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 | 502,997,636 | 5,661 | BHYCHIK | > We can use more efficient architecture dependent CRC32 implementation, see examples in zlib-ng library, especially "PCLMULQDQ" variant. You can also add a performance test, look at `tests/performance`
Now I use zlib-ng function. Performance improvement:
BEFORE OPTIMIZATION:
[
{
"hostname": "testserver.ru... | 2019-06-18T08:13:27 | 2019-06-18T08:13:27 | {} | NONE |
ClickHouse | ClickHouse | 503,030,171 | 5,648 | alexey-milovidov | > Same sample table, Ext4, SSD, select * from bad_partitions limit 1.
> About 4.5 times better in that case.
It because I had also lowered CPU usage. Before, the query was bounded by userspace CPU, not by syscalls. | 2019-06-18T09:41:42 | 2019-06-18T09:41:42 | {} | MEMBER |
ClickHouse | ClickHouse | 503,127,854 | 5,667 | BayoNet | When writing docs, it is worth to take into account who is a subject and what is an object of the action.
1. User inputs data into ClickHouse. ClickHouse accepts data for insertion in supported input formats. ClickHouse doesn't input data.
2. User retrieves data from ClickHouse. ClickHouse outputs data in supported... | 2019-06-18T13:09:08 | 2019-06-18T13:09:08 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 503,173,633 | 5,641 | 4ertus2 | Unexpected alias when parsing ColumnDescription
```
Syntax error (column parser): failed at position 38: AS a TTL d + toIntervalDay(1). Expected one of: LIKE, CODEC, GLOBAL NOT IN, IS, OR, QuestionMark, BETWEEN, NOT LIKE, AND, TTL, IN, NOT, COMMENT, token, NOT IN, GLOBAL IN
``` | 2019-06-18T14:56:16 | 2019-06-18T14:56:16 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 503,202,231 | 5,669 | kreuzerkrieg | and then it doesnt
> $ rm -rf *
> $ rm -rf .*
> rm: refusing to remove '.' or '..' directory: skipping '.'
> rm: refusing to remove '.' or '..' directory: skipping '..'
> $ cmake -DSPLIT_SHARED_LIBRARIES=ON ..
> -- The C compiler identification is GNU 7.4.0
> -- The CXX compiler identification is GNU 7.4.0
> ... | 2019-06-18T16:03:45 | 2019-06-18T16:03:45 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 503,227,605 | 5,669 | proller | Warning about incorrect options: https://github.com/yandex/ClickHouse/pull/5673 | 2019-06-18T17:12:27 | 2019-06-18T17:12:27 | {
"+1": 1
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 504,157,057 | 4,616 | bopohaa | We have the same problems with ClickHouse v19.8.3.8 and materialized views consuming from a tables with Kafka engine (protobuf): CH server crashing with segmentation fault | 2019-06-20T19:43:38 | 2019-06-20T19:43:38 | {} | NONE |
ClickHouse | ClickHouse | 504,173,788 | 5,582 | blind-oracle | @den-crane @Felixoid @alexey-milovidov @filimonov
Here's a *very dirty* script that does the job.
It inserts groups of 35k metrics (6 total) every 5 seconds with value 1 and then the same with value 3.
In the end each metric should average to 2 over 60 sec.
Tested on official Yandex .deb build 19.8.3.8
``... | 2019-06-20T20:37:24 | 2019-06-20T20:51:23 | {} | NONE |
ClickHouse | ClickHouse | 503,223,650 | 2,796 | abyss7 | @steezypunk according to the message from your log it's exactly the #5615 - will close for now as a duplicate. For further reference the initial problem with hanging shutdown in _any_ case was fixed in version 19.8+ | 2019-06-18T17:01:28 | 2019-06-18T17:01:28 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 503,784,877 | 5,679 | den-crane | Currently CH supports only in-memory JOIN, right table goes to in-memory hash table.
100 mil. table able to eat 200GB memory (depends of columns).
