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
507,164,684
5,804
alexey-milovidov
It's major usability issue when if you run clickhouse-server on a server with new tzdata and clickhouse-client on a server with old tzdata.
2019-07-01T08:16:58
2019-07-01T08:16:58
{ "+1": 2 }
MEMBER
ClickHouse
ClickHouse
507,223,610
5,744
filimonov
Хм. А раньше работало? `toYYYYMM(event_date)` возвращает число в формате 201906, которое нужно указывать без кавычек.
2019-07-01T11:20:26
2019-07-01T11:20:26
{}
CONTRIBUTOR
ClickHouse
ClickHouse
507,359,985
5,822
alexey-milovidov
It does not look like the right fix. Better to zero-initialize box? `Box box{}`
2019-07-01T17:43:36
2019-07-01T17:44:21
{}
MEMBER
ClickHouse
ClickHouse
507,679,587
5,438
alexey-milovidov
We need an integration test where the new code at least runs.
2019-07-02T13:35:24
2019-07-02T13:35:24
{}
MEMBER
ClickHouse
ClickHouse
507,154,994
5,734
filimonov
TRy to add some WHERE condition to limit the number of data scanned. Something like `WHERE timestamp > now() - 600` `WHERE date>=yesterday()` or similar Also - try 19.10 / 19.11 / latest master versions. There was some improvement in that recently: https://github.com/yandex/ClickHouse/pull/5648#issuecomment-50256238...
2019-07-01T07:47:03
2019-07-01T07:48:50
{}
CONTRIBUTOR
ClickHouse
ClickHouse
507,280,106
5,698
abyss7
Should almost fix #5615
2019-07-01T14:08:04
2019-07-01T14:08:04
{}
CONTRIBUTOR
ClickHouse
ClickHouse
507,758,164
5,833
den-crane
oops I misread as topK
2019-07-02T16:48:29
2019-07-02T16:52:12
{}
CONTRIBUTOR
ClickHouse
ClickHouse
507,771,199
5,836
kreuzerkrieg
and I definitely will once you leave me no choice :) meanwhile I prefer to work with default set of tools coming from official OS repo
2019-07-02T17:24:26
2019-07-02T17:24:26
{}
CONTRIBUTOR
ClickHouse
ClickHouse
507,793,861
5,681
filimonov
> Just thoughts, looks like it can be possible to implement auto increment with the same manner as MySQL does: > > 1. Support autoincrement_increment and autoincrement_offset in order to avoid intersection among shards > 2. For each table - keep an atomic counter for new data. When server restarts - try find maximu...
2019-07-02T18:28:49
2019-07-02T18:28:49
{}
CONTRIBUTOR
ClickHouse
ClickHouse
507,882,304
5,710
alexey-milovidov
TODO: It must work without `query_masking_rules` section in config (e.g. config file from old server). Make method `wipeSensitiveData` const; use const QueryMaskingRules in Context; return by const reference (or pointer?). Initialize eagerly, because it cannot be changed after startup, avoid getLock call.
2019-07-02T23:35:31
2019-07-02T23:35:31
{}
MEMBER
ClickHouse
ClickHouse
507,917,551
5,618
Aricliang
> > should not use the disk? Why still reporting memory limits? > > Because final aggregation result should fit into memory. > > So for group by SOME_HIGH_CARDINALITY_SET_OF_COLUMNS, ch will do intermediate partial aggregations on a disk, but in the end CH will gather the result into one part in memory. Yes, Y...
2019-07-03T02:54:10
2019-07-03T02:54:10
{}
NONE
ClickHouse
ClickHouse
507,217,071
4,265
alexey-milovidov
We have added better `rpm` packages but they still have some minor issues: https://github.com/yandex/ClickHouse/issues/5815
2019-07-01T10:57:31
2019-07-01T10:57:38
{}
MEMBER
ClickHouse
ClickHouse
507,283,986
5,821
alexey-milovidov
We also have nightly releases for .deb and .rpm in: https://repo.yandex.ru/clickhouse/deb/testing/main/ https://repo.yandex.ru/clickhouse/rpm/testing/x86_64/
2019-07-01T14:17:28
2019-07-01T14:17:28
{}
MEMBER
ClickHouse
ClickHouse
507,284,851
5,821
alexey-milovidov
> You need to install librdkafka No, you don't have to install rdkafka, because we have this library in our submodules.
