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 | 394,558,590 | 2,464 | alexey-milovidov | Almost Ok.
1. libsimd has no public license. You may add the same license as ClickHouse, because you are the owner of this library.
2. You can change padding in PODArray - it only affects in-memory data layout, no formats will change.
3. We need to compare this implementation versus auto-vectorization in gcc a... | 2018-06-05T02:15:58 | 2018-06-05T02:16:12 | {} | MEMBER |
ClickHouse | ClickHouse | 396,065,367 | 2,452 | tonylauCN | it's great work，thks！
i saw DELETE was released in
[https://github.com/yandex/ClickHouse/releases/tag/v1.1.54386-testing]()
if you give me some tips how to deal with that?
-------------------------------------
ALTER TABLE ontime_all
DELETE WHERE Year = 1987
Received exception fro... | 2018-06-10T17:12:10 | 2018-06-10T17:15:27 | {} | NONE |
ClickHouse | ClickHouse | 396,233,159 | 2,496 | filimonov | Duplicate: #1562 #2385 #1768 #1816
AFAIK the task is taken some time ago, so i hope it will be changed soon.
Workarounds are possible, try function toDateTimeOrZero / parseDateTimeBestEffort | 2018-06-11T12:54:33 | 2018-06-11T12:54:33 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 396,288,280 | 2,231 | bretthoerner | I just wanted to note that we ran into this issue too. It's not hard to handle on the writer-side but maybe if it's not going to be checked in ClickHouse this documentation should be changed:
https://clickhouse.yandex/docs/en/data_types/nested_data_structures/nested/
> During insertion, the system checks that they ... | 2018-06-11T15:40:02 | 2018-06-11T15:40:02 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 397,055,642 | 2,501 | alexey-milovidov | Almost perfect.
We only need to change the way how data is sent in INSERT query
(now it makes separate request for each block). | 2018-06-13T19:23:10 | 2018-06-13T19:23:10 | {} | MEMBER |
ClickHouse | ClickHouse | 397,218,782 | 2,467 | QSer29 | В логах ничего интересного не нашел, но причину обнаружил, может кому будет полезно.
Оказалось что при обновлении до версии 1.1.54385 не был выполнен рестарт серверов, в следствии чего на сервере с Distributed таблицей продолжала работать версия 1.1.54378, а на серверах с таблицами ReplicatedMergeTree - 1.1.54342. П... | 2018-06-14T08:43:23 | 2018-06-14T08:43:23 | {} | NONE |
ClickHouse | ClickHouse | 393,499,955 | 2,452 | lwolters | Yeah, that is also something we would like to know.... Keep on going the great work! | 2018-05-31T11:25:36 | 2018-05-31T11:25:36 | {} | NONE |
ClickHouse | ClickHouse | 393,612,868 | 2,362 | alexey-milovidov | There is no reason why we can't fill `min_date` and `max_date` with correct values if partition key have `Date` type. @ztlpn ? | 2018-05-31T17:36:21 | 2018-05-31T17:36:21 | {} | MEMBER |
ClickHouse | ClickHouse | 394,734,801 | 1,710 | krzyzakp | @alexey-milovidov any chance to use function, which you introduced here? Would be super useful for us. | 2018-06-05T14:39:40 | 2018-06-05T14:39:40 | {} | NONE |
ClickHouse | ClickHouse | 396,402,653 | 2,231 | alexey-milovidov | This bug is acknowledged. We should add missing checks. | 2018-06-11T22:06:45 | 2018-06-11T22:06:45 | {} | MEMBER |
ClickHouse | ClickHouse | 396,757,700 | 2,428 | alexey-milovidov | If we create a table with different partition key:
```
ENGINE = MergeTree
PARTITION BY tuple()
ORDER BY (Year, FlightDate)
```
the memory usage of clickhouse-server is much lower - less than 2 GB. | 2018-06-12T22:42:50 | 2018-06-12T22:42:50 | {} | MEMBER |
ClickHouse | ClickHouse | 397,047,045 | 2,503 | KochetovNicolai | ~~I think it's ok to use such sharding key if you aren't going to use more than 16 nodes.~~ | 2018-06-13T18:52:49 | 2018-06-13T23:09:53 | {} | MEMBER |
ClickHouse | ClickHouse | 393,513,933 | 2,448 | alexey-milovidov | It looks like the race codition still exist. There is a chance that materialized view will be added but streaming will not start. I'm right? | 2018-05-31T12:28:09 | 2018-05-31T12:28:09 | {} | MEMBER |
ClickHouse | ClickHouse | 396,038,664 | 1,851 | PhantomPhreak | I have the same problem after making a TCP balancer using keepalived | 2018-06-10T10:39:36 | 2018-06-10T10:39:36 | {} | NONE |
ClickHouse | ClickHouse | 396,752,018 | 2,428 | alexey-milovidov | It really needs slightly more than 8 GB of memory to load the data. | 2018-06-12T22:15:26 | 2018-06-12T22:15:26 | {} | MEMBER |
ClickHouse | ClickHouse | 396,954,645 | 2,502 | jonatasfreitasv | Sorry, I ignored copying jobs xml to Zookeeper. | 2018-06-13T14:19:50 | 2018-06-13T14:19:50 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 397,115,808 | 2,503 | alexey-milovidov | No, you shouldn't use `reinterpretAsUInt64` from hexadecimal string as a sharding key,
because you will get uneven distribution of shards.
