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 | 423,091,525 | 26 | blinkov | @dveselov in nearest plans are improvements for normal JOINs, the priority of correlated subqueries is not clear yet | 2018-09-20T08:28:42 | 2018-09-20T08:28:42 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 423,748,749 | 3,141 | alexey-milovidov | > Also, I noticed that you keep only very short list of previous version in apt repo and it makes rollback quite complicated
We started to keep all package versions since 1.1.54378. | 2018-09-22T14:41:18 | 2018-09-22T14:41:18 | {} | MEMBER |
ClickHouse | ClickHouse | 424,671,441 | 3,222 | filimonov | > May be I am doing smth wrong, but I can't get it.
The sample you show is array of arrays.
Please refer that sample:
```sql
:) CREATE TEMPORARY TABLE x ( id UInt32, vals Array(Int64) );
Ok.
0 rows in set. Elapsed: 0.003 sec.
:) insert into x(id,vals) VALUES (1, [1,2,3,4,5]), (2, [10000,10001]);
Ok.
... | 2018-09-26T10:50:48 | 2018-09-26T10:50:48 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 423,089,716 | 26 | dveselov | Do you guys have any plans to support correlated subqueries? | 2018-09-20T08:23:11 | 2018-09-20T08:23:11 | {} | NONE |
ClickHouse | ClickHouse | 423,208,862 | 3,168 | VadimPlh | #3182 | 2018-09-20T14:40:16 | 2018-09-20T14:40:16 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 423,367,220 | 3,184 | den-crane | ```
create table ttt_1(a Int64, b Int64) engine=MergeTree partition by b order by a;
insert into ttt_1 select number%2,number from numbers(100);
create table ttt_merge as ttt_1 engine=Merge(currentDatabase(), '^ttt');
```
select count() from (
select * from ttt_merge prewhere a = 0 and b=0 limit 100);
DB:... | 2018-09-20T23:34:37 | 2018-09-20T23:39:27 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 423,491,753 | 3,188 | alexey-milovidov | Non obvious. | 2018-09-21T10:45:50 | 2018-09-21T10:45:50 | {} | MEMBER |
ClickHouse | ClickHouse | 423,754,670 | 3,177 | e11it | Добавлю еще один пример неправильного cast:
19 символов помещается в Decimal64, вместо 18.
```bash
SELECT toDecimal64(toDecimal64('1234567890', 8), 9)
┌─toDecimal64(toDecimal64('1234567890', 8), 9)─┐
│ 1234567890.000000000 │
└──────────────────────────────────────────────┘
```
При ... | 2018-09-22T16:09:31 | 2018-09-22T16:09:31 | {} | NONE |
ClickHouse | ClickHouse | 423,810,561 | 3,141 | pavel-odintsov | That's great news! Thank you! | 2018-09-23T11:45:55 | 2018-09-23T11:45:55 | {} | NONE |
ClickHouse | ClickHouse | 424,565,649 | 3,210 | alexey-milovidov | > I propose to add API method to odbc/jdbc bridge, that will return value of escape character
Ok.
> maybe we should implement it for ODBC as well?
Yes. There is no reason, why it is not implemented.