2019-07-01T14:19:24
2019-07-01T14:19:24
{}
MEMBER
ClickHouse
ClickHouse
504,841,231
5,723
kagan770
Now I got the following: ``` FAILED: dbms/src/Functions/CMakeFiles/clickhouse_functions.dir/FunctionsGeo.cpp.o /usr/local/bin/g++-9 -DBOOST_SYSTEM_NO_DEPRECATED -DLZ4_DISABLE_DEPRECATE_WARNINGS=1 -DPOCO_STATIC -DPOCO_UNBUNDLED_ZLIB -DUNALIGNED_OK -DWITH_GZFILEOP -DX86_64 -DZLIB_COMPAT -Idbms/src/Functions/include ...
2019-06-24T03:13:55
2019-07-01T14:46:51
{}
NONE
ClickHouse
ClickHouse
507,702,130
5,286
abyss7
`topic`, `offset`, `key` columns are available in v19.11+
2019-07-02T14:27:32
2019-07-02T14:27:32
{}
CONTRIBUTOR
ClickHouse
ClickHouse
507,200,931
5,814
CurtizJ
It seems that it`s the same issue as #5515.
2019-07-01T10:02:25
2019-07-01T10:02:25
{}
MEMBER
ClickHouse
ClickHouse
507,201,671
5,809
maskshell
No, we got this error with just one ReplicatedReplacingMergeTree table, it has about 350 million rows. I will try to reCreate a local table and copy the data this night (GMT+08) to confirm it.
2019-07-01T10:04:54
2019-07-01T10:04:54
{}
NONE
ClickHouse
ClickHouse
507,274,317
5,812
den-crane
You need to upgrade at least to 19.1.14. https://github.com/yandex/ClickHouse/pull/4623
2019-07-01T13:53:51
2019-07-01T13:53:51
{}
CONTRIBUTOR
ClickHouse
ClickHouse
507,285,182
5,821
alexey-milovidov
To figure out, why Kafka support is not enabled, you can check cmake output for `kafka` substring.
2019-07-01T14:20:09
2019-07-01T14:20:19
{}
MEMBER
ClickHouse
ClickHouse
507,565,463
5,681
filimonov
> So, if we always insert 1 in some summing column it would be incremented by merges. But that will require reducing number of rows, right? With autoincrement the idea is to preserve all rows giving each the unique number. BTW (while talking about workarounds): did anybody try something like `xxx UInt64 DEFAUL...
2019-07-02T07:57:19
2019-07-02T07:58:38
{}
CONTRIBUTOR
ClickHouse
ClickHouse
507,568,209
5,681
4ertus2
> But that will require reducing number of rows, right? Yup, it’s autoincrement + unique PK. For real autoincrement with not uniq data we need timestamp + more complex aggregating function. But it’s still possible if we have the same sorting order each time (do not allow inserts in the past)
2019-07-02T08:05:39
2019-07-02T08:05:39
{}
CONTRIBUTOR
ClickHouse
ClickHouse
507,677,756
5,790
alesapin
Fix https://github.com/yandex/ClickHouse/pull/5832
2019-07-02T13:30:43
2019-07-02T13:30:43
{}
MEMBER
ClickHouse
ClickHouse
507,965,617
5,831
imdad1410
For example the yelp dataset in Json Format. I am trying to load it in clickhouse using `cat yelp_academic_dataset_business.json | clickhouse-client --input_format_skip_unknown_fields=1 --query="INSERT INTO dem FORMAT JSONEachRow` The error I get `Code: 27. DB::Exception: Cannot parse input: expected [ before: "Go...
2019-07-03T06:54:59
2019-07-03T06:54:59
{}
NONE
ClickHouse
ClickHouse
507,230,900
5,809
alesapin
It seems like this error doesn't connect with the final mark. It would be awesome if you can make a reproducible example.
