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
385,663,864
1,960
AlekSi
Documentation also states that "Graphite stores full data in ClickHouse" and make it sound like roll-ups are just read feature, the data on disk stays the same. It does not mention the fact that `OPTIMIZE` roll-ups data on disk.
2018-05-01T12:52:12
2018-05-01T12:52:12
{}
NONE
ClickHouse
ClickHouse
386,853,200
2,294
sundy-li
@zhang2014 Great to hear that. Let's work it together!!!
2018-05-06T04:33:24
2018-05-06T04:33:24
{}
CONTRIBUTOR
ClickHouse
ClickHouse
386,970,501
2,277
alexey-milovidov
Compiled expressions are invisible in `perf top`. How we can manage to make them visible (at least under some single name)?
2018-05-07T06:31:22
2018-05-07T06:31:22
{}
MEMBER
ClickHouse
ClickHouse
387,015,257
2,320
astudnev
Thank you, it works
2018-05-07T09:47:20
2018-05-07T09:47:20
{}
CONTRIBUTOR
ClickHouse
ClickHouse
387,035,966
2,277
alexey-milovidov
> The current tests should already do that with USE_EMBEDDED_COMPILER=1 since the compile_expressions setting is enabled by default. This is true for the gcc-7 Travis run, for instance. It's not obvious that compiled version is actually used. I've added a ProfileEvent for this purpose. You can add a test with `SELE...
2018-05-07T11:25:58
2018-05-07T11:25:58
{}
MEMBER
ClickHouse
ClickHouse
387,081,719
2,277
pyos
>Queries for hash function: Looks like pyos@0e09138 somehow makes the SQL version faster than the built-in function (400M rows/s vs. 360M). The only difference in the code generated before (when it used to produce ~310M rows/s) and after the commit are unconditional `add`s instead of `lea`+`test`+`cmovne`s. Huh.
2018-05-07T14:25:24
2018-05-07T14:28:26
{}
CONTRIBUTOR
ClickHouse
ClickHouse
387,202,926
2,277
pyos
>I've added a ProfileEvent for this purpose. You can add a test with SELECT ... FROM system.events (or propose a better solution). Such a test would need to be skipped if `USE_EMBEDDED_COMPILER` is 0. Is that possible?
2018-05-07T20:53:15
2018-05-07T20:53:15
{}
CONTRIBUTOR
ClickHouse
ClickHouse
388,059,579
2,341
yuragr76
Простого примера пока нет. Структура без лишних полей CREATE TABLE checks.tr_checks ( date_check Date , ... count_checks Int64 DEFAULT CAST(1 AS Int64), ... ) ENGINE = ReplicatedMergeTree('/clickhouse/tables/{layer}-{shard}/checks_tr_checks', '{replica}', date_check, (date_check_hash, time_check, date_check...
2018-05-10T13:52:16
2018-05-10T13:52:16
{}
NONE
ClickHouse
ClickHouse
388,062,116
2,341
yuragr76
Вот весь лог 2018.05.10 16:57:24.004025 [ 522 ] <Debug> executeQuery: (from [::ffff:10.176.1.126]:52652, user: grankin, query_id: 31872be7-4497-47e9-b9f2-3c2ce499f7e7) select sum(count_checks) as m0 from checks.tr_checks prewhere date_check >= addMont hs(toStartOfMonth(today()), -3) FORMAT TabSeparatedWithNamesAndT...
2018-05-10T14:00:48
2018-05-10T14:00:48
{}
NONE
ClickHouse
ClickHouse
388,202,741
2,294
alexey-milovidov
As I remember, the most efficient way to insert data is `LOAD DATA LOCAL INFILE`, but current solution seems good enough and I don't care.
2018-05-10T22:16:58
2018-05-10T22:16:58
{}
MEMBER
ClickHouse
ClickHouse
388,203,023
2,294
alexey-milovidov
> Support INSERT INTO mysql() SELECT * FROM table I see no reason why it shouldn't work automatically. > Add a new parameter at TableFunctionMySQL for ON DUPLICATE KEY And REPLACE. It can be postponed for the next commit.
2018-05-10T22:18:20
2018-05-10T22:18:20
{}
MEMBER
ClickHouse
ClickHouse
388,242,865
2,294
sundy-li
> I see no reason why it shouldn't work automatically. Because Table functions can only be specified in the FROM clause for select usage. > Add a new parameter at TableFunctionMySQL for ON DUPLICATE KEY The `ON DUPLICATE KEY` and `REPLACE INTO` clauses are only used in the MySQL insert query. But from now on `...
