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
348,564,442
1,504
ludv1x
It should be fixed in hotfix release 1.1.54318 https://github.com/yandex/ClickHouse/blob/master/CHANGELOG.md#clickhouse-release-1154318
2017-12-01T18:00:13
2017-12-01T18:00:13
{}
CONTRIBUTOR
ClickHouse
ClickHouse
349,000,732
1,579
Slind14
I see, thank you. Are you aware of anything public regarding the MaxMind geobase or some users I could ask? No need to reinvent the wheel.
2017-12-04T15:42:34
2017-12-04T15:42:34
{}
NONE
ClickHouse
ClickHouse
349,104,541
1,523
alexey-milovidov
Tnx! Sorry for the delay.
2017-12-04T21:03:27
2017-12-04T21:03:27
{}
MEMBER
ClickHouse
ClickHouse
350,200,450
1,565
alexey-milovidov
Ok, but the following modifications are useless: ``` Submodule googletest deleted from d175c8 dbms/src/Columns/ColumnConst.cpp dbms/src/Columns/ColumnConst.h dbms/src/Columns/ColumnNullable.h dbms/src/Columns/IColumn.h dbms/src/DataTypes/DataTypeNullable.cpp ``` 1. Accidentially deleted submodule. 2. New m...
2017-12-08T08:11:01
2017-12-08T08:11:01
{}
MEMBER
ClickHouse
ClickHouse
350,335,119
1,595
ztlpn
The crash occurs during the initialization of libmariadb client library. Did you compile ClickHouse yourself? Can you try the same query with libmysqlclient library?
2017-12-08T18:21:55
2017-12-08T18:21:55
{}
CONTRIBUTOR
ClickHouse
ClickHouse
350,865,423
1,246
alexey-milovidov
Fixed in master.
2017-12-11T21:33:52
2017-12-11T21:33:52
{}
MEMBER
ClickHouse
ClickHouse
351,390,023
1,625
tobad357
Sorry for one more request It would be great to be able to set the offset when creating the Kafka Table The default could be "latest" but also allow "earliest" and a specific value That way reprocessing of data would be to just drop the table and recreate with the correct offset
2017-12-13T13:24:56
2017-12-13T13:24:56
{}
NONE
ClickHouse
ClickHouse
351,516,229
1,628
alexey-milovidov
b3277408c
2017-12-13T20:44:02
2017-12-13T20:44:02
{}
MEMBER
ClickHouse
ClickHouse
348,988,567
1,576
ztlpn
This is a known deficiency. The problem is that the offending query is a scalar subquery. Scalar subqueries are executed during the query analysis phase, not during the execution phase. Therefore features like query cancellation and progress bars do not work.
2017-12-04T15:05:04
2017-12-04T15:05:04
{}
CONTRIBUTOR
ClickHouse
ClickHouse
350,273,386
1,599
antonio-antuan
Буду знать, спасибо.
2017-12-08T14:17:21
2017-12-08T14:17:21
{}
NONE
ClickHouse
ClickHouse
350,379,608
1,602
alexey-milovidov
А не смотрел, почему на стороне сервера не работает обычный max_network_bandwidth при отправке данных на клиент?
2017-12-08T21:43:29
2017-12-08T21:43:29
{}
MEMBER
ClickHouse
ClickHouse
350,420,560
1,602
kshvakov
Там `Throttler` выставляется на уровне `ClusterProxy` и ограничение работает только для работы внутри КХ, на сокет с клиентом только таймаут ставится. В общем случае это работает когда запрос простой, но если мы хотим лимитировать именно получение данных клиентом оно может работать не корректно в ряде случаев, наприм...
2017-12-09T03:28:40
2017-12-09T03:28:40
{}
CONTRIBUTOR
ClickHouse
ClickHouse
350,428,843
1,601
filimonov
If your name column contains not only 'X','Y','Z' then in first select all other values (out of that range) will be converted to 'Z' by ELSE clause in CASE WHEN .... May be that is the reason?