2019-07-01T11:45:41
2019-07-01T11:45:41
{}
MEMBER
ClickHouse
ClickHouse
507,250,491
4,402
akuzm
With test.sh, I can reproduce this on almost every run, on current master. Have no idea about the cause yet, will continue working on this.
2019-07-01T12:48:36
2019-07-01T12:48:36
{}
CONTRIBUTOR
ClickHouse
ClickHouse
507,318,992
4,631
abyss7
I have upgraded rdkafka library (as a suggested solution to #5680) in v19.11+ - since I still can't reproduce an issue, I would appreciate if someone will test 19.11 before it gets stable.
2019-07-01T15:43:25
2019-07-01T15:43:25
{ "+1": 3 }
CONTRIBUTOR
ClickHouse
ClickHouse
507,375,023
5,822
filimonov
> It does not look like the right fix. I agree, but i didn't find a better option. > Better to zero-initialize box? > `Box box{}` i've tried that, and also tried ```cpp auto box = boost::geometry::make_envelope<PointInPolygonWithGrid<CoordinateType>::Box>(polygon); ``` Both options doesn't help. :\...
2019-07-01T18:29:52
2019-07-01T19:30:04
{}
CONTRIBUTOR
ClickHouse
ClickHouse
507,556,229
5,681
4ertus2
It’s possible to get some magic here if we generalize Summing and Replacing merge tree. Then some columns would be summing and others would be replacing. So, if we always insert 1 in some summing column it would be incremented by merges. Of cause you need selects with FINAL there.
2019-07-02T07:29:48
2019-07-02T08:12:04
{}
CONTRIBUTOR
ClickHouse
ClickHouse
507,122,514
5,784
kriulkin
> uniqCombined(ring) AS uniq Thanks a lot! But is there any complete explanation about of memory usage by a query ?
2019-07-01T05:37:52
2019-07-01T05:37:52
{}
NONE
ClickHouse
ClickHouse
507,221,538
5,815
alexey-milovidov
On RHEL7 you can install packages with the following instruction: ``` cat > /etc/yum.repos.d/clickhouse.repo <<END [repo.yandex.ru_clickhouse_rpm_testing_x86_64] name=added from: https://repo.yandex.ru/clickhouse/rpm/testing/x86_64 baseurl=https://repo.yandex.ru/clickhouse/rpm/testing/x86_64 enabled=1 END yum...
2019-07-01T11:12:45
2019-07-01T11:12:45
{}
MEMBER
ClickHouse
ClickHouse
507,249,863
5,783
filimonov
When you execute the query on non-leader replica it get forwarded to leader using credentials provided in `<remote_servers>` configuration in your config.xml (or using user `default` with empty password if leader replica credentials are absent in `<remote_servers>`)
2019-07-01T12:46:53
2019-07-01T12:47:17
{}
CONTRIBUTOR
ClickHouse
ClickHouse
507,279,771
5,821
millin
You need to install librdkafka and clone this repo with submodules to build ClickHouse with Kafka engine.
2019-07-01T14:07:14
2019-07-01T14:07:14
{}
CONTRIBUTOR
ClickHouse
ClickHouse
507,280,990
5,615
abyss7
Situation improved for v19.9+
2019-07-01T14:10:14
2019-07-01T14:10:14
{}
CONTRIBUTOR
ClickHouse
ClickHouse
507,210,365
5,680
spyros87
@filimonov I have already seen that this PR has been merged. I am waiting for a stable 19.11 in order to validate. Thank you for sharing this info.
2019-07-01T10:33:35
2019-07-01T10:33:35
{}
CONTRIBUTOR
ClickHouse
ClickHouse
507,242,804
5,762
filimonov
Can you find what exactly happened just before it started? Didn't you call something like `system stop merges`?
2019-07-01T12:25:01
2019-07-01T12:25:14
{}
CONTRIBUTOR
ClickHouse
ClickHouse
507,308,932
5,823
bzon
I managed to fix the issue by adding the username and password in the `remote_servers` settings. From the previous (passwordless) ```xml <replica> <default_database>default</default_database> <host>ch-01-0.clickhouse.svc.cluster.local</host> ...
2019-07-01T15:19:06
2019-07-01T15:21:14
{}
NONE
ClickHouse
ClickHouse
507,207,553
5,809
alesapin
Do you have `LowCardinality` columns in the problem table?
