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 | 533,184,940 | 6,997 | sl4mmer | version 19.4.0.49 | 2019-09-19T15:29:00 | 2019-09-19T15:29:00 | {} | NONE |
ClickHouse | ClickHouse | 533,483,654 | 5,596 | excitoon | @alexey-milovidov I would like to ask to squash before merge, as there are too much trash commits in my branch. | 2019-09-20T09:36:15 | 2019-09-20T09:36:15 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 533,519,974 | 6,867 | akuzm | Fixed in #6951. | 2019-09-20T11:45:53 | 2019-09-20T11:45:53 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 533,536,736 | 6,956 | yingfeng | This issue should be the same with [5475](https://github.com/yandex/ClickHouse/issues/5475), it's a fault for my description that the filters are not executed in parallel, it's actually caused by the scalar subqueries---it has a long time to wait before the progress bar appears in the ClickHouse cli.
Another upda... | 2019-09-20T12:40:13 | 2019-09-20T12:40:13 | {} | NONE |
ClickHouse | ClickHouse | 533,591,822 | 7,018 | AntonSaykovsky | > > Возможно есть другой метод без движка Join выбрать все строки из левой таблицы с максимальной версией, если для одной версии может быть больше 1 строки?
>
> select argMax(tuple(*),ver) from test;
Спасибо, но видимо я не до конца был точен, поэтому не очень понятно, к примеру:
```
CREATE TABLE IF NOT EXIST... | 2019-09-20T15:06:21 | 2019-09-20T15:22:09 | {} | NONE |
ClickHouse | ClickHouse | 533,742,131 | 6,894 | azat | (fixed compilation error, due to missing `target_include_directories`) | 2019-09-20T23:21:05 | 2019-09-20T23:21:05 | {} | MEMBER |
ClickHouse | ClickHouse | 533,177,695 | 6,991 | alexey-milovidov | I think that allowing the user to write C++ code in SQL statements is not viable.
Better to only allow reading the code from files in configured location in filesystem.
We should not put all functions inside "udf" name. They should be registered with user defined names by `CREATE FUNCTION` query and appear in `syst... | 2019-09-19T15:12:47 | 2019-09-19T15:13:34 | {} | MEMBER |
ClickHouse | ClickHouse | 533,310,863 | 6,991 | filimonov | apache arrow gives and interesting toolset to build cross language udfs:
https://www.slideshare.net/ueshin/apache-arrow-and-pandas-udf-on-apache-spark | 2019-09-19T21:10:54 | 2019-09-19T21:15:31 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 533,510,289 | 5,475 | yingfeng | @alexey-milovidov @KochetovNicolai
The scalar subqueries' performance issue still exist even if I turn on the switch :
```
SET experimental_use_processors = 1
```
The version of ClickHouse is `v19.13.4.32-stable` and `v19.14.3.3-stable`. The cpu occupation is still kept 100% for a long time before the progress b... | 2019-09-20T11:07:47 | 2019-09-20T11:07:47 | {} | NONE |
ClickHouse | ClickHouse | 533,570,375 | 7,018 | AntonSaykovsky | Приведение типа например toUInt8(MAX(ver)) AS ver / toUInt64(MAX(ver)) AS ver также не помогает. | 2019-09-20T14:11:19 | 2019-09-20T14:23:20 | {} | NONE |
ClickHouse | ClickHouse | 533,577,512 | 7,018 | AntonSaykovsky | Попробовал, toUInt64() аналогично.
Возможно есть другой метод без движка Join выбрать все строки из левой таблицы с максимальной версией, если для одной версии (в рамках одного id) может быть больше 1 строки?
Варианты с подзапросом (... WHERE ver = (SELECT MAX(ver) FROM test AS t2 WHERE t1.id=t2.id))и обычным дж... | 2019-09-20T14:30:29 | 2019-09-20T15:20:49 | {} | NONE |
ClickHouse | ClickHouse | 533,692,213 | 7,009 | alexey-milovidov | @dujijun007 Why do you have `clickhosue-server` hostname instead of `clickhouse-server`? | 2019-09-20T20:06:48 | 2019-09-20T20:08:43 | {
"+1": 1
} | MEMBER |
ClickHouse | ClickHouse | 533,847,732 | 6,833 | ghost | @abyss7 Is there a way to configure or increase the global thread pool limit? | 2019-09-22T03:54:36 | 2019-09-22T03:54:36 | {} | NONE |
ClickHouse | ClickHouse | 533,225,706 | 6,981 | proller | No. its not normal.
Its worked before https://github.com/yandex/ClickHouse/commit/b4339f266df17c0cd873ef10249647409e66f229
git submodule update --init cant run on sources downloaded without git.
If cmake see no code on filesystem - this code bcome disabled and skipped from build.
Please just repair what you b... | 2019-09-19T17:14:15 | 2019-09-19T17:14:15 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 533,373,488 | 6,028 | shangshujie365 | > Ok.
>
> Now the next step:
> We have exact information about the number of rows in MergeTree tables. If the user do simple `SELECT count() ... FROM table` query, we can return the answer instantly.