For example, suppose you have 32 shards.
Let's look how random data will map to these shards:
```
:) SELECT reinterpretAsUInt64(hex(rand())) % 32 AS k, count() AS c FROM nu... | 2018-06-13T23:06:21 | 2018-06-13T23:06:21 | {} | MEMBER |
ClickHouse | ClickHouse | 393,546,165 | 2,348 | alexey-milovidov | Worthless right now, doesn't solve the original problem. | 2018-05-31T14:20:08 | 2018-05-31T14:20:08 | {} | MEMBER |
ClickHouse | ClickHouse | 394,808,473 | 1,710 | alexey-milovidov | I want to add a setting to switch between basic and "best effort" parsing in text formats.
This looks like relatively easy task but it's not in high priority right now. | 2018-06-05T18:13:59 | 2018-06-05T18:13:59 | {} | MEMBER |
ClickHouse | ClickHouse | 395,165,005 | 2,218 | jonatasfreitasv | Version 0.2.1-beta (Version name is reseted!)
https://github.com/HouseOps/HouseOps
- `Epic Feature` Alert in UI on new release or auto-update.
- `Feature` Additional serach in columns table results.
- `Feature` Button for open debug tools in production.
- `UI/UX` Fix column filter/search style (https://github.... | 2018-06-06T18:17:05 | 2018-06-06T18:17:05 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 395,302,472 | 2,439 | millin | I created the same using your statement and didn't see any errors in log.
What is your CH version and OS? | 2018-06-07T05:58:52 | 2018-06-07T05:58:52 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 395,386,705 | 2,478 | alex-zaitsev | As I mentioned in the Telegram channel, you are confusing partitions and parts. '200101_4_4_0' is a part name, partition name is '200101'. Documentation is pretty clear on this.
There is a bug here that you can not use FREEZE PARTITION '200101', it should work both in single quotas and without. | 2018-06-07T11:25:42 | 2018-06-07T11:25:42 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 395,861,966 | 1,710 | alexey-milovidov | Now you can use setting
`SET date_time_input_format = 'best_effort'`
to enable extended DateTime parsing.
Notes:
- it doesn't affect the behaviour of type conversion functions;
| 2018-06-08T19:16:59 | 2018-06-08T19:16:59 | {} | MEMBER |
ClickHouse | ClickHouse | 396,090,495 | 2,218 | jonatasfreitasv | Version 0.3.0-beta!
https://github.com/HouseOps/HouseOps
Feature In Processes List, recording feature to capture all processes.
Feature In Processes List, added line graphic to show pending processes.
Feature Query history.
Improvement Put http:// or https:// before host message in Database Settings.
UI/UX ... | 2018-06-10T23:29:58 | 2018-06-10T23:29:58 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 396,719,337 | 2,499 | alexey-milovidov | I know multiple users who run ClickHouse with Docker in production, including clustered setup.
Although I'm not an expert in Docker and I cannot confirm, whether it is a good practice. | 2018-06-12T20:17:08 | 2018-06-12T20:17:49 | {
"+1": 2
} | MEMBER |
ClickHouse | ClickHouse | 393,522,632 | 2,447 | alexey-milovidov | > P.S Can I use Clients sendQuery method for gathering data about dbs, tables?
This is perfectly Ok, but care should be taken when there are huge amount of tables.
We have servers with about million of small `StripeLog` tables for chunks.