> Currently, new implementation of ODBC, via unified bridge interface is registered via "idbc" name, in order... | 2018-09-26T02:41:42 | 2018-09-26T02:41:42 | {} | MEMBER |
ClickHouse | ClickHouse | 424,054,240 | 1,808 | dqminh | Sorry, this was a while ago and we couldnt replicate it even at the time IIRC. We havent seen the same error so far after a few version upgrades i think so I guess we can close this. | 2018-09-24T17:17:38 | 2018-09-24T17:17:38 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 424,084,500 | 3,203 | alexey-milovidov | Maybe I will get used to it in five minutes... | 2018-09-24T18:51:33 | 2018-09-24T18:51:33 | {} | MEMBER |
ClickHouse | ClickHouse | 424,687,311 | 3,218 | alexey-milovidov | Можно сделать так, что есть supertype, и это decimal, если он включает в себя всё множество значений соответствующего int-а. | 2018-09-26T11:53:58 | 2018-09-26T11:53:58 | {} | MEMBER |
ClickHouse | ClickHouse | 424,692,941 | 3,221 | 4ertus2 | Закомитил поддержку intDIv, intDivOrZero для decimal-ов в мастер. | 2018-09-26T12:14:35 | 2018-09-26T12:14:35 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 423,267,550 | 3,176 | alexey-milovidov | Test:
```
stateless.00411_long_accurate_number_comparison
Timeout
```
- known issue. | 2018-09-20T17:29:10 | 2018-09-20T17:29:10 | {} | MEMBER |
ClickHouse | ClickHouse | 423,438,978 | 3,185 | filimonov | Could you please provide the row sample? | 2018-09-21T07:20:35 | 2018-09-21T07:20:35 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 423,540,092 | 1,582 | muravjov | Same question. And I think tuple + max is fine in this situation:
```
select max((ts, size)) as tsSize, filter_field from table group by filter_field
┌─max(tuple(ts, size))─────────┬─filter_field─────┐
│ ('2018-09-18 08:04:45',1234) │ 105-project │
└──────────────────────────────┴─────────────┘
```
, isnt' it? | 2018-09-21T13:58:00 | 2018-09-21T13:58:00 | {} | NONE |
ClickHouse | ClickHouse | 423,572,707 | 3,192 | 4ertus2 | We need a way to represent our AST in tests and optimisations.
Example:
:) AST SELECT 1 AS a
(Explain ParsedAST
(SelectWithUnionQuery
(ExpressionList
(SelectQuery
(ExpressionList
(Literal UInt64 1 (Alias a))
)
)
)
)
) | 2018-09-21T15:25:40 | 2018-09-21T15:25:58 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 423,592,569 | 2,740 | wake-up-neo | Have the same issue on 1 of 3 nodes for ReplicatedReplacingMergeTree however replication is working fine. Could you please share your findings about this error? | 2018-09-21T16:21:19 | 2018-09-21T16:21:19 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 424,189,182 | 3,202 | zhang2014 | Done. | 2018-09-25T02:43:25 | 2018-09-25T02:43:25 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 424,543,354 | 3,167 | derekperkins | It really just depends on your dataset, so you're going down the right path with your tests. Always include the first columns of your primary key in the WHERE clause if possible.
You might try your query again but use `>= / <=` instead of `BETWEEN`. I feel like I tried it out a number of versions ago and `BETWEEN` h... | 2018-09-26T00:16:34 | 2018-09-26T00:16:34 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 422,989,948 | 3,143 | bocharov | We were affected by this issue as well, when trying to upgrade from 1.1.54388 to 18.12.14.
The exceptions we saw in the logs are:
```
<Error> executeQuery: Code: 241, e.displayText() = DB::Exception: Memory limit (total) exceeded: would use 30.00 GiB (attempt to allocate chunk of 1048591 bytes), maximum: 30.00 GiB, ... | 2018-09-19T23:22:34 | 2018-09-19T23:22:34 | {
"+1": 1
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 423,166,703 | 3,180 | alexey-milovidov | 86 kilobytes of memory doesn't looks too much. | 2018-09-20T12:36:36 | 2018-09-20T12:36:36 | {} | MEMBER |
ClickHouse | ClickHouse | 423,451,727 | 3,185 | filimonov | Did you change one of the following settings: `format_max_block_size`, `insert_format_max_block_size`, `max_block_size`, or some other settings?
```sql
select * from system.settings where changed = 1
```
Can you find the exact offset of that string in your source file?
```bash
grep --byte-offset '0,0,151699... | 2018-09-21T08:11:27 | 2018-09-21T08:11:27 | {
"+1": 1
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 424,536,684 | 3,167 | simPod | That helped a lot, thanks.
@derekperkins correct me if I wrong but
> but without all three, it has no way to optimize
Isn't it true that it optimizes even if one is included?
The docs says:
> If the primary key is (`a`, `b`), then adding another column `c will improve the performance if the following con... | 2018-09-25T23:39:17 | 2018-09-25T23:39:17 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 422,839,188 | 2,998 | KochetovNicolai | I should check it. Probably, compression size won't be less than for numeric column. | 2018-09-19T15:06:14 | 2018-09-19T16:20:04 | {} | MEMBER |
ClickHouse | ClickHouse | 423,140,814 | 3,179 | filimonov | This optimization is possible but not implemented yet.
If you can write C++ you can help the project by implementing that.
