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 | 358,282,662 | 1,608 | filimonov | > у нас используется эта БД
@wormen я догадываюсь, начало комментария было больше для @avsej - одного из разработчиков Couchbase, который видимо забрел сюда по ключевому слову Couchbase, и мы рады его тут видеть :) Поэтому постарался вкратце описать как Couchbase (может) использоваться в ClickHouse.
> в COUCHBAS... | 2018-01-17T11:55:01 | 2018-01-17T11:58:55 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 358,311,152 | 1,608 | filimonov | > по HTTP только КХ запрашивает данные, сами мы туда закинуть их не можем, если они изменились
Ну тут по аналогии с обычной жизнью - человек заглядывает в толковый словарь когда требуется, а не словарь будит человека среди ночи "эй, я такой новое слово узнал, слушай: жандобиться это согласно Далю означает заботиться... | 2018-01-17T13:54:52 | 2018-01-17T13:58:13 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 358,538,630 | 1,665 | cyfdecyf | @alexey-milovidov My application result differs using `readFloatTextFast` compared to `readFloatTextPrecise`, but only just a few places with a 1e-4 difference. The difference is much less compared to master branch. | 2018-01-18T05:03:33 | 2018-01-18T05:03:33 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 358,789,379 | 838 | alexey-milovidov | Our compression format is quite simple.
Look at `CompressedWriteBuffer` and `CompressedReadBufferBase` for reference.
In short, it is the following:
Compressed data is formed of blocks, one after the other. Blocks are completely independent.
Each block is:
```
/** The compressed block format is as follows:
... | 2018-01-18T21:35:58 | 2018-01-18T21:35:58 | {} | MEMBER |
ClickHouse | ClickHouse | 359,360,961 | 613 | mosquito | This not working anymore. The #1611 broke this. | 2018-01-22T08:58:31 | 2018-01-22T08:58:31 | {} | NONE |
ClickHouse | ClickHouse | 360,551,312 | 1,763 | zhang2014 | @alexey-milovidov please help review the changes | 2018-01-25T18:10:32 | 2018-01-25T18:10:32 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 358,091,082 | 1,665 | alexey-milovidov | Thank you!
It's still difficult to decide what is better. I'm really in doubt.
Probably I can make `readFloatTextFast` function to give exact result for all machine-generated inputs (inputs that was formatted with sprintf, double conversion, etc.), but with sacrificing 1..2 bits of precision for some unusual user... | 2018-01-16T20:17:13 | 2018-01-16T20:17:45 | {
"+1": 1
} | MEMBER |
ClickHouse | ClickHouse | 358,269,307 | 1,608 | filimonov | @avsej, с Kafka история немного другая. Kafka может используется как виртуальная таблица из которой можно читать (правда всего один раз) данные.
Вопрос @wormen - про подключение [внешних словарей](https://clickhouse.yandex/docs/ru/dicts/external_dicts.html) - это такая штука в Кликхаус, которую можно использовать к... | 2018-01-17T10:56:04 | 2018-01-17T11:07:15 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 358,274,838 | 1,608 | wormen | @filimonov
1) у нас используется эта БД
2) в COUCHBASE используются SQL запросы
про HTTP словарь, что-то даже не подумал, где по нему примеры можно глянуть? как его использовать | 2018-01-17T11:19:50 | 2018-01-17T11:19:50 | {} | NONE |
ClickHouse | ClickHouse | 358,281,163 | 1,608 | avsej | Да, у нас есть язык запросов N1QL, который очень похож на SQL, также в следующем релизе будет поддержка ещё одного диалекта SQL++, который позволяет делать запросы для аналитики более эффективно.
А примеры как сделать N1QL запрос на С++ с libcouchbase, можно посмотреть тут
https://github.com/couchbaselabs/devguide-... | 2018-01-17T11:48:30 | 2018-01-17T11:48:30 | {} | NONE |
ClickHouse | ClickHouse | 358,527,109 | 1,713 | shangshujie365 | May be ch should add some shardingkey-aware distributed plan | 2018-01-18T03:31:14 | 2018-01-18T03:31:14 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 358,737,727 | 1,786 | proller | https://github.com/yandex/ClickHouse/commit/90399b851d2c8c662a218de70e7394913e9ae7bc
| 2018-01-18T18:27:37 | 2018-01-18T18:27:37 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 359,559,216 | 1,802 | ben-motiv | Partially solved by taking advantage of the partitioning key support - thanks!
