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 | 409,902,013 | 2,790 | alesapin | Sorry, but I cannot reproduce your problem. I got correct merges at the end and begging of month even without optimize:
```
┌───────Date─┬────────────────Time─┬─Path───┬─Env─┬──────────────Value─┬─Version─┐
│ 2018-07-01 │ 2018-07-01 00:00:00 │ x_path │ │ 75.15138059161549 │ 2040 │
│ 2018-07-01 │ 2018-07-01 ... | 2018-08-02T11:59:56 | 2018-08-02T11:59:56 | {} | MEMBER |
ClickHouse | ClickHouse | 409,934,154 | 1,518 | adhamelia | Yeap @blinkov .. it works superb !!! also newer version has histogram function to achieve the same thing. | 2018-08-02T13:54:06 | 2018-08-02T13:54:06 | {} | NONE |
ClickHouse | ClickHouse | 410,193,135 | 2,790 | alesapin | Sorry, but even with your generation algorithm, I cannot reproduce the problem.
I have used this simple script in python https://pastebin.com/mLAeWrv1.
And got this data: https://pastebin.com/x8uq4X3C (distinct times), https://pastebin.com/JPrt0KRU (sample).
| 2018-08-03T09:04:55 | 2018-08-03T09:04:55 | {} | MEMBER |
ClickHouse | ClickHouse | 410,195,011 | 2,800 | alesapin | В этом случае лучше обратиться к меинтейнеру js-библиотеки. Тут врятли смогут помочь, потому что библиотек для разных языков [очень много](https://clickhouse.yandex/docs/ru/interfaces/third-party_client_libraries/). Для уверенности можно попробовать выполнить этот запрос через консольную утилиту `clickhouse-client`. | 2018-08-03T09:12:17 | 2018-08-03T09:12:17 | {} | MEMBER |
ClickHouse | ClickHouse | 410,241,498 | 2,790 | alesapin | I don't think so. Maybe you'll try my script and show resulting data in table? | 2018-08-03T12:41:44 | 2018-08-03T12:41:44 | {} | MEMBER |
ClickHouse | ClickHouse | 410,367,734 | 2,611 | alexey-milovidov | In master. | 2018-08-03T20:28:48 | 2018-08-03T20:28:48 | {} | MEMBER |
ClickHouse | ClickHouse | 410,496,823 | 2,094 | alexey-milovidov | Solved in different way.
https://github.com/yandex/ClickHouse/commit/c7e960b0a9aaa13d3f20898672c75f8fdccec567
https://github.com/yandex/ClickHouse/commit/114725b526a8db1d3baf757ab492d3878f3e50fc
https://github.com/yandex/ClickHouse/commit/5bce1d78b07c02c53f8552dd4337ef75ee8063e4
https://github.com/yandex/ClickHou... | 2018-08-05T05:25:41 | 2018-08-05T05:25:41 | {
"+1": 1
} | MEMBER |
ClickHouse | ClickHouse | 410,788,831 | 1,481 | blinkov | @bobrik can you provide a way to reproduce this? | 2018-08-06T17:35:52 | 2018-08-06T17:35:52 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 411,524,194 | 2,802 | alexey-milovidov | Nevertheless it's a good practice to have separate column with just Date, in addition to DateTime.
Because Date column has many repeated values and it will compress much better. In contrast, DateTime often have pseudorandom lower bits that correspond to seconds.
In total, reading and filtering by Date could be 10... | 2018-08-08T19:29:08 | 2018-08-08T19:30:33 | {} | MEMBER |
ClickHouse | ClickHouse | 409,891,330 | 2,790 | tomasdubec | no errors that look related in either /var/log/clickhouse-server/clickhouse-server.log and /var/log/clickhouse-server/clickhouse-server.err.log
I am in testing phase, so the "flow" was
* create table
* load it with dummy data (retrospectively)
* observe results of aggregation
Inserted about 300 million of rows... | 2018-08-02T11:11:28 | 2018-08-02T11:11:41 | {} | NONE |
ClickHouse | ClickHouse | 409,893,031 | 2,790 | alesapin | Please, try to optimize table with:
`OPTIMIZE TABLE cpu PARTITION 201806 FINAL` | 2018-08-02T11:19:21 | 2018-08-02T11:19:21 | {} | MEMBER |
ClickHouse | ClickHouse | 410,501,886 | 2,787 | zhang2014 | Almost done @alexey-milovidov @KochetovNicolai | 2018-08-05T07:34:51 | 2018-08-05T07:34:51 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 410,615,548 | 2,765 | wawanawna | Hello @alesapin . Do you mean method how to force MV reject insert data?