2017-12-09T06:56:26
2017-12-09T06:56:26
{}
CONTRIBUTOR
ClickHouse
ClickHouse
350,429,129
1,607
filimonov
Разумеется :) Так же как в нормальном SQL - `ALTER TABLE x ADD COLUMN ...` См. https://clickhouse.yandex/docs/ru/query_language/queries.html#id4 9 дек. 2017 г. 3:09 AM пользователь "Oleg Bogdanov" < notifications@github.com> написал: > Подскажите > возможно ли добавлять поля к таблице, когда таблица уже был...
2017-12-09T07:04:08
2017-12-09T07:04:08
{}
CONTRIBUTOR
ClickHouse
ClickHouse
349,338,964
1,562
filimonov
That sucks. Workarounds possible but they look really ugly. For that case you can get desired result with something like: `column_y % if(column_x = 0, if(column_y = 0, 1, column_y), column_x)` ``` SELECT if((number % 4) = 0, 0, 12) AS column_x, number AS column_y, column_y % if(column_x = 0, i...
2017-12-05T15:29:20
2017-12-05T15:29:20
{ "+1": 3 }
CONTRIBUTOR
ClickHouse
ClickHouse
349,351,685
1,562
StarWix
Thanks for the idea. Query can be more simple: ```sql SELECT if((number % 4) = 0, 0, 12) AS column_x, number AS column_y, column_y % if(column_x = 0, 1, column_x) AS column_y_modulo_column_x FROM system.numbers WHERE number < 60 LIMIT 60 ```
2017-12-05T16:06:17
2017-12-05T16:06:17
{ "+1": 2 }
NONE
ClickHouse
ClickHouse
350,338,372
1,591
ztlpn
There is no other way. The reason is that ALTERs do not go through ordinary replication queue - the replicas just see the latest version of table metadata and try to convert their version to match it. So in your case the stopped replica didn't detect any change. You can re-sync partition-by-partition though with ALT...
2017-12-08T18:36:16
2017-12-08T18:36:16
{}
CONTRIBUTOR
ClickHouse
ClickHouse
350,430,213
1,604
tacyuuhon
I already know. Thank you very much.
2017-12-09T07:31:43
2017-12-09T07:31:43
{}
NONE
ClickHouse
ClickHouse
350,944,240
1,458
bobrik
@velom, can you update the docs to clarify that 3.5 is not required?
2017-12-12T04:47:04
2017-12-12T04:47:04
{}
CONTRIBUTOR
ClickHouse
ClickHouse
351,046,922
1,211
tobad357
Thanks thats perfect for us
2017-12-12T13:11:01
2017-12-12T13:11:01
{}
NONE
ClickHouse
ClickHouse
349,063,423
1,588
alexey-milovidov
Thanks, @filimonov!
2017-12-04T18:46:35
2017-12-04T18:46:35
{}
MEMBER
ClickHouse
ClickHouse
349,957,965
127
dyomina-nn
Is any updates or workarounds?
2017-12-07T12:50:38
2017-12-07T12:50:38
{}
NONE
ClickHouse
ClickHouse
350,188,005
1,567
filimonov
I've just checked how clickhouse compress such a strings - in my case i have a `message` field in one of our tables. That is a string field, which currently have 25 different values of length 9-25 chars. In about 50% rows that field is empty. Also the same table have enum field `status` which can have 3 different va...
2017-12-08T06:53:46
2017-12-08T06:57:40
{}
CONTRIBUTOR
ClickHouse
ClickHouse
350,333,628
1,604
ztlpn
Formats support is independent of the Kafka engine. The list is available at https://clickhouse.yandex/docs/en/formats/index.html. Also recently support for the Cap'n proto format was added and there is work under way to support parquet and protobuf. There are no definite plans for the support of thrift or msgpack alth...
2017-12-08T18:15:29
2017-12-08T18:15:29
{}
CONTRIBUTOR
ClickHouse
ClickHouse
350,554,303
1,563
filimonov
> on v1.1.54304. ... please check on master Right now I have no possibility to test it on master (don't want to install lot of build stuff on old laptop), so please check if you can. Hope that most of cases will already work on master.