You can define a limit for maximum number of fetched tables. And probably do... | 2018-05-31T13:02:30 | 2018-05-31T13:02:30 | {} | MEMBER |
ClickHouse | ClickHouse | 393,561,978 | 2,454 | alexey-milovidov | No issue on recent ClickHouse versions. | 2018-05-31T15:07:44 | 2018-05-31T15:07:44 | {} | MEMBER |
ClickHouse | ClickHouse | 393,634,672 | 1,768 | alexey-milovidov | This issue is related to eager evaluation of conditional branches.
Workarounds:
- write `toDateTimeOrZero(x)` instead of `CAST(x AS DateTime)`;
- write `parseDateTimeBestEffort(x)` instead of `CAST(x AS DateTime)`; | 2018-05-31T18:40:50 | 2018-05-31T18:40:50 | {} | MEMBER |
ClickHouse | ClickHouse | 393,824,371 | 2,455 | millin | Для работы с executable словарями важно, чтобы STDOUT возвращал обратно такое же значение ключа, какое было передано при использовании функции dictGet*.
Например,
при использовании dictGetUInt64('geoip2', 'asn', tuple(ip)) и формата JSONEachRow:
для STDIN {"ip": "8.8.8.8"} приложение должен вернуть в STDOUT {"ip"... | 2018-06-01T09:21:51 | 2018-06-01T09:28:01 | {
"+1": 2
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 393,957,618 | 2,459 | alexey-milovidov | So, everything is perfect. | 2018-06-01T17:45:34 | 2018-06-01T17:45:34 | {} | MEMBER |
ClickHouse | ClickHouse | 396,070,575 | 2,260 | alexey-milovidov | @zhang2014
Thank you! This is definitely useful feature.
What to do further... For example, our users have a request to implement
`CREATE OR REPLACE` for tables and especially for views.
This should be done atomically (from the client point of view). It can be done under global (context) lock. | 2018-06-10T18:28:01 | 2018-06-10T18:29:47 | {} | MEMBER |
ClickHouse | ClickHouse | 396,737,592 | 2,498 | jonatasfreitasv | Thx @alexey-milovidov. | 2018-06-12T21:19:33 | 2018-06-12T21:19:33 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 396,753,972 | 2,428 | alexey-milovidov | But it needs to be explained. Maybe we can lower memory usage.
Dataset is 61.6 GB in CSV, it contains 166 million rows. So, each row is 360 bytes in average in CSV.
If we load in 1 048 576 rows per chunk (by default) it should take about 360 MB.
Data is loaded by clickhouse-client with "double buffering". It for... | 2018-06-12T22:24:31 | 2018-06-12T22:29:22 | {} | MEMBER |
ClickHouse | ClickHouse | 397,057,092 | 2,503 | alexey-milovidov | `reinterpretAsInt64('e0933a87-...`
will use first 64bits (8 bytes) of string: 'e0933a87' and this is effectively 32 bits.
~~If first component of UUID is assumed to be random (that's reasonable), then using reinterpretAsInt64 is Ok.~~
But it's much more obvious to use `cityHash64(uuid)`. | 2018-06-13T19:28:05 | 2018-06-13T23:10:04 | {} | MEMBER |
ClickHouse | ClickHouse | 393,632,131 | 2,147 | alexey-milovidov | We've the behaviour more strict in https://github.com/yandex/ClickHouse/blob/master/CHANGELOG.md#backward-incompatible-changes | 2018-05-31T18:34:32 | 2018-05-31T18:34:32 | {} | MEMBER |
ClickHouse | ClickHouse | 395,326,717 | 2,452 | yjygit | +1 | 2018-06-07T07:50:41 | 2018-06-07T07:50:41 | {} | NONE |
ClickHouse | ClickHouse | 395,541,744 | 2,026 | silviucpp | Hello @alexey-milovidov also this one was partially fixed. | 2018-06-07T19:44:08 | 2018-06-07T19:44:08 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 396,954,654 | 2,480 | KochetovNicolai | Segmentation fault is a bug, but I can't reproduce on the last version.
I've created table with the same structure on cluster. When I run alter query a get:
```
Code: 44. DB::Exception: Received from localhost:9004, ::1. DB::Exception: Cannot add column system.filesystem.files: column with this name already exists.
... | 2018-06-13T14:19:51 | 2018-06-13T14:19:51 | {} | MEMBER |
ClickHouse | ClickHouse | 393,530,868 | 2,444 | alexey-milovidov | :+1: | 2018-05-31T13:31:51 | 2018-05-31T13:31:51 | {} | MEMBER |
ClickHouse | ClickHouse | 394,139,445 | 2,457 | Slach | why you need ANY(event_timestamp) ?
you need random timestamp? or you just need last event_timestamp?
just try
```
select event_timestamp from logs order by event_timestamp limit 1
```
in your request you buffer 598 million values of event_timestamp in memory | 2018-06-03T06:02:58 | 2018-06-03T06:02:58 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 394,378,282 | 2,362 | alexey-milovidov | These fields (min_date, max_date) are left in `system.parts` table for compatibility with "classic" partitioning.
