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 | 401,273,049 | 2,192 | filimonov | > Where is it stated? I checked 'single page documentation' and didn't find this.
https://clickhouse.yandex/docs/en/formats/csv/
There is a plenty of options to do that without braking backward compatibility:
1) parametrize that behaviour
2) create separate format (like CSV_RFC)
3) detect quoting scheme once... | 2018-06-29T07:32:01 | 2018-06-29T07:32:19 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 401,444,120 | 1,999 | alexey-milovidov | Is it reproducible on the latest version?
I will need some minimal example to reproduce in our servers. | 2018-06-29T18:57:08 | 2018-06-29T18:57:08 | {} | MEMBER |
ClickHouse | ClickHouse | 401,843,876 | 2,494 | chris-baynes | I believe this is a bug, the simplest reproduction case is just:
```
:) select count(distinct null)
SELECT countDistinct(NULL)
┌─uniqExact(NULL)─┐
│ \N │
└─────────────────┘
```
`count` & `countDistinct` should always return an Int >= 0. This is currently an issue for us as it breaks an int... | 2018-07-02T15:30:46 | 2018-07-02T15:30:46 | {} | NONE |
ClickHouse | ClickHouse | 402,732,316 | 2,595 | filimonov | Hmmm... Should be checked. May be some improper escaping/deescaping of backslash when reading / writing alter to a zookeeper.
Possible workaround: try to use unhex('5C') as your split char (it's ascii code of backslash char).
And by the way: you can access last element of the array using negative index.
```
... | 2018-07-05T14:02:30 | 2018-07-05T14:02:30 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 402,876,654 | 379 | simPod | @girgitt doing it this way didn't come to my mind. Thanks.
Currently, for some cases I'm using this function https://docs.timescale.com/v0.10/api#time_bucket Would be nice if ClickHouse had something similar to improve query readability. | 2018-07-05T23:02:26 | 2018-07-05T23:02:26 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 402,248,216 | 2,588 | TimonKK | Спасибо, буду пробовать | 2018-07-03T18:16:52 | 2018-07-03T18:16:52 | {} | NONE |
ClickHouse | ClickHouse | 402,251,982 | 2,588 | TimonKK | Помогло:
1. запустил zookeeper cli (bin/zkCli.sh)
2. rmr <path_to_table>/mutations/XXXXXXXXXX | 2018-07-03T18:30:06 | 2018-07-03T18:30:06 | {} | NONE |
ClickHouse | ClickHouse | 402,482,215 | 2,591 | KochetovNicolai | Нет, для flat словарей так сделать не получится. Для flat словарей используется прямая адресация, из-за этого они быстро работают и имеют ограничение по памяти. В случае, если есть id со значением > 500000, можно:
* Сделать преобразование id руками так, чтобы они укладывались в требуемый интервал.
* Использовать hash... | 2018-07-04T13:39:12 | 2018-07-04T13:39:12 | {} | MEMBER |
ClickHouse | ClickHouse | 401,332,398 | 2,452 | vitalvas | Is it possible to add this functionality to the "MergeTree" engine?
```
ALTER TABLE log
DELETE WHERE ClientIPv4 = 0
Received exception from server (version 1.1.54388):
Code: 48. DB::Exception: Received from localhost:9000, 127.0.0.1. DB::Exception:
Mutations are not supported by storage MergeTree.
`... | 2018-06-29T11:55:29 | 2018-06-29T11:55:29 | {
"+1": 3
} | NONE |
ClickHouse | ClickHouse | 402,638,940 | 2,521 | ssmike | SELECT topK(0.2)(number) FROM (SELECT * FROM system.numbers LIMIT 50)
Received exception from server (version 1.1.54387):
Code: 69. DB::Exception: Received from localhost:9000, ::1. DB::Exception: Parameter 0 is illegal for aggregate function topK.
¯\\\_(ツ)\_/¯ | 2018-07-05T08:02:19 | 2018-07-05T08:02:54 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 402,877,716 | 1,645 | alexey-milovidov | This hypothetical feature shouldn't be recommended in typical usage scenarios, so configuration with `max_suspicious_broken_parts` and similar settings is enough. | 2018-07-05T23:08:57 | 2018-07-05T23:08:57 | {} | MEMBER |
ClickHouse | ClickHouse | 401,452,881 | 1,729 | alexey-milovidov | Released in 1.1.54388. See changelog. | 2018-06-29T19:33:49 | 2018-06-29T19:33:49 | {} | MEMBER |
ClickHouse | ClickHouse | 402,411,480 | 2,582 | filimonov | > In this case column y from table isn't calculated.