And nothing can be done. | 2019-06-19T23:38:13 | 2019-06-19T23:38:13 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 503,068,155 | 5,663 | sdlcwangsong | by the way, before crash, there was some exception:
DB::Exception: Too many simultaneous queries. Maximum: 100. Stack trace: | 2019-06-18T11:43:09 | 2019-06-18T11:43:09 | {} | NONE |
ClickHouse | ClickHouse | 503,225,671 | 4,736 | abyss7 | > can anyone confirm that the problem is fixed in 19.7? My test shows that data is still being lost in 19.7
Can you please provide the setup of your testing stand? I want to know the scenario to reproduce it. | 2019-06-18T17:07:00 | 2019-06-18T17:07:00 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 503,531,711 | 5,667 | blinkov | @akuzm btw doesn't the same rules apply for other table engines like URL or HDFS? | 2019-06-19T12:03:37 | 2019-06-19T12:03:37 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 503,551,340 | 5,674 | bgranvea | Do you think it is difficult to fix? I would like to avoid spending time in implementing workarounds for this :) | 2019-06-19T13:03:40 | 2019-06-19T13:03:40 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 503,574,409 | 5,615 | abyss7 | Probably caused by https://github.com/edenhill/librdkafka/issues/2077 - will try to implement any workaround. | 2019-06-19T14:02:26 | 2019-06-19T14:09:19 | {
"+1": 2
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 503,642,634 | 3,882 | blinkov | @dneuhaeuser-zalando could you try to reproduce on a recent release? | 2019-06-19T16:52:56 | 2019-06-19T16:52:56 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 503,677,814 | 3,882 | abraithwaite | I want to test a recent release, but wasn't able to build or find a docker image with the fix in it, so I opened this:
https://github.com/yandex/ClickHouse/issues/5675 | 2019-06-19T18:31:04 | 2019-06-19T18:31:04 | {} | NONE |
ClickHouse | ClickHouse | 503,905,083 | 5,646 | nshah14285 | Thanks.
It's working now. | 2019-06-20T07:07:24 | 2019-06-20T07:07:24 | {} | NONE |
ClickHouse | ClickHouse | 503,960,848 | 5,690 | filimonov | It's the expected behavior.
With string - it's unclear what you want to use: string comparison or number comparison.
With floats - see: https://github.com/yandex/ClickHouse/issues/5483#issuecomment-497322642
> If you do need strict values you use Decimal. If you do not need strict values you use Floats. If yo... | 2019-06-20T09:51:59 | 2019-06-20T09:51:59 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 504,012,531 | 5,349 | akuzm | There are two related bugs that still need to be fixed: #5666, #5653. | 2019-06-20T12:48:46 | 2019-06-20T12:48:46 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 503,036,701 | 5,662 | alexey-milovidov | @filimonov
`remote` table function works exactly as Distributed engine, because it creates temporary Distributed table to process the query.
@kreuzerkrieg
> but without the part when something executed locally and then sent to other shards
The intention is not quite clear...
Do you want to:
- execute a quer... | 2019-06-18T09:59:50 | 2019-06-18T10:00:08 | {} | MEMBER |
ClickHouse | ClickHouse | 503,417,407 | 5,618 | I-akshay | HI den-crane i was asking about distributed_aggregation_memory_efficient , i want to permanently update distributed_aggregation_memory_efficient to 1 ,
I do'nt want to set the property every time before the session
| 2019-06-19T05:54:44 | 2019-06-19T05:54:44 | {} | NONE |
ClickHouse | ClickHouse | 503,471,086 | 5,646 | nshah14285 | I have installed clickhouse 19.10.1.688 version. As per project requirement, I have to connect to clickhouse database from MYSQL client, so I tried below command
`mysql --protocol tcp -h localhost -P 9000 -u default --password -e "CREATE DATABASE x;"`
and
`mysql --protocol tcp -h localhost -P 9000 default -u d... | 2019-06-19T08:55:22 | 2019-06-19T08:55:22 | {} | NONE |
ClickHouse | ClickHouse | 503,472,812 | 4,631 | splichy | Hi, still happening even with kafka_skip_broken_messages=1 & kafka_num_consumers=1.
We have only standalone clickhouse instances(one for "production" and second for tests) each instance uses it's own Consumer Group. Topic is on 3 node kafka cluster, in 2 partitions.
Issue is on both CH instances. We are actually read... | 2019-06-19T09:00:08 | 2019-06-19T09:00:08 | {} | NONE |
ClickHouse | ClickHouse | 503,495,003 | 5,646 | yurriy | > I have installed clickhouse 19.10.1.688 version. As per project requirement, I have to connect to clickhouse database from MYSQL client, so I tried below command
>
> `mysql --protocol tcp -h localhost -P 9000 -u default --password -e "CREATE DATABASE x;"`
>
> and
>
> `mysql --protocol tcp -h localhost -P 900... | 2019-06-19T10:02:35 | 2019-06-19T10:20:25 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 504,024,552 | 5,700 | 4ertus2 | Test is needed
#5620 | 2019-06-20T13:25:08 | 2019-06-20T13:25:38 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 502,998,360 | 5,662 | filimonov | Aren't you trying to reinvent the `remote` table function?