With custom partitioning, every table may have different partition keys and we cannot choose single type for these columns. Probably we can add string columns with serialized value of minimums and maximu... | 2018-06-04T14:39:04 | 2018-06-04T14:39:04 | {} | MEMBER |
ClickHouse | ClickHouse | 394,942,993 | 2,439 | millin | Could you show your CREATE TABLE requests of both mq and tab_view? | 2018-06-06T05:17:44 | 2018-06-06T05:17:44 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 395,042,013 | 2,376 | AbdealiLoKo | Sorry for the delay ... never saw this.
EDIT: I pulled the latest `stable` branch and got this cmake output:
```
$ cmake .. -DCMAKE_CXX_COMPILER=`which g++-8` -DCMAKE_C_COMPILER=`which gcc-8`
-- The C compiler identification is GNU 8.1.0
-- The CXX compiler identification is GNU 8.1.0
-- Checking whether C co... | 2018-06-06T11:52:47 | 2018-06-06T11:57:18 | {} | NONE |
ClickHouse | ClickHouse | 395,965,482 | 2,490 | alexey-milovidov | Also need to test when query is in URL parameter, but there are external tables in multipart/form-data. | 2018-06-09T12:26:32 | 2018-06-09T12:26:32 | {} | MEMBER |
ClickHouse | ClickHouse | 396,502,696 | 2,266 | pavel-odintsov | Thank you for help!
I think it works fine :)
```
root@armlab2:/var/log/clickhouse-server# clickhouse-client
ClickHouse client version 1.1.54386.
Connecting to localhost:9000 as user default.
Connected to ClickHouse server version 1.1.54386.
localhost :) show databases
SHOW DATABASES
┌─name────┐
│ d... | 2018-06-12T08:04:17 | 2018-06-12T08:04:17 | {
"+1": 1
} | NONE |
ClickHouse | ClickHouse | 396,743,071 | 2,498 | alexey-milovidov | So, why you get these errors? | 2018-06-12T21:38:14 | 2018-06-12T21:38:14 | {} | MEMBER |
ClickHouse | ClickHouse | 394,296,456 | 2,414 | ztlpn | It is rather difficult to predict what the bottleneck is because we haven't spent much time optimizing for the case of many small queries. For example last time someone checked this case the bottleneck was surprisingly in generating `query_id` strings.
This means you need to profile. Try applying the load and determ... | 2018-06-04T09:43:30 | 2018-06-04T09:43:30 | {
"+1": 1
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 395,362,605 | 2,439 | LiuYangkuan | PR [#2448](https://github.com/yandex/ClickHouse/pull/2448)
It is not a critical bug. And table can work fine, but an Exception in log. | 2018-06-07T09:52:50 | 2018-06-07T09:52:50 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 396,591,641 | 2,455 | filimonov | Ещё из "побочных" эффектов - запрос будет выполняться в несколько потоков, но доступ к словарю будет только одним потоком. Т.е. это выглядит примерно так: все данные быстро и эффективно обрабатываются, ну скажем в 16 потоков, используя все можности процессора, а потом приходит момент когда нужно "заглянуть в словарь" и... | 2018-06-12T13:37:13 | 2018-06-12T13:37:13 | {
"+1": 2
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 396,919,710 | 2,494 | KochetovNicolai | In your case you can use `assumeNotNull` function:
```
:) SELECT if(countDistinct(subType) IS NULL, '0 Items', concat(CAST(assumeNotNull(countDistinct(subType)) AS Char), ' Items')) FROM default.nullableCountTest GROUP BY subType
SELECT if(isNull(countDistinct(subType)), '0 Items', concat(CAST(assumeNotNull(count... | 2018-06-13T12:28:07 | 2018-06-13T12:28:07 | {} | MEMBER |
ClickHouse | ClickHouse | 397,287,943 | 2,414 | silviucpp | I personally have different results than yours:
I created the following tables on 3 nodes (VM with 2 cores, 8 GB ram, no SSD):
```sql
CREATE DATABASE test;
CREATE TABLE test.collection_src (timestamp DateTime, username Nullable(String))
ENGINE = MergeTree
PARTITION BY toYYYYMM(timestamp)
ORDER BY... | 2018-06-14T13:03:37 | 2018-06-14T13:03:37 | {
"+1": 1
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 397,495,007 | 2,509 | AlexMAS | Thank you for your reply. Yes, I assumed that it was intentional changes. I've added processing for this case to my code so it isn't critical for me. I think we can close this issue. | 2018-06-15T02:28:48 | 2018-06-15T03:02:33 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 394,286,388 | 2,218 | perez1987 | Testing in development environment :-) | 2018-06-04T09:08:49 | 2018-06-04T09:08:49 | {
"+1": 1
} | NONE |
ClickHouse | ClickHouse | 395,398,666 | 2,478 | silviucpp | Also in documentation :
Creates a local backup of one or multiple partitions. The name can be the full name of the partition (for example, 201403), or its prefix (for example, 2014) - then the backup will be created for all the corresponding partitions.