2019-07-01T10:23:56
2019-07-01T10:23:56
{}
MEMBER
ClickHouse
ClickHouse
507,274,058
5,545
abyss7
If an issue still persists, then can you provide any related logs to see actual error messages?
2019-07-01T13:53:13
2019-07-01T13:53:13
{}
CONTRIBUTOR
ClickHouse
ClickHouse
507,291,341
5,790
alesapin
Telegram: @Alesapin Email: alesap@yandex-team.ru
2019-07-01T14:35:19
2019-07-01T14:35:19
{}
MEMBER
ClickHouse
ClickHouse
507,635,480
5,681
alex-krash
Just thoughts, looks like it can be possible to implement auto increment with the same manner as MySQL does: 1. Support autoincrement_increment and autoincrement_offset in order to avoid intersection among shards 2. For each table - keep an atomic counter for new data. When server restarts - try find maximum value fr...
2019-07-02T11:27:26
2019-07-02T11:27:26
{ "+1": 3, "eyes": 1 }
CONTRIBUTOR
ClickHouse
ClickHouse
507,825,395
958
alexey-milovidov
Ok, I understand the idea of this fix. But it does not help.
2019-07-02T20:06:32
2019-07-02T20:06:32
{}
MEMBER
ClickHouse
ClickHouse
507,188,233
5,809
alesapin
1. Yes `write_final_mark` still not in stable release yes. It'll be available in 19.10. 2. As I understood you correctly: You just update clickhouse to 19.6.2 and got this error with each ReplicatedReplacingMergeTree table?
2019-07-01T09:24:46
2019-07-01T09:24:46
{}
MEMBER
ClickHouse
ClickHouse
507,221,916
4,456
k0st1an
The error occurs when you restart the ClickHouse server or restart the server itself.
2019-07-01T11:14:05
2019-07-01T11:14:05
{}
NONE
ClickHouse
ClickHouse
507,265,441
5,472
den-crane
There is no such entity like a partition on physical level in CH. It just an abstraction based on parts properties.
2019-07-01T13:31:00
2019-07-01T13:31:00
{}
CONTRIBUTOR
ClickHouse
ClickHouse
507,282,055
5,790
alexey-milovidov
It's Ok. You can send via email or Telegram if you want.
2019-07-01T14:12:59
2019-07-01T14:12:59
{}
MEMBER
ClickHouse
ClickHouse
507,325,356
5,790
alexey-milovidov
@alesapin has found the root cause of the issue.
2019-07-01T15:59:06
2019-07-01T15:59:06
{}
MEMBER
ClickHouse
ClickHouse
507,334,370
5,821
filimonov
Most probably non linux platform (macos?), rdkafka build is not working properly on other platforms currently and is disabled in cmake.
2019-07-01T16:23:52
2019-07-01T16:23:52
{}
CONTRIBUTOR
ClickHouse
ClickHouse
507,755,653
5,835
den-crane
Please share cluster description ( XML ). And a result of select value from system.settings where name like 'max_parallel_replicas';
2019-07-02T16:41:42
2019-07-02T16:41:42
{}
CONTRIBUTOR
ClickHouse
ClickHouse
507,772,498
5,840
kreuzerkrieg
Thanks Alexey
2019-07-02T17:27:56
2019-07-02T17:27:56
{}
CONTRIBUTOR
ClickHouse
ClickHouse
507,876,106
5,842
alexey-milovidov
BTW, I cannot reproduce hang with a simple test.
2019-07-02T23:04:39
2019-07-02T23:04:39
{}
MEMBER
ClickHouse
ClickHouse
507,998,192
5,456
jangorecki
2 weeks ago there was a PR merged that mentioned this issue, what is the current status of this?
2019-07-03T08:37:57
2019-07-03T08:37:57
{}
NONE
ClickHouse
ClickHouse
507,159,011
5,681
filimonov
For distributed scenarios it's hard to guarantee 'fair' autoincrement between shards/replicas (but with some limitations and zookeeper coordination probably doable).