In our case we mapped MV to disk in RAM and when there is not enough space problems happened.
I may suggest simplest way, change dir with MV access rights, pretty sure we will see same behaviour. | 2018-08-06T07:35:10 | 2018-08-06T07:35:10 | {} | NONE |
ClickHouse | ClickHouse | 411,283,979 | 2,640 | Jingyi244 | +1. | 2018-08-08T04:47:49 | 2018-08-08T04:47:49 | {} | NONE |
ClickHouse | ClickHouse | 411,493,112 | 2,806 | alexey-milovidov | We have added filtering of ODBC connection string (for security reasons, to disallow custom user specified path to `.so` etc), but it appeared to be overly aggressive. https://github.com/yandex/ClickHouse/pull/2772
For example, it wraps all parameter values to `{}` (example: `Database={test}`) but drivers don't unde... | 2018-08-08T17:47:17 | 2018-08-08T17:47:17 | {
"hooray": 2
} | MEMBER |
ClickHouse | ClickHouse | 409,776,448 | 2,766 | lamberken | @alesapin, hi, my point is as follows
first, the difference between ck and other dbs is that when using the `ReplicatedMergeTree` engine, each block will create a znode node. when the data source has a large amount of data, it'll create many znodes at a moment.
second, it can't be controlled on the client side. F... | 2018-08-02T01:33:21 | 2018-08-02T01:33:21 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 409,918,422 | 2,743 | adhamelia | Thanks @alexey-milovidov :+1:
I will do some bench marking and identify the performance gap. | 2018-08-02T13:03:31 | 2018-08-02T13:03:31 | {} | NONE |
ClickHouse | ClickHouse | 410,204,218 | 2,777 | alesapin | External dictionaries reload will be fixed in this pr https://github.com/yandex/ClickHouse/pull/2779.
| 2018-08-03T09:48:31 | 2018-08-03T09:48:31 | {
"+1": 1
} | MEMBER |
ClickHouse | ClickHouse | 410,371,721 | 1,393 | alexey-milovidov | Missing field visitor ToString that is required for information logging in `CollapsingSortingBlockInputStream` for `CollapsingMergeTree` was implement by Šimon Podlipský in #2618. | 2018-08-03T20:46:15 | 2018-08-03T20:46:38 | {} | MEMBER |
ClickHouse | ClickHouse | 410,497,110 | 2,805 | zhang2014 | > Don't understand exactly how it will work? In what cases it would not require to modify the query as in previous point?
I'm not sure if it's necessary, but here's an example:
```
CREATE TABLE merge_tree(date Date, k1 Int32, v1 Int32) ENGINE = MergeTree(date, (k1, date), 8192);
CREATE MATERIALIZED VIEW materia... | 2018-08-05T05:33:13 | 2018-08-05T05:33:13 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 411,538,559 | 2,829 | champtar | workaround for now: quote everything
```
import csv
with open('GeoLite2-City-Locations-en.csv', 'rb') as csvfilein:
with open('GeoLite2-City-Locations-en-quoted.csv', 'w') as csvfileout:
csvreader = csv.DictReader(csvfilein, delimiter=',')
writer = csv.DictWriter(csvfileout, fieldnames=csvread... | 2018-08-08T20:19:16 | 2018-08-08T20:19:16 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 409,906,647 | 2,790 | tomasdubec | `select * from test.cpu_graphite_real where Date between '2018-06-28' and '2018-07-03' and Path='router_1500' order by Time` :
```
┌───────Date─┬────────────────Time─┬─Path────────┬─Env──┬──────────────Value─┬─Version─┐
│ 2018-06-28 │ 2018-06-28 00:00:00 │ router_1500 │ prod │ 2.9559166666666665 │ 1 │
│ 2018-... | 2018-08-02T12:19:15 | 2018-08-02T12:19:15 | {} | NONE |
ClickHouse | ClickHouse | 409,964,406 | 2,447 | Tezd | @alexey-milovidov updated PR according to information from new system tables. | 2018-08-02T15:19:50 | 2018-08-02T15:19:50 | {} | NONE |
ClickHouse | ClickHouse | 410,485,167 | 2,768 | alexey-milovidov | @ztlpn Probably the error comes from mutations after RENAME TABLE. | 2018-08-04T23:52:54 | 2018-08-04T23:52:54 | {} | MEMBER |
ClickHouse | ClickHouse | 410,771,456 | 2,615 | xmariachi | Got it, in the example it's the `queue` table that should free the cgroup when dropped. Tested, it works fine. Thanks. | 2018-08-06T16:40:46 | 2018-08-06T16:40:46 | {} | NONE |
ClickHouse | ClickHouse | 411,523,079 | 2,802 | alexey-milovidov | > Don't know if there is any impact, but if you use only DateTime and not Date, system.parts min_date and max_date is 0
`min_date` and `max_date` columns in `system.parts` are for convenience and don't affect query performance. If you have DateTime column as partition key, the table will track min and max value of t... | 2018-08-08T19:25:10 | 2018-08-08T19:25:22 | {} | MEMBER |
ClickHouse | ClickHouse | 410,418,905 | 2,804 | filimonov | Currently doable with one db per shard.