2018-05-11T02:40:40
2018-05-11T02:40:40
{}
CONTRIBUTOR
ClickHouse
ClickHouse
388,268,207
2,343
sundy-li
It seems like `--output-format=CSV` did't work, all test output are TabSeparated.
2018-05-11T05:59:29
2018-05-11T05:59:29
{}
CONTRIBUTOR
ClickHouse
ClickHouse
386,946,157
2,277
alexey-milovidov
compileFunctions is executed multiple times for the same expression. For example, for query: ``` SELECT count() FROM system.numbers_mt WHERE NOT ignore(number + number) ``` it is executed 8 times.
2018-05-07T03:13:54
2018-05-07T03:13:54
{}
MEMBER
ClickHouse
ClickHouse
387,019,258
2,320
ztlpn
Support for applying ALTER PARTITION commands directly to materialized views was added in 1.1.54380
2018-05-07T10:04:07
2018-05-07T10:04:07
{}
CONTRIBUTOR
ClickHouse
ClickHouse
387,202,603
2,201
alexey-milovidov
NaNs and NULLs are last (with NaNs before NULLs). It is described in `IColumn::compareAt` (look at `nan_direction_hint` parameter).
2018-05-07T20:52:12
2018-05-07T20:52:12
{}
MEMBER
ClickHouse
ClickHouse
387,359,243
2,326
ztlpn
As stated in [the docs](https://clickhouse.yandex/docs/en/query_language/queries/#into-outfile-clause), the output file is created on the client side. It is pretty hard to do via the HTTP interface :) So INTO OUTFILE is disabled in this case. You can use `clickhouse-client` for that.
2018-05-08T10:31:30
2018-05-08T10:31:30
{}
CONTRIBUTOR
ClickHouse
ClickHouse
387,387,019
2,321
KochetovNicolai
ClickHouse has fast but not very precise parser in case you insert data. It works similar to: ``` SELECT 1.23456789 * 100000000. ┌─multiply(1.23456789, 100000000.)─┐ │ 123456788.99999999 │ └──────────────────────────────────┘ ```
2018-05-08T12:33:31
2018-05-08T12:33:31
{}
MEMBER
ClickHouse
ClickHouse
387,665,377
2,331
migue
I've tried `1.1.54380` and seems that the issue has been already solved. Sorry for the noise
2018-05-09T08:36:07
2018-05-09T08:36:07
{}
NONE
ClickHouse
ClickHouse
387,772,822
2,332
den-crane
>TCP Request. Address: 1.2.3.4:58719 It's really strange IP -- 1.2.3.4, usually it's been used for icmp hole punching. Check your firewall settings, it seems you allow tcp connections from the public network.
2018-05-09T15:11:36
2018-05-09T15:11:46
{}
CONTRIBUTOR
ClickHouse
ClickHouse
388,231,123
2,338
zhang2014
Maybe related to #2226
2018-05-11T01:17:22
2018-05-11T01:17:22
{}
CONTRIBUTOR
ClickHouse
ClickHouse
386,736,791
2,317
alexey-milovidov
It looks like false positive. We can filter it with `pragma GCC diagnostic`. I doubt about possible performance differences after this change. We can test with the cases when many simple Field objects are created. Like `sumMap` aggregate function. Although usage of multiple small Field objects is already subopt...
2018-05-04T21:13:23
2018-05-04T21:13:59
{}
MEMBER
ClickHouse
ClickHouse
387,144,513
2,261
atris
I have raised a PR for the same. https://github.com/yandex/ClickHouse/pull/2325
2018-05-07T17:41:06
2018-05-07T17:41:06
{}
CONTRIBUTOR
ClickHouse
ClickHouse
387,876,610
2,332
filimonov
May be some health checks? I.e. for example loadbalancer checking that port is opened?
2018-05-09T21:09:54
2018-05-09T21:10:53
{ "+1": 1 }
CONTRIBUTOR
ClickHouse
ClickHouse
387,880,760
2,332
derekperkins
@filimonov Thanks for the suggestion, that is absolutely it, I can't believe I didn't realize it sooner. Right now, the health check is opening a TCP connection to port 9000. Is there a better way to make sure that ClickHouse is working properly that won't throw errors?
2018-05-09T21:26:54
2018-05-09T21:26:54
{}
CONTRIBUTOR
ClickHouse
ClickHouse
387,958,353
2,335
derekperkins
This is also the root cause of #2295. For some reason after the server shuts down here, it uses a full cpu core while not logging anything. I don't know what it could possibly be doing, but it seems like it probably gets stuck in a deadlock, which may explain why the http server doesn't also shut down. cc @alex-zait...