OK, you right. That sample was not good, use that instead:
```
SELECT x, sleepEachRow(0.1)+pow(x,2) as y FROM aliases_lazyness WHERE x=1;
```
As you can see that perfectly works without calculating `y` for whole table. And that behavior seems logical.
The ... | 2018-07-04T08:52:34 | 2018-07-04T09:17:16 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 402,778,965 | 2,598 | blinkov | Yep, see https://clickhouse.yandex/docs/en/agg_functions/combinators/#-foreach | 2018-07-05T16:26:19 | 2018-07-05T16:26:19 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 402,880,922 | 2,383 | alexey-milovidov | It's totally Ok for `CREATE DATABASE` (the only issue is unclean error message)
but it's not Ok for `CREATE DATABASE IF NOT EXISTS`. | 2018-07-05T23:29:56 | 2018-07-05T23:29:56 | {} | MEMBER |
ClickHouse | ClickHouse | 403,031,806 | 2,478 | ztlpn | ```sql
ALTER TABLE my_project.my_collection_1_src FREEZE PARTITION '200101_4_4_0';
```
won't work because `'200101_4_4_0'` is a part, not a partition (each partition consists of several parts with the same value of the partition key).
Next, because in your case the partition key is integer-valued, you can't use s... | 2018-07-06T13:19:16 | 2018-07-06T13:19:16 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 401,615,971 | 2,578 | lamberken | It's similar to #827. how to solve it now. thanks.
I understand why it was designed. my solution is each row has a `uuid` column.
Whether or not the parameter can also be achieved?
| 2018-07-01T16:02:52 | 2018-07-01T16:20:05 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 401,911,559 | 2,571 | alexey-milovidov | #2584 | 2018-07-02T19:39:06 | 2018-07-02T19:39:06 | {} | MEMBER |
ClickHouse | ClickHouse | 402,132,027 | 2,582 | KochetovNicolai | ```SQL
SELECT x, sleepEachRow(0.1) as y FROM aliases_lazyness WHERE x = 1;
```
In this case column y from table isn't calculated. What's calculated is constant column y with value `sleepEachRow(0.1)`. For some reason it's calculated 4 times during reprocessing, and it's ok.
In other words, query works exactly the ... | 2018-07-03T12:06:42 | 2018-07-03T12:06:42 | {} | MEMBER |
ClickHouse | ClickHouse | 402,438,135 | 2,589 | berghauz | I'm sorry, it's my fault. I'm checking another table when comparing size...
btw, any chance to support ON CLUSTER in future? | 2018-07-04T10:30:00 | 2018-07-04T10:30:00 | {} | NONE |
ClickHouse | ClickHouse | 402,533,557 | 2,591 | antonio-antuan | Благодарю за объяснения | 2018-07-04T17:45:51 | 2018-07-04T17:45:51 | {} | NONE |
ClickHouse | ClickHouse | 402,570,068 | 2,521 | alexey-milovidov | TODO: allocate all state inplace, get rid of `WeightedValue *` pointer.
Look for `uniqUpTo` aggregate function as an example. | 2018-07-04T23:08:17 | 2018-07-04T23:08:17 | {} | MEMBER |
ClickHouse | ClickHouse | 402,640,006 | 2,521 | ssmike | Ok, I've fixed limits. But I don't understand what you expect to see in histogram for one value. | 2018-07-05T08:06:31 | 2018-07-05T09:52:39 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 402,727,531 | 2,596 | filimonov | Why do you expect it should use another day?
From documentation:
> `toDate()`
> Converts a date with time to a certain fixed date, while preserving the time."
1970-01-02 is definitely 'certain fixed date', isn't it? :)
1970-01-01 has a special meaning in clickhouse. It's 'out of the range' date, printed as ... | 2018-07-05T13:47:59 | 2018-07-05T13:48:23 | {
"+1": 1
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 402,878,693 | 2,111 | alexey-milovidov | The behaviour is determined by types of arguments, regardless to values of constants.