2017-12-10T15:01:41
2017-12-10T15:02:41
{}
CONTRIBUTOR
ClickHouse
ClickHouse
350,861,983
1,378
alexey-milovidov
You can try current master.
2017-12-11T21:21:13
2017-12-11T21:21:13
{}
MEMBER
ClickHouse
ClickHouse
350,862,128
1,373
alexey-milovidov
Fixed in master.
2017-12-11T21:21:47
2017-12-11T21:21:47
{}
MEMBER
ClickHouse
ClickHouse
350,862,700
1,370
alexey-milovidov
Of course. Qualified names should not be confused with simple aliases.
2017-12-11T21:24:00
2017-12-11T21:24:00
{}
MEMBER
ClickHouse
ClickHouse
351,061,836
1,028
silviucpp
@alexey-milovidov this was fixed in the last master right ? I guess you can close it
2017-12-12T14:10:09
2017-12-12T14:10:09
{}
CONTRIBUTOR
ClickHouse
ClickHouse
351,428,495
1,598
silviucpp
I just added some steps to replicate the crash for groupUniqArray that's taking place after this commits ```sql CREATE DATABASE db1; CREATE TABLE db1.tx_src ( partition Date, timestamp DateTime, user_id Nullable(UInt32), counter_id Nullable(UInt32), description Nullable(String) ) ENGINE =...
2017-12-13T15:39:07
2017-12-13T15:39:30
{}
CONTRIBUTOR
ClickHouse
ClickHouse
348,544,826
1,571
alexey-milovidov
Tnx!
2017-12-01T16:43:22
2017-12-01T16:43:22
{}
MEMBER
ClickHouse
ClickHouse
348,545,835
1,575
alexey-milovidov
This patch is critical (functional tests fail under UBSan and ASan).
2017-12-01T16:46:56
2017-12-01T16:47:56
{}
MEMBER
ClickHouse
ClickHouse
348,992,866
1,579
ztlpn
Yandex geo dictionaries are proprietary and are not distributed freely. There have been reports of people successfully using MaxMind geobase with ClickHouse so you could try that. But most likely you would have to prepare the `regions_hierarchy.txt` file yourself.
2017-12-04T15:19:09
2017-12-04T15:19:27
{}
CONTRIBUTOR
ClickHouse
ClickHouse
349,286,831
1,562
StarWix
In common case I would like execute query like this: `SELECT column_x == 0 ? 0 : modulo(column_y, column_x) FROM table`
2017-12-05T12:14:06
2017-12-05T12:14:57
{}
NONE
ClickHouse
ClickHouse
349,976,251
127
dyomina-nn
unfortunately my MATERIALIZED VIEWs dont work if they have subqueries or JOIN, but with AS keyword they work normal. And for me using long and ugly selects isnt very good, there are too many big and long selects.... At that VIEW works fine, analogic MATERIALIZED VIEW doesnt work...
2017-12-07T14:05:58
2017-12-07T15:03:09
{}
NONE
ClickHouse
ClickHouse
350,213,270
1,598
silviucpp
Another issue involving nullable is that ALTER is not working properly. you can see https://github.com/yandex/ClickHouse/issues/1322 for simple steps to replicate.
2017-12-08T09:13:31
2017-12-08T09:13:31
{}
CONTRIBUTOR
ClickHouse
ClickHouse
350,333,448
1,567
valyala
> Sorry, but it sounds like reinventing the wheel. If you don't know the properties of the strings - cardinality, entropy etc. - the best universal way of making strings smaller is called "compression" :) Most of compression algorithms use approach similar to what you descibe. ClickHouse uses LZ4 (or optionally zstd) w...
2017-12-08T18:14:42
2017-12-08T18:14:42
{}
NONE
ClickHouse
ClickHouse
350,867,494
1,190
alexey-milovidov
We have this task in internal tracker.
2017-12-11T21:41:34
2017-12-11T21:41:34
{}
MEMBER
ClickHouse
ClickHouse
348,546,513
1,575
alexey-milovidov
Also please change `Poco::DirectoryIterator` and `boost::filesystem` to `std::filesystem` everywhere, (probably without `boost::filesystem::recursive_directory_iterator` if it is still not implemented in libstdc++, libc++).