However this still is a problem - I've got a lot of dictionary lookups that I would like to group by. IE primary key includes "device" and a dictionary may have a field keyed to that as "name". I want to be able to group by "name" and hav... | 2018-01-22T20:47:11 | 2018-01-22T20:47:11 | {} | NONE |
ClickHouse | ClickHouse | 359,696,743 | 1,052 | shkrid | Bash implementation for CH backups https://github.com/jetbrains-infra/clickhouse-tasks/blob/master/backup/entrypoint.sh that I have found. | 2018-01-23T07:12:16 | 2018-01-23T07:12:16 | {
"+1": 1
} | NONE |
ClickHouse | ClickHouse | 359,957,444 | 1,815 | pavel-odintsov | Hello!
That's awesome! Also would be nice to show it in "DESCRIBE TABLE" because it's more script friendly.
Very useful thing for schema migration! | 2018-01-23T22:45:33 | 2018-01-23T22:45:33 | {} | NONE |
ClickHouse | ClickHouse | 358,704,355 | 1,782 | sanchezzzhak | yes, use
```select * from table where id > 10 limit 5 FORMAT JSON```
view keys `totals`, `rows_before_limit_at_least`
| 2018-01-18T16:38:50 | 2018-01-19T08:10:17 | {} | NONE |
ClickHouse | ClickHouse | 359,978,131 | 1,665 | alexey-milovidov | Can you give us one or a few rows for an example?
How to do it:
- find out a batch that you cannot insert;
- save it somewhere; check that the issue still reproduce, when you try to insert it again;
- leave just one row where the issue still reproduce. | 2018-01-24T00:23:28 | 2018-01-24T00:23:28 | {} | MEMBER |
ClickHouse | ClickHouse | 358,295,720 | 1,608 | avsej | То есть, если по какой-то причине, внешний источник падает, а передача данных не завершена, clickhouse будет выкачивать заново? | 2018-01-17T12:55:37 | 2018-01-17T12:55:37 | {} | NONE |
ClickHouse | ClickHouse | 358,305,728 | 1,608 | avsej | Понятно, в таком случае нет смысла использовать словари для Couchbase. А судя по описанию работы, Couchbase тоже можно было бы интегрировать как Kafka.
В обычном режиме, просто стримились бы данные, а с materialized view, можно использовать ReplacingMergeTree, чтобы получить то, что реально содержится в Couchbase (б... | 2018-01-17T13:34:34 | 2018-01-17T13:34:34 | {} | NONE |
ClickHouse | ClickHouse | 357,952,896 | 1,553 | ludv1x | Note that `CLEAR COLUMN IN PARTITION` currently has an inefficient implementation.
If there are active INSERTs or background merges in the same partition, then the replication is stuck while `CLEAR COLUMN IN PARTITION` is executing. | 2018-01-16T13:03:37 | 2018-01-17T21:27:23 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 359,310,801 | 1,803 | Slach | my mistake
i used golang maps for table definition and "create table" statement on different nodes had different order of table fields | 2018-01-22T02:47:43 | 2018-01-22T02:47:43 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 360,248,029 | 1,818 | akonyaev90 | I use zookeeper 3.4.5. May be this problem due with zookeeper version? | 2018-01-24T19:34:35 | 2018-01-24T19:34:35 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 360,511,681 | 1,254 | Jackge1979 | The columns in your CSV file contain "," and you can include the column with "". For example: "abc", "123", "a, b, c" | 2018-01-25T16:01:38 | 2018-01-25T16:01:38 | {
"+1": 1
} | NONE |
ClickHouse | ClickHouse | 357,859,202 | 1,665 | cyfdecyf | @alexey-milovidov I've tested with the latest commit in your branch, by changing `readFloatText` to use `readFloatTextPrecise`, my application are getting expected results.
My simple performance test reduced to only 0.8 seconds when using `readFloatTextFast`, just half of the origin float parsing implementation, `re... | 2018-01-16T05:41:04 | 2018-01-16T05:41:04 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 357,939,661 | 1,553 | ludv1x | > ALTER TABLE test.clear_column CLEAR COLUMN num IN PARTITION '201612';
>
> SELECT data_compressed_bytes, data_uncompressed_bytes FROM system.columns WHERE (database = 'test') AND (table = 'clear_column') AND (name = 'str');
>
> // expect: 28, 2 as before previous insert, currently have 56, 4
There is an error... | 2018-01-16T12:01:56 | 2018-01-16T12:01:56 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 358,257,949 | 1,608 | avsej | @wormen, может быть "любая" в смысле [ODBC](https://github.com/yandex/ClickHouse/blob/master/dbms/src/Storages/StorageODBC.h), которые поддерживаются [Poco::Data]( https://pocoproject.org/docs/Poco.Data.ODBC.html). Для Kafka они писали отдельный коннектор [StorageKafka.cpp](https://github.com/yandex/ClickHouse/blob/mas... | 2018-01-17T10:11:40 | 2018-01-17T10:11:40 | {} | NONE |
ClickHouse | ClickHouse | 358,308,212 | 1,608 | wormen | ... в принципе, я получил ответ, малость не тот который ожидал, но тоже сойдет ...
