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 | 411,691,982 | 2,818 | alexey-milovidov | Just run: `sudo perf top`
(To install run `sudo apt-get install linux-tools-$(uname -r)`)
http://www.brendangregg.com/perf.html | 2018-08-09T09:12:21 | 2018-08-09T09:12:21 | {
"+1": 1
} | MEMBER |
ClickHouse | ClickHouse | 411,954,444 | 2,821 | zhang2014 | This should be a version-incompatible feature, but is not recorded in the change log @alexey-milovidov
https://github.com/yandex/ClickHouse/blob/master/CHANGELOG.md#new-features-1
```
Support for adding arbitrary message separators when reading from Kafka Amos Bird.
``` | 2018-08-10T02:14:04 | 2018-08-10T02:14:04 | {
"+1": 1
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 412,586,758 | 2,856 | alexey-milovidov | @filimonov This pull request just filters content of system tables, that contain information about another tables. All system tables remain available, but only with information about tables that the user is allowed to see. | 2018-08-13T16:51:41 | 2018-08-13T16:51:41 | {} | MEMBER |
ClickHouse | ClickHouse | 412,587,240 | 2,856 | alexey-milovidov | @zhang2014 We can't filter system.processes, system.query_log, because they don't contain information about database. Even more, the query can use multiple databases. Lets just leave it as is. | 2018-08-13T16:53:14 | 2018-08-13T16:53:14 | {} | MEMBER |
ClickHouse | ClickHouse | 412,851,941 | 2,828 | alesapin | Single failed test in Autotests and Travis is
```
00687_top_and_offset:
Code: 406. DB::Exception: Can not use TOP and LIMIT together
``` | 2018-08-14T12:17:02 | 2018-08-14T12:30:16 | {} | MEMBER |
ClickHouse | ClickHouse | 413,830,391 | 2,883 | 4ertus2 | I don't think 'SELECT * FROM target' is a good solution on big data. You probably have a date column with partitioning on it. Would it be enough to update just a time interval including some old records? If so, it should be much faster. Cause if you have partitioned table in ClickHouse and date predicate in select it w... | 2018-08-17T10:56:10 | 2018-08-17T10:56:10 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 414,161,255 | 2,890 | alexey-milovidov | We can implement `system.licenses` table with the list of components that was actually linked into the server. This is like to `about:license` page in Firefox and `chrome://credits` in Chromium.
Also it will be nice to have `system.contributors` with the list of all contributors from the git log. | 2018-08-19T22:42:54 | 2018-08-19T22:42:54 | {} | MEMBER |
ClickHouse | ClickHouse | 412,072,979 | 2,156 | Ouriel | This post seems related and add some info:
https://stackoverflow.com/questions/50350466/distinct-select-with-where-filter-on-clickhouse-distributed-table-failed/51786576#51786576 | 2018-08-10T12:51:44 | 2018-08-10T12:51:44 | {} | NONE |
ClickHouse | ClickHouse | 412,520,809 | 2,858 | hashbash | Сейчас у нас уже есть отдельный кластер для аналитиков.
Но здесь есть две проблемы для аналитиков:
- запросы работают медленнее
- большие словари не помещаются в память
Обе проблемы следствие того, что в аналитическом кластере меньше железок, а дублировать мощности не хочется. Отсюда и необходимость такой фичи. | 2018-08-13T13:37:58 | 2018-08-13T14:12:59 | {} | NONE |
ClickHouse | ClickHouse | 412,072,340 | 2,156 | Ouriel | I have the same or a similar issue on 1.1.54388
I can't post the full table definition but it's a distributed table with userId as sharding key (with cityhash function) and date as partition for the mergeTree table under.