2017-12-01T16:49:32
2017-12-01T16:51:28
{}
MEMBER
ClickHouse
ClickHouse
350,032,939
1,567
valyala
See a simple program for evaluation potential win when converting string column blocks into dictionary-indexed int column blocks - https://play.golang.org/p/zLK9z6FMOK . It generates `count` strings containing `uniqCount` unique values. The average string length is `avgLen`. Then it constructs three blocks from thes...
2017-12-07T17:10:19
2017-12-07T17:10:19
{}
NONE
ClickHouse
ClickHouse
350,250,094
1,600
PhantomPhreak
Would like to have this feature to make dumps from Clickhouse
2017-12-08T12:18:14
2017-12-08T12:18:14
{}
NONE
ClickHouse
ClickHouse
350,861,504
1,393
alexey-milovidov
Visitors are completely not implemented for UInt128. That's ridiculous.
2017-12-11T21:19:21
2017-12-11T21:19:21
{}
MEMBER
ClickHouse
ClickHouse
350,866,430
1,211
alexey-milovidov
BTW, it uses HALF_EVEN (banker's rounding) by default.
2017-12-11T21:37:33
2017-12-11T21:37:33
{}
MEMBER
ClickHouse
ClickHouse
350,943,559
1,474
vavrusa
This is not a libpoco or ClickHouse bug, it's a bug with gcc-7 on High Sierra, see https://github.com/open-mpi/ompi/issues/4373 and https://www.mail-archive.com/gcc-bugs@gcc.gnu.org/msg544669.html
2017-12-12T04:41:22
2017-12-12T04:41:22
{}
CONTRIBUTOR
ClickHouse
ClickHouse
349,036,993
1,576
filimonov
But if it can be killed by timeout, can't the same approach be used for regular kill?
2017-12-04T17:26:39
2017-12-04T17:26:39
{}
CONTRIBUTOR
ClickHouse
ClickHouse
349,857,391
1,488
flowbehappy
@alexey-milovidov Hi alexey, I installed llvm 5.0 with your script today on my mac. But when I tried to compile ClickHouse with it errors come: ``` ClickHouse/dbms/src/Common/StringUtils.h:3:10: fatal error: 'string' file not found #include <string> ^~~~~~~~ 1 error generated. ``` Looks like it c...
2017-12-07T04:09:26
2017-12-07T04:10:23
{}
CONTRIBUTOR
ClickHouse
ClickHouse
350,870,243
1,622
filimonov
Check that https://github.com/Altinity/clickhouse-rpm probably they solved that by building openssl in newer version
2017-12-11T21:51:52
2017-12-12T09:56:54
{}
CONTRIBUTOR
ClickHouse
ClickHouse
351,159,855
1,626
silviucpp
@alexey-milovidov this issue can be replicated also on master
2017-12-12T19:13:05
2017-12-12T19:13:05
{}
CONTRIBUTOR
ClickHouse
ClickHouse
348,994,721
1,576
filimonov
That's wierd. Any plans to change that?
2017-12-04T15:24:38
2017-12-04T15:24:38
{}
CONTRIBUTOR
ClickHouse
ClickHouse
349,095,459
1,551
alexey-milovidov
Now it's time to remove "lazy" calls to `readPrefix` inside `readImpl`...
2017-12-04T20:30:37
2017-12-04T20:30:37
{}
MEMBER
ClickHouse
ClickHouse
349,309,696
1,589
filimonov
Very unreadable response, but generally you're trying to insert Array data inside string field. Compare ` "proxied": []` in json vs `proxied String` in schema.
2017-12-05T13:51:27
2017-12-05T13:52:15
{ "+1": 3, "hooray": 2 }
CONTRIBUTOR
ClickHouse
ClickHouse
349,929,441
1,254
Nina07
I got the same error for my CSV file. Tried checking from online tool for CSV formatting issue, [CSV Lint](http://csvlint.io/) . I got this error while I was trying to format my CSV to add single quotes to string data using LibreOffice Calc ( Ubuntu ) and the reason I got was ; **Structural problem: Non-standard L...