спасибо за информацию | 2018-01-17T13:43:55 | 2018-01-17T13:43:55 | {} | NONE |
ClickHouse | ClickHouse | 358,820,297 | 1,763 | alexey-milovidov | autotests | 2018-01-18T23:48:08 | 2018-01-18T23:48:08 | {} | MEMBER |
ClickHouse | ClickHouse | 359,421,082 | 1,632 | yurchenkosv | @kshvakov Да, спасибо, после переезда на 1.1.54318-stable, проблема решилась, clickhouse сейчас использует ~600мб и замечательно отдает память. | 2018-01-22T13:19:56 | 2018-01-22T13:19:56 | {} | NONE |
ClickHouse | ClickHouse | 360,256,698 | 1,820 | bobrik | This may be related to #1821, as it happened not long after. | 2018-01-24T20:04:24 | 2018-01-24T20:04:24 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 360,439,169 | 1,816 | KochetovNicolai | > if it matches the given regex (dd:dd:dd) it should be able to subset the string, or not?
No, it isn't true for `if` function. ClickHouse evaluates both branches of `if`. You can rewrite query like:
```sql
SELECT
if(match(duration, '\\d\\d:\\d\\d:\\d\\d'), duration, NULL) AS d,
(((toUInt8(concat('0'... | 2018-01-25T11:24:24 | 2018-01-25T11:24:24 | {} | MEMBER |
ClickHouse | ClickHouse | 360,447,693 | 1,817 | KochetovNicolai | Такой запрос действительно должен отрабатывать с константной памятью.
Единственное, есть подозрение на чтение из таблицы на основе словаря - в таком случае все ключи из словаря записываются в память (кроме cache словарей, для них берем только ключи из кеша), и если словарь большой, памяти может не хватить из-за размер... | 2018-01-25T12:05:05 | 2018-01-25T12:05:05 | {} | MEMBER |
ClickHouse | ClickHouse | 358,300,671 | 1,608 | filimonov | > Хотя вон Kafka у вас бесконечно в цикле читает данные ... Или это уже не dictionary?
Повторюсь, с Kafka - история совсем другая. Это не словарь, а "виртуальная" таблица.
Kafka - это очередь сообщений. Поэтому из неё по определению читают один раз, чтобы что-то с этими сообщениями сделать - переложить в перманен... | 2018-01-17T13:15:36 | 2018-01-17T13:26:51 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 358,976,432 | 167 | menvil | @alexey-milovidov Отличная новость. Давайте очень нужно. | 2018-01-19T14:12:04 | 2018-01-19T14:12:04 | {} | NONE |
ClickHouse | ClickHouse | 358,981,524 | 1,763 | alexey-milovidov | stateless.00195_shard_union_all_and_global_in
stateless.00506_global_in_union | 2018-01-19T14:31:49 | 2018-01-19T14:31:49 | {} | MEMBER |
ClickHouse | ClickHouse | 359,399,379 | 1,788 | KochetovNicolai | Probably, it's a problem with optimization: some necessary columns are removed from subquery. Workaround is:
```sql
SELECT
x,
one
FROM
(
SELECT
x,
1 AS one
FROM test
ALL INNER JOIN
(
SELECT
count(),
1 AS one
FROM t... | 2018-01-22T11:40:57 | 2018-01-22T11:40:57 | {} | MEMBER |
ClickHouse | ClickHouse | 359,842,061 | 1,805 | vavrusa | Hi @vicchen7, what’s your clickhouse version? Can you reproduce that on latest master if you’re running a version from package? The Kafka engine performance should be fast enough at least, depending on your expectations. I’m getting 100-200k rows/s per consumer on my laptop. | 2018-01-23T16:13:56 | 2018-01-23T16:13:56 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 359,985,820 | 1,665 | cyfdecyf | @miggi I use clickhouse-client to import CSV which works fine. Maybe the problem only shows up in jdbc related code. | 2018-01-24T01:05:14 | 2018-01-24T01:05:14 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 360,076,129 | 1,813 | mikhailov-jet | No timeouts, but there are several HTTPHandler: Code: 24, e.displayText() = DB::Exception: Cannot write to ostream
```
