11-12T19:41:01 | 2020-11-12T19:41:01 | {} | MEMBER |
ClickHouse | ClickHouse | 726,365,186 | 16,946 | filimonov | 1) for segfault - please look for a query which lead to the error - grep 71a0dc3b-9e55-4d49-9312-b8a1
de6a027e
2) for netlink error - may be you run in docker w/o CAP_NET_ADMIN (but Afair it should not be a fatal error)
3) about non-starting clickhouse - it seems there is an issue with loading one of the tables.... | 2020-11-12T21:58:34 | 2020-11-12T21:58:34 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 726,418,899 | 16,957 | den-crane | https://clickhouse.tech/docs/en/operations/settings/query-complexity/#settings-max_bytes_before_external_group_by
`set max_bytes_before_external_group_by = 20000000000;` | 2020-11-13T00:08:36 | 2020-11-13T00:08:36 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 725,924,356 | 16,372 | TCeason | Could you provide you table structure under the mysql database？ | 2020-11-12T08:28:31 | 2020-11-12T08:28:31 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 725,982,600 | 16,910 | dependabot-preview[bot] | OK, I won't notify you again about this release, but will get in touch when a new version is available. If you'd rather skip all updates until the next major or minor version, let me know by commenting `@dependabot ignore this major version` or `@dependabot ignore this minor version`.
If you change your mind, just re-... | 2020-11-12T10:15:16 | 2020-11-12T10:15:16 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 726,054,383 | 16,926 | Zhile | @filimonov @oandrew Do you know how can we drop these caches? | 2020-11-12T12:42:10 | 2020-11-12T12:42:10 | {} | NONE |
ClickHouse | ClickHouse | 726,149,266 | 16,933 | filimonov | I guess you are not talking about plain selects? Rather `insert ... select` or other long-running queries?
Because for plain select we can't store result on the server waiting while the client will come to get that. | 2020-11-12T15:28:05 | 2020-11-12T15:29:30 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 726,167,762 | 16,621 | filimonov | BTW
hadoop has now official native c++ client which looks quite active
https://github.com/apache/hadoop/tree/trunk/hadoop-hdfs-project/hadoop-hdfs-native-client/src/main/native/libhdfspp
and libhdfs3 doesn't look like actively supported: https://github.com/erikmuttersbach/libhdfs3/commits/apache-rpc-9
Seems... | 2020-11-12T15:57:34 | 2020-11-12T16:07:14 | {
"+1": 1
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 726,002,140 | 16,636 | alexey-milovidov | @hombit Need some adjustments in tests. | 2020-11-12T10:52:36 | 2020-11-12T10:52:36 | {} | MEMBER |
ClickHouse | ClickHouse | 726,182,147 | 16,890 | filimonov | It seems fixed in 20.8.4 and newer... | 2020-11-12T16:20:50 | 2020-11-12T16:20:50 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 726,514,477 | 16,957 | l1t1 | thanks | 2020-11-13T04:56:30 | 2020-11-13T04:56:30 | {} | NONE |
ClickHouse | ClickHouse | 725,913,325 | 15,235 | cronzevid | Seeing the same issue, used clickhouse-copier v20.11.2.1
```
2020.11.12 06:53:53.336132 [ 5343 ] {} <Debug> ClusterCopier: DROP TABLE query was successfully executed on 4 nodes.
2020.11.12 06:53:53.336148 [ 5343 ] {} <Information> ClusterCopier: Process table task target_cluster.data.stats_hourly_1 with 1 shards, ... | 2020-11-12T08:08:01 | 2020-11-12T08:36:24 | {} | NONE |
ClickHouse | ClickHouse | 725,982,605 | 16,911 | dependabot-preview[bot] | OK, I won't notify you again about this release, but will get in touch when a new version is available. If you'd rather skip all updates until the next major or minor version, let me know by commenting `@dependabot ignore this major version` or `@dependabot ignore this minor version`.
If you change your mind, just re-... | 2020-11-12T10:15:16 | 2020-11-12T10:15:16 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 725,982,606 | 16,912 | dependabot-preview[bot] | OK, I won't notify you again about this release, but will get in touch when a new version is available. If you'd rather skip all updates until the next major or minor version, let me know by commenting `@dependabot ignore this major version` or `@dependabot ignore this minor version`.