Neither
`SELECT [-9223372036854775808, 9223372036854775807]`
`SELECT [toInt64(0), toUInt64(0)]`
`SELECT [number, -number] FROM numbers(10)`
work by the same reason. | 2018-07-05T23:14:37 | 2018-07-05T23:14:37 | {} | MEMBER |
ClickHouse | ClickHouse | 402,524,564 | 2,591 | KochetovNicolai | Поиск по атрибутам в словаре будет работать одинаково (неоптимально) как для словарей типа flat, так и для hashed. В таком подзапросе сначала будет прочитан весь словарь (для hashed это может работать даже чуть быстрее), а затем сформировано множество из id (которое тоже хеш-таблица).
Для поиска по атрибутам использ... | 2018-07-04T16:43:26 | 2018-07-04T16:43:26 | {} | MEMBER |
ClickHouse | ClickHouse | 402,540,647 | 2,589 | berghauz | Thank you! | 2018-07-04T18:38:32 | 2018-07-04T18:38:32 | {} | NONE |
ClickHouse | ClickHouse | 402,561,604 | 2,574 | alexey-milovidov | Or it became a little better for unknown reason (no need to investigate):
```
$ for i in {1..100}; do (time ./clickhouse-master local --input-format CSV --structure 'a String, b String, c UInt64' --query="SELECT count() FROM table" < test.csv) 2>&1 | grep user; done | sort | head -n10
user 0m0.116s
user 0m0... | 2018-07-04T21:27:20 | 2018-07-04T21:28:14 | {} | MEMBER |
ClickHouse | ClickHouse | 402,746,314 | 2,590 | Imaskar | Reference query works as expected with these changes.
```
┌──────────event_time─┬─status1─┬─status2─┬─delta1─┬─delta2─┬─delta3─┬─delta4─┐
│ 2018-07-05 17:37:29 │ 1 │ ᴺᵁᴸᴸ │ 0 │ ᴺᵁᴸᴸ │ 0 │ 0 │
│ 2018-07-05 17:37:36 │ ᴺᵁᴸᴸ │ 1 │ ᴺᵁᴸᴸ │ 0 │ 7 │ 0 │
│ 2018-07-05 17:3... | 2018-07-05T14:44:24 | 2018-07-05T14:44:24 | {} | NONE |
ClickHouse | ClickHouse | 402,780,587 | 2,598 | chamb | Thank you.
I see **sumForEach** does exactly what I want. | 2018-07-05T16:32:02 | 2018-07-05T16:32:02 | {} | NONE |
ClickHouse | ClickHouse | 402,838,866 | 2,599 | alexey-milovidov | It should work in about the same way.
There are some considerations in usability. When you execute an OPTIMIZE query it will create many merges in the replication queue and merges of new parts won't be created for a long period of time. If you have inserts in your table, they will end up refused with the message "To... | 2018-07-05T20:11:02 | 2018-07-05T20:11:29 | {} | MEMBER |
ClickHouse | ClickHouse | 402,104,741 | 2,585 | ztlpn | Thanks Amy! | 2018-07-03T10:46:27 | 2018-07-03T10:46:27 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 402,566,301 | 2,521 | alexey-milovidov | ```
SELECT
CounterID,
count(),
histogramIf(10)(SendTiming, SendTiming != -1)
FROM hits_1000m_single
GROUP BY CounterID
ORDER BY count() DESC
LIMIT 20
```
Looks like the results are wrong:
```
232261 │ 12148031 │ [(0,2.2250738585072014e-308,12148031)]
``` | 2018-07-04T22:17:56 | 2018-07-04T22:17:56 | {} | MEMBER |
ClickHouse | ClickHouse | 401,340,671 | 2,226 | ztlpn | Seems that #2083 is still not fixed in this PR. | 2018-06-29T12:33:18 | 2018-06-29T12:33:34 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 401,433,988 | 2,575 | alexey-milovidov | Привет.
> Можно ли в Click House создавать собственные движки для таблиц, как дополнительный динамический плагин?
Архитектура ClickHouse позволяет добавлять свои движки таблиц, но не динамически. Они должны находиться в кодовой базе ClickHouse, хотя в то же время, их код довольно изолирован.
Для примера смотри... | 2018-06-29T18:18:22 | 2018-06-29T18:18:39 | {} | MEMBER |
ClickHouse | ClickHouse | 401,446,015 | 2,568 | alexey-milovidov | @filimonov :+1: | 2018-06-29T19:04:33 | 2018-06-29T19:04:33 | {} | MEMBER |
ClickHouse | ClickHouse | 401,457,776 | 2,571 | alexey-milovidov | #1804 | 2018-06-29T19:55:19 | 2018-06-29T19:55:19 | {} | MEMBER |
ClickHouse | ClickHouse | 402,495,796 | 2,579 | igor-sh8 | Hi.