2018-05-10T05:46:08
2018-05-10T05:46:08
{}
CONTRIBUTOR
ClickHouse
ClickHouse
388,058,879
2,341
KochetovNicolai
Нет ли простого примера, на котом падение воспроизводится? Какая структура таблицы checks.tr_checks, является ли она Distributed или Replicated?
2018-05-10T13:50:00
2018-05-10T13:50:00
{}
MEMBER
ClickHouse
ClickHouse
388,337,361
2,294
zhang2014
I created a issues[#2346] to track subsequent changes.
2018-05-11T11:27:11
2018-05-11T11:27:11
{}
CONTRIBUTOR
ClickHouse
ClickHouse
386,654,340
2,277
alexey-milovidov
> always inlining compilable subexpressions completely undoes all of common subexpression eliminator's work — using a heuristic, e.g. thresholding based on the size of the subexpression's graph multiplied by the number of times it is used, might be better overall; That's Ok. The effect is like to be neglible because...
2018-05-04T16:28:20
2018-05-04T16:28:20
{}
MEMBER
ClickHouse
ClickHouse
387,753,284
1,686
dqminh
Another case for implementing aggressive merge to grab more small parts at once. We had some MVs that created many small parts from a bigger NULL table firehose, and they do not merge very well, causing huge number of parts that will be merged, replicated, deleted repeatedly. ``` SELECT database, ta...
2018-05-09T14:15:01
2018-05-10T17:22:00
{ "+1": 3 }
CONTRIBUTOR
ClickHouse
ClickHouse
386,970,033
2,277
alexey-milovidov
Queries for hash function: ``` WITH bitXor(number, 0x4CF2D2BAAE6DA887) AS x0, bitXor(x0, bitShiftRight(x0, 33)) AS x1, x1 * 0xff51afd7ed558ccd AS x2, bitXor(x2, bitShiftRight(x2, 33)) AS x3, x3 * 0xc4ceb9fe1a85ec53 AS x4, bitXor(x4, bitShiftRight(x4, 33)) AS x5 SELECT x5, intHash64(nu...
2018-05-07T06:28:51
2018-05-07T06:28:51
{}
MEMBER
ClickHouse
ClickHouse
387,241,402
2,277
alexey-milovidov
> Such a test would need to be skipped if USE_EMBEDDED_COMPILER is 0. Is that possible? Add "compile" word to the test name (we have two existing tests of that kind). (As far as I remember, they aren't disabled automatically, but will be easily disabled.)
2018-05-07T23:53:58
2018-05-07T23:53:58
{}
MEMBER
ClickHouse
ClickHouse
387,471,857
2,321
trolley813
@KochetovNicolai it still can be a concern, since JDBC driver (for example) uses simple `Double.toString()` ([see here](https://github.com/yandex/clickhouse-jdbc/blob/24f1a40fb85a29c5d86b8567b7dda0666163e94f/src/main/java/ru/yandex/clickhouse/ClickHousePreparedStatementImpl.java#L240)) on batch insertion, so it can lea...
2018-05-08T17:02:01
2018-05-08T17:02:01
{}
NONE
ClickHouse
ClickHouse
387,736,706
2,334
filimonov
Use direct http connections if you definitely want to go on lower level.
2018-05-09T13:22:53
2018-05-09T13:22:53
{}
CONTRIBUTOR
ClickHouse
ClickHouse
388,081,556
2,339
KochetovNicolai
Такой настройки нет. ClickHouse ожидает, что неактивная реплика будет когда-нибудь запущена, и данные не будут потеряны. Если реплика не будет восстановлена, то должно помочь удаление ноды из зукипера, соответствующей реплике.
2018-05-10T15:04:30
2018-05-10T15:04:30
{}
MEMBER
ClickHouse
ClickHouse
388,115,955
2,339
alxark
@KochetovNicolai спасибо за ответ. Это может быть проблематично в сильно динамических окружениях. Можно ли ожидать появления такой настройки в будущем? Чистка нод как я понимаю тоже необходимо, потому что они могут подыхать с завидной регулярностью в больших инсталляциях.
2018-05-10T16:58:38
2018-05-10T16:58:38
{}
NONE
ClickHouse
ClickHouse
386,106,526
2,310
alexey-milovidov
Sure. We have an intent to complete it. But for now there is no time schedule for this feature - we will try to do it "in between" of another features.