See docs: https://clickhouse.yandex/docs/en/query_language/table_functions/remote/
And samples in tests: https://github.com/yandex/ClickHouse/search?l=SQL&q=remote
P.S. Also there is quite similar `cluster` table function, which accepts cluster name in... | 2019-06-18T08:15:28 | 2019-06-18T08:16:22 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 503,134,781 | 5,669 | kreuzerkrieg | define clean... I've deleted the content of `build` folder when rebuilding
submodules - is `git submodule update --init --recursive` good enough?
The normal build looks to be be building fine, but I didnt finish it, just stopped it after 2/3 of the build, when I sure was beyond point of failure. But I will recheck it... | 2019-06-18T13:27:03 | 2019-06-18T13:27:03 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 503,199,468 | 5,669 | kreuzerkrieg | ok...
when it builds
> $ rm -rf *
> $ ll
> total 1220
> drwxr-xr-x 2 user user 4096 Jun 18 16:28 ./
> drwxr-xr-x 15 user user 4096 Jun 18 14:53 ../
> -rw-r--r-- 1 user user 1080360 Jun 18 16:01 .ninja_deps
> -rw-r--r-- 1 user user 156565 Jun 18 16:01 .ninja_log
> $ rm -rf .*
> rm: refusing to remo... | 2019-06-18T15:56:59 | 2019-06-18T15:56:59 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 503,272,062 | 5,674 | 4ertus2 | Related #4904 #4731 | 2019-06-18T19:11:22 | 2019-06-18T19:11:22 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 503,344,841 | 5,612 | den-crane | seems you are looking for -- iF(isNaN(x) or abs(x) == inf, Null, x)
select iF(isNaN(x) or abs(x) == inf, Null, x) from (select 1/0 x union all select 0/0 union all select 1/1)
```
select
id,
iF(isNaN(avgIf(value, id == 3) as average) or abs(average) == inf, 0, average) avg_fix,
maxIf(value, id == 3) as m... | 2019-06-18T23:09:57 | 2019-06-18T23:09:57 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 503,968,790 | 5,662 | filimonov | Just to understand the problem better - it seems like you still need to do one query per server, so why don't you execute the query directly on needed server? | 2019-06-20T10:15:50 | 2019-06-20T10:15:50 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 504,018,716 | 5,582 | blind-oracle | @alexey-milovidov @filimonov
That would be nice if we can fix it. I went through the Graphite merging functions, but without deeper knowledge about CH it's hard to find where the problem is... | 2019-06-20T13:07:19 | 2019-06-20T13:07:19 | {} | NONE |
ClickHouse | ClickHouse | 504,102,987 | 5,694 | den-crane | > Have you seen such solutions for other DBMS?
I've seen :( . AWS RedShift
copy sales
from 'emr://j-SAMPLE2B500FC/myoutput/part-*'
| 2019-06-20T16:56:12 | 2019-06-20T16:56:12 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 503,129,397 | 5,668 | 4ertus2 | #5670 | 2019-06-18T13:13:21 | 2019-06-18T13:13:21 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 503,224,367 | 5,382 | abyss7 | Implements #5286 | 2019-06-18T17:03:25 | 2019-06-18T17:03:25 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 503,428,482 | 5,436 | zhang2014 | @alexey-milovidov
We also have similar requirements. The following is my discussion with @TCeason , If you think it is ok, maybe we can try to complete this feature : )
First configure the following in the configuration file:
``` xml
<CustomHttpURLs>
<CustomHttpURL>
<URL>xxx/{database_name}/{tabl... | 2019-06-19T06:40:35 | 2019-06-19T06:40:35 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 503,962,923 | 5,695 | filimonov | Most of 18.6 issues were fixed in later versions. Please update to latest clickhouse version. | 2019-06-20T09:58:04 | 2019-06-20T09:58:04 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 503,984,985 | 5,662 | kreuzerkrieg | why execute 150 queries on 150 servers instead to let the machine do that?
And I still want to see progress of all 150 queries combined. In one client. | 2019-06-20T11:12:40 | 2019-06-20T11:12:40 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 504,002,324 | 5,582 | Felixoid | Unfortunately, I don't know where to look in this case. For me, it looks as well as a bug and we need help from some real engineers here.