My data folder looks like:
200101_4_4_0
200101_6_6_0
20... | 2018-06-07T12:13:33 | 2018-06-07T12:13:33 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 395,411,778 | 1,655 | andrusha | If you make your `Dynamic.X` a nested table, then you could do `arrayElement(indexOf(` to put specific key as a column manually. | 2018-06-07T13:01:20 | 2018-06-07T13:01:20 | {} | NONE |
ClickHouse | ClickHouse | 395,446,634 | 2,477 | alexey-milovidov | Thanks! | 2018-06-07T14:43:41 | 2018-06-07T14:43:41 | {} | MEMBER |
ClickHouse | ClickHouse | 396,235,285 | 2,467 | filimonov | В логах при выполнении этого запроса не остаётся чего-нибудь интересного? | 2018-06-11T13:01:54 | 2018-06-11T13:01:54 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 397,404,121 | 2,509 | alexey-milovidov | This is intended behaviour.
In previous version you get empty result for this query.
This was changed in version 1.1.54356:
https://github.com/yandex/ClickHouse/blob/master/CHANGELOG.md#new-features-3
You can turn back previous behaviour with setting "empty_result_for_aggregation_by_empty_set".
But current beh... | 2018-06-14T19:02:02 | 2018-06-14T19:04:45 | {
"+1": 1
} | MEMBER |
ClickHouse | ClickHouse | 393,513,465 | 2,452 | blinkov | ALTER DELETEs are almost ready, there has been a working demo on last ClickHouse meetup in Moscow few weeks ago.
ALTER UPDATEs are coming straight after, using the same approach.
The roadmap is kind of outdated at the moment, we'll deal with that soon. | 2018-05-31T12:25:59 | 2018-05-31T12:25:59 | {
"+1": 8
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 394,171,118 | 2,362 | den-crane | Wondering why it (min_date) has type Date since partition key can be any type including tuple.
Maybe it should be replaced with min_value ~ toString (tuple( (min(partK0), min(partK1), ..., min(partKn) )) (the same way as partition column). | 2018-06-03T15:46:37 | 2018-06-03T15:56:06 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 395,167,881 | 1,729 | alexey-milovidov | We decided that we don't need to apply any escaping in Pretty formats. | 2018-06-06T18:26:57 | 2018-06-06T18:26:57 | {} | MEMBER |
ClickHouse | ClickHouse | 395,178,534 | 1,729 | alexey-milovidov | The only issue is that you wouldn't be able to differentiate between NULL and string with 'NULL' value.
I'm still in doubt. | 2018-06-06T19:02:22 | 2018-06-06T19:02:22 | {} | MEMBER |
ClickHouse | ClickHouse | 394,893,333 | 2,452 | salisbury-espinosa | zlzforever
https://github.com/yandex/ClickHouse/blob/master/dbms/src/Parsers/ASTAlterQuery.cpp#L170
```c++
else if (p.type == ASTAlterQuery::DELETE)
{
settings.ostr << (settings.hilite ? hilite_keyword : "") << indent_str << "DELETE WHERE " << (settings.hilite ? hilite_none : "");
... | 2018-06-05T23:34:02 | 2018-06-10T15:12:42 | {} | NONE |
ClickHouse | ClickHouse | 396,412,941 | 2,266 | pavel-odintsov | Hello!
Thank you!
Can you slightly fix build script to suppress SSE features for ARM 64 platform?