2019-07-01T07:59:18
2019-07-01T07:59:35
{}
CONTRIBUTOR
ClickHouse
ClickHouse
507,229,821
5,790
alesapin
Which engine do you use? Isn't `ReplicatedReplacingMergeTree`?
2019-07-01T11:41:50
2019-07-01T11:41:50
{}
MEMBER
ClickHouse
ClickHouse
507,285,342
5,264
bzon
This issue happened to me as well after updating our clickhouse default user credentials. From passwordless to having a new password. Query to a distributed table now returns this error. **Clickhouse version:** 19.4.1.3 ``` 2019.07.01 14:14:53.654539 [ 47 ] {68558650-7e64-4c30-b5c8-53ec893d861d} <Error> exe...
2019-07-01T14:20:35
2019-07-01T14:20:35
{ "+1": 1 }
NONE
ClickHouse
ClickHouse
507,159,982
5,472
filimonov
What do you mean be part creation time? New part is created after each merge. Merges happen all the time. Probably you expect something different?
2019-07-01T08:02:24
2019-07-01T08:02:24
{}
CONTRIBUTOR
ClickHouse
ClickHouse
507,161,142
5,182
filimonov
Cross-linking: https://gist.github.com/alexey-milovidov/d62d73222d83b9319dc519cbb13aeff6
2019-07-01T08:06:08
2019-07-01T08:06:08
{}
CONTRIBUTOR
ClickHouse
ClickHouse
507,395,796
5,823
den-crane
All of this is supposed behavior. >Is there a better way to secure this though? * leave password `<password>password123</password>` * return back to passwordless user and allow default to connect only from IP addreses of CH nodes.
2019-07-01T19:34:35
2019-07-01T19:34:35
{ "+1": 3 }
CONTRIBUTOR
ClickHouse
ClickHouse
507,470,448
5,828
alexey-milovidov
A test: ``` docker run -it --rm --volume=/home/milovidov/work/ClickHouse/build:/build ubuntu:disco /bin/bash apt update apt-get install tzdata </dev/null /build/dbms/programs/clickhouse local --query "SELECT timezone()" ```
2019-07-02T00:17:45
2019-07-02T00:17:45
{}
MEMBER
ClickHouse
ClickHouse
507,520,056
5,612
CAspeling
Any help asap would be greatly appreciated.
2019-07-02T05:08:38
2019-07-02T05:08:38
{}
NONE
ClickHouse
ClickHouse
507,274,329
5,820
alexey-milovidov
A test: `WITH (SELECT stochasticLinearRegressionState(1, 2, 3)) AS model SELECT evalMLMethod(model, toFloat64(1), toFloat64(1));`
2019-07-01T13:53:53
2019-07-01T13:53:53
{}
MEMBER
ClickHouse
ClickHouse
507,278,415
4,736
abyss7
First attempt to reproduce was unsuccessful - no data lost. > After consumption finished How did you make sure that consumption is finished on all tables? Can you provide tail (~1000 lines) of the logs to the moment when you do a check on number of messages? Can you try the same scenario with kafka setting `kafka_...
2019-07-01T14:03:48
2019-07-01T14:03:48
{}
CONTRIBUTOR
ClickHouse
ClickHouse
507,281,634
5,681
filimonov
> @filimonov K-sortable identifiers Yep. Using time based identifiers is one of the possible workaround / 'limitations' (while it's not equal to 'autoincement'). > would need to touch ZK only after node reboots to ensure clock didn't go backwards irc. ntpd sounds much better (one-time sync is always bad due to...
2019-07-01T14:11:55
2019-07-01T14:13:20
{ "+1": 1 }
CONTRIBUTOR
ClickHouse
ClickHouse
507,586,363
5,601
anindo-bandyopadhyay
@alexey-milovidov we are using confluent KSQL as our ETL tool on top of Kafka. We are pushing data to Kafka from multiple transaction databases using CDC, then we enrich information by combining data in different Kafka topics using KSQL. Confluent recommends using AVRO format, since it not only saves space, it is also ...