@alexey-milovidov and @filimonov, could you consider this as a moderate/major bug? | 2019-06-20T12:15:04 | 2019-06-20T12:15:04 | {} | MEMBER |
ClickHouse | ClickHouse | 504,088,733 | 5,694 | 4ertus2 | Have you seen such solutions for other DBMS?
It's looks like your service could be 10-lines bash script & cron. If you do not need guaranties. If you need guaranties you probably need something more, ex. kafka or your own service. | 2019-06-20T16:13:26 | 2019-06-20T16:13:26 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 504,102,467 | 5,030 | proller | @YiuRULE Can you show how you exactly generate list from https://publicsuffix.org/list/public_suffix_list.dat ?
I want to make this process reproducible. | 2019-06-20T16:54:34 | 2019-06-20T16:55:33 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 504,109,155 | 5,582 | blind-oracle | @den-crane I'll try to create a reproducible setup and will get back.
Probably it will involve pushing some fake metrics into CH directly or though `carbon-clickhouse` at around the same rate as we do...
Because I have the feeling that the issue is dependent on the ingest rate (in fact on the frequency of backgroun... | 2019-06-20T17:15:15 | 2019-06-20T17:17:39 | {} | NONE |
ClickHouse | ClickHouse | 503,128,496 | 5,669 | 4ertus2 | Is it a clean build? Have you updated git submodules? Does normal build fail? | 2019-06-18T13:10:58 | 2019-06-18T13:10:58 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 503,272,224 | 5,674 | 4ertus2 | FYI @abyss7 | 2019-06-18T19:11:53 | 2019-06-18T19:11:53 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 503,360,139 | 5,612 | den-crane | OK. Understood now.
it's really wrong behavior in case of max()/avg group by 1 on zero rows.
```
select max(a),avg(a) from (select 5 a) where 1=0 settings empty_result_for_aggregation_by_empty_set=0;
┌─max(a)─┬─avg(a)─┐
│ 0 │ nan │
└────────┴────────┘
psql > select max(a), avg(a) from (select... | 2019-06-19T00:29:44 | 2019-06-19T01:02:14 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 503,809,969 | 5,612 | CAspeling | I don't understand what the last examples meant. We expect maxIf to give back nan instead of zero when avg would give back nan. This allows us to handle those nan's in different ways depending on what graph type we render. | 2019-06-20T01:56:58 | 2019-06-20T01:56:58 | {} | NONE |
ClickHouse | ClickHouse | 504,050,864 | 5,682 | den-crane | in users.xml in default profile
<enable_optimize_predicate_expression>0</enable_optimize_predicate_expression>
please reopen this issue. It's real bug and should be fixed. | 2019-06-20T14:35:33 | 2019-06-20T14:35:33 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 502,966,076 | 5,590 | hczhcz | TODO: Fix param handling | 2019-06-18T06:28:40 | 2019-06-18T06:28:40 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 503,078,315 | 2,796 | filimonov | > Face similar issues with 19.8.3.8, on drop kafka table engine got this error:
> `%3|1560495915.155|cppkafka|ClickHouse 19.8.3.8#consumer-3| Failed to close consumer [ClickHouse 19.8.3.8#consumer-3]: Local: Broker handle destroyed`
>
> The query than hung up.
Was the engine active before unsuccessful DROP attem... | 2019-06-18T12:16:31 | 2019-06-18T12:16:31 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 503,157,780 | 5,595 | inv2004 | > I have questions in https://t.me/joinchat/B44Vi1T-PQA97S02SBwXIw
Functions : groupArrayMovingSum and groupArrayMovingAvg
| 2019-06-18T14:21:42 | 2019-06-18T15:27:10 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 503,367,462 | 5,671 | liuluo129 | > > one of the server load become very high
>
> Do you have the issue with the same one server every time?
> CH version?
> How many columns in your table?
there is 6 server, and every server have the problem, but every time only one server has problem
my CH version is v18.14.10-testing.
there is 16 column... | 2019-06-19T01:13:04 | 2019-06-19T01:20:22 | {} | NONE |
ClickHouse | ClickHouse | 503,557,249 | 5,680 | spyros87 | This issue can be fixed if librdkafka can be upgraded to v1.0.1 according to https://github.com/edenhill/librdkafka/issues/2371 | 2019-06-19T13:19:40 | 2019-06-19T13:19:40 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 503,560,121 | 5,674 | 4ertus2 | I think it will take some time. It's not clear how much cause there're some more critical issues now.