```
SELECT DISTINCT userId
FROM test.events
WHERE date >= '2018-07-15'
LIMIT 1
Receive... | 2018-08-10T12:48:58 | 2018-08-18T13:07:07 | {} | NONE |
ClickHouse | ClickHouse | 411,864,846 | 2,821 | amosbird | Yeah, that update turns
```
Kafka(kafka_broker_list, kafka_topic_list, kafka_group_name, kafka_format
[, kafka_schema, kafka_num_consumers])
```
into
```
Kafka(kafka_broker_list, kafka_topic_list, kafka_group_name, kafka_format
[, kafka_row_delimiter, kafka_schema, kafka_num_consumers])
```
... | 2018-08-09T19:09:38 | 2018-08-09T19:09:38 | {} | COLLABORATOR |
ClickHouse | ClickHouse | 412,094,003 | 2,828 | alesapin | I don't understand why travis fails. Seems that the main reason is internal timeouts. | 2018-08-10T14:10:18 | 2018-08-10T14:11:28 | {} | MEMBER |
ClickHouse | ClickHouse | 412,118,537 | 2,840 | zhang2014 | So why do we need it? I'm just curious : ) | 2018-08-10T15:29:57 | 2018-08-10T15:31:54 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 412,124,940 | 2,840 | VadimPlh | 
| 2018-08-10T15:50:36 | 2018-08-10T15:50:36 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 412,591,035 | 2,856 | alexey-milovidov | @zhang2014 The only missing part is integration test. | 2018-08-13T17:04:43 | 2018-08-13T17:04:43 | {} | MEMBER |
ClickHouse | ClickHouse | 412,817,217 | 2,862 | alexey-milovidov | Thank you!
> Supports only those time zones that never had the time differ from UTC for a partial number of hours (without leap seconds) over the entire time range you will be working with.
BTW, this is not true, we actually have support for time zones with fractional offset. | 2018-08-14T09:49:08 | 2018-08-14T09:49:08 | {} | MEMBER |
ClickHouse | ClickHouse | 413,565,287 | 2,876 | konkov-alexey | yep, by default php csv-parser use backslash as escape symbol instead of double-quote. changing escape symbol to double quote solves the problem.
===
по-умолчанию пхп-парсер использует эскейп-символом обратный слэш. если поставить эскейп-символом двойные кавычки, то всё работает корректно. | 2018-08-16T14:28:33 | 2018-08-16T14:28:33 | {} | NONE |
ClickHouse | ClickHouse | 414,185,728 | 2,744 | alexey-milovidov | It doesn't work if you have another comment after query:
```
SELECT x; -- { serverError 47 }
-- Hello, world
SELECT 1;
``` | 2018-08-20T03:11:03 | 2018-08-20T03:11:03 | {} | MEMBER |
ClickHouse | ClickHouse | 414,189,056 | 2,892 | alexey-milovidov | The list of external files is tracked and updated on the fly. | 2018-08-20T03:39:59 | 2018-08-20T03:39:59 | {} | MEMBER |
ClickHouse | ClickHouse | 413,091,806 | 2,873 | zhang2014 | details of error: https://travis-ci.org/yandex/ClickHouse/jobs/416204943#L1889 | 2018-08-15T04:35:25 | 2018-08-15T04:35:25 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 414,185,981 | 2,744 | alexey-milovidov | ```
size_t pos = query.find("--");
```
And this is absolutely wrong:
consider `SELECT '--'` | 2018-08-20T03:13:04 | 2018-08-20T03:13:04 | {} | MEMBER |
ClickHouse | ClickHouse | 412,182,512 | 2,848 | proller | because current method with type parsing does not work n complex queries like
`SELECT *, (CASE WHEN (number == 1) THEN 'o' WHEN (number == 2) THEN 'two long string' WHEN (number == 3) THEN 'r' WHEN (number == 4) THEN NULL ELSE '-' END) FROM system.numbers LIMIT 6` | 2018-08-10T19:29:30 | 2018-08-10T19:30:05 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 413,631,910 | 2,883 | DeanThompson | I had tried the following SQL
```sql
SELECT target.* FROM target ALL LEFT JOIN staging USING id WHERE staging.id IS NULL
UNION ALL
SELECT * FROM staging
```
But ClickHouse said
```
Received exception from server (version 18.6.0):
Code: 47. DB::Exception: Received from localhost:9000, 127.0.0.1. DB:... | 2018-08-16T17:57:03 | 2018-08-16T17:57:03 | {} | NONE |
ClickHouse | ClickHouse | 413,770,850 | 2,883 | DeanThompson | I found a work around, but wondering if there is a better solution.
```sql
SELECT *
FROM target
WHERE NOT (id IN
(
SELECT id
FROM staging
))
UNION ALL
SELECT *
FROM staging
``` | 2018-08-17T06:29:13 | 2018-08-17T06:29:13 | {} | NONE |
ClickHouse | ClickHouse | 414,244,257 | 2,887 | amosbird | Hmm, isn't this doable by CRDT tricks? a simple test shows that the plain SQL routine outperforms this UDAF.