If you change your mind, just re-... | 2020-11-12T10:15:16 | 2020-11-12T10:15:16 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 726,058,748 | 16,621 | filimonov | > 1. I do not see any evidence that libhdfs3 can use cyrus
Yeah, it looks you're right. I was checking it some time ago, but probably i've looked libhdfs++ by mistake...
https://issues.apache.org/jira/browse/HDFS-10450
So nevermind, and sorry for disturbing :)
| 2020-11-12T12:51:22 | 2020-11-12T12:51:39 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 726,124,671 | 16,907 | kmatt | ClickHouse server version 20.9.3 revision 54439.
Will test with 20.11.2.1-2 | 2020-11-12T14:49:47 | 2020-11-12T14:49:47 | {} | NONE |
ClickHouse | ClickHouse | 726,153,091 | 16,926 | filimonov | I guess there is no way for now inside clickhouse. It needs to be checked/fixed in ClickHouse source code.
As a temporary workaround you can install some local proxy (nginx / haproxy / whatever) on the same host as clickhouse, and use `localhost:...` as `format_avro_schema_registry_url`
| 2020-11-12T15:34:08 | 2020-11-12T15:34:08 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 726,185,476 | 16,925 | den-crane | Wondering, your table is partitioned `PARTITION BY toYYYYMMDD(timestamp)` and sorted by `ORDER BY (name, timestamp,`.
Why do you need additional `INDEX timestamp_idx timestamp TYPE minmax GRANULARITY 4` ?
Does it help at all?
>Bad size of marks file '/var/lib/clickhouse/store/0a0/0a0b87aa-c65c-4c7c-a3ce-cedd7... | 2020-11-12T16:26:39 | 2020-11-12T16:26:39 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 726,489,883 | 16,957 | den-crane | 1. because these 2% the second stage , it combines results from several buckets into a single one and heavily uses disk.
2. 20000000000 -> 20 000 000 000 20 GB
20000000000/1024/1024/1024 = 18 GB
| 2020-11-13T03:45:57 | 2020-11-13T03:45:57 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 725,908,057 | 16,902 | jus1096 | Perfect!
**assumeNotNull**
it solved my problem. | 2020-11-12T07:58:56 | 2020-11-12T07:58:56 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 725,930,996 | 16,372 | lostsnow | >
>
> Could you provide you table structure under the mysql database？
Over 70 tables, and I don't know which one's causing the problems. | 2020-11-12T08:41:07 | 2020-11-12T08:41:07 | {} | NONE |
ClickHouse | ClickHouse | 726,001,824 | 13,894 | alexey-milovidov | Fixed in version 20.12-testing when the setting `aggregate_functions_null_for_empty` is enabled. | 2020-11-12T10:51:55 | 2020-11-12T10:51:55 | {} | MEMBER |
ClickHouse | ClickHouse | 726,073,695 | 16,587 | KochetovNicolai | ```
SELECT sumMap(sm)
FROM
(
SELECT sumMap([1], [1]) AS sm
FROM remote('127.0.0.{1,1}', default.my_table)
GROUP BY
col1,
col2
)
Query id: 385ec036-1adf-482f-964a-8c442a48cea4
┌─sumMap(sm)───────┐
│ ([1],[10000000]) │
└──────────────────┘
1 rows in set. Elapsed: 0.190 sec... | 2020-11-12T13:21:55 | 2020-11-12T13:21:55 | {} | MEMBER |
ClickHouse | ClickHouse | 726,495,559 | 16,937 | satanson | @akuzm some test results has been added | 2020-11-13T04:09:49 | 2020-11-13T04:09:49 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 751,194,173 | 18,494 | TszKitLo40 | So we need to do the follow things:
1. Add the setting `log_comment` and implement the corresponding behavior.
2. The corresponding behavior which you mentioned in the issue should be implemented in `executeQuery.h/cpp` and `QueryLog.h`.
Do you think I have understood clearly? | 2020-12-25T07:14:43 | 2020-12-25T07:15:26 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 751,199,875 | 18,511 | alexey-milovidov | We don't support dates before unix epoch. | 2020-12-25T07:45:20 | 2020-12-25T07:45:20 | {
"-1": 2
} | MEMBER |
ClickHouse | ClickHouse | 751,224,574 | 18,074 | olgarev | @filimonov Hi, this PR is waiting for your review. | 2020-12-25T10:01:23 | 2020-12-25T10:01:23 | {} | CONTRIBUTOR |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.