2018.01.22 17:32:01.451810 [ 37 ] <Error> executeQuery: Code: 24, e.displayText() = DB::Exception: Cannot write to ostream, e.what() = DB::Exception (from 127.0.0.1:36238) (in query: SELECT * FROM ... | 2018-01-24T09:45:48 | 2018-01-24T09:45:48 | {} | NONE |
ClickHouse | ClickHouse | 358,410,532 | 1,758 | kmatt | @ztlpn The server is only logging files processed to other nodes, the node I'm running ETL on is not mentioned:
```
[/var/log/clickhouse-server:#] awk -F'/' '/DirectoryMonitor/ { s[$7]++ } END { for (i in s) print i, s[i] }' clickhouse-server.log
default@xxx%2Exxx%2E7%2E123:9000 140764
default@xxx%2Exxx%2E7%2E122... | 2018-01-17T19:12:08 | 2018-01-17T19:12:08 | {} | NONE |
ClickHouse | ClickHouse | 358,411,801 | 1,758 | ztlpn | This is as expected - data that is intended to be inserted into the local node is inserted immediately, without first writing it to disk. | 2018-01-17T19:16:14 | 2018-01-17T19:16:14 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 358,472,713 | 1,784 | ben-motiv | Further testing - using ReplacingMergeTree on a single non-replicated machine I found the optimize table <tablename> works, but only if run a number of times in succession. Rapidly calling it ~10 times from tabix will see it successful.
Logs don't suggest any reason for this. Each time I run it I see it do some tidy... | 2018-01-17T22:35:59 | 2018-01-17T22:35:59 | {} | NONE |
ClickHouse | ClickHouse | 358,522,804 | 1,781 | Mcgeey | @meshpaul what do you mean? cancel? | 2018-01-18T03:01:11 | 2018-01-18T03:01:11 | {} | NONE |
ClickHouse | ClickHouse | 359,834,611 | 1,809 | ludv1x | Incorrect task number.
Reopened PR https://github.com/yandex/ClickHouse/pull/1814 | 2018-01-23T15:53:15 | 2018-01-23T15:53:15 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 360,190,829 | 1,816 | filimonov | Clickhouse alway calculate ALL the branches both good and bad, and only after that choose one.
Problem similar to #1562 #1768
There are plans to chabge that: see https://www.altinity.com/blog/2018/1/8/clickhouse-roadmap-2018 | 2018-01-24T16:27:46 | 2018-01-24T16:27:46 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 357,680,231 | 1,768 | volvoski | that's not the case. i use the length function for string type already.
Main goal is to select data from table_1 and insert it to table_2.
While inserting i want to transform field_2 (string) according to its length.
if length = 10 then CAST(toDate(field_2) AS DateTime)
if length = 19 then CAST(field_2 AS DateTime... | 2018-01-15T13:17:36 | 2018-01-15T13:17:36 | {} | NONE |
ClickHouse | ClickHouse | 357,702,461 | 1,751 | silviucpp | @sanchezzzhak Ohh nice :) I'll test this out ! | 2018-01-15T14:46:07 | 2018-01-15T14:46:07 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 357,723,232 | 1,761 | SaltTan | No, it does not | 2018-01-15T16:00:36 | 2018-01-15T16:00:36 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 357,818,042 | 1,758 | kmatt | Config: https://gist.github.com/kmatt/8e705cd388a9b841be015764b631ac11
> Some pointers: if you didn't restart ClickHouse, numbers in the names of temporary .bin files should be unique.
Yes, Clickhouse engine was restarted between loads, so the sequence number would have restarted.
> Another strange thing: 6350... | 2018-01-16T00:17:04 | 2018-01-16T00:32:17 | {} | NONE |
ClickHouse | ClickHouse | 358,296,930 | 1,608 | avsej | Хотя вон Kafka у вас бесконечно в цикле читает данные, пока не остановят.
https://github.com/yandex/ClickHouse/blob/master/dbms/src/Storages/StorageKafka.cpp#L396-L427
Мы могли бы и из Couchbase так данные получать. Или это уже не dictionary?