```
dell123 :) SELECT count(x), countIf(y, abs(y) < 1 ) FROM (SELECT sum(1) x, sum( (date = '2018-08-06') * 2 - 1 ) / x y FROM retention_test WHERE date IN ('2018-08-06', '2018-08-08') GROUP BY uid);
SE... | 2018-08-20T08:49:45 | 2018-08-20T08:49:45 | {
"+1": 2
} | COLLABORATOR |
ClickHouse | ClickHouse | 414,250,450 | 2,892 | pavel-odintsov | Hello!
Wow! That's awesome! It works for me! Thank you! | 2018-08-20T09:12:01 | 2018-08-20T09:12:10 | {} | NONE |
ClickHouse | ClickHouse | 411,956,016 | 2,821 | zhang2014 | Sorry, I didn't see the instructions
https://github.com/yandex/ClickHouse/blob/master/CHANGELOG.md#backward-incompatible-changes | 2018-08-10T02:24:07 | 2018-08-10T02:24:07 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 412,039,760 | 2,738 | blinkov | @den-crane do you have any further questions? | 2018-08-10T10:09:38 | 2018-08-10T10:09:38 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 412,155,648 | 2,841 | alexey-milovidov | :+1:
A few notes:
- commits were squashed because there are commits from previous PRs; it's better to merge with upstream before PR;
- note that each table can have individual settings (SETTINGS clause in CREATE TABLE) and currently there is still no way to show (except `SHOW CREATE TABLE`) or change (except manu... | 2018-08-10T17:45:00 | 2018-08-10T17:45:00 | {} | MEMBER |
ClickHouse | ClickHouse | 412,500,167 | 2,857 | filimonov | Same for FULL OUTER and RIGHT joins.
https://github.com/yandex/ClickHouse/blob/master/dbms/tests/queries/0_stateless/00203_full_join.sql | 2018-08-13T12:23:47 | 2018-08-13T12:23:47 | {
"+1": 2
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 412,558,909 | 2,856 | filimonov | I think system database should be generally available for all the users - as it contain system.number, system.one and other very handy tables. Only few tables can be considered as potentially unsafe: system.query_log, system.zookeeper. System.processes can be limited to current user only for users with limited access. | 2018-08-13T15:28:46 | 2018-08-13T15:28:46 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 413,299,051 | 2,231 | MasterGroosha | Can reproduce on 18.6.0 and 18.10.0. Any news on this? | 2018-08-15T18:55:40 | 2018-08-16T07:25:58 | {} | NONE |
ClickHouse | ClickHouse | 413,987,538 | 2,890 | filimonov | > Please clarify the requirement according to specific licenses (and quote the definition of "final software distibution").
Generally most of licenses just force you to add copyright notice to "all copies or substantial portions of Software". So if you distribute some software in binary form (separate from sources) ... | 2018-08-17T21:03:04 | 2018-08-17T21:21:48 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 414,188,734 | 2,892 | alexey-milovidov | Sure. You can add files for each user inside `/etc/clickhouse-server/users.d` directory.