2017-12-07T10:39:20
2017-12-07T10:39:20
{}
NONE
ClickHouse
ClickHouse
350,381,108
1,575
alexey-milovidov
What about patch for boost::filesystem?
2017-12-08T21:50:54
2017-12-08T21:50:54
{}
MEMBER
ClickHouse
ClickHouse
349,091,645
1,551
alexey-milovidov
I will add at least one clarifying comment...
2017-12-04T20:16:34
2017-12-04T20:16:34
{}
MEMBER
ClickHouse
ClickHouse
349,524,560
1,496
alexey-milovidov
I expect it will be available in next stable release (you will see it in changelog). There is also minor chance that the next release will be "bugfix only" - then you will see it in one release after.
2017-12-06T03:53:41
2017-12-06T03:53:41
{}
MEMBER
ClickHouse
ClickHouse
350,202,136
1,598
silviucpp
Hello, is this suppose to fix https://github.com/yandex/ClickHouse/issues/1028 ?
2017-12-08T08:20:20
2017-12-08T08:20:20
{}
CONTRIBUTOR
ClickHouse
ClickHouse
350,401,668
1,601
dmitsolodkiy
It is understood it should work this way, but that was not the question. Perhaps, I formulated it in rather a strange way, sorry) The problem is that the result of 1 and 2 query (i mean, count value) for Z should be the same anyway (you can clearly see it if you create some mickey-mouse dataset or... if you just perfor...
2017-12-08T23:53:37
2017-12-08T23:58:30
{}
NONE
ClickHouse
ClickHouse
350,430,159
1,607
wormen
спасибо ... видимо не внимательно смотрел ...
2017-12-09T07:30:22
2017-12-09T07:30:22
{}
NONE
ClickHouse
ClickHouse
351,024,960
1,543
KochetovNicolai
Hi, I've published a short [tutorial](https://github.com/yandex/clickhouse-presentations/blob/master/tutorials/catboost_with_clickhouse_ru.md), which shows how to use trained CatBoost models in ClickHouse. It's in Russian, but I'll translate it soon. Short retelling: 1. Trained model from [Amazon Employee Access Ch...
2017-12-12T11:30:53
2017-12-12T11:30:53
{ "+1": 1 }
MEMBER
ClickHouse
ClickHouse
350,217,791
1,598
silviucpp
Ah ok ! great !
2017-12-08T09:34:55
2017-12-08T09:34:55
{}
CONTRIBUTOR
ClickHouse
ClickHouse
350,553,754
1,563
silviucpp
@filimonov this functions are crashing on master as well ? I'm asking because master have major changes for Nullable
2017-12-10T14:53:46
2017-12-10T14:53:46
{}
CONTRIBUTOR
ClickHouse
ClickHouse
350,864,197
1,275
alexey-milovidov
We have added support for multidimensional arrays and combinations of arrays and tuples in tables. It is available in latest master (still yet to be released).
2017-12-11T21:29:45
2017-12-11T21:29:45
{}
MEMBER
ClickHouse
ClickHouse
350,865,993
1,456
alexey-milovidov
This fixes #1243
2017-12-11T21:35:53
2017-12-11T21:35:53
{}
MEMBER
ClickHouse
ClickHouse
350,868,884
1,036
alexey-milovidov
Now it is much better: ``` Received exception from server (version 1.1.54325): Code: 386. DB::Exception: Received from localhost:9000, 127.0.0.1. DB::Exception: There is no common type for types Float64, Int64 because some of them are integers and some are floating point, but there is no floating point type, that ...