2018-05-02T20:15:10
2018-05-02T20:15:10
{ "+1": 3 }
MEMBER
ClickHouse
ClickHouse
386,399,997
2,273
bocharov
@alexey-milovidov @dqminh It seems that problem is not with materialized view, but with the `Nested` data type, it's broken in new version 1.1.54380. To reproduce create such or similar table: ``` CREATE TABLE r0.tmp_requests_1m ( date Date, zoneId UInt32, timeslot DateTime, requests UInt64, ...
2018-05-03T18:53:14
2018-05-03T18:53:14
{}
CONTRIBUTOR
ClickHouse
ClickHouse
386,649,764
2,277
ztlpn
autotests
2018-05-04T16:11:18
2018-05-04T16:11:18
{}
CONTRIBUTOR
ClickHouse
ClickHouse
386,735,055
2,317
alexey-milovidov
How exactly does it complain?
2018-05-04T21:05:58
2018-05-04T21:05:58
{}
MEMBER
ClickHouse
ClickHouse
386,852,979
2,294
zhang2014
I'm also interested in this task. Maybe we can work together to make it better? I've listed several improvements that we might be able to do together : ) : - [ ] Build write to mysql using prepared_statement - [ ] Support `INSERT INTO mysql() SELECT * FROM table` - [ ] Add a new parameter at `TableFunctionMySQL` f...
2018-05-06T04:26:24
2018-05-06T04:26:24
{}
CONTRIBUTOR
ClickHouse
ClickHouse
387,369,837
2,278
filimonov
We also encountered really slow startups - several minutes of server downtime during restart doesn't look nice.
2018-05-08T11:17:40
2018-05-08T11:18:18
{}
CONTRIBUTOR
ClickHouse
ClickHouse
387,523,599
1,999
harlinb
I encountered the same issue after doing an ANY LEFT JOIN version 1.1.54343 ``` 2018.05.08 19:42:12.059772 [ 54126 ] <Error> BaseDaemon: ######################################## 2018.05.08 19:42:12.059794 [ 54126 ] <Error> BaseDaemon: (from thread 54125) Received signal Segmentation fault (11). 2018.05.08 19:42...
2018-05-08T19:57:56
2018-05-08T21:20:59
{}
NONE
ClickHouse
ClickHouse
387,834,541
2,332
derekperkins
After running without any external access for a few hours, there was no difference in the log errors, so it wasn't just rogue bots trying to connect. It must be internal.
2018-05-09T18:35:16
2018-05-09T18:46:05
{}
CONTRIBUTOR
ClickHouse
ClickHouse
385,643,575
1,660
AlekSi
This is done now: https://clickhouse.yandex/docs/en/table_engines/graphitemergetree/
2018-05-01T10:37:02
2018-05-01T10:37:02
{}
NONE
ClickHouse
ClickHouse
386,946,324
2,277
alexey-milovidov
We need at least one functional test that proves that compilation actually works.
2018-05-07T03:15:13
2018-05-07T03:15:13
{}
MEMBER
ClickHouse
ClickHouse
387,700,334
1,469
zhangguiyu
Wow! thanks for the heads-up! I will try this out.
2018-05-09T10:49:54
2018-05-09T10:49:54
{}
NONE
ClickHouse
ClickHouse
386,027,065
79
abdrakhmanov
i have similar problem. For example: ``` CREATE TABLE `t1` ( `c0` UInt8, `c1` Nullable(Date), `c4` Date ) ENGINE = MergeTree(c4, (c4), 8192) ``` Insertion in tabix work fine: ``` INSERT INTO t1 VALUES (1,NULL,'2018-04-01') ``` Insert query in command-line doesn't work: ``` #cat c.csv 1,NULL,'2018-0...
2018-05-02T15:53:28
2018-05-09T17:09:29
{}
CONTRIBUTOR
ClickHouse
ClickHouse
387,910,457
2,332
den-crane
try to use 8123 instead of 9000 $ curl http://localhost:8123 Ok.
2018-05-10T00:00:51
2018-05-10T00:01:42
{ "+1": 1 }
CONTRIBUTOR
ClickHouse
ClickHouse
388,394,184
2,342
den-crane
Another order in partition by drop table test.testy; CREATE TABLE test.testy(d Int64, n Int64) ENGINE = MergeTree Partition by (n,ceil(d/2)) Order by tuple(); insert into test.testy select 1, 33 n from numbers(1); insert into test.testy select 2, 33 n from numbers(10000000); insert into test.testy select 1...