| 2018-08-20T03:36:55 | 2018-08-20T03:36:55 | {} | MEMBER |
ClickHouse | ClickHouse | 414,188,968 | 2,892 | alexey-milovidov | ```
$ cat /etc/clickhouse-server/users.d/alice.xml
<yandex>
<users>
<alice>
<profile>analytics</profile>
<networks>
<ip>::/0</ip>
</networks>
<password_sha256_hex>...</password_sha256_hex>
<quota>analytics</quota>
</alice... | 2018-08-20T03:38:58 | 2018-08-20T03:39:10 | {
"+1": 1
} | MEMBER |
ClickHouse | ClickHouse | 412,126,413 | 2,840 | VadimPlh | @zhang2014
For TOP ... WITH TIES | 2018-08-10T15:55:38 | 2018-08-10T15:55:38 | {
"+1": 1
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 412,513,269 | 2,858 | blinkov | На данный момент см. последний абзац в https://clickhouse.yandex/docs/ru/operations/access_rights/
Для таких аналитиков можно завести отдельный небольшой кластер с readwrite доступом и настроить там distributed таблицы, которые бы «смотрели» на основной кластер. Заодно это обезопасит основной кластер от неожиданно з... | 2018-08-13T13:13:39 | 2018-08-13T13:13:39 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 412,821,943 | 2,581 | VadimPlh | #2863 | 2018-08-14T10:07:17 | 2018-08-14T10:07:17 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 413,715,822 | 1,661 | Slind14 | @blinkov do you have any idea/suggestion on how we can improve this? Sampling doesn't work either because the joined table is still massive. | 2018-08-16T23:39:22 | 2018-08-16T23:39:22 | {} | NONE |
ClickHouse | ClickHouse | 413,985,264 | 2,890 | filimonov | | | | |
|-----------------------------|-------------------------------------... | 2018-08-17T20:52:45 | 2018-08-17T20:52:45 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 414,373,491 | 2,888 | alexey-milovidov | Now it is better:
```
milovidov-Pro-P30 :) CREATE TABLE test.null (x UInt8) ENGINE = Null
CREATE TABLE test.null
(
x UInt8
)
ENGINE = Null
Ok.
0 rows in set. Elapsed: 0.012 sec.
milovidov-Pro-P30 :) INSERT INTO test.null SELECT NULL
INSERT INTO test.null SELECT NULL
Received exception fro... | 2018-08-20T16:11:36 | 2018-08-20T16:11:36 | {} | MEMBER |
ClickHouse | ClickHouse | 412,056,366 | 2,840 | VadimPlh | @zhang2014 You are right) It seems like a `LIMIT`. It's the new syntax | 2018-08-10T11:32:06 | 2018-08-10T11:33:55 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 412,151,307 | 2,840 | derekperkins | Does this solve this issue in https://github.com/yandex/ClickHouse/issues/2198? | 2018-08-10T17:29:19 | 2018-08-10T17:29:19 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 412,447,791 | 2,855 | sundy-li | Such as
```
SELECT quantileExactIf(number, number > 100)
FROM numbers(90)
┌─quantileExactIf(number, greater(number, 100))─┐
│ 0 │
└───────────────────────────────────────────────┘
SELECT quantileExactIf(toFloat64(number), number > 100)
FROM numbers(90)
┌─quanti... | 2018-08-13T08:39:47 | 2018-08-13T08:40:39 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 412,621,189 | 2,738 | den-crane | How it looks on production now:
```
┌─formatReadableSize(sum(data_compressed_bytes))─┬─formatReadableSize(sum(data_uncompressed_bytes))─┐
│ 703.49 MiB │ 2.41 GiB │
└────────────────────────────────────────────────┴────────────────────────... | 2018-08-13T18:39:53 | 2018-08-13T18:40:09 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 413,881,900 | 2,888 | 4ertus2 | There's no such information in context of error.
I've looked at the code and do not see an easy way to make error more informative. | 2018-08-17T14:22:36 | 2018-08-17T14:22:36 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 413,960,582 | 2,890 | alexey-milovidov | Please clarify the requirement according to specific licenses (and quote the definition of "final software distibution"). | 2018-08-17T19:03:33 | 2018-08-17T19:03:33 | {} | MEMBER |
ClickHouse | ClickHouse | 411,736,277 | 2,502 | mudarisov566 | Up! I have this is problem too! | 2018-08-09T12:12:58 | 2018-08-09T12:12:58 | {} | NONE |
ClickHouse | ClickHouse | 411,732,033 | 2,818 | bocharov | @alexey-milovidov I usually use CLion and it uses debug build by default.
I've built in release mode and here are the numbers:
Just datetime components, without concat: 75.79 million rows/s.
```
select count() from (select toString(toYear(t)), toString(toMonth(t)), toString(toDayOfMonth(t)), toString(toHour(t)), ... | 2018-08-09T11:55:03 | 2018-08-10T15:14:09 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 412,190,650 | 2,848 | alexey-milovidov | Incorrect. | 2018-08-10T20:02:59 | 2018-08-10T20:02:59 | {} | MEMBER |
ClickHouse | ClickHouse | 412,371,950 | 2,853 | blinkov | Likely it was just some inactive parts being deleted in background, try adding `...where active` to that query. | 2018-08-12T21:10:29 | 2018-08-12T21:10:29 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 412,427,228 | 2,853 | fizerkhan | @blinkov I see following definitions in the doc
```
active (UInt8) – Indicates whether the part is active. If a part is active, it is used in a table;
otherwise, it will be deleted. Inactive data parts remain after merging.