And we can get 'rows' from system.parts, it can be used for query 'select count() from table where partition = xx... | 2019-09-20T02:05:14 | 2019-09-20T02:05:14 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 533,568,200 | 7,018 | den-crane | > Ваша проблема в функции max() она вернет Float64, который не поддерживается в движке Join (в качестве ключей)
```
CREATE TABLE IF NOT EXISTS max_versions_test
ENGINE = Join(ANY, INNER, id, ver) AS
SELECT
id,
MAX(ver) AS ver
FROM test
GROUP BY id
Ok.
0 rows in set. Elapsed: 0.864 sec.
:) des... | 2019-09-20T14:05:34 | 2019-09-20T14:06:38 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 533,623,338 | 7,018 | AntonSaykovsky | Видимо пока придётся загнать id и max(ver) под хэш функцию, а в основной таблице id/ver и джойнить по результату. Какая из hash(int,isn’t):UInt64 на текущий момент самая быстрая не подскажите? | 2019-09-20T16:29:09 | 2019-09-20T16:29:09 | {} | NONE |
ClickHouse | ClickHouse | 533,843,844 | 6,894 | alexey-milovidov | BTW The only way to provide proper performance tests (integrated in CI) is to finish https://github.com/yandex/ClickHouse/pull/4178 first. | 2019-09-22T02:12:17 | 2019-09-22T02:12:25 | {} | MEMBER |
ClickHouse | ClickHouse | 533,327,232 | 6,994 | 4ertus2 | According to trace the crash is on Nullable(Float64) column. If only you do not use join_use_nulls=1 on some node. You may try to select only Float64 columns.
Also check that there’s no mix of configs affecting joins on nodes. | 2019-09-19T22:07:48 | 2019-09-19T22:07:48 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 533,520,052 | 6,946 | akuzm | Fixed in #6951. | 2019-09-20T11:46:09 | 2019-09-20T11:46:09 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 533,609,639 | 6,615 | alexey-milovidov | @nicelulu And now it's assigned to you. | 2019-09-20T15:49:09 | 2019-09-20T15:49:09 | {} | MEMBER |
ClickHouse | ClickHouse | 533,718,962 | 6,781 | tavplubix | Failed tests and known issues:
- `00915_simple_aggregate_function` and `00725_ipv4_ipv6_domains` - `CAST('string literal', 'IPv4')` doesn't work because it parses string literal as `UInt32`, not as `IPv4`
- `00748_insert_array_with_null` - different error code (seems to be ok)
- `00700_decimal_arithm`, `00700_decima... | 2019-09-20T21:44:22 | 2019-09-20T21:44:41 | {} | MEMBER |
ClickHouse | ClickHouse | 533,847,166 | 6,833 | ghost | Sorry for the delay on responding to this.
Just got into this state again and `show processlist` is unable to run since a task cannot be scheduled. After restarting, `show processlist` shows nothing. | 2019-09-22T03:40:45 | 2019-09-22T03:40:45 | {} | NONE |
ClickHouse | ClickHouse | 533,879,948 | 6,894 | azat | > I don't care too much about the number of commits.
I don't care about number of commits either, I'm care about history, but until you are fine with this, I'm okay.
>merged
Thanks! | 2019-09-22T12:59:27 | 2019-09-22T12:59:27 | {} | MEMBER |
ClickHouse | ClickHouse | 533,271,377 | 6,998 | 4ertus2 | Could you reproduce it on 19.13 or 19.14? There was a fix in this place that is hard to backport. Most probably it's a duplicate for #5859 | 2019-09-19T19:19:23 | 2019-09-19T19:21:46 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 533,406,443 | 7,006 | slimtom95 | How do you create those tables, and what's your cluster configuration?
Read the doc carefully, it should work fine | 2019-09-20T05:01:42 | 2019-09-20T05:03:07 | {} | NONE |
ClickHouse | ClickHouse | 533,440,033 | 6,994 | 4ertus2 | If you set join_use_nills=1 JOIN could make nullable column from not nullable one. According to trace there were a Nullable column, so a source column is nullable or join_use_nulls was set.
I think if we set this flag on one node and do not set on another we could get the same troubles: there would be columns with dif... | 2019-09-20T07:20:02 | 2019-09-20T07:20:33 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 533,444,694 | 5,747 | kainever7 | @blinkov Kerberos support for ZK and HDFS on 28 Jun ---- Which version？ | 2019-09-20T07:36:06 | 2019-09-20T07:36:06 | {} | NONE |
ClickHouse | ClickHouse | 533,481,273 | 5,747 | blinkov | @kainever7 as far as I remember it's still missing | 2019-09-20T09:29:10 | 2019-09-20T09:29:10 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 533,585,186 | 7,018 | den-crane |
> Возможно есть другой метод без движка Join выбрать все строки из левой таблицы с максимальной версией, если для одной версии может быть больше 1 строки?
select argMax(tuple(*),ver) from test; | 2019-09-20T14:49:48 | 2019-09-20T14:49:48 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 533,615,438 | 7,022 | 4ertus2 | Closes #5254 | 2019-09-20T16:04:59 | 2019-09-20T16:07:58 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 533,664,013 | 7,009 | filimonov | Interesting. Can you show create table statement? Engine, Keys (order by, partition by, skipping index) , and columns used in query should be enough.