I don't know any workaround for it. | 2019-06-19T13:26:58 | 2019-06-19T13:26:58 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 503,785,286 | 5,618 | den-crane | ```
cat /etc/clickhouse-server/conf.d/memory_usage.xml
<?xml version="1.0" ?>
<yandex>
<profiles>
<default>
<max_memory_usage_for_all_queries>101310968832</max_memory_usage_for_all_queries>
<max_bytes_before_external_group_by>50655484416</max_bytes_before_external_group_by>
... | 2019-06-19T23:40:33 | 2019-06-19T23:40:33 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 503,967,888 | 5,695 | peaksnail | when I set enable_optimize_predicate_expression=1 in config.xml ,then i select * from system.settings where name='enable_optimize_predicate_expression' , the value =0 ,It not work, is there something wrong? | 2019-06-20T10:13:06 | 2019-06-20T10:13:06 | {} | NONE |
ClickHouse | ClickHouse | 504,044,509 | 5,695 | den-crane | enable_optimize_predicate_expression should be users.xml not in config.xml
enable_optimize_predicate_expression became stable in 19.5 and it's still has issues with SQL results. | 2019-06-20T14:19:54 | 2019-06-20T14:19:54 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 502,964,535 | 5,618 | I-akshay | @filimonov
can we add distributed_aggregation_memory_efficient property in /memory_usage.xml | 2019-06-18T06:22:31 | 2019-06-18T06:22:31 | {} | NONE |
ClickHouse | ClickHouse | 503,008,849 | 5,662 | kreuzerkrieg | Hm... it works on top of `Distributed` which works with `DistributedBlockOutputStream` and (I guess) `DirectoryMonitor` which in turn somewhat confusing for me at this moment, but I will dig into the code and check it right away.
Now the question, what if table function is not good enough for me, and I want to hide al... | 2019-06-18T08:44:50 | 2019-06-18T08:44:50 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 503,155,296 | 5,671 | den-crane | >one of the server load become very high
Do you have the issue with the same one server every time?
CH version?
How many columns in your table? | 2019-06-18T14:16:09 | 2019-06-18T14:16:26 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 503,187,732 | 5,618 | den-crane | > @filimonov
> can we add distributed_aggregation_memory_efficient property in /memory_usage.xml
yes you can
check that settings are applied to your session
select * from system.settings where name = 'max_bytes_before_external_group_by' | 2019-06-18T15:29:17 | 2019-06-18T15:29:41 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 503,344,633 | 5,030 | alexey-milovidov | Merged. Sorry for the delay. | 2019-06-18T23:08:59 | 2019-06-18T23:08:59 | {} | MEMBER |
ClickHouse | ClickHouse | 503,377,665 | 5,677 | den-crane | ZK is in-memory database.
So lets check ZK database size
echo mntr | nc localhost 2181|grep zk_approximate_data_size
zk_approximate_data_size 68723698
~68MB
The same size as snapshot size
du -sh /var/lib/zookeeper/version-2/snapshot.4350f30c19
77M /var/lib/zookeeper/version-2/snapshot.4350f30c19
So i... | 2019-06-19T02:10:41 | 2019-06-19T02:10:41 | {
"+1": 1
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 504,077,950 | 5,582 | den-crane | @blind-oracle Need to reproduce your issue somehow. I tried but failed. Everything worked fine. Maybe because my data is too fake. | 2019-06-20T15:44:21 | 2019-06-20T15:44:21 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 503,643,458 | 5,480 | kolsys | v19.7.5.27 same issue. | 2019-06-19T16:55:13 | 2019-06-19T16:55:13 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 503,941,160 | 5,646 | nshah14285 | I am able to connect ClickHouse from mysql client but when I tried to connect from node application using [MYSQL](https://www.npmjs.com/package/mysql) module it throws an error
> max_packet_size: 0
character_set: !
user: default
auth_response length: 20
auth_response: ;t#3
�O1�cfk�Xkva�
dat... | 2019-06-20T08:55:52 | 2019-06-20T08:55:52 | {} | NONE |
ClickHouse | ClickHouse | 503,974,085 | 5,648 | filimonov | > 19.4.3.11. HDD raid 10
Checked that on the same version & another station (19.4.3.11, i7 9750H, NVMe SSD):
```
echo 'select * from bad_partitions limit 1' | clickhouse-benchmark -i 20 -d 0
Loaded 1 queries.