```
When will the part become active or inactive?
In ou... | 2018-08-13T07:07:55 | 2018-08-13T07:07:55 | {} | NONE |
ClickHouse | ClickHouse | 413,558,055 | 2,876 | konkov-alexey | 1. create table Temp (date Date, id Int8, s String, i Int8) ENGINE = MergeTree(date, (id, date) , 8192);
2. insert into Temp values(today(), 1, 'some text\\', 0);
3. select * from Temp;
`2018-08-16 │ 1 │ some text\\ │ 0`
4. select * from Temp format CSV;
`"2018-08-16",1,"some text\",0`
and result of selecting ... | 2018-08-16T14:07:50 | 2018-08-16T14:10:33 | {} | NONE |
ClickHouse | ClickHouse | 414,253,191 | 2,887 | zhang2014 | > Hmm, isn't this doable by CRDT tricks? a simple test shows that the plain SQL routine outperforms this UDAF.
👍 Great idea, But it gets more and more complicated as you add states.
| 2018-08-20T09:22:09 | 2018-08-20T09:22:09 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 414,254,940 | 2,887 | sundy-li | @amosbird That is really smart... 👍
But `count(x)` is wrong answer, it includes the visitors only in `2018-08-08` | 2018-08-20T09:28:25 | 2018-08-20T10:01:07 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 411,574,494 | 2,818 | bocharov | @alexey-milovidov please have another look.
I've optimized it to use just `strftime()` and almost no allocations.
These are benchmarks I've run on my laptop:
Initial version of formatDateTime before optimization - 425.70 thousand rows/s:
```
select count() from (select formatDateTime(t, '%m/%d/%Y %H:%M:%S') ... | 2018-08-08T22:31:48 | 2018-08-08T22:31:48 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 412,259,034 | 379 | SudhakarSA | > We don't have "fill" option for GROUP BY, so you need to fill missing points on client. Currently, we considering for implementation something like GROUP BY ... WITH FILL.
@alexey-milovidov We are trying to create a timeseries chart using highcharts and for this, we are grouping the data every five minutes.. the m... | 2018-08-11T07:58:03 | 2018-08-11T07:58:03 | {
"+1": 14
} | NONE |
ClickHouse | ClickHouse | 412,449,404 | 2,855 | alexey-milovidov | A test? | 2018-08-13T08:46:02 | 2018-08-13T08:46:02 | {} | MEMBER |
ClickHouse | ClickHouse | 413,158,336 | 2,874 | alexey-milovidov | > git submodule update --init --recursive | 2018-08-15T10:31:41 | 2018-08-15T10:31:41 | {} | MEMBER |
ClickHouse | ClickHouse | 413,485,582 | 2,881 | zhang2014 | 00646_url_engine is not a stable test, I cannot reproduce Travis's problems in my development environment | 2018-08-16T09:36:33 | 2018-08-16T09:37:34 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 413,532,237 | 2,880 | den-crane | This is a feature as documented:
Fields are searched for on any nesting level, indiscriminately. If there are multiple matching fields, the first occurrence is used.
just re-design your json to something like {"first":{"second_nested":{"value":"null"}},"second":"target"}
https://clickhouse.yandex/docs/en/query_lan... | 2018-08-16T12:42:35 | 2018-08-16T12:55:21 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 414,276,787 | 1,661 | Slind14 | Atm. just one server, yes. Using FINAL has been super slow as well, I could give it another shot but I doubt that it is much faster than what I'm currently doing. It wasn't back then. Especially since I'm not reporting on these metrics every time and final would then be needed for every view. | 2018-08-20T10:54:09 | 2018-08-20T10:55:20 | {} | NONE |
ClickHouse | ClickHouse | 411,595,285 | 2,818 | alexey-milovidov | Are you sure that you build with release mode (it is by default)?
Performance numbers that you show are like the numbers from debug build.
(I have debug build on my laptop and get almost the same numbers.)
All performance comparisons in debug build are plain useless.