Check that comment: https://github.com/yandex/ClickHouse/issues/1287#issuecomment-406180820
@alexey-milovidov what do you think about database-dependant macros (described in [comment](https://github.com/yandex/ClickHouse/issues/1287#issuecomment-406180820)) or suffixes? | 2018-08-04T03:16:55 | 2018-08-04T03:16:55 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 410,485,321 | 2,798 | alexey-milovidov | NULLs are supported. This is documentation issue.
You can insert NULLs from JSONEachRow if you have Nullable data type in your table: by writing them as null or by omitting the key and value pair. The latter is treated as inserting default value. | 2018-08-04T23:57:39 | 2018-08-04T23:57:39 | {} | MEMBER |
ClickHouse | ClickHouse | 410,493,875 | 2,777 | alexey-milovidov | Added removing clusters. | 2018-08-05T03:56:19 | 2018-08-05T03:56:19 | {
"+1": 1
} | MEMBER |
ClickHouse | ClickHouse | 409,677,892 | 2,781 | alexey-milovidov | Спасибо! Действительно, так гораздо лучше. | 2018-08-01T18:35:39 | 2018-08-01T18:35:39 | {} | MEMBER |
ClickHouse | ClickHouse | 409,904,209 | 2,792 | alesapin | Currently ClickHouse has problems caused by new versioning. They will be fixed with new releases, some of them are fixed inside [this commit](https://github.com/yandex/ClickHouse/commit/d72ba2701b90541f916d5e030e42617b09d5c6d7). | 2018-08-02T12:09:08 | 2018-08-02T12:09:08 | {} | MEMBER |
ClickHouse | ClickHouse | 409,921,468 | 2,790 | tomasdubec | system.merges is empty (0 rows in set).
there's no point in providing the data, as they are randomly generated. I can describe them:
* Path is "router_1" to "router_10000" incremented by 1
* Time is incremented by 300 (5minutes granularity)
* Env is not important, just some string
* Value is randomly generated f... | 2018-08-02T13:13:41 | 2018-08-02T13:13:41 | {} | NONE |
ClickHouse | ClickHouse | 410,350,151 | 2,804 | champtar | Hi @alexey-milovidov
Erasure coding would be awesome, but I was really thinking about something simpler (my reference to RAID5 might have mislead you), even if shard placement has to be done manually (with ansible ;) )
I see 2 simple (tm) solutions:
1) at the distributed table level
If I call my table `logs`, I ... | 2018-08-03T19:15:30 | 2018-08-03T19:15:30 | {
"+1": 1
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 410,465,352 | 2,691 | Amper | @xmariachi Getting the key is not a problem. But I mean, I didn't know how to work with the key should look like from the point of view of the syntax of creating a table with Kafka engine.