Thanks for the info.
So as I understand, we can not use the same table. We need to create a new table in the new cluster, then copy data to this table and then drop old table (if we need it) ?
| 2018-07-04T14:31:35 | 2018-07-04T14:31:35 | {} | NONE |
ClickHouse | ClickHouse | 402,565,723 | 2,521 | alexey-milovidov | Simple test for performance and precision:
```
SELECT arrayJoin(histogram(10)(number)) AS res, res.3 / (res.2 - res.1) AS density, bar(density * 1000, 0, 1000, 80) AS bar FROM numbers(1000000000)
```
it's about 8.5 million rows/sec per single core.
The bottleneck is `compress` function and, consequently, priorit... | 2018-07-04T22:11:14 | 2018-07-04T22:11:14 | {} | MEMBER |
ClickHouse | ClickHouse | 402,761,336 | 2,595 | filimonov | > But there is a problem with character escaping ... Or not?
Reproducable test should be created to answer that question. May be it's a bug.
What CH versions / distribution do you have?
| 2018-07-05T15:28:36 | 2018-07-05T15:29:50 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 402,992,349 | 2,600 | igor-sh8 | Great.
Thanks for the detailed explanation. | 2018-07-06T10:14:44 | 2018-07-06T10:14:44 | {} | NONE |
ClickHouse | ClickHouse | 401,451,499 | 2,460 | alexey-milovidov | Need a bit more info to reproduce. | 2018-06-29T19:27:55 | 2018-06-29T19:27:55 | {} | MEMBER |
ClickHouse | ClickHouse | 401,455,053 | 2,507 | alexey-milovidov | It the `datacenter` cluster contains local shard? | 2018-06-29T19:43:33 | 2018-06-29T19:43:33 | {} | MEMBER |
ClickHouse | ClickHouse | 401,876,686 | 2,582 | derekperkins | I'm also very interested in adding dict aliases if they are equally performant with putting it into the query. | 2018-07-02T17:27:21 | 2018-07-02T17:27:21 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 402,570,538 | 2,521 | alexey-milovidov | This should throw an exception:
`SELECT histogram(1.5)(1)` | 2018-07-04T23:15:15 | 2018-07-04T23:15:15 | {} | MEMBER |
ClickHouse | ClickHouse | 402,860,941 | 2,281 | alexey-milovidov | First, we can add random part for wait intervals. For example, instead of 10 seconds it can be resheduled in an interval, uniformly distributed in 8..12 seconds.
Next we can review all places when some task is rescheduled unconditionally.
> So I'm wonder if a better approach will be to reschedule the tasks only:
... | 2018-07-05T21:41:37 | 2018-07-05T22:29:06 | {
"+1": 1
} | MEMBER |
ClickHouse | ClickHouse | 401,596,624 | 2,507 | lamberken | @alexey-milovidov , yes.
the name of cluster is `datacenter`.
the name of database is `datacenter` too.
#### cluster's config
```
SELECT *
FROM system.clusters
┌─cluster────┬─shard_num─┬─shard_weight─┬─replica_num─┬─host_name──────┬─port─┬─is_local─┬─user────┬─default_database─┐
│ datacenter │ 1 ... | 2018-07-01T10:00:47 | 2018-07-01T10:00:47 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 402,404,300 | 2,589 | ztlpn | For each data part ALTER DELETE will prepare a mutated data part and substitute the original part for it in the active part set. But the original part will linger in the inactive state before the cleaner process deletes it (by default this happens after 8 minutes).
Does the problem persist after ~10 minutes? You can... | 2018-07-04T08:27:14 | 2018-07-04T08:27:14 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 402,421,130 | 2,582 | filimonov | Another demo showing that existing approach in not correct:
```sql
create table aliases_lazyness3 (x UInt32, y ALIAS runningDifference(x)) Engine=MergeTree ORDER BY x;
insert into aliases_lazyness3 select * from numbers(20);
select x,y from aliases_lazyness3 where x in (5,10,15);
┌──x─┬─y─┐
│ 5 │ 1 │
│ 10 │ 1... | 2018-07-04T09:26:18 | 2018-07-04T09:32:22 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 402,503,024 | 2,589 | ztlpn | ON CLUSTER should work, the syntax is (just as with other ALTER commands):
```sql
ALTER TABLE [db].table ON CLUSTER 'cluster' DELETE WHERE expr
``` | 2018-07-04T14:59:33 | 2018-07-04T14:59:33 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 402,583,171 | 2,592 | den-crane | BTW, it will break current functionality, group by [expression]
select count() from system.one
group by 1
order by 1+0 | 2018-07-05T01:45:27 | 2018-07-05T01:46:03 | {
"confused": 1
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 402,686,233 | 2,592 | blinkov | Yep, this feature is questionable, though I know a bunch of people who are big fans of it.