`stringstream` is the canonical example of s... | 2018-08-09T00:15:09 | 2018-08-09T00:21:24 | {} | MEMBER |
ClickHouse | ClickHouse | 412,151,516 | 2,840 | alexey-milovidov | > Does this solve this issue in #2198?
No. The task #2198 is still unassigned.
| 2018-08-10T17:30:01 | 2018-08-10T17:30:01 | {} | MEMBER |
ClickHouse | ClickHouse | 411,760,273 | 1,981 | PolinaIP | I am seeing this too. What I do wrong?
```
lickhouse-server.service - ClickHouse Server (analytic DBMS for big data)
Loaded: loaded (/etc/systemd/system/clickhouse-server.service; enabled; vendor preset: enabled)
Active: activating (auto-restart) (Result: exit-code) since Thu 2018-08-09 13:35:27 UTC; 18s... | 2018-08-09T13:38:41 | 2018-08-09T13:38:41 | {} | NONE |
ClickHouse | ClickHouse | 412,007,663 | 146 | fizerkhan | +1 Do you have any milestone for this feature? | 2018-08-10T08:00:55 | 2018-08-10T08:01:08 | {} | NONE |
ClickHouse | ClickHouse | 412,042,294 | 146 | blinkov | @fizerkhan workaround is INSERT+SELECT or clickhouse-copier data into a new table with properly named columns | 2018-08-10T10:21:05 | 2018-08-10T10:21:05 | {
"+1": 1
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 412,432,377 | 2,853 | blinkov | It happens in background automatically: basically ClickHouse chooses some active parts, merges the data from them into one larger part, then atomically replaces old parts with the new one. New one becomes active, while old ones become inactive to be deleted later (their data is also in the new larger part, so they are ... | 2018-08-13T07:33:12 | 2018-08-13T07:33:12 | {
"+1": 2,
"heart": 1
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 413,295,012 | 1,966 | faucct | > В ClickHouse используются строгие типы данных. Конечно можно просто все столбцы записать как строки, но тогда это будет работать менее эффективно, чем если заранее для числовых стобцов указать подходящий тип: https://clickhouse.yandex/docs/ru/data_types/int_uint/ , указать в каких столбцах хранятся даты и т.д. Сделат... | 2018-08-15T18:41:55 | 2018-08-15T18:41:55 | {} | NONE |
ClickHouse | ClickHouse | 413,850,783 | 2,827 | 4ertus2 | 18.10.3
drop table if exists t;
create table t
(
a Int32,
b Int32
) engine = MergeTree partition by (a,b) order by (a);
insert into t values (1, 1);
alter table t add column c Int32;
select b from t prewhere a < 1000; -- OK
select c from t where a < 1000; -- OK
select c from t prewhere a < 1000... | 2018-08-17T12:31:11 | 2018-08-17T12:33:39 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 414,270,945 | 2,883 | blinkov | @DeanThompson consider using https://clickhouse.yandex/docs/en/operations/table_engines/collapsingmergetree/ for this, if used properly it'll allow to get rid of a staging table in the first place. | 2018-08-20T10:29:49 | 2018-08-20T10:29:49 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 411,609,179 | 2,818 | sundy-li |
```
21.76% clickhouse [.] cctz::detail::impl::n_sec
21.68% clickhouse [.] std::string::append
19.61% clickhouse [.] cctz::detail::format
6.89% clickhouse [.] memcpy
4.42% libc-2.19.so [.] __strchr_sse2
4.20% clickhouse [.] ... | 2018-08-09T01:37:33 | 2018-08-09T01:37:33 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 412,009,605 | 146 | blinkov | @fizerkhan unfortunately, there's no ETA at the moment. | 2018-08-10T08:09:16 | 2018-08-10T08:09:16 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 412,016,999 | 146 | fizerkhan | @blinkov Thanks for the update. Do you have any other method to do it? | 2018-08-10T08:39:16 | 2018-08-10T08:39:16 | {} | NONE |
ClickHouse | ClickHouse | 413,138,771 | 2,447 | Tezd | @alexey-milovidov sorry to bother, but just want to know if there are any more changes that are needed? | 2018-08-15T09:06:20 | 2018-08-15T09:06:20 | {} | NONE |
ClickHouse | ClickHouse | 414,277,820 | 1,661 | blinkov | So have you rewritten your select queries so they return correct result on CollapsingMergeTree **without** FINAL? If not, it's definitely worth giving a shot. | 2018-08-20T10:58:52 | 2018-08-20T10:58:52 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 413,851,915 | 2,887 | sundy-li | Currently there are two main sql scripts that can calculate the retention, but they run very slowly.