| 2018-08-04T17:33:19 | 2018-08-04T17:33:19 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 410,497,259 | 2,805 | alexey-milovidov | Looks redundant, because
```
ALTER TABLE merge_tree AS SELECT date, k1, toString(v1) FROM merge_tree;
```
will be more obvious. | 2018-08-05T05:37:36 | 2018-08-05T05:37:36 | {} | MEMBER |
ClickHouse | ClickHouse | 410,975,151 | 2,810 | BigBlueWhale0 | Ok. I installed this package:
$ rpm -qa | grep clickhouse
clickhouse-common-static-1.1.54385-1.el7.x86_64
**clickhouse-debuginfo-1.1.54385-1.el7.x86_64**
clickhouse-client-1.1.54385-1.el7.x86_64
clickhouse-server-1.1.54385-1.el7.x86_64
clickhouse-server-common-1.1.54385-1.el7.x86_64
What should I do next? addr... | 2018-08-07T08:20:47 | 2018-08-07T08:21:29 | {} | NONE |
ClickHouse | ClickHouse | 411,013,813 | 2,810 | BigBlueWhale0 | Moreover, If I add addiction flags I get this result:
```# addr2line -sfiCe /usr/lib/debug/usr/bin/clickhouse.debug 0x46c7585 0x46c7ace
DB::ReplacingSortedBlockInputStream::insertRow(std::vector<COWPtr<DB::IColumn>::mutable_ptr<DB::IColumn>, std::allocator<COWPtr<DB::IColumn>::mutable_ptr<DB::IColumn> > >&, unsigne... | 2018-08-07T10:39:50 | 2018-08-07T10:39:50 | {} | NONE |
ClickHouse | ClickHouse | 411,018,734 | 2,813 | 4ertus2 | Fixed in master: 8fde28ded
Thank you for you reply. | 2018-08-07T11:01:22 | 2018-08-07T11:01:22 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 411,335,409 | 1,368 | blinkov | ok, thanks | 2018-08-08T08:52:54 | 2018-08-08T08:52:54 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 409,887,852 | 2,790 | alesapin | Are there any errors in logs for this period of time or maybe later? Have you done any changes with this table like alters, deletes by hand or something like this? Also, provide your ClickHouse version please.
I see that this problem occurs at the last day of the month and the first day of the next month. Did this r... | 2018-08-02T10:55:48 | 2018-08-02T11:03:45 | {} | MEMBER |
ClickHouse | ClickHouse | 410,367,470 | 2,618 | alexey-milovidov | The implementation is still incomplete, but it is better than nothing and now #2611 is fixed. | 2018-08-03T20:27:41 | 2018-08-03T20:27:41 | {} | MEMBER |
ClickHouse | ClickHouse | 410,738,743 | 2,581 | VadimPlh | Не думаю, что баг смог сам пройти) Буду благодарен, если расскажете, проявился ли баг опять при переходе на новую партицию. | 2018-08-06T15:01:22 | 2018-08-06T15:01:22 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 410,780,143 | 2,810 | BigBlueWhale0 | Thank you for reply
Unfortunately, I get that result:
```
$ addr2line -iCe /usr/bin/clickhouse 0x46c7585 0x46c7ace
??:?
??:?
``` | 2018-08-06T17:09:12 | 2018-08-06T17:09:12 | {} | NONE |
ClickHouse | ClickHouse | 411,100,910 | 2,809 | simPod | Maybe resolved by https://github.com/yandex/ClickHouse/pull/2618 as it's not happening to other types. | 2018-08-07T15:37:04 | 2018-08-07T15:37:04 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 411,279,586 | 2,795 | alexey-milovidov | Fixed in master. | 2018-08-08T04:19:30 | 2018-08-08T04:19:30 | {} | MEMBER |
ClickHouse | ClickHouse | 411,284,001 | 2,818 | alexey-milovidov | Ok. The only drawback is performance.
```
SELECT count() FROM hits_1000m WHERE NOT ignore(formatDateTime(EventTime, '%m/%d/%Y %H:%M:%S'))
```
33.70 million rows/s.
(It is just 2.46 million rows/s. per single core, that is painful slow.)