Please check that issue: https://github.com/yandex/ClickHouse/issues/1344 | 2018-09-20T10:58:37 | 2018-09-20T10:58:48 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 423,217,009 | 3,174 | Jack012a | Hello filimonov,
I do have some issues on understanding the meaning of the following config.
<merge_tree>
/** If table contains at least that many active parts, artificially slow down insert into table. */
<parts_to_delay_insert>300</parts_to_delay_insert>
/** If more than this number active par... | 2018-09-20T15:02:14 | 2018-09-20T15:08:34 | {
"+1": 5
} | NONE |
ClickHouse | ClickHouse | 423,452,689 | 3,187 | filimonov | You can try to export data to CSV / TSV from Hive [like that](https://stackoverflow.com/a/19730477/1555175)
And import it to ClickHouse via clickhouse-client.
Or just directly with pipe (without files):
```
hive -e 'select * from some_table where ... ' | clickhouse-clent --query='INSERT INTO table FORMAT TSV'
... | 2018-09-21T08:15:11 | 2018-09-21T08:17:25 | {
"+1": 2
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 423,453,872 | 3,185 | swyucZhang | > Did you change one of the following settings: `format_max_block_size`, `insert_format_max_block_size`, `max_block_size`, or some other settings?
>
> ```sql
> select * from system.settings where changed = 1
> ```
>
> Can you find the exact offset of that string in your source file?
>
> ```shell
> grep --by... | 2018-09-21T08:19:40 | 2018-09-21T08:19:40 | {} | NONE |
ClickHouse | ClickHouse | 424,589,979 | 3,208 | vavrusa | I could add it as a setting, but I'm not sure what would be a use case for that. Do you have a preference @ztlpn @alexey-milovidov ? | 2018-09-26T05:31:35 | 2018-09-26T05:31:35 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 423,074,882 | 3,167 | filimonov | Please check that: https://clickhouse.yandex/docs/en/operations/table_engines/mergetree/#primary-keys-and-indexes-in-queries | 2018-09-20T07:39:56 | 2018-09-20T07:39:56 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 423,106,480 | 3,157 | datayy | dpkg-reconfigure tzdata | 2018-09-20T09:14:58 | 2018-09-20T09:14:58 | {
"hooray": 3
} | NONE |
ClickHouse | ClickHouse | 423,143,297 | 3,180 | blinkov | @thanhthai3010 20 indeed sounds like too many, please refer to https://clickhouse.yandex/docs/en/operations/table_engines/mergetree/#primary-keys-and-indexes-in-queries and https://github.com/yandex/clickhouse-presentations/blob/master/percona2018/internals.pdf | 2018-09-20T11:07:44 | 2018-09-20T11:07:44 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 423,175,419 | 3,180 | thanhthai3010 | Thank a lot for your help. | 2018-09-20T13:05:08 | 2018-09-20T13:05:08 | {} | NONE |
ClickHouse | ClickHouse | 423,686,213 | 2,987 | yvlasenko | @AndriiOpryshko from the command above it doesn't look like you provided CH config file. It should be mounted from the host machine. | 2018-09-21T22:20:33 | 2018-09-21T22:20:33 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 423,931,046 | 3,191 | filimonov | ClickHouse doesn't support transactions for now.
Usually, insertions can be considered atomic, i.e. if you will insert both blocks at once you should get whole data inserted (or some error happen for the whole block).
At that moment there are only 2 ways of deleting data from ClickHouse: be partitions (so you c... | 2018-09-24T10:21:49 | 2018-09-24T10:21:49 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 424,377,160 | 3,144 | veloman-yunkan | @champtar Please send the curl post example to my email address. | 2018-09-25T14:54:29 | 2018-09-25T14:54:29 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 424,679,199 | 75 | myy2829 | +1 to fix | 2018-09-26T11:20:39 | 2018-09-26T11:20:39 | {} | NONE |
ClickHouse | ClickHouse | 423,074,171 | 3,174 | filimonov | The main requirement about insert to Clickhouse: you should never send too many INSERT statements per second. Ideally - one insert per second / per few seconds.
So you can insert 100K rows per second but only with one big bulk INSERT statement. When you send hundreds / thousands insert statements per second to *Merg... | 2018-09-20T07:37:40 | 2018-09-20T09:13:38 | {
"+1": 27
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 424,022,704 | 3,203 | filimonov | It's quite easy to change the [output format](https://github.com/yandex/ClickHouse/blob/master/libs/libdaemon/src/OwnPatternFormatter.cpp) of Clickhouse logs.