1.use `groupUniqArrayIf`
```
SELECT
r1,
r2
FROM
(
SELECT
groupUniqArrayIf(uid, date = '2018-08-13') AS users1,
groupUniqArrayIf(uid, date = '2018-08-14') AS users2,
l... | 2018-08-17T12:36:01 | 2018-08-20T10:59:12 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 414,283,113 | 1,661 | Slind14 | How should it work without using final? Doesn't this inflate the data when it returns real time data which hasn't been merged yet? | 2018-08-20T11:23:08 | 2018-08-20T11:23:08 | {} | NONE |
ClickHouse | ClickHouse | 412,000,617 | 1,981 | PolinaIP | the problem is solved! I just had to change the user to root
in /etc/systemd/system/clickhouse-server.service | 2018-08-10T07:30:45 | 2018-08-10T07:30:45 | {
"+1": 4,
"hooray": 1
} | NONE |
ClickHouse | ClickHouse | 412,153,339 | 2,840 | alexey-milovidov | > So why do we need it? I'm just curious : )
LIMIT OFFSET is required for compatibility.
This syntax exists in PostgreSQL and was standardized in SQL 2008.
Some tools will generate queries using this syntax (instead of `LIMIT n, m`).
It also was added in MySQL with the same motivation.
And `SELECT TOP` is a bo... | 2018-08-10T17:36:45 | 2018-08-10T17:49:05 | {} | MEMBER |
ClickHouse | ClickHouse | 412,515,834 | 2,803 | TimonKK | Спасибо. А есть хоть примерные сроки попадания фикса в stable? | 2018-08-13T13:21:58 | 2018-08-13T13:21:58 | {} | NONE |
ClickHouse | ClickHouse | 412,654,732 | 2,803 | alexey-milovidov | Несколько недель. | 2018-08-13T20:35:14 | 2018-08-13T20:35:14 | {
"+1": 1
} | MEMBER |
ClickHouse | ClickHouse | 414,024,402 | 2,888 | alexey-milovidov | See `ConvertingBlockInputStream::readImpl()` method.
You can add try ... catch ... rethrow there. | 2018-08-18T01:49:51 | 2018-08-18T01:49:51 | {
"+1": 1
} | MEMBER |
ClickHouse | ClickHouse | 414,207,075 | 2,887 | sundy-li | 1. I got that `yandex.Metrica` use `uniq` function to get the unique visitors, but it has some disadvantages.
2. `uniq` function returns approximate results, yet `retention` returns the exactly results.
3. `uniq` needs `global in` to process distribute queries.
4. Last but not least, `retention` could be u... | 2018-08-20T06:10:57 | 2018-08-20T06:29:49 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 414,273,960 | 2,887 | amosbird | @sundy-li nice catch. the fix is trivial. Just replace `count(x)` with `countIf(y, y > -1)`
@zhang2014 https://github.com/yandex/ClickHouse/issues/1646 would be one way to alleviate the complication. Or else this umbrella issue https://github.com/yandex/ClickHouse/issues/11 | 2018-08-20T10:43:05 | 2018-08-20T10:43:05 | {} | COLLABORATOR |
ClickHouse | ClickHouse | 412,321,094 | 2,852 | simPod | Just for the further reference, it indeed works as expected. It is just needed to wait for some point in time when merge occurs. | 2018-08-12T06:08:47 | 2018-08-12T06:08:47 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 412,461,475 | 2,855 | alexey-milovidov | Thank you! | 2018-08-13T09:33:02 | 2018-08-13T09:33:02 | {} | MEMBER |
ClickHouse | ClickHouse | 412,478,553 | 2,855 | alesapin | It seems like this PR failed test 00606_quantiles_and_nans https://travis-ci.org/yandex/ClickHouse/jobs/415344194. | 2018-08-13T10:44:25 | 2018-08-13T10:52:19 | {} | MEMBER |
ClickHouse | ClickHouse | 414,274,242 | 1,661 | blinkov | @Slind14 I see you have tried CollapsingMergeTree earlier, but had some issues with FINAL. Meanwhile the recommended approach is to rewrite queries in a way that FINAL is not needed, like replacing COUNT(*) with SUM(Sign), have you managed to do that?