I got number of error messages about it:
```
Feature record: CXX_FEATURE:0cxx_variadic_templates
Performing C++ SOURCE FILE Test HAVE_POPCNT succeeded with the following output:
Change Dir: /build/clickhouse-... | 2018-06-11T22:58:53 | 2018-06-11T22:58:53 | {
"+1": 1
} | NONE |
ClickHouse | ClickHouse | 396,716,868 | 2,498 | alexey-milovidov | Every package (either testing and stable) pass all tests before we publish it on `repo.yandex.ru`.
You should look at test results. It may be:
- different configuration that prevents some tests to run;
- different build of clickhouse.
| 2018-06-12T20:08:26 | 2018-06-12T20:08:26 | {} | MEMBER |
ClickHouse | ClickHouse | 396,737,344 | 2,499 | jonatasfreitasv | Thx a lot @alexey-milovidov. | 2018-06-12T21:18:39 | 2018-06-12T21:18:39 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 396,906,664 | 2,500 | KochetovNicolai | Can you provide
* the way how you have created tables
* your cluster configuration
* structure of your tables (show create table)
?
Also, you may check that all replicas has the same schema (run show create table on each node), and that all cluster configs are the same.
| 2018-06-13T11:35:47 | 2018-06-13T11:35:47 | {} | MEMBER |
ClickHouse | ClickHouse | 393,518,386 | 2,218 | meshpaul | @sundy-li anything wrong with this one https://github.com/yandex/clickhouse-jdbc. We use it and WOW what a difference 75% decrease on our GC during massive heavy loads when we switched from Oracle | 2018-05-31T12:46:19 | 2018-05-31T12:46:19 | {} | NONE |
ClickHouse | ClickHouse | 393,935,588 | 2,459 | alexey-milovidov | Why we have
options `SYSLOG_CONS | SYSLOG_PID`
and `LOG_DAEMON` facility
for local syslog, but
`LOG_USER` facility
for rsyslog? | 2018-06-01T16:30:26 | 2018-06-01T16:30:26 | {} | MEMBER |
ClickHouse | ClickHouse | 395,442,819 | 2,466 | den-crane | >На сколько я понимаю, должна быть 1 суммированная запись?
только если final мердж сольется в одну партицию, что не гарантировано и неизвестно когда случится.
схлопывание это часть процесса слияние (merge) оно не гарантировано, и не работает через несколько партиций, поэтому запросы надо всегда делать либо selec... | 2018-06-07T14:33:09 | 2018-06-07T14:37:27 | {
"+1": 1
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 395,853,105 | 2,484 | alexey-milovidov | Rough performance comparison.
When parsing a value of type Date, there is no performance difference with master.
When parsing DateTime with default (`basic`) method, the performance is slightly worse:
```
for i in {1..25}; do
T1=$((time ~/clickhouse-master local --structure 'd DateTime' --query="SELECT cou... | 2018-06-08T18:44:46 | 2018-06-08T18:45:24 | {} | MEMBER |
ClickHouse | ClickHouse | 394,168,387 | 2,457 | Jingyi244 | @Slach
Thanks for ur reply.
I am using ANY(event_timestamp) in the upper sql just to reduce the output to 1 line, for test purpose. Sorry for the misleading.
And i am doing user funnel analysis, as shown in Page 56 of the slide [Creating marketing funnels or how to calculate complex metrics in ClickHouse]
[http... | 2018-06-03T15:03:26 | 2018-06-03T15:03:26 | {} | NONE |
ClickHouse | ClickHouse | 395,841,937 | 2,486 | alexey-milovidov | It's not possible to update already inserted rows in Join table.
Yes, you can recreate a table. | 2018-06-08T18:05:16 | 2018-06-08T18:05:16 | {} | MEMBER |
ClickHouse | ClickHouse | 393,943,746 | 2,459 | alex-krash | 1. rsyslog does'not support SYSLOG_CONS|SYSLOG_PID options (for local, these option's combination has been taken from existing one implementation)
2. According to facility, I've picked up existing for local syslog, and used Poco's default for rsyslog. Possibly, to simplify and unify usage, we may use LOG_DAEMON for rs... | 2018-06-01T16:59:37 | 2018-06-01T16:59:37 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 394,050,161 | 2,218 | jonatasfreitasv | Version 1.1.0-beta!!!
Waiting feedbacks!!!
[Epic Feature] Process List Screen
[Epic Feature] Server Settings section
[Epic Feature] Database Graph (alpha)
[Feature] Database Alias
[Feature] Abort query
[Feature] Select database for run query
[Feature] Copy to clipboard results in CSV format
[... | 2018-06-02T02:23:56 | 2018-06-02T02:23:56 | {
"+1": 1
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 394,354,421 | 635 | sanjayraja | Any idea when this would be available ?