```
SELECT count() FROM hits_1000m WHERE NOT ignore(toString(toYear(Ev... | 2018-08-08T04:47:59 | 2018-08-08T04:49:43 | {} | MEMBER |
ClickHouse | ClickHouse | 409,894,217 | 2,790 | tomasdubec | OK, I did..no change. Should this work retrospectively? or only for newly aggregated data? | 2018-08-02T11:24:23 | 2018-08-02T11:24:23 | {} | NONE |
ClickHouse | ClickHouse | 410,239,855 | 2,790 | tomasdubec | Weird...so could it be the amount of data I generated? Seems to be the only difference. My testing consisted of such generated random data, for past 6 months.. | 2018-08-03T12:34:40 | 2018-08-03T12:34:40 | {} | NONE |
ClickHouse | ClickHouse | 410,262,320 | 2,802 | alesapin | No it doesn't. You have tried the new syntax https://clickhouse.yandex/docs/en/operations/table_engines/custom_partitioning_key/ with custom partitioning ability. | 2018-08-03T13:58:35 | 2018-08-03T13:58:35 | {} | MEMBER |
ClickHouse | ClickHouse | 410,304,978 | 2,798 | den-crane | try to omit null attributes, JSONEachRow treat {} the same as {"voucher":null} | 2018-08-03T16:22:46 | 2018-08-03T16:22:46 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 403,788,280 | 2,618 | simPod | 1) it's implemented in https://github.com/yandex/ClickHouse/blob/master/dbms/src/Functions/FunctionsCoding.h#L893, couldn't it be somehow leveraged?
2) hm, I'm afraid I'm out of skill for that, first time editing C code. Maybe someone can cherrypick and continue the work here? | 2018-07-10T11:19:23 | 2018-08-05T13:12:06 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 410,660,026 | 2,581 | vasyaabr | Сейчас мне тоже не удалось воспроизвести эту ошибку, более того, теперь при консистентном чтении в результатах выборки снова данные всех партиций (т.е. ошибка через некоторое продолжительное время (несколько дней) излечилась сама).
Текущая версия сервера: 18.4.54399 (обновились)
Сегодня-завтра у нас произойдет п... | 2018-08-06T10:16:13 | 2018-08-06T10:16:13 | {} | NONE |
ClickHouse | ClickHouse | 410,806,618 | 2,810 | alexey-milovidov | To get the result you should also install `clickhouse-debuginfo` package https://packagecloud.io/altinity/clickhouse | 2018-08-06T18:30:06 | 2018-08-06T18:30:06 | {} | MEMBER |
ClickHouse | ClickHouse | 411,335,203 | 1,368 | jaykelin | @blinkov not appeared in version v1.1.54342 | 2018-08-08T08:52:01 | 2018-08-08T08:52:01 | {} | NONE |
ClickHouse | ClickHouse | 411,494,006 | 2,806 | alexey-milovidov | According to git log, it is not in 18.5.1. Minimum version with regression is 18.6.0. | 2018-08-08T17:50:14 | 2018-08-08T17:50:14 | {} | MEMBER |
ClickHouse | ClickHouse | 409,937,774 | 2,795 | SaltTan | This one works:
drop table if exists test_mv_group;
create MATERIALIZED VIEW test_mv_group (x String)
ENGINE = MergeTree Partition by tuple() order by tuple()
AS SELECT x FROM test.test group by x;
select * from test_mv_group
┌─x─────┐
│ stest │
└───────┘
1 rows in set. Elapsed: 0.001 sec.
| 2018-08-02T14:04:23 | 2018-08-02T14:04:23 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 410,165,426 | 2,691 | xmariachi | Kafka Messages have a Key-Value format.
In many cases, where there is no need for a key, the message is sent in the value and the key is left null.
But that is not necessarily the case.
So, I understand the Kafka Engine tries to read from the value only, and not the key, to deserialize the message as a record into... | 2018-08-03T07:08:26 | 2018-08-03T07:10:42 | {} | NONE |
ClickHouse | ClickHouse | 410,167,982 | 2,799 | alesapin | Hi,
these settings should be specified in `users.xml` for each user (or in default section), and you should connect to ClickHouse with the same user. From example config:
```xml
<yandex>
<!-- Profiles of settings. -->
<profiles>
<!-- Default settings. -->
<default>
<!-- Maxim... | 2018-08-03T07:20:48 | 2018-08-03T09:05:54 | {
"+1": 2
} | MEMBER |
ClickHouse | ClickHouse | 410,497,569 | 2,805 | zhang2014 | Ok, I understand | 2018-08-05T05:46:47 | 2018-08-05T05:46:47 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 410,784,895 | 2,782 | alexey-milovidov | You forgot about config.xml | 2018-08-06T17:23:39 | 2018-08-06T17:23:39 | {} | MEMBER |
ClickHouse | ClickHouse | 411,437,769 | 2,827 | den-crane | Though
```
SELECT *
FROM y
PREWHERE a < 1000
┌─a─┬─b─┬──────────z─┬─c──┬─d──┐
│ 1 │ 0 │ 2018-01-01 │ [] │ [] │
└───┴───┴────────────┴────┴────┘
┌─a─┬─b─┬──────────z─┬─c──┬─d──┐
│ 2 │ 0 │ 2018-02-01 │ [] │ [] │
└───┴───┴────────────┴────┴────┘
``` | 2018-08-08T15:01:42 | 2018-08-08T15:01:42 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 411,488,286 | 2,802 | champtar | Don't know if there is any impact, but if you use only `DateTime` and not `Date`, `system.parts` `min_date` and `max_date` is 0
```
DESCRIBE TABLE testdatetime
┌─name───────┬─type─────────────┬─default_type─┬─default_expression─┐
│ datetime │ DateTime │ │ │
│ clientipv... | 2018-08-08T17:31:32 | 2018-08-08T17:31:32 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 411,519,189 | 1,567 | champtar | PR #2830 | 2018-08-08T19:11:15 | 2018-08-08T19:11:15 | {
"+1": 2
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 409,831,448 | 2,775 | fizerkhan | @vpanfilov Thanks for the detailed answer. I can understand about Array functions.