Queries executed: 20.
QPS: 0.876, RPS: 0.876, MiB/s: 0.000, result RPS: 0.876, result MiB/s: 0.000.... | 2019-06-20T10:34:06 | 2019-06-20T10:34:06 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 503,399,356 | 4,736 | wawaka | Setup is very simple.
I had a large topic in kafka with jsons.
I have created 4 identical tables in clickhouse (1 standalone server) with corresponding kafka tables and mat views that feed from mentioned kafka topic.
After consumption finished I observed gaps in data inserted into those 4 tables in random places. (-... | 2019-06-19T04:17:09 | 2019-06-19T04:17:09 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 503,405,469 | 5,677 | ghost | @den-crane
The use_minimalistic_checksums_in_zookeeper is enabled.
```
:) select * from system.merge_tree_settings
...
use_minimalistic_checksums_in_zookeeper │ 1 │ 0 │
...
```
My ZooKeeper's zk_approximate_data_size is ```1219607498.```
I will monitor how this value g... | 2019-06-19T04:53:18 | 2019-06-19T04:53:18 | {} | NONE |
ClickHouse | ClickHouse | 503,437,520 | 5,648 | jackpgao | - 3T * 12 Raid 5 with Raid Card.
## my test tables
```
SELECT
database,
table,
partition,
sum(((data_compressed_bytes / 1024) / 1024) / 1024) AS size_GB
FROM system.parts
WHERE active = 1
GROUP BY
database,
table,
partition
HAVING size_GB > 100
ORDER BY size_GB DESC
LIMI... | 2019-06-19T07:13:23 | 2019-06-19T07:13:23 | {} | NONE |
ClickHouse | ClickHouse | 503,506,222 | 5,667 | akuzm | Please see the updated text.
> All the information about specific table engine must be in the documentation for this engine, not in the top of Formats topic.
Formatting the data for inserts and selects, and reading external data files with the File engine are the only two things the file formats are used for. The... | 2019-06-19T10:37:56 | 2019-06-19T10:37:56 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 503,908,277 | 5,674 | bgranvea | I was thinking on pushing the criteria myself on the left table:
```
SELECT ts, id, id_b, b.ts, b.id, id_c
FROM (SELECT ts, id, id_b FROM A) AS a
ALL LEFT JOIN B AS b ON b.id = a.id_b
WHERE a.ts <= toDateTime('1970-01-01 03:00:00');
=>
SELECT ts, id, id_b, b.ts, b.id, id_c
FROM (SELECT ts, id, id_b FROM A... | 2019-06-20T07:18:32 | 2019-06-20T07:18:32 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 503,915,635 | 5,678 | zhang2014 | Performance info:
``` SQL
CREATE TABLE bf_performance_test
(
`order_key` Int64,
`bf_index` String,
`other_column_1` UInt64,
`other_column_2` UInt32,
`other_column_3` String,
`other_column_4` Float64,
INDEX idx bf_index TYPE bloom_filter GRANULARITY 1
)
ENGINE = MergeTree
O... | 2019-06-20T07:42:41 | 2019-06-20T07:43:25 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 503,965,353 | 5,693 | filimonov | Distributed table doesn't store any data itself. You need to drop partitions from underlying tables.
Use `ON CLUSTER` clause to run DDL queries across the shards. | 2019-06-20T10:05:10 | 2019-06-20T10:05:26 | {
"+1": 1
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 503,967,523 | 5,691 | filimonov | `visitParam...` functions support only some subset of JSON. Namely:
>The following assumptions are made:
>...
>4. The JSON doesn't have space characters outside of string literals.
https://clickhouse.yandex/docs/en/query_language/functions/json_functions/
You have a space after colon char.