2019-07-02T08:57:50
2019-07-02T08:59:33
{}
NONE
ClickHouse
ClickHouse
507,655,319
5,257
alesapin
Trace for this error (v19.3.9.12): ```` #0 0x000000000a9a496d in __cxa_throw () #1 0x000000000343916e in Poco::Bugcheck::nullPointer (ptr=0xaba822e "_pInstance", file=0xaa3f6a0 "../contrib/poco/Util/include/Poco/Util/Application.h", line=457) at /usr/include/c++/8/ext/new_allocator.h:86 #2 0x00000000061e80a8 in ...
2019-07-02T12:29:19
2019-07-02T12:30:15
{}
MEMBER
ClickHouse
ClickHouse
507,771,727
5,836
alexey-milovidov
#5840
2019-07-02T17:25:48
2019-07-02T17:25:48
{}
MEMBER
ClickHouse
ClickHouse
507,152,379
5,804
simPod
Ok, thanks!
2019-07-01T07:38:42
2019-07-01T07:38:42
{}
CONTRIBUTOR
ClickHouse
ClickHouse
507,166,233
5,681
nvartolomei
@filimonov K-sortable identifiers, would need to touch ZK only after node reboots to ensure clock didn't go backwards irc.
2019-07-01T08:21:30
2019-07-01T08:21:30
{}
CONTRIBUTOR
ClickHouse
ClickHouse
507,258,710
5,681
den-crane
Auto-increments are useless. In multi-shard they could be implemented only with cache. So each shard increment id+100m(b)il. in ZK and uses this cache. But cache makes id uncoordinated with time of insert through all shards.
2019-07-01T13:12:43
2019-07-01T13:12:43
{}
CONTRIBUTOR
ClickHouse
ClickHouse
507,271,490
4,498
filimonov
You can use official build (ubuntu/docker), or look for librdkafka in cmake output, maybe some adjustment can be required for your platform. Please check: https://github.com/yandex/ClickHouse/blob/master/cmake/find_rdkafka.cmake
2019-07-01T13:46:48
2019-07-01T13:46:48
{}
CONTRIBUTOR
ClickHouse
ClickHouse
507,960,880
5,831
filimonov
What do you mean by 'special characters'? Can you provide some minimal sample?
2019-07-03T06:37:47
2019-07-03T06:37:47
{}
CONTRIBUTOR
ClickHouse
ClickHouse
507,197,029
5,472
mcarbonneaux
partition are made generaly based on somefing based on time, but not directly on the time... generaly they whant to remove the oldest created partition (not the modification time of the partition)... generaly all information inserted or merged in a partition are related to the time range of the partition key... ...
2019-07-01T09:50:22
2019-07-01T09:51:45
{}
NONE
ClickHouse
ClickHouse
507,224,829
5,815
alexey-milovidov
CentOS 6: init script does not work: ``` docker run --rm -it centos:6 /bin/bash ``` ``` cat > /etc/yum.repos.d/clickhouse.repo <<END [repo.yandex.ru_clickhouse_rpm_testing_x86_64] name=added from: https://repo.yandex.ru/clickhouse/rpm/testing/x86_64 baseurl=https://repo.yandex.ru/clickhouse/rpm/testing/x86_...
2019-07-01T11:24:46
2019-07-01T11:28:43
{}
MEMBER
ClickHouse
ClickHouse
507,272,345
5,790
SaltTan
The engine is ReplicatedMergeTree. Do you need the data file as well? It is quite big. ``` 131921876 Jun 25 16:08 unexpected_201906_8555_9352_5/ml_sm_ep1.bin 501 Jun 25 16:08 unexpected_201906_8555_9352_5/ml_sm_ep1.dict.bin 1831392 Jun 25 16:08 unexpected_201906_8555_9352_5/ml_sm_ep1.dict.mrk 1831392 Ju...
2019-07-01T13:48:59
2019-07-01T13:48:59
{}
CONTRIBUTOR
ClickHouse
ClickHouse
507,280,751
5,821
den-crane
>I have build ClickHouse from source with the latest github version. So you've compiled CH without Kafka. Why do you need to build CH by yourself? You can download latest builds https://github.com/yandex/ClickHouse/commits/master press green check sign -> details on 'ClickHouse build check' and download 'bundl...