My question is about better pattern for ClickHouse. Let me give an example:
I have a route `/Checkout` and I am going to write `response time` of the this route to ClickHouse.
I will get billions of events per month.
My Schema ... | 2018-08-02T07:21:17 | 2018-08-02T07:22:20 | {} | NONE |
ClickHouse | ClickHouse | 410,367,648 | 1,770 | alexey-milovidov | In master. | 2018-08-03T20:28:25 | 2018-08-03T20:28:25 | {} | MEMBER |
ClickHouse | ClickHouse | 410,626,649 | 2,790 | tomasdubec | I tried your code and got https://pastebin.com/kvxAUnhR which looks ok.
One observation - I was loading the data in a bit different way. I generated CSV and pasted it into clickhouse-client like this:
```./gen | clickhouse-client --query="INSERT INTO cpu FORMAT CSV"```
(which was way faster then spawning new click... | 2018-08-06T08:18:13 | 2018-08-06T08:18:13 | {} | NONE |
ClickHouse | ClickHouse | 411,068,882 | 2,811 | lamberken | Hi, reduce the parallelism of sink which inserts datas into CK server. | 2018-08-07T14:07:18 | 2018-08-07T14:07:18 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 409,884,229 | 2,786 | zhang2014 | I am currently trying to fix this problem, but it is not compatible with the version. | 2018-08-02T10:40:03 | 2018-08-02T10:40:03 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 410,476,692 | 2,805 | alexey-milovidov | That's exactly what we need!
> Support CREATE OR REPLACE VIEW Syntax
Should be rather straightforward. Only have to take care on locking (operation should be atomic from the view of client). It's ok if it will still be non-atomic in case of unexpected errors in the middle.
> Support ALTER VIEW view_name() AS n... | 2018-08-04T20:48:42 | 2018-08-04T20:51:06 | {} | MEMBER |
ClickHouse | ClickHouse | 410,506,987 | 2,615 | zhang2014 | This should be expected.
The consumer table is not the `kafka consumer`, Because we have a group name in the `kafka engine`. So it's correct to remove `kafka consumer` only when the `kafka engine` is dropped. | 2018-08-05T09:19:40 | 2018-08-05T09:19:40 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 410,974,004 | 1,468 | blinkov | @bobrik can you provide a way to reproduce this? | 2018-08-07T08:16:35 | 2018-08-07T08:16:35 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 411,099,629 | 2,820 | blinkov | You need to set the readonly setting of this user to value 2. Such user still won't be able to change data in tables, but clients which try to silently override settings will start working.
See https://clickhouse.yandex/docs/en/operations/settings/query_complexity/#readonly | 2018-08-07T15:33:30 | 2018-08-07T15:33:30 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 411,516,513 | 2,802 | simPod | Yes, it's mentioned in this issue https://github.com/yandex/ClickHouse/issues/2362 | 2018-08-08T19:01:49 | 2018-08-08T19:01:49 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 409,733,444 | 2,775 | vpanfilov | Read these section of documentation:
- https://clickhouse.yandex/docs/en/query_language/agg_functions/combinators/
- https://clickhouse.yandex/docs/en/query_language/functions/higher_order_functions/
There are many ways for working with arrays in Clickhouse: using aggregating functions with `-Array` combinator, ... | 2018-08-01T21:41:08 | 2018-08-01T21:51:03 | {} | NONE |
ClickHouse | ClickHouse | 409,839,276 | 2,775 | vpanfilov | Clickhouse compress data by default, so de-normalized scheme should work good enough in terms of disk space. But you should carefully choose the primary key for sorting rows in MergeTree engine (table is stored in pieces sorted by PK).