But it is also not a big issue to parse that optional query_id with `grok` filter of logstash (it uses usual regular expressions, optional groups are supported)... | 2018-09-24T15:44:50 | 2018-09-24T15:57:20 | {
"+1": 1
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 424,084,688 | 3,201 | alexey-milovidov | Yes, in case of DNS errors, it fails too early and no failover is performed. | 2018-09-24T18:52:11 | 2018-09-24T18:52:11 | {} | MEMBER |
ClickHouse | ClickHouse | 424,521,036 | 2,868 | alex-zaitsev | Looks fixed in 18.12.13
https://github.com/yandex/ClickHouse/pull/3087 | 2018-09-25T22:19:51 | 2018-09-25T22:19:51 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 424,407,062 | 3,177 | 4ertus2 | И сложно по реализации и расходится с текущим поведением при приведении Decimal-ов друг к другу (там лишние знаки отбрасываются). В добавок, с round есть особенности типа: обычное ожидается или банковское. Текущая реализация предполагает, что trunc(x, 0) = toIntNN(x), trunc(x, s) = toDecimalNN(x, s).
#3211 | 2018-09-25T16:12:52 | 2018-09-26T05:32:20 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 423,125,686 | 3,179 | SnailFastGo | @filimonov
Thank you very much for your reply！
However，In my opinion, althoug parallel processing, every part can reserve topk records .finally all of parts are sorted with topk records their own. Then the result is also right. Or could you give me a counterexample? | 2018-09-20T10:09:30 | 2018-09-20T10:09:30 | {} | NONE |
ClickHouse | ClickHouse | 423,209,074 | 3,011 | VadimPlh | #3182 #3168 | 2018-09-20T14:40:51 | 2018-09-20T14:40:51 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 423,475,567 | 3,151 | alesapin | https://github.com/yandex/ClickHouse/pull/3188 Fixing PR | 2018-09-21T09:39:13 | 2018-09-21T09:39:13 | {
"+1": 1
} | MEMBER |
ClickHouse | ClickHouse | 423,515,296 | 3,187 | panda403 | Wow！Thanks，I tried your suggestion,It's seems the most efficient way I've found,Thank you very much!@filimonov | 2018-09-21T12:31:48 | 2018-09-21T12:31:48 | {} | NONE |
ClickHouse | ClickHouse | 423,908,916 | 1,861 | filimonov | #2202 should fix that, but according to reports from Telegram channel problem still exists. | 2018-09-24T08:42:28 | 2018-09-24T08:42:28 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 424,030,797 | 3,203 | alexey-milovidov | Maybe something like `{}` in absense of query context will be appropriate solution? | 2018-09-24T16:07:18 | 2018-09-24T16:07:18 | {} | MEMBER |
ClickHouse | ClickHouse | 424,090,558 | 3,201 | alexey-milovidov | This will help: https://github.com/yandex/ClickHouse/commit/17b8e209221061325ad7ba0539f03c6e65f87f29 | 2018-09-24T19:10:46 | 2018-09-24T19:10:46 | {} | MEMBER |
ClickHouse | ClickHouse | 424,143,896 | 3,179 | filimonov | @alexey-milovidov but i think it still do sorting inside EACH block ignoring the fact that requested sorting order is PK, and it's enough to read few blocks from each part and merge them. Or it's not true? | 2018-09-24T22:17:04 | 2018-09-24T22:17:04 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 424,282,497 | 3,144 | veloman-yunkan | Hi @champtar,
I reviewed my changes and couldn't spot anything that could break existing functionality (unless the old code contained hidden bugs).