> Kafka история немного другая. Kafka может используется как виртуал... | 2018-01-17T13:00:41 | 2018-01-17T13:00:41 | {} | NONE |
ClickHouse | ClickHouse | 359,975,488 | 1,665 | miggi | Hi @cyfdecyf
**" DB::Exception: Cannot read floating point value: (at row 1)"** we're still getting the errors when inserting large "flat" column set. Actually we've broke our production after upgrade and related issue. How to overcome this problem ?
We use jdbc-clickouse batching and it's hard to figure out what'... | 2018-01-24T00:08:21 | 2018-01-24T00:08:21 | {} | NONE |
ClickHouse | ClickHouse | 360,156,809 | 1,818 | Yashin32 | Confirming this issue, had the same error after updating to 54337. Table engine is ReplicatedSummingMergeTree + Distributed. Downgrade to 54327 also helped to solve issue. Stack trace: https://pastebin.com/NpY3WdtC | 2018-01-24T14:43:20 | 2018-01-24T14:43:20 | {} | NONE |
ClickHouse | ClickHouse | 360,393,972 | 1,818 | misja | I'm on 3.4.8, don't really have the impression this is zookeeper related. | 2018-01-25T08:21:12 | 2018-01-25T08:21:12 | {} | NONE |
ClickHouse | ClickHouse | 357,958,351 | 1,769 | volvoski | hit the same bug;
field_1: Nullable(String)
table_1 count: 6 billion
```
select count() from table_1 where isNull(field_1);
``` | 2018-01-16T13:24:49 | 2018-01-16T13:24:49 | {} | NONE |
ClickHouse | ClickHouse | 358,284,763 | 1,608 | avsej | А есть в clickhouse такой режим, в котором мы можем уведомлять его об изменениях (удаление и добавление ключей) непрерывно? Что-то типа репликации. У нас есть специальный протокол, который я мог бы использовать, чтобы выгрузить ключи и необходимые данные для словаря в clickhouse и поддерживать в актуальном стостоянии. ... | 2018-01-17T12:05:00 | 2018-01-17T12:05:00 | {} | NONE |
ClickHouse | ClickHouse | 358,298,194 | 1,608 | avsej | libcouchbase пока ещё не умеет DCP (это протокол репликации, который я упоминал выше), но я работаю над прототипом, который в конечном итоге войдёт в libcouchbase | 2018-01-17T13:05:43 | 2018-01-17T13:05:43 | {} | NONE |
ClickHouse | ClickHouse | 358,652,645 | 1,758 | ztlpn | Ok, so apparently this is a duplicate of #1404 (that was fixed in a4636f86563482ae37a6d92dfa5ff1361458b4b3 and released in 1.1.54337).
The bug is triggered when there is a shard with the single replica which is local and `internal_replication` is true (just as in your case). Do `.bin` files reside directly in the `/... | 2018-01-18T13:53:50 | 2018-01-18T13:54:44 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 357,863,229 | 936 | ivikramsahu | ↘ Progress: 538.05 million rows, 4.84 GB (2.46 million rows/s., 22.15 MB/s.) Received exception from server:
Code: 240. DB::Exception: Received from localhost:9000, 127.0.0.1. DB::Exception: Allocator: Cannot mremap., errno: 12, strerror: Cannot allocate memory.
can anyone help me resolve this seems creepy..? | 2018-01-16T06:12:10 | 2018-01-16T06:12:10 | {} | NONE |
ClickHouse | ClickHouse | 358,307,440 | 1,608 | wormen | > В определенном смысле это уже реализовано, ведь HTTP работает "по TCP сокетам".
да, но у HTTP нет двустороннего обмена+по HTTP только КХ запрашивает данные, сами мы туда закинуть их не можем, если они изменились | 2018-01-17T13:41:03 | 2018-01-17T13:41:03 | {} | NONE |
ClickHouse | ClickHouse | 358,308,142 | 1,781 | meshpaul | Cancel
On Jan 17, 2018 07:24, "Mcgeey" <notifications@github.com> wrote:
> I used clickhouse's kafka engine, but there was an exception.
> When I send data to kafka not artificially add \ n, this error occurs.
> Code: 27. DB::Exception: Received from localhost:9000, 127.0.0.1.