I don't know if there is any benchmarks for storing metrics in arrays. Many peopl... | 2018-08-02T07:52:59 | 2018-08-02T07:52:59 | {} | NONE |
ClickHouse | ClickHouse | 409,848,487 | 2,766 | alesapin | Unfortunately the answer is no, ClickHouse doesn't have this setting. | 2018-08-02T08:27:50 | 2018-08-02T08:27:50 | {} | MEMBER |
ClickHouse | ClickHouse | 410,048,997 | 2,787 | alexey-milovidov | Could you please also provide a setting to turn on old behaviour for compatibility?
Something like `asterisk_left_columns_only` (= 0 by default)
| 2018-08-02T19:57:29 | 2018-08-02T19:57:29 | {} | MEMBER |
ClickHouse | ClickHouse | 410,275,249 | 2,786 | TimonKK | [It](https://github.com/yandex/ClickHouse/issues/2803) is same problem | 2018-08-03T14:42:26 | 2018-08-03T14:42:26 | {} | NONE |
ClickHouse | ClickHouse | 410,338,530 | 2,804 | alexey-milovidov | "Erasure coding". It's not implemented.
Main reason is implementation complexity.
Things to consider:
- modern reconstruction codes like [LRC](https://github.com/baishancloud/lrc-erasure-code) require to use multiple chunks (example: 12) - it requires sophisticated methods to spread chunks across servers;
- it's ... | 2018-08-03T18:28:31 | 2018-08-03T18:28:31 | {} | MEMBER |
ClickHouse | ClickHouse | 410,770,303 | 2,691 | xmariachi | Suggestion: When referencing the fields, you could reference
`key.field` or `value.field`.
e.g. define fields like `key.field as field, value.otherValue as otherValue`.
For backwards compatibility, not specifying a prefix would by default mean `value.field`. | 2018-08-06T16:36:42 | 2018-08-06T16:36:42 | {} | NONE |
ClickHouse | ClickHouse | 411,564,464 | 2,802 | champtar | If you are already partitioned by day, Date should be useless (but maybe it's not right now, haven't tested) | 2018-08-08T21:52:43 | 2018-08-08T21:52:43 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 410,477,645 | 2,804 | alexey-milovidov | > what do you think about database-dependant macros (described in comment) or suffixes?
If I understand correctly, we may introduce special `{database}` macro, that will always expand to database name. And you can use something like `r1-{database}` as replica name.
Sounds Ok. | 2018-08-04T21:06:22 | 2018-08-04T21:06:22 | {} | MEMBER |
ClickHouse | ClickHouse | 411,119,135 | 2,820 | stiv-yakovenko | thank you this helps
| 2018-08-07T16:29:52 | 2018-08-07T16:29:52 | {} | NONE |
ClickHouse | ClickHouse | 411,311,057 | 2,821 | honavar-sohan | English Translation from google translate
The kafka table created with the clickhouse 390 version, json format. The upgraded new version is not supported, and the separator of the kafka message is added, resulting in the original kakfk table being unavailable after the upgrade. | 2018-08-08T07:18:53 | 2018-08-08T07:18:53 | {} | NONE |
ClickHouse | ClickHouse | 411,559,850 | 2,829 | alexey-milovidov | For cached dictionaries, settings should override query settings. | 2018-08-08T21:34:08 | 2018-08-08T21:34:08 | {} | MEMBER |
ClickHouse | ClickHouse | 410,421,780 | 2,775 | filimonov | > In Schema 1, I will write every single route access as single row in ClickHouse.
> In Schema 2, I will batch the 100 route access(response times in array) in a single row.
If you will actively use that column, and most of heavy selects will require ARRAY JOIN - just prefer Scheme 1. Simple integers are more effe... | 2018-08-04T04:10:31 | 2018-08-04T04:10:31 | {
"+1": 1
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 410,497,592 | 1,828 | alexey-milovidov | Merged.