2019-07-01T14:09:42
2019-07-01T14:10:30
{ "+1": 1 }
CONTRIBUTOR
ClickHouse
ClickHouse
507,284,301
5,472
filimonov
@mcarbonneaux when merging two parts: one created at 2019-01-01 and second created at 2019-01-31 (monthly partitioning) - which creation date you expect to use? It sounds like you need a column like `creation_time DateTime DEFAULT now()` and TTL based on that column.
2019-07-01T14:18:10
2019-07-01T14:20:27
{}
CONTRIBUTOR
ClickHouse
ClickHouse
507,722,212
5,835
rogue-squid
My hypothesis of it choosing random replicas on each servers is correct. When I for example add the sum of shard 1, shard 2 and shard 2 again (as if the wrong database on server 3 had been picked) the sum is one of the ones also occuring in the distributed query results.
2019-07-02T15:15:44
2019-07-02T15:15:44
{}
NONE
ClickHouse
ClickHouse
507,221,693
5,809
maskshell
No we don't have this data type. ```SQL CREATE TABLE report.amg_wms_stock_period_local ( `key_id` String, `brand_id` Int64, `storage_id` Int64, `spu_id` Int64, `skc_id` Int64, `sku_id` Int64, `biz_type` Int32, `business_date` Date, `pre_num` Int64, `pre_cost_money` Int64, `pre_sale_money` Int64...
2019-07-01T11:13:19
2019-07-01T11:13:19
{}
NONE
ClickHouse
ClickHouse
507,230,352
5,790
alesapin
Can you provide broken column files (both `.bin` and `.mrk`)?
2019-07-01T11:43:43
2019-07-01T11:43:43
{}
MEMBER
ClickHouse
ClickHouse
507,238,114
5,744
AlexAkulov
До версии v19.4.4.33 включительно, всё работало. Пробовал указать число или toYYYYMM('201810'), тоже не работет. ``` 09724b9d69a8 :) ALTER TABLE testdb.table1 FREEZE PARTITION ID '201810' ALTER TABLE testdb.table1 FREEZE PARTITION ID '201810' Received exception from server (version 19.9.2): Code: 368. ...
2019-07-01T12:10:04
2019-07-01T12:13:31
{}
NONE
ClickHouse
ClickHouse
507,249,325
5,789
alesapin
Market has another problem.
2019-07-01T12:45:12
2019-07-01T12:45:12
{}
MEMBER
ClickHouse
ClickHouse
507,255,738
5,762
den-crane
it could be mutations canceled by max_execution_time
2019-07-01T13:04:11
2019-07-01T13:04:11
{}
CONTRIBUTOR
ClickHouse
ClickHouse
507,328,400
5,744
AlexAkulov
@den-crane У меня таблицы созданы старым способом.
2019-07-01T16:06:58
2019-07-01T16:06:58
{}
NONE
ClickHouse
ClickHouse
507,446,440
5,472
mcarbonneaux
They sound like that... In fact the date of first insert in the partition... Or the first insert in parts.system table of the parts information ... With now() time... It's only useful when the partition key are based on time... But it's generally the case... OK It's indirect way to remove oldest partition in si...
2019-07-01T22:20:23
2019-07-01T22:20:23
{}
NONE
ClickHouse
ClickHouse
507,466,540
5,781
aycyang
Thank you, that solves my use case!
2019-07-01T23:57:41
2019-07-01T23:57:41
{}
NONE
ClickHouse
ClickHouse
507,631,197
4,663
pbuxtonccuk
TH-HA > Hello , i've also used logstash to do that , but it lack of flexbility on the clickhouse export.Also logstash is very heavy and not really user friendly when developping. > Instead to handle both case , i would suggest to use Apache NIFI. > Nifi is able to query on MSSQL transform your data and push them t...
2019-07-02T11:14:15
2019-07-02T11:14:15
{}
NONE
ClickHouse
ClickHouse
507,770,106
5,836
alexey-milovidov
Ok, I will make a patch. BTW, releases are build with gcc-8 and we are going to drop support for gcc-7. And it does not matter what compiler is in OS packages as you can always install newer compiler.
2019-07-02T17:21:40
2019-07-02T17:21:40
{}
MEMBER
ClickHouse
ClickHouse
507,822,005
958
alexey-milovidov
I don't understand how it helps and why the issue was present. Will revert.