Did you observe this problem with a clean build?
| 2018-09-25T10:03:46 | 2018-09-25T10:03:46 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 424,380,981 | 3,210 | alex-krash | Currently, new implementation of ODBC, via unified bridge interface is registered via "idbc" name, in order to being able to perform comparison between them. | 2018-09-25T15:03:49 | 2018-09-25T15:03:49 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 422,970,380 | 3,171 | alesapin | This is wrong. | 2018-09-19T21:49:27 | 2018-09-19T21:49:27 | {} | MEMBER |
ClickHouse | ClickHouse | 423,516,142 | 3,188 | alesapin | Test failed becase of timeout. | 2018-09-21T12:35:23 | 2018-09-21T12:35:23 | {} | MEMBER |
ClickHouse | ClickHouse | 423,886,937 | 3,198 | simross | Thank you for response. Changed kafka table creation to:
`CREATE TABLE default.k ( EventDate Date) ENGINE = Kafka SETTINGS kafka_broker_list = 'localhost:9092,localhost:9093,localhost:9094', kafka_topic_list = 'a', kafka_group_name = 'group1', kafka_format = 'JSONEachRow', kafka_num_consumers = 1`
Unfortunately, prob... | 2018-09-24T06:38:40 | 2018-09-24T06:38:40 | {} | NONE |
ClickHouse | ClickHouse | 424,084,034 | 3,203 | alexey-milovidov | I've tried it, but it looks weird:
```
2018.09.24 21:49:16.045919 [ 19 ] {} <Debug> default.one_table (Data): Loading data parts
2018.09.24 21:49:16.048664 [ 19 ] {} <Debug> default.one_table (Data): Loaded data parts (1 items)
2018.09.24 21:49:16.050706 [ 19 ] {} <Debug> default.t (Data): Loading data parts
201... | 2018-09-24T18:50:07 | 2018-09-24T18:50:07 | {} | MEMBER |
ClickHouse | ClickHouse | 423,366,488 | 3,184 | den-crane | ```
create table ttt_1(a Int64, b Int64) engine=MergeTree partition by tuple() order by a;
insert into ttt_1 select number%2,number from numbers(100);
create table ttt_merge as ttt_1 engine=Merge(currentDatabase(), '^ttt');
```
select count() from ttt_merge prewhere a = 0;
DB::Exception: Cannot find column a... | 2018-09-20T23:30:06 | 2018-09-20T23:30:06 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 423,452,178 | 2,866 | qza1800 | I got same issue, but pointInPolygon returns 1 whilst expected 0
SELECT pointInPolygon((106.677085876465,10.7744951248169), [(106.667161868227,10.7675128359024),(106.667165727127,10.7675059912261),(106.667170817563,10.7674904752629),(106.667229225265,10.7672278502066),(106.667231193621,10.7672115129572),(106.667229912... | 2018-09-21T08:13:17 | 2018-09-21T08:14:18 | {} | NONE |
ClickHouse | ClickHouse | 423,754,805 | 3,177 | e11it | Ну и первый пример это скорее BUG. | 2018-09-22T16:11:36 | 2018-09-22T16:11:36 | {} | NONE |
ClickHouse | ClickHouse | 423,929,220 | 3,187 | filimonov | Please kindly close the issue if you have no more questions regarding that case. | 2018-09-24T10:13:03 | 2018-09-24T10:13:03 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 424,076,547 | 3,177 | 4ertus2 | В первом кейсе действетльно, можно кинуть исключение.
Про второй кейс есть в документации в git-е (пока не доехал до веб-а):
https://github.com/yandex/ClickHouse/blob/master/docs/ru/data_types/decimal.md
> Внутри данные представляются как знаковые целые числа, соответсвующей разрядности. Реальные диапазоны, хранящ... | 2018-09-24T18:26:47 | 2018-09-24T18:27:19 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 424,095,883 | 3,179 | alexey-milovidov | @filimonov We already use almost the same optimization.
We keep top N ordered records for every processed block. We collect all that filtered blocks and finally extract top N records from all of them by merging.
This algorithm (partial sort in mini-batches before merging) is more CPU efficient than usual priority... | 2018-09-24T19:28:27 | 2018-09-24T19:28:27 | {} | MEMBER |
ClickHouse | ClickHouse | 424,453,304 | 3,144 | champtar | done | 2018-09-25T18:33:15 | 2018-09-25T18:33:15 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 422,871,435 | 3,171 | alesapin | Unfortunately it doesn't work. AddingDefaultBlockOutputStream add columns to end of block, but SquashingBlockOutputStream expects columns in "header" order. | 2018-09-19T16:35:57 | 2018-09-19T16:35:57 | {} | MEMBER |
ClickHouse | ClickHouse | 423,095,101 | 3,159 | zhang2014 | Done. | 2018-09-20T08:39:22 | 2018-09-20T08:39:22 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 423,646,960 | 3,190 | alexey-milovidov | Thank you! | 2018-09-21T19:29:32 | 2018-09-21T19:29:32 | {} | MEMBER |
ClickHouse | ClickHouse | 424,032,516 | 3,203 | WouldYouKindly | @alexey-milovidov yes, it would completely solve our problem.