> DB::Exception: Cannot parse input: exp... | 2018-01-17T13:43:39 | 2018-01-17T13:43:39 | {} | NONE |
ClickHouse | ClickHouse | 358,313,304 | 1,608 | avsej | ну это да, Couchbase не брокер сообщений, но его данные можно трансформировать в поток изменений, что может быть удобно для анализа. Тот же протокол используется для построения различных индексов в остальных сервисах самого Couchbase, да и внешних инструментов. В ту же Kafka можно писать всё что происходит в Couchbase.... | 2018-01-17T14:02:42 | 2018-01-17T14:02:42 | {} | NONE |
ClickHouse | ClickHouse | 359,243,380 | 1,197 | Imaskar | If your table is MergeTree(_cx, (_c5, _c6, _c10, _c11), 8192) and you query on _c6, the index probably won't be used at all. You don't iterate columns you need indexes on in the 2nd argument, you build one complex index where rows firs sorted on _c5, then on _c6 and so on. Try to add condition on _c5 in your query. | 2018-01-21T12:04:48 | 2018-01-21T12:04:48 | {
"+1": 1
} | NONE |
ClickHouse | ClickHouse | 357,688,933 | 1,768 | filimonov | Simpler test case:
```
SELECT
CASE
WHEN length(field_2) = 10 THEN cast(toDate(field_2) as DateTime)
WHEN length(field_2) = 19 THEN cast(field_2 as DateTime)
ELSE cast('0000-00-00 00:00:00' as DateTime)
END
FROM
( SELECT toString(now()) AS field_2
UNION ALL
SELECT toString(today())
)... | 2018-01-15T13:53:55 | 2018-01-15T13:53:55 | {
"+1": 1
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 357,943,936 | 1,553 | filimonov | Yes, it works correctly. Sorry.
Actually I had that problem when was clearing production data.
But i can't reproduce it now, may be i just did something wrong, don't remember now. | 2018-01-16T12:22:42 | 2018-01-16T12:22:42 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 358,431,196 | 1,758 | kmatt | Of 15,586,402,301 lines in the source files, 15,584,535,754 appear in a COUNT(*) in Clickhouse.
Close enough that I know that with 155GB per node on a 4 server cluster, the 200GB of remaining .bin files in the path for the Distributed table were largely ingested.
They just are not removed, and do not appear in t... | 2018-01-17T20:17:55 | 2018-01-17T20:17:55 | {} | NONE |
ClickHouse | ClickHouse | 358,839,575 | 1,763 | zhang2014 | Would you please let me see the details of the error? for example: "https://travis-ci.org/apache/incubator-servicecomb-saga/builds/329913893?utm_source=github_status&utm_medium=notification" | 2018-01-19T01:35:46 | 2018-01-19T01:35:46 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 359,241,195 | 1,567 | Imaskar | Separate dictionary for each block? Wouldn't it take twice the time?
What about auto-expanding enum? That would add new value automatically when encountered. mapd does that. | 2018-01-21T11:24:01 | 2018-01-21T11:24:01 | {} | NONE |
ClickHouse | ClickHouse | 358,008,730 | 1,553 | filimonov | Good to know.
But rather not a problem for my case, because i olny clean columns in really old partitions - to remove some unneeded debug details for old records. | 2018-01-16T15:55:47 | 2018-01-16T15:55:47 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 358,082,996 | 1,758 | kmatt | Since I'm running under CentOS, and not using the Altinity RPMs due to issues with OpenSSL version compatibility, is there anything that could be wrong with the way Clickhouse is started on these nodes using the init.d script above?
Source builds are not using any special configuration parameters:
git clone -... | 2018-01-16T19:48:51 | 2018-01-16T19:50:58 | {} | NONE |
ClickHouse | ClickHouse | 358,242,173 | 1,756 | hamsterready | Okay, it explains this:
```
SELECT toDate('1823-12-12')
┌─toDate(\'1823-12-12\')─┐
│ 0000-00-00 │
└────────────────────────┘
1 rows in set. Elapsed: 0.035 sec.
```
(which is not a big problem to me)
And it is already in the docs: https://clickhouse.yandex/docs/en/single/index.html#date... | 2018-01-17T09:11:05 | 2018-01-17T09:11:05 | {} | NONE |
ClickHouse | ClickHouse | 358,075,344 | 1,776 | skaurus | And I'm back.
First, let's create a `ReplicatedSummingMergeTree` table and a `Buffer` (or it will be painfully slow):
```
CREATE TABLE break_me (
aid UInt32,
bid UInt32,
cid UInt32,
event_date Date,
event_type Enum8('show' = 1, 'view' = 2, 'click' = 3),
event_revenue Float32
) ENGINE = Replicate... | 2018-01-16T19:23:42 | 2018-01-17T17:23:14 | {} | NONE |
ClickHouse | ClickHouse | 358,818,959 | 1,676 | alexey-milovidov | + two another places:
9f57a1f7a | 2018-01-18T23:40:42 | 2018-01-18T23:40:42 | {} | MEMBER |
ClickHouse | ClickHouse | 359,464,342 | 1,782 | chaitenyay | Thanks, it works | 2018-01-22T15:46:08 | 2018-01-22T15:46:08 | {} | NONE |
ClickHouse | ClickHouse | 360,076,873 | 1,813 | mikhailov-jet | Looks that this issue might be related to the issue #1730
Our queries have replace_running_query=1.