2018-05-11T15:14:50
2018-05-11T15:16:57
{}
CONTRIBUTOR
ClickHouse
ClickHouse
386,104,316
2,311
alexey-milovidov
We already have it! Look at: ``` <!-- Uncomment if use part_log <part_log> <database>system</database> <table>part_log</table> <flush_interval_milliseconds>7500</flush_interval_milliseconds> </part_log> --> ``` in config.xml. If you enable it, the table `syste...
2018-05-02T20:07:37
2018-05-02T20:07:37
{}
MEMBER
ClickHouse
ClickHouse
387,030,117
2,277
pyos
>compileFunctions is executed multiple times for the same expression. For example, for query: It is executed as many times as `ExpressionActions` is constructed. If that's a problem, perhaps `ExpressionActions` should be reused instead of walking the AST every time. This is probably outside the scope of this PR. ...
2018-05-07T10:56:13
2018-05-07T11:01:45
{}
CONTRIBUTOR
ClickHouse
ClickHouse
387,693,522
1,469
tuedelue
Just as a followup: I have seen in recent changelogs, that there is the `arrayCumSum` method, which can be used for getting at least the accumulate values. (Edit: You of course may use any other array operation, the following describes sort of a grouping workaround) I am not sure whether this performs well enough on...
2018-05-09T10:20:25
2018-05-09T10:55:03
{ "+1": 7, "rocket": 3 }
NONE
ClickHouse
ClickHouse
388,017,440
2,338
zzl-wy
SELECT datasource_id AS dimension_datasource_id, sum(if_in_pkts) AS metric_in_pkt_pps FROM merge(ncompass, '^(data_17_1508831948520041_3600_20180510_all)$') WHERE dimension_datasource_id = 1525680565993917 GROUP BY dimension_datasource_id ORDER BY metric_in_pkt_pps DESC LIMIT 0, 50 ┌─_table───────────...
2018-05-10T10:34:42
2018-05-10T10:34:42
{}
NONE
ClickHouse
ClickHouse
388,073,838
2,096
Jingyi244
@sundy-li I have just built clickhouse with the patch. exploring windowFunnel now... Thanks for the repo, i will try it later.
2018-05-10T14:40:04
2018-05-10T14:40:04
{}
NONE
ClickHouse
ClickHouse
388,297,364
2,338
zzl-wy
I have found a new problem.when I used group by clause the _table column would be added to the group condition caused the result not correct.The sql and result as follow picture in the url. http://chuantu.biz/t6/309/1526026570x-1376440078.png
2018-05-11T08:26:43
2018-05-11T08:26:43
{}
NONE
ClickHouse
ClickHouse
388,353,956
2,348
ludv1x
I am not sure about the necessity of these changes.
2018-05-11T12:49:36
2018-05-11T12:49:36
{}
CONTRIBUTOR
ClickHouse
ClickHouse
386,803,284
2,266
proller
you always can build on any linux with gcc7+ or clang5+. also now pbuilder build fixed for arm. Example for my board: ```env DIST=bionic DISABLE_PARALLEL=1 EXTRAPACKAGES="clang-6.0 libstdc++-8-dev" DEB_CC=clang-6.0 DEB_CXX=clang++-6.0 CMAKE_FLAGS=" -DNO_WERROR=1 " ./release```
2018-05-05T12:46:22
2018-05-05T12:46:22
{}
CONTRIBUTOR
ClickHouse
ClickHouse
386,968,182
1,152
kszucs
We can close this now, ibis now integrates with clickhouse pretty well.
2018-05-07T06:18:24
2018-05-07T06:18:24
{}
CONTRIBUTOR
ClickHouse
ClickHouse
387,353,946
2,245
StarWix
@alexey-milovidov it would be nice to have this setting.
2018-05-08T10:11:46
2018-05-08T10:11:46
{}
NONE
ClickHouse
ClickHouse
387,382,553
2,291
filimonov
My fault: folder was not writable for clickhouse user.
2018-05-08T12:15:39
2018-05-08T12:15:39
{}
CONTRIBUTOR
ClickHouse
ClickHouse
387,382,951
2,327
filimonov
Documented: https://clickhouse.yandex/docs/en/data_types/date/ > A date. Stored in two bytes as the number of days since 1970-01-01 (unsigned). Allows storing values from just after the beginning of the Unix Epoch to the upper threshold defined by a constant at the compilation stage (currently, this is until the ye...