Use cases they mentioned were like trying to aggregate roughly similar stuff on different sets of columns. And this syntax allows to avoid typing those column names twice, while retaining original labels.
Probably some day ... | 2018-07-05T11:00:41 | 2018-07-05T11:00:41 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 402,749,274 | 2,595 | perez1987 | @filimonov Works! But there is a problem with character escaping when a field with a default value is inserted into tables with ReplicatedMergeTree. Or not? | 2018-07-05T14:52:56 | 2018-07-05T14:52:56 | {} | NONE |
ClickHouse | ClickHouse | 402,787,508 | 2,596 | alexey-milovidov | We use `1970-01-02` because `1970-01-01 00:00:00 MSK` has unix timestamp of `-10800` that is before unix epoch. | 2018-07-05T16:56:57 | 2018-07-05T16:57:07 | {} | MEMBER |
ClickHouse | ClickHouse | 401,443,388 | 2,001 | alexey-milovidov | You can run `clickhouse-client` under gdb:
`gdb --args clickhouse-client ...`
and type `r`
Then try to reproduce this issue and type `bt` to get a stack trace.
Please post stack trace here. | 2018-06-29T18:54:26 | 2018-06-29T18:54:46 | {} | MEMBER |
ClickHouse | ClickHouse | 401,652,565 | 2,578 | zhang2014 | Maybe check `insert_deduplicate` setting | 2018-07-02T02:05:31 | 2018-07-02T02:05:31 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 402,083,869 | 1,435 | jackpgao | I've got the same error, but not caused by the max open files, but the memory.
As ClickHouse uses too much page cache, but the OS is too 'lazy' to release the page cache when the free memory is too little, and with a lot of queries, the memory is out, and ClickHouse met the error 'Resource temporarily unavailabl'.
... | 2018-07-03T09:49:35 | 2018-07-03T09:49:35 | {} | NONE |
ClickHouse | ClickHouse | 402,515,612 | 1,569 | hakman | @wawanawna and me found the above workaround to be working quite well. We used tmpfiles.d to recreate _format_version.txt_ at startup:
```
> cat /etc/tmpfiles.d/clickhouse.conf
/mnt/local/clickhouse/data/default/table/format_version.txt 0644 clickhouse clickhouse - 1
```
Also, tried to create a more permane... | 2018-07-04T15:54:53 | 2018-07-04T15:54:53 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 402,679,368 | 2,592 | filimonov | Not sure if it should be ported to Clickhouse.
It's available is MySQL and some other RDBMS, but it's is **very** controversial, and it's definitely a bad practice. It was removed from SQL standard in 1999. The only plus of that it that it allows you to gain few second on typing SQL, but all other aspects are minuse... | 2018-07-05T10:30:57 | 2018-07-05T11:10:06 | {
"+1": 1
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 402,770,387 | 2,578 | lamberken | @zhang2014 :100: | 2018-07-05T15:57:22 | 2018-07-05T15:57:22 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 402,954,135 | 2,111 | silviucpp | OK I thought best approach is to try to find the data type that fit's the values or in case of a cast to use the data type from the cast. | 2018-07-06T07:41:34 | 2018-07-06T07:41:34 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 401,451,994 | 2,458 | alexey-milovidov | Also geohash support will be useful:
https://github.com/yinqiwen/geohash-int/blob/master/geohash.c | 2018-06-29T19:29:58 | 2018-06-29T19:29:58 | {} | MEMBER |
ClickHouse | ClickHouse | 401,552,369 | 2,571 | alexey-milovidov | You can mitigate it with `SET use_index_for_in_with_subqueries = 0` | 2018-06-30T16:43:50 | 2018-06-30T16:43:50 | {
"+1": 1
} | MEMBER |
ClickHouse | ClickHouse | 401,773,171 | 2,576 | ztlpn | Thanks for noticing, I've rebuilt it.