Going to check if removing replace_running_query helps. | 2018-01-24T09:48:29 | 2018-01-24T09:48:29 | {} | NONE |
ClickHouse | ClickHouse | 360,137,536 | 1,818 | misja | I can confirm this, had exactly the same issue when setting up a replica, downgrading from 1.1.54342 to 1.1.54327 solved this. | 2018-01-24T13:42:01 | 2018-01-24T13:42:01 | {} | NONE |
ClickHouse | ClickHouse | 358,296,941 | 1,608 | wormen | еще вопрос, примерно в ту же сторону, что и у @avsej
планируется ли добавить подгрузку данных словарей, по TCP сокетам? думаю так будет быстрее и и проще, для обновления новых/измененных данных в КХ | 2018-01-17T13:00:43 | 2018-01-17T13:00:43 | {} | NONE |
ClickHouse | ClickHouse | 358,314,216 | 1,608 | filimonov | > ну это да, Couchbase не брокер сообщений, но его данные можно трансформировать в поток изменений, что может быть удобно для анализа. Тот же протокол используется для построения различных индексов в остальных сервисах самого Couchbase, да и внешних инструментов.
Интересно. Не знал, почитаю.
В общем случая для ст... | 2018-01-17T14:06:00 | 2018-01-17T14:06:00 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 358,371,287 | 1,665 | cyfdecyf | @alexey-milovidov Thanks for your explanation. Great points and I agree differ in 1 bit won't be a problem for most applications. I'll check my application results with `readFloatTextFast` tomorrow.
If you choose the faster way by default, just make sure to mention the problem in document, and better to give user an... | 2018-01-17T17:01:47 | 2018-01-17T17:01:47 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 358,591,737 | 1,164 | StarWix | Query doesn't work
```sql
SELECT
AccountId
FROM
(
SELECT
AccountId,
Currency
FROM Accounts
LIMIT 2 BY Currency
)
```
I understood that all columns from `limit by` sections should be used in outer query.
Workaround:
```sql
SELECT
AccountId,
ignore(Curr... | 2018-01-18T09:41:28 | 2018-01-18T09:49:05 | {} | NONE |
ClickHouse | ClickHouse | 358,819,661 | 1,676 | alexey-milovidov | BTW, TEMPORARY tables cannot be renamed at all. | 2018-01-18T23:44:46 | 2018-01-18T23:44:46 | {} | MEMBER |
ClickHouse | ClickHouse | 358,948,587 | 1,771 | tobad357 | @dvahram great!
issue #1219 can be closed once this is merged | 2018-01-19T12:09:58 | 2018-01-19T12:09:58 | {} | NONE |
ClickHouse | ClickHouse | 359,122,446 | 1,800 | kevinnlei | closing, jsoneach row doesnt support complex types. | 2018-01-19T23:57:36 | 2018-01-19T23:57:36 | {} | NONE |
ClickHouse | ClickHouse | 357,683,908 | 1,768 | filimonov | Aha, I see. You've used field_2 in two meanings so I misunderstand the problem.
In that case the problem is in that Clickhouse alway calculate ALL the branches both good and bad, and only after that choose one.