As of hardware, it sounds like you are using just one server, rig... | 2018-08-20T10:44:18 | 2018-08-20T10:44:18 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 414,340,597 | 2,896 | champtar | #2830 #2869 | 2018-08-20T14:42:27 | 2018-08-20T14:42:27 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 411,694,316 | 666 | RustRw |
can i use more than one include_froms ， for example： macros in different host is different.
@ztlpn
i fould if i use more than one include_froms, only first is useful. | 2018-08-09T09:20:14 | 2018-08-09T09:35:59 | {} | NONE |
ClickHouse | ClickHouse | 412,055,553 | 2,840 | zhang2014 | What does this do? It looks like it's the same as the `limit` | 2018-08-10T11:27:44 | 2018-08-10T11:27:44 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 412,458,034 | 2,856 | zhang2014 | I don't know how to check `system.processes` table | 2018-08-13T09:19:59 | 2018-08-13T09:20:28 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 412,954,826 | 2,868 | SaltTan | The issue is present in 18.10.3
```
:) SELECT arrayFilter(x -> x IN (select 42), [42]) from remote('localhost,some_other_host', system.one)
SELECT arrayFilter(x -> (x IN
(
SELECT 42
)), [42])
FROM remote('localhost,some_other_host', system.one)
┌─arrayFilter(lambda(tuple(x), in(x, _subquery1... | 2018-08-14T17:37:18 | 2018-08-14T17:37:50 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 413,001,382 | 2,870 | den-crane | There are two issues:
1. a table with engine=Dictionary or database=Dictionary breaks dictionaries_lazy_load
2. a dictionary stays empty forever because it failed to load on CH startup (with lifetime=0) because CH table was not ready. | 2018-08-14T20:16:15 | 2018-08-14T20:18:46 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 413,029,315 | 2,015 | alexey-milovidov | TODO: verify that all test passed when the setting is enabled globally. | 2018-08-14T21:56:34 | 2018-08-14T21:56:34 | {} | MEMBER |
ClickHouse | ClickHouse | 413,541,270 | 2,876 | 4ertus2 | Can't reproduce on master branch:
$ ./clickhouse-client -nq 'DROP TABLE t; CREATE TABLE t (id UInt32, str String, int Int32) engine = Memory; INSERT INTO t FORMAT CSV' < ~/clickhouse/test/2876.csv
$ ./clickhouse-client -q 'SELECT * FROM t'
1 some text \\\\ 0
$ cat ~/clickhouse/test/2876.csv
1,"some text... | 2018-08-16T13:14:50 | 2018-08-16T13:22:27 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 413,905,406 | 2,770 | alex-krash | @alexey-milovidov , I've implemented missing formats, but faced performance problems (my implementation has weaker performance, comparing to reference query, you've provided in https://github.com/yandex/ClickHouse/pull/2818):
```
SELECT count()
FROM
(
SELECT replaceRegexpOne(toString(t), '^(\\d\\d\\d\\d)-(\... | 2018-08-17T15:40:26 | 2018-08-17T15:40:26 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 414,055,216 | 2,156 | wake-up-neo | The same for **LIMIT** on **FINAL** query for **ReplicatedCollapsingMergeTree**
SELECT * FROM table_name FINAL LIMIT 1
> Code: 171. DB::Exception: Received from localhost:9002, ::1. DB::Exception: Block structure mismatch in UNION stream: different number of columns
The column which it shows different is cityH... | 2018-08-18T12:39:11 | 2018-08-18T12:42:49 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 414,194,213 | 2,887 | alexey-milovidov | What about the following query?
```
SELECT
uniq(uid) AS r1,
uniqIf(uid, date = '2018-08-14') AS r2
FROM events
WHERE date IN ('2018-08-13', '2018-08-14')
AND uid IN (SELECT uid FROM events WHERE date = '2018-08-13')
``` | 2018-08-20T04:29:04 | 2018-08-20T04:29:44 | {} | MEMBER |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.