The previous build failed due to some problems with the ubuntu keyserver. | 2018-07-02T11:23:18 | 2018-07-02T11:23:18 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 402,435,138 | 975 | filimonov | You can easy archive needed result with higher-order functions.
I've created sample here: https://gist.github.com/filimonov/2ff5f083c2f874eceebde8877721afc4 | 2018-07-04T10:18:00 | 2018-07-04T10:18:00 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 402,570,284 | 2,521 | alexey-milovidov | The type of parameter is not checked:
```
SELECT CounterID, count(), histogramIf(-1)(SendTiming, SendTiming > 0) FROM hits_1000m_single GROUP BY CounterID ORDER BY count() DESC LIMIT 20
```
```
Memory limit (for query) exceeded: would use 128.02 GiB
``` | 2018-07-04T23:11:43 | 2018-07-04T23:11:43 | {} | MEMBER |
ClickHouse | ClickHouse | 402,570,499 | 2,521 | alexey-milovidov | Incorrect error message:
```
:) SELECT histogram(0.5)(1)
SELECT histogram(0.5)(1)
Received exception from server (version 1.1.54388):
Code: 36. DB::Exception: Received from localhost:9000, ::1. DB::Exception: Bin count should be positive.
``` | 2018-07-04T23:14:46 | 2018-07-04T23:14:46 | {} | MEMBER |
ClickHouse | ClickHouse | 402,753,740 | 379 | Girgitt | @simPod what exactly do you think is missing from the available groupping by subdivisions?
Considering such example of a table:
CREATE TABLE default.analog_archives ( d Date DEFAULT CAST(0 AS Date), ts UInt32, ids UInt32, val Float32)..
you can specify any reasonable (ok - down to intervals rounded to 5 min ... | 2018-07-05T15:05:39 | 2018-07-05T15:08:11 | {
"+1": 2
} | NONE |
ClickHouse | ClickHouse | 402,881,129 | 2,560 | alexey-milovidov | Confirmed. | 2018-07-05T23:31:03 | 2018-07-05T23:31:03 | {} | MEMBER |
ClickHouse | ClickHouse | 403,024,912 | 2,111 | alexey-milovidov | We have two functions:
- creation of array;
- CAST.
These functions are executed independently (context-free).
The issue is related just to creation of array. | 2018-07-06T12:51:19 | 2018-07-06T12:51:46 | {} | MEMBER |
ClickHouse | ClickHouse | 402,212,186 | 2,588 | ztlpn | Можно вручную удалить команду мутации из ZooKeeper и тогда сервер её пропустит. Для этого нужно узнать ID мутации из таблицы `system.mutations`, соответствующая znode находится в ZK по пути `<path_to_table>/mutations/<id>`. | 2018-07-03T16:14:06 | 2018-07-03T16:14:06 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 402,508,992 | 2,591 | antonio-antuan | // telegram:
словари используем в подзапросах, типа того:
`select ... from some_table where obj_id in (select id from dict where dict_attr = 'value')`
можно ли предположить из слова hash, что поиск по атрибутам словаря, а не по ключам, будет работать так себе?
Плюс, словарь будет расти.
Верно ли я другое предпо... | 2018-07-04T15:24:35 | 2018-07-04T15:24:35 | {} | NONE |
ClickHouse | ClickHouse | 402,566,785 | 2,521 | alexey-milovidov | ```
:) SELECT histogram(10)(-2.0)
SELECT histogram(10)(-2.)
┌─histogram(10)(-2.)───────────────┐
│ [(-2,2.2250738585072014e-308,1)] │
└──────────────────────────────────┘
``` | 2018-07-04T22:23:57 | 2018-07-04T22:23:57 | {} | MEMBER |
ClickHouse | ClickHouse | 402,720,788 | 2,596 | soulne4ny | It is implementation, however quite unexpected.
https://github.com/yandex/ClickHouse/blob/13762471da910d5b7ab37c1510616cac24e34ab2/dbms/src/Functions/FunctionsDateTime.h#L191
| 2018-07-05T13:25:22 | 2018-07-05T13:25:22 | {} | NONE |
ClickHouse | ClickHouse | 402,888,112 | 2,407 | alexey-milovidov | > I am not sure that ClickHouse have to close all connection on SIGSTOP.
The process cannot handle SIGSTOP. | 2018-07-06T00:18:23 | 2018-07-06T00:18:23 | {} | MEMBER |
ClickHouse | ClickHouse | 402,990,476 | 2,600 | filimonov | `in_order` - defines a strategy of choosing replica to query. I.e. when you're sending select to Distibuted table it should choose one replica from each shard to ask for a data. in_order says that it should always prefer first listed in config.