@filimonov thank you. We have to use grok at the moment, but it would be much nicer to have a predetermined list of fields. Easier to read and faster to parse (IIRC grok is considered to be much slower than dissect). | 2018-09-24T16:12:19 | 2018-09-24T16:12:19 | {} | NONE |
ClickHouse | ClickHouse | 424,089,470 | 3,203 | alexey-milovidov | In master. | 2018-09-24T19:07:02 | 2018-09-24T19:07:02 | {
"heart": 1
} | MEMBER |
ClickHouse | ClickHouse | 424,146,135 | 3,066 | champtar | Fixed in 18.14 | 2018-09-24T22:27:01 | 2018-09-24T22:27:01 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 423,164,272 | 3,174 | Jack012a | Hello filimonov,
I use the following command to insert the data into ClickHouse.
cat *.clickhouse | clickhouse-client --query="INSERT INTO default.MyTable(....")...
Since the number of files are large, I don't think many insert statement can be triggered.
Question> Based on the above statement, how many ins... | 2018-09-20T12:27:32 | 2018-09-20T12:27:49 | {} | NONE |
ClickHouse | ClickHouse | 423,330,998 | 3,159 | alexey-milovidov | Something is wrong: https://travis-ci.org/yandex/ClickHouse/builds/430907161#L8332 | 2018-09-20T20:58:27 | 2018-09-20T20:58:27 | {} | MEMBER |
ClickHouse | ClickHouse | 423,407,878 | 3,159 | zhang2014 | fixed test failure and `test_merge_table_over_distributed` | 2018-09-21T04:12:47 | 2018-09-21T04:13:04 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 423,757,371 | 329 | orloffv | Any news? | 2018-09-22T16:50:15 | 2018-09-22T16:50:15 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 423,949,916 | 3,199 | MPocahontas | Thanks. It solved my problem. | 2018-09-24T11:55:17 | 2018-09-24T11:55:17 | {} | NONE |
ClickHouse | ClickHouse | 424,093,126 | 3,177 | e11it | Нельзя ли реализовать так, чтобы результат приведение типа decimal к Int не зависел от размерности дробной части? Например присваивать целую часть, а затем добавлять результат округления дробной части. | 2018-09-24T19:19:25 | 2018-09-24T19:19:25 | {} | NONE |
ClickHouse | ClickHouse | 424,152,537 | 3,206 | alexey-milovidov | To restore from the backup, you should use `ALTER ... ATTACH PARTITION` query. It will select active parts and attach them.
If you attach parts one by one by issuing `ALTER ... ATTACH PART` query, you will select both active and inactive parts with duplicate data. Inactive parts are the parts that are kept for some ... | 2018-09-24T22:56:57 | 2018-09-24T22:56:57 | {} | MEMBER |
ClickHouse | ClickHouse | 424,292,488 | 3,177 | e11it | Может быть лучше сделать так, чтобы не было неочевидных поведений:
a) Приведение к Int вызывает round(x, 0) - дорогая операция.
b) Две дополнительные функции для быстрого получения целой и дробной части для типа decimal. | 2018-09-25T10:38:40 | 2018-09-25T10:38:40 | {} | NONE |
ClickHouse | ClickHouse | 423,114,203 | 3,179 | filimonov | Because of ORDER BY. Due to parallel processing of different parts and the fact that different parts can have intersecting PK ranges clickhouse really need to do ORDER BY on all the resulting rows, and after that apply the LIMIT. | 2018-09-20T09:37:56 | 2018-09-20T09:37:56 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 423,417,834 | 3,184 | zhang2014 | #3165 fixes this problem.
BTW. Using `_table` virtual column in `prewhere` is not allowed | 2018-09-21T05:30:14 | 2018-09-21T05:30:14 | {
"+1": 1
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 423,455,057 | 3,185 | swyucZhang | > > Did you change one of the following settings: `format_max_block_size`, `insert_format_max_block_size`, `max_block_size`, or some other settings?