We are also trying to use Clickhouse with Mondrian - but the generated SQL uses joins with alias which is not supported in Clickhouse | 2018-06-04T13:28:34 | 2018-06-04T13:28:34 | {
"+1": 1
} | NONE |
ClickHouse | ClickHouse | 394,738,125 | 2,414 | agaurav | @ztlpn It seems to be cpu bound as IO hardly crossed 4% (both cpu & disk IO) while cpu was 100% on all cores and load avg also peaked to 125.
I will try to profile the execution and come up with some code level insights. | 2018-06-05T14:48:22 | 2018-06-05T14:48:22 | {} | NONE |
ClickHouse | ClickHouse | 394,844,005 | 2,471 | alexey-milovidov | Squashed manually:
https://github.com/yandex/ClickHouse/commit/51044279e5921dba4bc7cb70ab203589cc7fb126 | 2018-06-05T20:11:18 | 2018-06-05T20:11:18 | {} | MEMBER |
ClickHouse | ClickHouse | 395,262,745 | 2,475 | alexey-milovidov | Also look at https://github.com/yandex/ClickHouse/blob/master/utils/link-validate/link-validate.sh | 2018-06-07T01:21:26 | 2018-06-07T01:21:26 | {} | MEMBER |
ClickHouse | ClickHouse | 395,391,246 | 2,478 | silviucpp | @alex-zaitsev you are right. I consulted the documentation again and seems you cannot FREEZE only specific parts.
But 2 bugs still remains here:
1. The one you mentioned : FREEZE PARTITION '200101', it should work both in single quotas and without
2. Documentation doesn't say anything regarding WITH NAME syntax... | 2018-06-07T11:44:33 | 2018-06-07T11:44:33 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 397,514,235 | 2,500 | sundy-li | Checked , there are some replicas with different schema, thank you. | 2018-06-15T05:01:45 | 2018-06-15T05:01:45 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 393,546,651 | 2,411 | alexey-milovidov | Thank you! | 2018-05-31T14:21:44 | 2018-05-31T14:21:44 | {} | MEMBER |
ClickHouse | ClickHouse | 393,791,649 | 2,448 | LiuYangkuan | Yes, you got it! The kafka StreamThread only judge the dependencies is not enough.
Always the second time, the dependency tables--materialized views is ready.
So this is not a critical bug. | 2018-06-01T07:31:41 | 2018-06-01T07:31:41 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 394,081,665 | 2,453 | otisg | Note there is a mailing list/group for questions. | 2018-06-02T11:53:34 | 2018-06-02T11:53:34 | {} | NONE |
ClickHouse | ClickHouse | 394,164,544 | 2,445 | the-tinderbox | I think this issue is related with https://github.com/yandex/ClickHouse/issues/2231 | 2018-06-03T14:03:37 | 2018-06-03T14:03:37 | {} | NONE |
ClickHouse | ClickHouse | 394,317,760 | 2,417 | ztlpn | Millions of old parts is definitely not normal. During normal operation there should be about `old_parts_lifetime * <rate of inserts>` old parts. With the default value of old_parts_lifetime = 480 and recommended rate of insert ~ 1 per second we get ~1000 old parts.
Possible reasons that could cause that many parts ... | 2018-06-04T11:08:50 | 2018-06-04T11:08:50 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 395,533,832 | 2,010 | silviucpp | Hello @alexey-milovidov any idea when this bug will be fixed ?
| 2018-06-07T19:15:49 | 2018-06-07T19:15:49 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 395,539,885 | 2,010 | alexey-milovidov | It was left unfinished, but thank you for reminder, I'll try to keep up! | 2018-06-07T19:37:13 | 2018-06-07T19:37:13 | {} | MEMBER |
ClickHouse | ClickHouse | 395,955,283 | 2,490 | alexey-milovidov | Please add a test. Look at `dbms/tests/queries` directory. You can use any test of `sh` type as an example. | 2018-06-09T09:37:37 | 2018-06-09T09:37:37 | {} | MEMBER |
ClickHouse | ClickHouse | 396,165,238 | 2,260 | zhang2014 | @alexey-milovidov Thank for you review, I have learned a lot.