Problem similar to https://github.com/yandex/ClickHouse/issues/1562
You can use (ugly) workaround
... | 2018-01-15T13:33:07 | 2018-01-15T13:35:37 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 357,858,002 | 1,675 | lamberken | I test the config last week, yes, it's ok | 2018-01-16T05:31:18 | 2018-01-16T05:31:18 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 358,478,832 | 1,784 | ben-motiv | Further - Trying again on a ReplicatedReplacingMergeTree across two servers. Optimize returns OK on one machine, but doesn't do anything, and on the 2nd machine it complains that the user password is not correct (use the same users.xml, and all DDL 'on cluster' statements work fine, and the data has replicated, so not ... | 2018-01-17T23:01:54 | 2018-01-17T23:01:54 | {} | NONE |
ClickHouse | ClickHouse | 358,516,519 | 1,784 | ben-motiv | Temporary solution:
Select
partition,
partCount
from(
SELECT
partition,
count(*) as partCount
FROM
system.parts
WHERE
active
AND (table = 'TABLE')
group by
partition
)
where
partCount > 1
And run optimize table <TABLE> partition <PARTITION>... | 2018-01-18T02:20:44 | 2018-01-18T02:20:44 | {} | NONE |
ClickHouse | ClickHouse | 358,819,281 | 1,676 | alexey-milovidov | TEMPORARY table cannot be renamed from readonly user? | 2018-01-18T23:42:33 | 2018-01-18T23:42:33 | {} | MEMBER |
ClickHouse | ClickHouse | 358,948,186 | 1,690 | tobad357 | Tested on a new build and seems to work as expected now
No dropped messages | 2018-01-19T12:07:58 | 2018-01-19T12:07:58 | {} | NONE |
ClickHouse | ClickHouse | 359,866,064 | 1,813 | proller | can you grep log, last 60 seconds before crash, is here some "timeout" messages or something suspicious? | 2018-01-23T17:25:08 | 2018-01-23T17:25:08 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 357,696,059 | 1,751 | sanchezzzhak | i did export table 45GB or 780 billion rows (16 columns). make dump < 10 min | 2018-01-15T14:22:27 | 2018-01-15T14:23:11 | {
"+1": 5
} | NONE |
ClickHouse | ClickHouse | 357,705,975 | 1,758 | ztlpn | I couldn't reproduce it. Can you try to produce a minimal example that I can test? Also, please share your cluster config.
Some pointers: if you didn't restart ClickHouse, numbers in the names of temporary .bin files should be unique. Thus it is not possible that two data loads produce files with the same name. Mayb... | 2018-01-15T14:58:52 | 2018-01-15T14:58:52 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 357,940,018 | 1,732 | f1yegor | generally speaking yes, in Rainbird we are talking about pre-aggregated data, on the other hand, Clickhouse allows to work with raw data. | 2018-01-16T12:03:26 | 2018-01-16T12:03:26 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 358,281,128 | 1,776 | skaurus | Tested `ReplicatedMergeTree` with the same script. Works fine. | 2018-01-17T11:48:23 | 2018-01-17T11:48:23 | {} | NONE |
ClickHouse | ClickHouse | 358,293,996 | 1,608 | filimonov | > А есть в clickhouse такой режим, в котором мы можем уведомлять его об изменениях (удаление и добавление ключей) непрерывно? Что-то типа репликации.
Нет, в настоящее время нет. Полагаю, отчасти потому, что используемые до сих пор источники данных для внешних словарей не позволяли осуществлять такой мониторинг. Ну ... | 2018-01-17T12:48:04 | 2018-01-17T12:48:04 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 358,310,969 | 1,608 | avsej | ну так clickhouse не инициирует обновление внешних словарей и даже кафка в режиме только чтения, я прав, @filimonov? | 2018-01-17T13:54:14 | 2018-01-17T13:54:14 | {} | NONE |
ClickHouse | ClickHouse | 359,812,759 | 1,813 | mikhailov-jet | Error appears after loading approximately 100 000 rows in 100 batches of 1000 rows each | 2018-01-23T14:47:17 | 2018-01-23T14:47:17 | {} | NONE |
ClickHouse | ClickHouse | 360,219,497 | 1,820 | bobrik | Memory usage of merges on 3 machines that decided to do this:
| 2018-01-24T17:57:33 | 2018-01-24T17:57:33 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 358,384,633 | 838 | mattbostock | I'd like to give this a go but I'm new to the project. Would anyone more familiar with it be able to give me some pointers on the implementation, specifically adding a custom compression codec per column? | 2018-01-17T17:45:20 | 2018-01-17T17:45:20 | {} | NONE |
ClickHouse | ClickHouse | 358,817,129 | 1,676 | alexey-milovidov | There is a regression: if `default` user is readonly, the server cannot start.
(Not your fault. I will fix it.) | 2018-01-18T23:30:34 | 2018-01-18T23:30:34 | {} | MEMBER |
ClickHouse | ClickHouse | 360,469,878 | 1,816 | inkrement | Thanks for the input. Quite interesting that both parts are evaluated, I wasn't aware of this! Now, I wrote it like this (Although, it is not so clean as @KochetovNicolai s version, it also works nicely):
```sql
SELECT if(toUInt8OrZero(extract(duration,'(\\d\\d):\\d\\d:'))*60*60 + toUInt8OrZero(extract(duration,'\\... | 2018-01-25T13:43:24 | 2018-01-25T13:43:24 | {} | NONE |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.