2019-07-02T19:55:40
2019-07-02T19:55:40
{}
MEMBER
ClickHouse
ClickHouse
507,878,627
5,042
alexey-milovidov
@CurtizJ will do this task after another task (rewriting DNS Cache).
2019-07-02T23:17:14
2019-07-02T23:17:14
{}
MEMBER
ClickHouse
ClickHouse
507,254,606
5,744
den-crane
19.10.1.669 create table foo(D Date, S String) engine=MergeTree partition by toYYYYMM(D) order by S ALTER TABLE foo FREEZE PARTITION ID '201810' Ok.
2019-07-01T13:01:00
2019-07-01T13:01:00
{}
CONTRIBUTOR
ClickHouse
ClickHouse
507,270,503
5,809
alesapin
Seems to be fixed here https://github.com/yandex/ClickHouse/pull/5720. We will backport this fix to 19.8 and 19.9.
2019-07-01T13:44:05
2019-07-01T13:44:29
{}
MEMBER
ClickHouse
ClickHouse
507,162,033
5,680
filimonov
Version of rdkafka updated in v19.11 (or latest master): https://github.com/yandex/ClickHouse/pull/5699
2019-07-01T08:08:56
2019-07-01T13:47:37
{}
CONTRIBUTOR
ClickHouse
ClickHouse
507,273,638
5,820
alexey-milovidov
CC @Quid37 @ilyshnikova
2019-07-01T13:52:12
2019-07-01T13:52:12
{}
MEMBER
ClickHouse
ClickHouse
507,336,389
5,744
filimonov
А откуда вообще взялось это `ID` в команде? Кажется стандартный синтаксис и документация используют единственный способ указания партиций - без всяких ID: `ALTER TABLE testdb.table1 FREEZE PARTITION 201810`. Или я что-то пропустил?
2019-07-01T16:29:52
2019-07-01T16:29:52
{}
CONTRIBUTOR
ClickHouse
ClickHouse
507,357,927
5,744
AlexAkulov
Ну, как все знают в документации КХ многих вещей не написано) Я вначале пробовал без ID по имени, там вообще всё очень неочевидно работает, особенно при восстановлении. Невозможно понять, что востановилось, а что нет.
2019-07-01T17:37:05
2019-07-01T17:37:05
{}
NONE
ClickHouse
ClickHouse
507,571,922
5,681
filimonov
> For real autoincrement with not uniq data we need timestamp + more complex aggregating function. I don't think it's a good way. The simplest solution is @nvartolomei proposal: create some function like `ksuid()` ([see](https://github.com/segmentio/ksuid)) or similar, and use it as a DEFAULT.
2019-07-02T08:16:35
2019-07-02T08:16:35
{ "+1": 1 }
CONTRIBUTOR
ClickHouse
ClickHouse
507,605,497
5,827
alexey-milovidov
Merged in #5828
2019-07-02T09:50:43
2019-07-02T09:50:43
{}
MEMBER
ClickHouse
ClickHouse
507,619,311
5,681
4ertus2
My point is it's possible to implement autoincrement as selected information, not as an index. And I agree that it's not user generally wants. The feature request is for autoincremented column in index. And it probably will not be implemented ever. But if someone needs autoincrement outside of index, there could be som...
2019-07-02T10:34:31
2019-07-02T10:37:12
{}
CONTRIBUTOR
ClickHouse
ClickHouse
507,632,956
5,681
filimonov
> And it probably will not be implemented ever. Actually @andyyzh is [contributor](https://github.com/yandex/ClickHouse/pulls?utf8=%E2%9C%93&q=is%3Apr+author%3Aandyyzh). I'm not sure if he don't plan to add some variant of autoincrement? :)
2019-07-02T11:19:35
2019-07-02T11:19:35
{}
CONTRIBUTOR
ClickHouse
ClickHouse
514,077,638
6,031
yurriy
Maybe yes, I will look into that. Sorry for the inconvenience with php client, it will be possible to connect with default options in near future.
2019-07-23T06:49:04
2019-07-23T06:49:04
{}
CONTRIBUTOR