2018-05-08T12:17:16
2018-05-08T12:17:16
{}
CONTRIBUTOR
ClickHouse
ClickHouse
387,951,351
79
filimonov
In TSV Nulls are represened as \N Usually you can check exactly which format clickhouse expect to see by writing few rows in that format by clickhouse. I.e. ``` SELECT * FROM t1 FORMAT TSV; ```
2018-05-10T04:51:30
2018-05-10T04:51:30
{ "+1": 1 }
CONTRIBUTOR
ClickHouse
ClickHouse
388,610,365
2,352
sundy-li
In 2017, there was a wonderful [share](https://events.yandex.com/lib/talks/247/) by `Mariya Mansurova` who taught us how to calculate complex funnel metrics in ClickHouse by `sequenceMatch`. But it was less efficient (pattern matching) and it lacks the sliding timestamp window support, which is very important o...
2018-05-13T08:30:53
2018-05-13T08:39:28
{ "+1": 13 }
CONTRIBUTOR
ClickHouse
ClickHouse
385,682,750
2,290
den-crane
dict example with incl https://gist.github.com/den-crane/d6e0f1c7849a0bc44c27eab2de51c911#file-ch-dictionary-source-included
2018-05-01T14:23:15
2018-05-01T14:23:15
{}
CONTRIBUTOR
ClickHouse
ClickHouse
386,498,619
2,070
LiuYangkuan
I have fix this BUG. https://github.com/yandex/ClickHouse/pull/1556 But this is not a commended method to use kafka engine to put stream data into distributed engine table. You can only use kafka engine to put data into local table in every server node, and read from distributed table:)
2018-05-04T03:39:20
2018-05-04T03:39:20
{}
CONTRIBUTOR
ClickHouse
ClickHouse
386,735,283
2,317
pyos
``` /home/pyos/ClickHouse/dbms/src/Core/Field.h: In member function ‘virtual bool DB::ParserSetQuery::parseImpl(DB::IParser::Pos&, DB::ASTPtr&, DB::Expected&)’: /home/pyos/ClickHouse/dbms/src/Core/Field.h:364:34: error: ‘*((void*)&<anonymous> +8)’ may be used uninitialized in this function [-Werror=maybe-uninitialize...
2018-05-04T21:07:03
2018-05-04T21:07:03
{}
CONTRIBUTOR
ClickHouse
ClickHouse
386,821,664
2,260
zhang2014
I'm done. Please help review the changes. @alexey-milovidov @ztlpn
2018-05-05T17:26:23
2018-05-05T17:26:23
{}
CONTRIBUTOR
ClickHouse
ClickHouse
386,867,850
2,277
alexey-milovidov
TODO: - implement for bitwise (done), comparison, conditional functions with numeric arguments if not already; - compare performance of ADD/OR chains of different sizes; - compare performance of integer division by constant with current implementation (that is using libdivide); - compare performance of embedded sim...
2018-05-06T09:59:24
2018-05-07T03:09:01
{}
MEMBER
ClickHouse
ClickHouse
387,379,113
2,326
vixa2012
I made it via the shell ``` ssh user@192.198.0.1 /bin/bash << EOF clickhouse-client --user=admin --password=111111 --query="SELECT * FROM test.\"stats\" AS t WHERE date = '2018-04-07' INTO OUTFILE '/user/reports/1525777158845.csv' FORMAT CSVWithNames" EOF ``` My nodejs solution ``` const { exec } = r...
2018-05-08T12:00:03
2018-05-08T12:05:02
{ "+1": 1 }
NONE
ClickHouse
ClickHouse
387,947,085
2,332
derekperkins
That worked great, thanks!
2018-05-10T04:15:12
2018-05-10T04:15:12
{}
CONTRIBUTOR
ClickHouse
ClickHouse
387,321,505
2,325
atris
@alexey-milovidov Please review the second version
2018-05-08T08:11:10
2018-05-08T08:11:10
{}
CONTRIBUTOR
ClickHouse
ClickHouse
387,354,509
2,281
silviucpp
@alexey-milovidov any thought on this ?
2018-05-08T10:13:40
2018-05-08T10:13:40
{}
CONTRIBUTOR
ClickHouse
ClickHouse
387,976,583
2,336
filimonov
Can you provide more information? For example - query which caused that?