2017-12-11T21:46:47
2017-12-11T21:46:47
{}
MEMBER
ClickHouse
ClickHouse
351,275,687
1,614
lppsuixn
just try this [https://github.com/jaykelin/clickhouse-hdfs-loader](url)
2017-12-13T04:01:21
2017-12-13T04:01:46
{}
NONE
ClickHouse
ClickHouse
351,428,220
1,028
silviucpp
Steps to replicate : ```sql CREATE DATABASE db1; CREATE TABLE db1.tx_src ( partition Date, timestamp DateTime, user_id Nullable(UInt32), counter_id Nullable(UInt32), description Nullable(String) ) ENGINE = ReplicatedMergeTree('/clickhouse/tables/{shard}/db1/tx_src', '{replica}', partition...
2017-12-13T15:38:11
2017-12-13T15:38:11
{}
CONTRIBUTOR
ClickHouse
ClickHouse
348,975,532
1,582
ztlpn
Hi @badvir ! You can filter only the newest records by adding a time predicate to the WHERE clause (e.g. `time >= now() - 3600` will filter the latest hour of records and then ORDER BY if you need strict chronological ordering). Do you store data in the tables of MergeTree family? I assume yes, because you mentio...
2017-12-04T14:21:50
2017-12-04T14:21:50
{}
CONTRIBUTOR
ClickHouse
ClickHouse
349,034,341
1,576
ztlpn
The reason behind the weirdness is that it allows using the index with the results of a scalar subquery. E.g. if you write `WHERE index_column IN (<scalar subquery>)`, the index will be used. Note that non-scalar subqueries are executed during the main execution phase and for the query `WHERE index_column IN (<a query ...
2017-12-04T17:19:20
2017-12-04T17:19:20
{}
CONTRIBUTOR
ClickHouse
ClickHouse
350,060,602
1,567
filimonov
Sorry, but it sounds like reinventing the wheel. If you don't know the properties of the strings - cardinality, entropy etc. - the best *universal* way of making strings smaller is called "compression" :) Most of compression algorithms use approach similar to what you descibe. ClickHouse uses LZ4 (or optionally zstd)...
2017-12-07T18:54:38
2017-12-08T06:59:46
{}
CONTRIBUTOR
ClickHouse
ClickHouse
350,429,645
1,601
dmitsolodkiy
Ah, yes, forgot, that filtering with WHERE will also be applied after. Ok, thank you and sorry for taking so much of your time!
2017-12-09T07:17:56
2017-12-09T07:17:56
{}
NONE
ClickHouse
ClickHouse
350,861,770
1,388
alexey-milovidov
Fixed a long time ago.
2017-12-11T21:20:20
2017-12-11T21:20:20
{}
MEMBER
ClickHouse
ClickHouse
351,252,558
1,625
tobad357
@vavrusa would that also fix the issue where the last messages never get flushed? Because if I understand your change correctly it will fix the issue where a shutdown causes some messages to get lost but I'm not sure it fixes messages that end up never being flushed?
2017-12-13T01:33:32
2017-12-13T01:33:32
{}
NONE
ClickHouse
ClickHouse
348,993,838
1,582
filimonov
Due to the fact that last inserted data is most probably was not merged yet, the selects with partitioning key go fast. In my case - for realtime data from console i usually do something like ``` SELECT * FROM table_name WHERE part_key = today() and timestamp > now() - 600 ORDER BY timestamp DESC LIMIT...
2017-12-04T15:22:03
2017-12-04T16:00:12
{ "+1": 3 }
CONTRIBUTOR
ClickHouse
ClickHouse
350,205,605
1,598
alexey-milovidov
This fixes tons of another issues, but #1028 is still not fixed. Maybe I'll look at it at the same time.
2017-12-08T08:37:40
2017-12-08T08:37:40
{}
MEMBER
ClickHouse
ClickHouse
350,280,655
1,601
filimonov
Your `WHERE name` in first case works on the alias created `(CASE ... ) as name`, and in second case - on table column `name`. Same for group by. Just remember one simple trick when working with clickhouse - never create aliases with names which already exist in table, or used in that select. Otherwise it's not alwa...
2017-12-08T14:47:01
2017-12-08T14:48:30
{}
CONTRIBUTOR
ClickHouse
ClickHouse
351,148,195
1,625
vavrusa
It's batching 16 messages at a time currently. I'm reworking this to track offsets externally and commit only when it's flushed.