@proller You've failed. | 2018-08-05T05:47:36 | 2018-08-05T05:47:36 | {} | MEMBER |
ClickHouse | ClickHouse | 410,502,460 | 2,733 | zhang2014 | Any questions about that ?
If not, I will complete a preliminary implementation as soon as possible.
The code may be more conducive to discussion :) | 2018-08-05T07:46:53 | 2018-08-05T07:46:53 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 411,146,348 | 2,806 | The-Alchemist | Can confirm I see this problem in 18.6 as well, and downgrading to 18.1.0 fixes it.
After some testing, I didn't have to downgrade all the way to 18.1.0; downgrading to 18.5.1 was enough.
I'm using the Docker images, so it's possible that 18.5.1 and 18.6 use different base images, so may have different versions o... | 2018-08-07T17:57:33 | 2018-08-07T17:57:33 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 409,855,086 | 2,778 | dqminh | @alesapin @alexey-milovidov its unfortunate. I will rethink our upgrade path then. It would also be great to know where exactly this bug is coming from (i havent seen anything similar in changelog or issue tracker) | 2018-08-02T08:51:55 | 2018-08-02T08:51:55 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 409,872,645 | 2,786 | alesapin | Yes, this is weird, but this behaviour is [documented](https://clickhouse.yandex/docs/en/query_language/select/#join-clause) and works according to documentation:
> All columns that are not needed for the JOIN are deleted from the subquery.
`subquery` here is right part of join. Currently you have to specify all ... | 2018-08-02T09:54:04 | 2018-08-02T09:54:04 | {} | MEMBER |
ClickHouse | ClickHouse | 410,193,592 | 2,765 | alesapin | Maybe you'll provide any method to reproduce the problem? | 2018-08-03T09:06:42 | 2018-08-03T09:06:42 | {} | MEMBER |
ClickHouse | ClickHouse | 410,196,105 | 2,792 | alesapin | Also, you got 18.4 instead 18.1 inside docker image because of accidental error in release cycle. This issue applies only to the 18.1 image. | 2018-08-03T09:16:44 | 2018-08-03T09:16:44 | {} | MEMBER |
ClickHouse | ClickHouse | 410,305,526 | 2,802 | alexey-milovidov | Need to update the docs. | 2018-08-03T16:24:40 | 2018-08-03T16:24:40 | {
"+1": 1
} | MEMBER |
ClickHouse | ClickHouse | 410,359,659 | 2,779 | alexey-milovidov | You should add integration test (or at least estimate how hard it would be to add integration test for this feature). | 2018-08-03T19:54:52 | 2018-08-03T19:54:52 | {} | MEMBER |
ClickHouse | ClickHouse | 410,974,543 | 1,368 | blinkov | @jaykelin so have this issue still persisted on newer ClickHouse builds? | 2018-08-07T08:18:31 | 2018-08-07T08:18:31 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 411,559,524 | 2,829 | alexey-milovidov | It can be solved if we will allow to specify custom settings for dictionaries:
```
<dictionaries>
<dictionary>
<name>geoname</name>
...
<settings>
...
```
It is not implemented but it looks quite easy to do. | 2018-08-08T21:32:53 | 2018-08-08T21:32:53 | {
"+1": 1
} | MEMBER |
ClickHouse | ClickHouse | 409,915,021 | 2,790 | alesapin | Version is written by user. It is used for automatic deletion of values at the same time.
> Version -- Determines which value of the metric with the same Path and Time will remain in the database.
In your output even 300 seconds interval aggregation doesn't work. It is very strange.
Can you provide
`select ... | 2018-08-02T12:51:01 | 2018-08-02T12:51:01 | {} | MEMBER |
ClickHouse | ClickHouse | 410,631,882 | 2,581 | VadimPlh | Здравствуйте, а Вы не могли бы, пожалуйста, рассказать чуть подробнее (интересует порядок ваших действий). Повторить эту багу мне пока не удалось. Спасибо! | 2018-08-06T08:37:13 | 2018-08-06T08:37:13 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 410,767,486 | 2,810 | alexey-milovidov | Could you please run:
`addr2line -iCe /usr/bin/clickhouse 0x46c7585 0x46c7ace` | 2018-08-06T16:27:47 | 2018-08-06T16:27:47 | {} | MEMBER |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.