2018-05-10T07:33:27
2018-05-10T07:33:27
{}
CONTRIBUTOR
ClickHouse
ClickHouse
388,393,257
2,342
den-crane
drop table test.testy; CREATE TABLE test.testy(d Int64, n Int64) ENGINE = MergeTree Partition by (ceil(d/2), n) Order by tuple(); insert into test.testy select 1, 33 n from numbers(1); insert into test.testy select 2, 33 n from numbers(10000000); insert into test.testy select 1, 34 n from numbers(1); SE...
2018-05-11T15:11:35
2018-05-11T15:16:51
{}
CONTRIBUTOR
ClickHouse
ClickHouse
388,395,965
2,342
den-crane
Changed the order of inserts: drop table test.testy; CREATE TABLE test.testy(d Int64, n Int64) ENGINE = MergeTree Partition by (ceil(d/2),n) Order by tuple(); insert into test.testy select 2, 33 n from numbers(10000000); insert into test.testy select 1, 33 n from numbers(1); insert into test.testy select 1...
2018-05-11T15:20:39
2018-05-11T15:45:22
{}
CONTRIBUTOR
ClickHouse
ClickHouse
386,038,771
2,264
meshpaul
:+1:. Agreed but the way one can monitor understand these is put some load and see if something changes and relates to CH degrade or error rate increase.
2018-05-02T16:29:15
2018-05-02T16:29:15
{ "laugh": 1 }
NONE
ClickHouse
ClickHouse
387,619,385
2,317
alexey-milovidov
Fixed in separate PR.
2018-05-09T04:51:45
2018-05-09T04:51:45
{}
MEMBER
ClickHouse
ClickHouse
388,348,308
2,294
alexey-milovidov
> Because Table functions can only be specified in the FROM clause for select usage. We have `INSERT INTO TABLE FUNCTION`.
2018-05-11T12:23:18
2018-05-11T12:23:18
{}
MEMBER
ClickHouse
ClickHouse
388,548,425
2,260
zhang2014
How can I rerun the test?
2018-05-12T11:23:51
2018-05-12T11:23:51
{}
CONTRIBUTOR
ClickHouse
ClickHouse
386,139,090
2,277
ztlpn
autotests
2018-05-02T22:18:50
2018-05-02T22:18:50
{}
CONTRIBUTOR
ClickHouse
ClickHouse
386,413,989
2,314
den-crane
The problem is that 0 is a constant, need to replace it with any non-reducible (to constant) expression. select dictGetString('dim', 'dim_id', x) from (select tuple(toInt64(rand64()*0)) as x); select dictGetString('dim', 'dim_id', x) from (select tuple(toInt64(blockSize()=0)) as x); select dictGetString('dim...
2018-05-03T19:45:21
2018-05-03T19:48:04
{}
CONTRIBUTOR
ClickHouse
ClickHouse
387,152,799
2,245
atris
Is this issue relevant? I would like to fix this if it makes sense
2018-05-07T18:08:48
2018-05-07T18:08:48
{}
CONTRIBUTOR
ClickHouse
ClickHouse
387,202,887
2,201
alexey-milovidov
``` SELECT arraySort([1, nan, 2, NULL, 3, nan, 4, NULL]) ┌─arraySort([1, nan, 2, NULL, 3, nan, 4, NULL])─┐ │ [1,2,3,4,nan,nan,NULL,NULL] │ └───────────────────────────────────────────────┘ 1 rows in set. Elapsed: 0.083 sec. SELECT arrayReverseSort([1, nan, 2, NULL, 3, nan, 4, NULL]) ┌─...
2018-05-07T20:53:08
2018-05-07T20:53:08
{}
MEMBER
ClickHouse
ClickHouse
387,271,695
2,245
atris
@alexey-milovidov Please let me know
2018-05-08T03:16:10
2018-05-08T03:16:10
{}
CONTRIBUTOR
ClickHouse
ClickHouse
387,418,961
379
simPod
I think this issue can be reopened as IMHO it would be handy to have such grouping function in CH. Have anyone found any workaround for this except for computing it on client side which is not a very good solution?
2018-05-08T14:20:45
2018-05-08T14:20:51
{}
CONTRIBUTOR
ClickHouse
ClickHouse
388,117,709
1,215
proller
fixed
2018-05-10T17:04:11
2018-05-10T17:04:11
{}
CONTRIBUTOR
ClickHouse
ClickHouse
386,637,854
2,273
bocharov
@alexey-milovidov Minimal reproducible example is: ``` CREATE TABLE test ( date Date, key UInt32, testMap Nested( k UInt16, v UInt64) ) ENGINE = SummingMergeTree(date, (date, key), 1); INSERT INTO test SELECT today(), rand(), [toUInt16(number)], [rand()] FROM...