2017-12-12T18:42:44
2017-12-12T18:42:44
{}
CONTRIBUTOR
ClickHouse
ClickHouse
349,047,590
1,588
filimonov
1. Supported regexp syntax: https://github.com/google/re2/wiki/Syntax No lookaheads as you see. :( 2. Documented: > If the regex doesn’t contain subpatterns, it takes the fragment that matches the entire regex. Otherwise, it takes the fragment that matches the first subpattern. In that case regexp CONTAINs 2...
2017-12-04T17:57:55
2017-12-04T17:59:52
{}
CONTRIBUTOR
ClickHouse
ClickHouse
349,328,868
1,589
Boolman
Yes, thanks! That was the case.
2017-12-05T14:57:41
2017-12-05T14:57:41
{}
NONE
ClickHouse
ClickHouse
349,610,599
1,464
AlexMAS
@Slach Yes, I know. I mentioned your answer in my question above but it is not convenient. In my case I solved the problem by using the [`named volume`](https://docs.docker.com/compose/compose-file/#volume-configuration-reference). Maybe it more appropriate and right way when we talk about Docker. Nevertheless I expect...
2017-12-06T11:22:23
2017-12-06T11:22:23
{}
CONTRIBUTOR
ClickHouse
ClickHouse
349,945,022
1,254
filimonov
"Expected end of line" usually means that Clickhouse expects less columns that appears in CSV file. So for example your table / or your input statement expect 3 columns, but your CSV have 4.
2017-12-07T11:49:19
2017-12-07T11:49:51
{ "+1": 6, "-1": 1 }
CONTRIBUTOR
ClickHouse
ClickHouse
349,972,503
127
filimonov
Generally VIEWS do work nice. My workaround is to avoid views, and generate long and ugly selects.
2017-12-07T13:52:00
2017-12-07T13:52:00
{}
CONTRIBUTOR
ClickHouse
ClickHouse
350,215,075
1,598
alexey-milovidov
That's more complicated. The reason of #1322 is the same as ``` SELECT CAST(NULL AS Nullable(String)) AS x, CAST(x AS Nullable(UInt8)) ... Cannot parse UInt8 from String, because value is too short. ``` and ``` SELECT CAST(NULL AS Nullable(UInt8)) AS x, intDiv(10, x) ... Division by zero. ``` and related t...
2017-12-08T09:22:08
2017-12-08T09:22:32
{}
MEMBER
ClickHouse
ClickHouse
350,487,078
1,602
alexey-milovidov
Ok. Ещё одна проблема - так как в клиенте можно задать все настройки, передающиеся на сервер (включая `max_network_bandwidth`), то название для параметра `network_bandwidth` не является адекватным - слишком легко перепутать. Можно назвать `max_client_network_bandwidth`.
2017-12-09T16:38:34
2017-12-09T16:38:34
{}
MEMBER
ClickHouse
ClickHouse
350,866,059
1,243
alexey-milovidov
Fixed in master.
2017-12-11T21:36:05
2017-12-11T21:36:05
{}
MEMBER
ClickHouse
ClickHouse
351,257,012
1,622
kmatt
Removing openssl-static and installing OpenSSL v1.1.0g from source allows successful builds on CentOS 6.9 and 7.4
2017-12-13T01:57:37
2017-12-13T01:57:37
{}
NONE
ClickHouse
ClickHouse
351,351,605
1,630
silviucpp
Problem with building with gcc-7 can be solved by brew remove gcc-6 if it's installed as well. for some reason this is the problem. Clang still not sure how can be fixed
2017-12-13T10:35:26
2017-12-13T10:35:26
{}
CONTRIBUTOR
ClickHouse
ClickHouse
348,954,519
1,402
leoluk
> an additional ENGINE parameter with a file containing librdkafka settings That's what I would do, just put all broker settings in a XML configuration file just as with external dictionaries, and then reference them in the ENGINE config. That will make them easy to re-use and modify.