Proabably `JS... | 2019-06-20T10:11:54 | 2019-06-20T10:11:54 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 504,111,899 | 5,646 | yurriy | > I am able to connect ClickHouse from mysql client but when I tried to connect from node application using [MYSQL](https://www.npmjs.com/package/mysql) module it throws an error
>
[MYSQL](https://www.npmjs.com/package/mysql) library does not support authentication plugins other than mysql_native_password, which mak... | 2019-06-20T17:23:27 | 2019-06-20T17:23:27 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 504,217,331 | 5,582 | den-crane | I've got the same
```
┌─Value─┬────toDateTime(Time)─┐
│ 1 │ 2019-06-20 21:02:00 │
│ 2 │ 2019-06-20 21:03:00 │
│ 2 │ 2019-06-20 21:04:00 │
│ 2 │ 2019-06-20 21:05:00 │
│ 2 │ 2019-06-20 21:06:00 │
│ 1 │ 2019-06-20 21:07:00 │
│ 1 │ 2019-06-20 21:08:00 │
│ 2 │ 2019-06-20 21:09:00 ... | 2019-06-20T22:07:53 | 2019-06-20T22:07:53 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 502,996,893 | 5,609 | kreuzerkrieg | > Personally I would prefer to answer the ClickHouse question, not a question about where to ask questions ;)
Here you go :)
https://github.com/yandex/ClickHouse/issues/5662
| 2019-06-18T08:11:23 | 2019-06-18T08:11:23 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 503,004,066 | 5,582 | Felixoid | One more thing, could you maybe try to test CH older than 19.4.0.49? To exclude #4426
E.g. v19.3.6-stable from here https://github.com/yandex/ClickHouse/tags?after=v19.1.13-testing | 2019-06-18T08:31:17 | 2019-06-18T08:31:17 | {} | MEMBER |
ClickHouse | ClickHouse | 503,213,517 | 5,669 | kreuzerkrieg | Hmmm… shouldn’t it be mutually exclusive? Checking…
Sent from Mail for Windows 10
From: proller
Sent: Tuesday, June 18, 2019 19:29
To: yandex/ClickHouse
Cc: kreuzerkrieg; Author
Subject: Re: [yandex/ClickHouse] SPLIT_SHARED_LIBRARIES build is broken?(#5669)
You should use two options:
-DUSE_STATIC_LIBRARIES=0 -DSPL... | 2019-06-18T16:34:07 | 2019-06-18T16:34:07 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 503,515,615 | 5,653 | akuzm | Turns out the Tuple type is irrelevant. The core problem is that we can't determine the size of a Block that has no columns. We run into this problem at some point, when all the columns of a block are to be filled with defaults.
We have not received any complaints about this issue, and the way we work with Blocks is... | 2019-06-19T11:09:43 | 2019-06-19T11:09:43 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 503,816,575 | 5,682 | onine007 | thank you @den-crane
how to change enable_optimize_predicate_expression = 0 to clickhouse , not session | 2019-06-20T02:32:15 | 2019-06-20T03:08:28 | {} | NONE |
ClickHouse | ClickHouse | 502,957,248 | 5,618 | I-akshay | > set in the session using [ set max_bytes_before_external_group_by = .... ]
> or users' profiles.
>
> ```
> cat /etc/clickhouse-server/conf.d/memory_usage.xml
> <?xml version="1.0" ?>
> <yandex>
> <profiles>
> <default>
> <max_memory_usage_for_all_queries>101310968832</max_memory_usag... | 2019-06-18T05:50:59 | 2019-06-18T05:50:59 | {} | NONE |
ClickHouse | ClickHouse | 503,129,256 | 5,648 | den-crane | 19.4.3.11. HDD raid 10
INSERT INTO bad_partitions SELECT * FROM numbers(10000)
```
echo 'select * from bad_partitions limit 1' | clickhouse-benchmark --database dw -i 20 -d 0
Loaded 1 queries.
Queries executed: 20.
QPS: 16.449, RPS: 16.449, MiB/s: 0.000, result RPS: 16.449, result MiB/s: 0.000.
0.000% 0... | 2019-06-18T13:12:57 | 2019-06-18T13:12:57 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 503,417,340 | 5,212 | andyyzh | All previous tests have passed. I don't think the failing tests are caused by my change.