2018-05-04T15:30:34
2018-05-04T15:30:34
{}
CONTRIBUTOR
ClickHouse
ClickHouse
387,144,591
2,325
atris
Issue: https://github.com/yandex/ClickHouse/issues/2261
2018-05-07T17:41:21
2018-05-07T17:41:21
{}
CONTRIBUTOR
ClickHouse
ClickHouse
387,735,375
2,334
filimonov
When using clickhouse jdbc client you can't change the format of data interchange between client and server (otherwise jdbc client will not be able to parse out columns with data).
2018-05-09T13:18:30
2018-05-09T13:18:30
{}
CONTRIBUTOR
ClickHouse
ClickHouse
387,999,268
2,334
Joozty
@filimonov Thanks for quick answer.
2018-05-10T09:12:19
2018-05-10T09:12:19
{}
NONE
ClickHouse
ClickHouse
388,005,426
2,096
Jingyi244
@sundy-li thanks 4 the share. https://github.com/analysys/olap/tree/master/%E5%BC%80%E6%BA%90%E7%BB%84%E7%AC%AC%E4%B8%80%E5%90%8D we are following the toutrial on the upper site, try to compile clickHouse. but failed on **git checkout ab7672f329f7736756542268178e6f9f7e32325a** could u release a new patch? ...
2018-05-10T09:39:17
2018-05-10T09:39:17
{}
NONE
ClickHouse
ClickHouse
388,043,453
2,096
sundy-li
@Jingyi244 That code is not maintained by me , you could try the latest windowFunnel code [here](https://github.com/sundy-li/ClickHouse/tree/feature/funnelWindow),based on commit 12905f5c6f621e7e1051e818f04bb4d56cbf06bb
2018-05-10T12:48:23
2018-05-10T12:48:23
{}
CONTRIBUTOR
ClickHouse
ClickHouse
388,072,321
2,341
yuragr76
вот такой запрос работает select count_checks + 1 from checks.tr_checks prewhere date_check >= addMonths(toStartOfMonth(today()), -3) limit 100 а вот такой ругается select sum(count_checks + 1) from checks.tr_checks prewhere date_check >= addMonths(toStartOfMonth(today()), -3) limit 100 ClickHouse exception, co...
2018-05-10T14:35:01
2018-05-10T14:35:01
{}
NONE
ClickHouse
ClickHouse
386,262,925
2,311
bocharov
@alexey-milovidov it appears, that `system.part_log` doesn't log all new parts, maybe it's omitting ones which came from materialized views. We're observing many tiny parts when working with materialized views and they are visible in `system.parts` and on disk. It's related to https://github.com/yandex/ClickHouse/issue...
2018-05-03T11:20:51
2018-05-03T11:20:51
{}
CONTRIBUTOR
ClickHouse
ClickHouse
387,010,387
2,320
filimonov
Each materialized view create underlying table with .inner. suffix to store the data. Try drop partition from that underlying table.
2018-05-07T09:26:58
2018-05-07T09:26:58
{}
CONTRIBUTOR
ClickHouse
ClickHouse
385,925,765
2,264
ThomasBarach
+1
2018-05-02T09:55:31
2018-05-02T09:55:31
{}
NONE
ClickHouse
ClickHouse
386,413,118
2,280
simPod
Thanks for reply!
2018-05-03T19:42:08
2018-05-03T19:42:08
{}
CONTRIBUTOR
ClickHouse
ClickHouse
387,350,173
2,245
alexey-milovidov
The behaviour is correct. It is mostly compatible to the SQL standard and to MySQL, Postgres, SQLite, MS SQL: http://sqlfiddle.com/#!9/0a28f18/105 http://sqlfiddle.com/#!17/0a28f/82 http://sqlfiddle.com/#!5/0a28f18/1 http://sqlfiddle.com/#!18/9eecb/16802 In contrast, NaNs are not exist in standard SQL and we ...
2018-05-08T09:57:01
2018-05-08T09:58:05
{}
MEMBER
ClickHouse
ClickHouse
387,787,272
2,332
derekperkins
> It's really strange IP -- 1.2.3.4, usually it's been used for icmp hole punching. I took out my public IP and replaced it with `1.2.3.4` when reporting this issue. > Check your firewall settings, it seems you allow tcp connections from the public network. That's a good idea, I'll try closing my load balancer...
2018-05-09T15:55:38
2018-05-09T15:55:46
{}
CONTRIBUTOR