2017-12-04T12:55:53
2017-12-04T12:55:53
{}
CONTRIBUTOR
ClickHouse
ClickHouse
349,070,688
1,580
ztlpn
DateTime operations support a range of years from 1970 to 2105. When results fall outside this range, the result is unspecified.
2017-12-04T19:07:00
2017-12-04T19:07:00
{}
CONTRIBUTOR
ClickHouse
ClickHouse
349,519,254
1,496
lppsuixn
@ztlpn hi,if this will be merged in next stable?
2017-12-06T03:17:58
2017-12-06T03:17:58
{}
NONE
ClickHouse
ClickHouse
349,571,541
51
osbertngok
May I know if this option is implemented?
2017-12-06T08:39:01
2017-12-06T08:39:01
{}
NONE
ClickHouse
ClickHouse
350,212,177
1,598
alexey-milovidov
https://github.com/yandex/ClickHouse/commit/e2cd0272a48b9bfa9b2639bc4421283075036dc9
2017-12-08T09:08:37
2017-12-08T09:08:37
{}
MEMBER
ClickHouse
ClickHouse
350,270,702
1,599
ztlpn
Да, неудачно получилось. Партицию всегда можно дропнуть по ID (это первая часть имени куска). Для куска `19700101_0_0_0` это `19700101`: ``` ALTER TABLE test_sh3_rep2.test DROP PARTITION ID '19700101' ```
2017-12-08T14:04:49
2017-12-08T14:04:49
{}
CONTRIBUTOR
ClickHouse
ClickHouse
350,380,699
1,578
alexey-milovidov
Ok, but the second part is implemented not enough carefully.
2017-12-08T21:48:59
2017-12-08T21:48:59
{}
MEMBER
ClickHouse
ClickHouse
351,073,630
1,028
silviucpp
Seems even worst now. It's crashing the server on last master: ``` 2017.12.12 14:46:26.537735 [ 37 ] <Error> BaseDaemon: ######################################## 2017.12.12 14:46:26.537822 [ 37 ] <Error> BaseDaemon: (from thread 36) Received signal Segmentation fault (11). 2017.12.12 14:46:26.537843 [ 37 ] <Error...
2017-12-12T14:51:07
2017-12-12T14:51:07
{}
CONTRIBUTOR
ClickHouse
ClickHouse
351,275,523
1,621
lppsuixn
you can try ReplicatedMergeTree table engine,because MergeTree is not able to guarantee data consistency on all replicas
2017-12-13T03:59:59
2017-12-13T03:59:59
{}
NONE
ClickHouse
ClickHouse
349,091,520
1,551
alexey-milovidov
Looks good!
2017-12-04T20:16:08
2017-12-04T20:16:08
{}
MEMBER
ClickHouse
ClickHouse
349,104,191
1,447
alexey-milovidov
One question remains.
2017-12-04T21:02:09
2017-12-04T21:02:09
{}
MEMBER
ClickHouse
ClickHouse
350,548,811
1,563
filimonov
List of function calls which produce segfaults on v1.1.54304. Mostly connected with NULL and Nullables, please check on master. ``` toNullable(CAST('\0' AS Nullable(String))) assumeNotNull(CAST('\0' AS Nullable(String))) tan(NULL) sin(NULL) lgamma(NULL) erfc(NULL) erf(NULL) sqrt(NULL) log10(NULL) tgamma(NU...
2017-12-10T13:34:32
2017-12-10T13:37:31
{ "+1": 1 }
CONTRIBUTOR
ClickHouse
ClickHouse
350,942,207
1,567
vavrusa
I think it would be perhaps easier and more general solution to implement support for zstd dictionaries. Merging phase could train dictionary for each column and part, which would be used for compression and decompression. See the API https://github.com/facebook/zstd/blob/dev/lib/dictBuilder/zdict.h#L54 The decompr...
2017-12-12T04:30:28
2017-12-12T04:30:45
{ "+1": 1 }
CONTRIBUTOR
ClickHouse
ClickHouse
351,515,515
1,628
alexey-milovidov
4d4087913
2017-12-13T20:41:05
2017-12-13T20:41:05
{}
MEMBER