@alexey-milovidov If you have time, take a look. | 2019-06-19T05:54:25 | 2019-06-19T05:54:25 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 503,603,549 | 5,689 | 4ertus2 | Reproduced on master | 2019-06-19T15:12:43 | 2019-06-19T15:12:43 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 503,011,267 | 5,582 | blind-oracle | @Felixoid
Recreated the cluster from scratch using 19.3.6, will see by the end of day if it changed anything :) | 2019-06-18T08:51:17 | 2019-06-18T08:51:17 | {} | NONE |
ClickHouse | ClickHouse | 503,042,664 | 5,662 | kreuzerkrieg | @alexey-milovidov
I think none of above... I will provide an example and correct me if I wrong
I have something like:
```
INSERT INTO T2
SELECT * FROM T1
```
Where T1 is, lets say, a real MergeTree table on each cluster node, and T2 is created using some kind of table engine which in turn looks like following... | 2019-06-18T10:17:41 | 2019-06-18T10:17:41 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 503,211,725 | 5,669 | proller | You should use two options:
-DUSE_STATIC_LIBRARIES=0 -DSPLIT_SHARED_LIBRARIES=1 | 2019-06-18T16:29:06 | 2019-06-18T16:29:06 | {
"+1": 1
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 503,393,620 | 5,669 | kreuzerkrieg | > You should use two options:
> -DUSE_STATIC_LIBRARIES=0 -DSPLIT_SHARED_LIBRARIES=1
This solved the problem. Thanks | 2019-06-19T03:40:46 | 2019-06-19T03:40:46 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 503,660,498 | 5,682 | den-crane | set enable_optimize_predicate_expression = 0 & query1 == 2 rows
set enable_optimize_predicate_expression = 1 & query1 == 0 rows
| 2019-06-19T17:42:38 | 2019-06-19T17:42:38 | {
"+1": 1
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 503,830,130 | 5,612 | CAspeling | Here is an example straight from our database for your perusal. At any point in time, we will query on a time frame, where a result for a specific ID may or may not occur. `maxIf()` is not behaving correctly when no result comes back, while `avgIf()` returns null/ nothing, not even NAN.
I have tried copying the emp... | 2019-06-20T03:47:37 | 2019-06-20T03:47:58 | {} | NONE |
ClickHouse | ClickHouse | 503,964,156 | 5,667 | akuzm | > @akuzm btw doesn't the same rules apply for other table engines like URL or HDFS?
Good point, added a mention of these as well. | 2019-06-20T10:01:37 | 2019-06-20T10:01:37 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 503,071,414 | 5,648 | filimonov | > > Same sample table, Ext4, SSD, select * from bad_partitions limit 1.
>
> > About 4.5 times better in that case.
>
> It because I had also lowered CPU usage. Before, the query was bounded by userspace CPU, not by syscalls.
Ok. The important thing that it's now faster both with help of stat and cpu optimizati... | 2019-06-18T11:54:22 | 2019-06-18T12:02:24 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 503,159,707 | 5,582 | blind-oracle | @Felixoid
Nope, same behavior in 19.3.6, two rows with same timestamp:
```
│ 1560866760 │ 2019-06-18 16:06:00 │ 64 │
│ 1560866760 │ 2019-06-18 16:06:00 │ 1160 │
└────────────┴─────────────────────┴───────┘
```
Then the row with value 1160 is discarded:
```
│ 1560866700 │ 2019-06-18 16:05:00 │ 64 │
... | 2019-06-18T14:26:12 | 2019-06-18T14:26:12 | {} | NONE |
ClickHouse | ClickHouse | 503,353,286 | 5,612 | CAspeling | How will we know what is a real max of 0 and what isn't?
We are drawing graphs of incoming data for IoT devices that report states.
Sometimes that state is 0 and sometimes it is a value higher or lower than
0.
We ask the database for all values according to an aggregation selection eg
max or avg etc for specific time... | 2019-06-18T23:52:37 | 2019-06-18T23:52:37 | {} | NONE |
ClickHouse | ClickHouse | 503,866,412 | 5,612 | CAspeling | Here is a comparison:
`avg(null) = null` or "empty string" as it is called in the documentation
`avgIf(null, id == 1) = null` or "empty string"
`max(null) = null` or "empty string"
`maxIf(null, id == 1) = 0`
**Expected:**
`maxIf(null, id == 1) = null` or "empty string"
___
![image](https://user-images.git... | 2019-06-20T04:55:30 | 2019-06-20T05:02:27 | {} | NONE |
ClickHouse | ClickHouse | 503,999,218 | 5,582 | blind-oracle | @Felixoid @den-crane
Guys, any ideas which direction to look? Tested 19.8, same result of course as Graphite* part haven't changed.
This issue manifests all over the place in my CH servers. It's not much visible if the metric does not change much over the averaging period (like bandwidth which stays more or less... | 2019-06-20T12:04:40 | 2019-06-20T12:05:00 | {} | NONE |
ClickHouse | ClickHouse | 504,092,086 | 5,695 | 4ertus2 | Need reproduction on newer version. | 2019-06-20T16:23:03 | 2019-06-20T16:23:03 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 515,745,369 | 6,190 | alexey-milovidov | CC @sundy-li @zhang2014 | 2019-07-28T08:55:43 | 2019-07-28T08:55:43 | {} | MEMBER |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.