If necessary, I will continue to follow up and fix these problems, But it may be a few days later, The [TCP ClickHouse JDBC Driver](https://github.com/houseflys/ClickHouse-Native-JDBC) we implemented will be released recently : )
cc @sundy-li | 2018-06-11T08:34:48 | 2018-06-11T08:44:41 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 396,416,982 | 2,266 | pavel-odintsov | I slightly changed configuration to disable SSE offloads to:
```
env DIST=bionic DISABLE_PARALLEL=1 CMAKE_FLAGS="-DHAVE_SSE41=0 -DHAVE_SSE42=0 -DHAVE_POPCNT=0 -DHAVE_SSE2_INTRIN=0 -DSSE2~AG=' ' -DHAVE_PCLMULQDQ_INTRIN=0 -DPCLMULFLAG=' '" EXTRAPACKAGES="clang-6.0 libstdc++-8-dev" DEB_CC=clang-6.0 DEB_CXX=clang++-6.0 C... | 2018-06-11T23:21:06 | 2018-06-11T23:21:06 | {} | NONE |
ClickHouse | ClickHouse | 397,282,522 | 975 | ervivekmehta |
RunningDifference(x) for multiple x values
Ask Question
up vote
0
down vote
favorite
My Table tbl_data(event_time, monitor_id,type,event_date,status)
select status,sum(runningDifference(event_time)) as delta from (SELECT status,event_date,event_time FROM tbl_data WHERE event_date >= '2018-05-01' AND monito... | 2018-06-14T12:48:48 | 2018-06-14T12:48:48 | {} | NONE |
ClickHouse | ClickHouse | 393,507,964 | 2,452 | erdebee | +1 | 2018-05-31T12:01:59 | 2018-05-31T12:01:59 | {} | NONE |
ClickHouse | ClickHouse | 395,043,002 | 2,376 | AbdealiLoKo | I pulled the latest "stable" branch and retried this. now it gives me:
```
15%] Building CXX object contrib/capnproto/c++/src/kj/CMakeFiles/kj.dir/main.c++.o
In file included from /usr/local/Cellar/gcc/8.1.0/include/c++/8.1.0/map:60,
from /Users/abdealijk/Documents/oss/yandex/ClickHouse/contrib/ca... | 2018-06-06T11:56:42 | 2018-06-06T11:59:07 | {} | NONE |
ClickHouse | ClickHouse | 395,258,538 | 2,439 | LiuYangkuan | Here it is:
Create table mq (id UInt64, name String, addr String) ENGINE=Kafka(‘Kafka-server:9092’, ’topic_name’, ‘group0',’TSV’);
Create materialized view tab_view (id UInt64, name String, addr String) ENGINE=Log as select * from mq;
> 在 2018年6月6日，下午1:17，Alexander Millin <notifications@github.com <mailto:notificati... | 2018-06-07T00:51:47 | 2018-06-07T00:51:47 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 395,305,486 | 2,439 | LiuYangkuan | My CH version is 1.1.54383, OS is Centos 7.2.
My bug fix patch is merge into master branch. | 2018-06-07T06:16:08 | 2018-06-07T06:16:08 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 395,341,185 | 2,439 | millin | You mean your pr #2398?
I think your changes are the source of this issue. In stable branch it works fine. | 2018-06-07T08:41:07 | 2018-06-07T08:41:07 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 395,990,444 | 2,260 | alexey-milovidov | I would like to prefer different method of implementation:
There is default implementation for truncate, that simply do drop and create table under lock.
All tables will use default implementation except ReplicatedMergeTree.
Cons:
- the lock will be in global scope, not for table scope;
- `metadata_modificatio... | 2018-06-09T18:44:04 | 2018-06-09T18:44:14 | {} | MEMBER |
ClickHouse | ClickHouse | 397,116,246 | 2,503 | alexey-milovidov | @KochetovNicolai And your statement is also wrong. Here are results for 16, 8 and 4 shards:
```
:) SELECT reinterpretAsUInt64(hex(rand())) % 16 AS k, count() AS c, bar(c, 0, 1000000, 100) AS b FROM numbers(1000000) GROUP BY k ORDER BY k
SELECT
reinterpretAsUInt64(hex(rand())) % 16 AS k,
count() AS ... | 2018-06-13T23:08:52 | 2018-06-13T23:12:27 | {} | MEMBER |
ClickHouse | ClickHouse | 397,222,189 | 2,384 | TimonKK | Такое повторяется и для такого запроса
SELECT
_id
FROM
users u
WHERE
(
u.createdLang IN ('ru')
AND u.isAdmin = 0
AND u.role = 'guest'
)
Poco::Exception. Code: 1000, e.code() = 1241, e.displayText() = mysqlxx::BadQuery: Operand shou... | 2018-06-14T08:54:31 | 2018-06-14T08:54:31 | {} | NONE |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.