So it defines how Distributed table work.
Leader in clickhouse is co... | 2018-07-06T10:06:54 | 2018-07-06T10:06:54 | {
"+1": 6
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 402,212,302 | 2,588 | ztlpn | Для предотвращения такого в дальнейшнем, надо усилить валидацию команд мутации. | 2018-07-03T16:14:30 | 2018-07-03T16:14:30 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 402,557,412 | 2,566 | alexey-milovidov | Thanks!
What to do next:
- check that all existing tests passed if you replace AggregatingMergeTree to SummingMergeTree. | 2018-07-04T20:50:18 | 2018-07-04T20:50:18 | {} | MEMBER |
ClickHouse | ClickHouse | 402,557,915 | 2,574 | alexey-milovidov | TODO:
- rename settings to more descriptive names;
- check maximum performance difference in extreme case; | 2018-07-04T20:54:24 | 2018-07-04T20:54:24 | {} | MEMBER |
ClickHouse | ClickHouse | 402,871,875 | 1,259 | alexey-milovidov | The behaviour is correct: dots in column names are reserved for Nested types.
It is still possible to define columns with dot in name and with non-Array type, but the prefix before dot should not clash with name of any other column. | 2018-07-05T22:33:06 | 2018-07-05T22:33:06 | {} | MEMBER |
ClickHouse | ClickHouse | 402,905,510 | 2,334 | HadoopChen | ## SQL
select * from default.tag limit 1 ;
## ERROR
Error running query: Code: 62, e.displayText() = DB::Exception: Syntax error (Multi-statements are not allowed): failed at position 187 (end of query): ; FORMAT JSON. , e.what() = DB::Exception
## Beg For Help !
1. http://host:port/default
2. It is ok when ... | 2018-07-06T02:18:23 | 2018-07-06T03:42:08 | {} | NONE |
ClickHouse | ClickHouse | 402,945,946 | 2,595 | perez1987 | @filimonov thank you.
We have tested it and we can not create it with the command I said in my first comment. With your workaround, it works. I think it's a bug when creating a column in tables of the Replicated type | 2018-07-06T07:03:09 | 2018-07-06T07:03:09 | {} | NONE |
ClickHouse | ClickHouse | 401,271,791 | 2,192 | Imaskar | > In clickhouse as documentation says "Both double and single quotes are supported."
Where is it stated? I checked 'single page documentation' and didn't find this. If it is explicitly stated, then changing it to RFC may break backward compatibility. | 2018-06-29T07:26:16 | 2018-06-29T07:26:16 | {} | NONE |
ClickHouse | ClickHouse | 401,304,427 | 2,441 | zhang2014 | @ludv1x @Sindbag Is there any new progress about this feature you can share with us? We really need this :) | 2018-06-29T09:40:03 | 2018-06-29T09:40:03 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 401,608,439 | 2,368 | valyala | @dqminh, we successfully use `chproxy` during maintenance / upgrade procedure as a workaround for the described issue.
[Chproxy](https://github.com/Vertamedia/chproxy/blob/master/README.md) automatically routes each query to the least loaded healthy replica. `Least loaded` means the replica with the minimum number o... | 2018-07-01T13:54:42 | 2018-07-01T13:54:42 | {} | NONE |
ClickHouse | ClickHouse | 401,634,089 | 1,569 | socketpair | Seems, ClickHouse doesn't use correct sequence of atomic operations when working with files. It should do them in the following way:
```
1. create temporary file in same DIR named format_version.txt.XXXX
2. fchmod()
3. write somewhat to it
4. flush()
5. fdatasync()
6. close()
7. rename( format_version.txt.XXX... | 2018-07-01T21:21:55 | 2018-07-01T21:21:55 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 402,135,065 | 2,579 | KochetovNicolai | Yes, you can reshard your data with clickhouse-copier. What's you need is to properly configure your task. Write your current cluster as `<source_cluster>` and cluster with new replicas as `<destination_cluster>`, `<table_pull>` is your current table and `<table_push>` is a table you will reshard to.
| 2018-07-03T12:17:09 | 2018-07-03T12:17:09 | {} | MEMBER |
ClickHouse | ClickHouse | 402,435,244 | 1,569 | wawanawna | We also have a problems with this issue.