> > ```sql
> > select * from system.settings where changed = 1
> > ```
> >
> >
> > Can you find the exact offset of that string in your source file?
> > ```shell
... | 2018-09-21T08:23:59 | 2018-09-21T08:23:59 | {} | NONE |
ClickHouse | ClickHouse | 424,092,980 | 3,185 | alexey-milovidov | DateTime in text format can be read in two following forms:
1. `YYYY-MM-DD hh:mm:ss` (preferred form).
2. A number with unix timestamp (supported as an exception), with the following considerations:
```
/** Read 10 characters, that could represent unix timestamp.
* Only unix timestamp of 5-10 characters ... | 2018-09-24T19:18:57 | 2018-09-24T19:18:57 | {} | MEMBER |
ClickHouse | ClickHouse | 423,124,176 | 3,179 | shangshujie365 | but i think, ch can use topk to optimise this query, each thread just keep the top 2000 records and merge them finally | 2018-09-20T10:05:48 | 2018-09-20T10:05:48 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 423,195,300 | 3,174 | filimonov | Sorry, didn't follow your first post.
You have opposite problem - you're trying to put 200Gb of data with one very huge insert.
Can't you just insert your files one by one with tiny delays?
```bash
for i in *.clickhouse; do
[ -f "$i" ] || break;
echo "Processing $i"
cat $i | clickhouse-client --query="IN... | 2018-09-20T14:03:10 | 2018-09-20T14:06:54 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 423,627,801 | 3,168 | Hitvz | thx，i will try it later | 2018-09-21T18:18:31 | 2018-09-21T18:18:31 | {
"+1": 1
} | NONE |
ClickHouse | ClickHouse | 423,928,612 | 3,199 | filimonov | Just give access for your user to some third dummy database which will be used only for connecting and will always exists (for example to the `default`).
```xml
<allow_databases>
<database>default</database>
<database>testDb</database>
</allow_databases>
```
Use that `default` database in connection settin... | 2018-09-24T10:10:08 | 2018-09-24T10:10:08 | {
"+1": 1
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 424,331,301 | 3,144 | champtar | Hi @veloman-yunkan
It's with a clean build
I can share an example curl post that work before but not after with you if you want
(sadly I can't share it in public & I was not able to do a short repro case) | 2018-09-25T12:54:41 | 2018-09-25T12:54:41 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 424,658,155 | 3,218 | 4ertus2 | Привет,
Текущее поведение соответствует ожиданиям. Когда делали обобщение - предполагалось, что объединять decimal-ы и не-decimal-ы не потребуется.
Обсудим в ближайшее время, можно ли найти тут воркэраунд или нужно менять поведение. | 2018-09-26T10:00:16 | 2018-09-26T10:04:03 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 422,836,128 | 2,998 | KochetovNicolai | Hello.
Sorry for late answer. I think it's a great idea to promote low cardinality columns (and, probably, any expression with LowCardinality type at result) to PREWHERE. I haven't planned this optimization yet, but I'll add it at least for case `LowCardinality = const` if it's easy enough. | 2018-09-19T14:57:26 | 2018-09-19T15:01:40 | {} | MEMBER |
ClickHouse | ClickHouse | 422,837,874 | 2,998 | alexey-milovidov | @KochetovNicolai Why it should not work automatically due to small compressed size of column? | 2018-09-19T15:02:22 | 2018-09-19T15:02:22 | {} | MEMBER |
ClickHouse | ClickHouse | 423,428,784 | 3,179 | SnailFastGo | Unfortunately！I am not good at writing C++. Howerve I will keep attention to clickhouse | 2018-09-21T06:33:30 | 2018-09-21T06:33:30 | {} | NONE |
ClickHouse | ClickHouse | 423,441,525 | 3,185 | swyucZhang | > Could you please provide the row sample?
create table SQL:
CREATE TABLE *** ( id String, ...... , tag_998 DateTime , tag_999 DateTime , record_date Date ) ENGINE = MergeTree(record_date, (id), 8192)
the row is too long and this is the error part
 = 0, NULL, intDivOrZero(toDecimal64(0, 4), 0))
```
Сейчас оно ругнется, что операция не поддерживается. Чуть позже сделаю... | 2018-09-26T11:29:58 | 2018-09-26T11:29:58 | {} | CONTRIBUTOR |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.