We have MaterializedView which we map to RAM (dir with MV table)
And each time after restart we see
> 2018.07.04 09:51:36.334149 [ 1 ] <Error> Application: DB::Exception: Cannot create table from metadata file /mnt/local/clickhouse/metadata/default//XXX.sql, error: DB::Exce... | 2018-07-04T10:18:28 | 2018-07-04T10:18:28 | {} | NONE |
ClickHouse | ClickHouse | 402,453,316 | 873 | filimonov | https://clickhouse.yandex/docs/en/roadmap/
In roadmap on Q4 of 2018 (but it's just a roadmap, not a hard schedule). | 2018-07-04T11:38:53 | 2018-07-04T11:38:53 | {
"+1": 7,
"laugh": 1
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 402,561,256 | 2,574 | alexey-milovidov | Performance is Ok.
```
seq 1 1000000 | sed -r -e "s/^.+$/'',hello,'\0'/" > test.csv
for i in {1..100}; do (time ./clickhouse-master local --input-format CSV --structure 'a String, b String, c UInt64' --query="SELECT count() FROM table" < test.csv) 2>&1 | grep user; done | sort | head -n10
for i in {1..100}; d... | 2018-07-04T21:23:47 | 2018-07-04T21:23:47 | {} | MEMBER |
ClickHouse | ClickHouse | 402,569,457 | 2,521 | alexey-milovidov | The values don't look real on this query:
```
┌─CounterID─┬──count()─┬─quantilesTimingIf(0, 0.1, 0.2, 0.3, 0.4, 0.5, 0.6, 0.7, 0.8, 0.9, 1)(SendTiming, greater(SendTiming, 0))─┬─histogramIf(10)(SendTiming, greater(SendTiming, 0))─────────────────────────────────────────────────────────────────────────────────────────... | 2018-07-04T22:59:03 | 2018-07-04T22:59:16 | {} | MEMBER |
ClickHouse | ClickHouse | 402,951,635 | 2,281 | silviucpp | Hello,
Thanks for your feedback ! I'll take a look soon on this and add support for exponential backoff. Should I add the interval in a config setting ? or I can hardcode between 8 and 12 ?
For my personal knowledge can you please elaborate a bit why "It's important to select merges even when there are no writes ... | 2018-07-06T07:29:54 | 2018-07-06T07:29:54 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 401,533,325 | 2,460 | giuliaScalaberni | Hi, I've tried to reproduce this database with engine MergeTree. The problem still exist. The Database is made of two tables, each with only fields varchar type (a part for one type date for the engine). I use 8192 for index granularity. I insert the rows by an ETL flow by mETL software.
In a PostgreSQL db not on the ... | 2018-06-30T10:48:27 | 2018-06-30T10:48:27 | {} | NONE |
ClickHouse | ClickHouse | 401,615,095 | 2,578 | lamberken | @alexey-milovidov, hi, can you help me, thank you.
| 2018-07-01T15:48:19 | 2018-07-01T15:48:19 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 402,451,738 | 873 | dveselov | Any progress on this? | 2018-07-04T11:31:40 | 2018-07-04T11:31:40 | {} | NONE |
ClickHouse | ClickHouse | 402,566,882 | 2,521 | alexey-milovidov | https://en.cppreference.com/w/cpp/types/numeric_limits/min | 2018-07-04T22:25:11 | 2018-07-04T22:25:11 | {} | MEMBER |
ClickHouse | ClickHouse | 402,570,392 | 2,521 | alexey-milovidov | As we preallocate memory for maximum size of histogram, better to limit its size by 256 (or maybe 1000). | 2018-07-04T23:13:18 | 2018-07-04T23:13:18 | {} | MEMBER |
ClickHouse | ClickHouse | 402,814,055 | 2,599 | alexey-milovidov | You have implemented it only for MergeTree, not for Replicated variants. | 2018-07-05T18:33:40 | 2018-07-05T18:33:40 | {} | MEMBER |
ClickHouse | ClickHouse | 402,827,567 | 2,599 | amosbird | Ah, is it straghtforward to apply the merge loop for Replicated variants? | 2018-07-05T19:24:53 | 2018-07-05T19:24:53 | {} | COLLABORATOR |
ClickHouse | ClickHouse | 403,012,703 | 2,602 | ztlpn | That was fixed in #2400
The fix made it into the 1.1.54388 release | 2018-07-06T11:56:43 | 2018-07-06T11:56:43 | {
"+1": 1
} | CONTRIBUTOR |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.