Also please repeat both queries in clickhouse-client, with `set send_logs_level='trace' `and send the results. | 2019-09-20T18:32:14 | 2019-09-20T18:32:14 | {
"+1": 1
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 533,705,497 | 6,968 | alexey-milovidov | It's Ok. Added a comment. | 2019-09-20T20:53:46 | 2019-09-20T20:53:46 | {} | MEMBER |
ClickHouse | ClickHouse | 533,181,738 | 6,972 | qMBQx8GH | Hi 4ertus2!
Here is .sql's from metadata directory:
[b24stat.zip](https://github.com/yandex/ClickHouse/files/3631998/b24stat.zip)
| 2019-09-19T15:21:38 | 2019-09-19T15:21:38 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 533,411,764 | 2,416 | yuzhichang | @alexey-milovidov I noticed that `WITH clause` speedups my query which involves a big bitmap(cardinality 99M ) filter from 394s to 93s. Could you explain why it happen? Thanks!
The original query:
```
SELECT
toDate ( order_complete_time ) AS rowData,
city_tier AS columnData,
count( DISTINCT mid_seq... | 2019-09-20T05:30:20 | 2019-09-20T05:30:58 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 533,491,335 | 6,681 | audricbizouarn | +1 | 2019-09-20T09:59:35 | 2019-09-20T09:59:35 | {} | NONE |
ClickHouse | ClickHouse | 533,516,562 | 6,876 | stavrolia | This PR moved to #7016 | 2019-09-20T11:31:54 | 2019-09-20T11:31:54 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 533,706,741 | 6,957 | alexey-milovidov | The name is not descriptive enough (it does not explain what subset we extract).
Does `bitmapSmallestValuesFrom` look better? | 2019-09-20T20:57:55 | 2019-09-20T20:58:03 | {} | MEMBER |
ClickHouse | ClickHouse | 533,711,887 | 6,993 | den-crane | ignored_ parts are inactive parts not deleted because of CH spontaneous reboot.
When you attaching them you making duplicates of existing data. | 2019-09-20T21:17:42 | 2019-09-20T21:17:42 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 533,775,111 | 6,957 | yuzhichang | I think `bitmapSubsetLimit` has better relationship with `bitmapSubsetInRange` than `bitmapSmallestValuesFrom`.
Maybe `bitmapSubsetWithLimit`? | 2019-09-21T07:12:34 | 2019-09-21T07:12:34 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 533,150,688 | 5,417 | amosbird | Yeah, that was a difficulty when I built this hash map, which leads me to disable consecutive caches for certain maps. | 2019-09-19T14:15:22 | 2019-09-19T14:15:22 | {} | COLLABORATOR |
ClickHouse | ClickHouse | 533,616,725 | 6,991 | alexey-milovidov | I thought about something like
```
CREATE FUNCTION f ... ENGINE = SQL
```
Where engine is mapped to something like `IUserDefinedFunctionCreator` with the main responsibility is to register a new function in FunctionFactory.
PS. C++ UDFs should be the last thing that we implement.
The main reason is debug-ab... | 2019-09-20T16:08:37 | 2019-09-20T16:08:56 | {
"+1": 2
} | MEMBER |
ClickHouse | ClickHouse | 533,708,365 | 6,997 | den-crane | it's unsolvable without reproducible example.
How do you insert the data? Go / native ? | 2019-09-20T21:03:51 | 2019-09-20T21:03:51 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 533,825,438 | 6,894 | azat | > could you publish the benchmark so that I can reproduce your results?
For now it is here:
https://github.com/azat-archive/hashtable-bench.git
(although it requires some #ifdef cleanups)
> Maybe it even makes sense to save it for the future in Dictionary/tests
Hm, this will requires even more submodules, but ... | 2019-09-21T19:47:01 | 2019-09-21T19:47:01 | {} | MEMBER |
ClickHouse | ClickHouse | 533,392,295 | 6,993 | flyfoxCI | attach part seems not work. I found the partition_id in system.detached_parts and execute alter table attach partition id ... at last the data recover! thanks you | 2019-09-20T03:41:02 | 2019-09-20T03:41:02 | {} | NONE |
ClickHouse | ClickHouse | 533,424,016 | 6,994 | ghuname | I don't understand the sentence: "If only you do not use join_use_nulls=1 on some node".
Can you please rephrase it?
Anyway, all 10 nodes have setting set join_use_nulls = 0.
I found what caused an error. This is the case.
At the time t0, base tables are created with Nullable columns.
After creation of the bas... | 2019-09-20T06:21:27 | 2019-09-20T06:21:27 | {} | NONE |
ClickHouse | ClickHouse | 533,605,230 | 6,615 | alexey-milovidov | > Sorry, I didn't see it when I was doing it.
That's all right. The only thing that matters is that the task is done :) | 2019-09-20T15:37:57 | 2019-09-20T15:37:57 | {} | MEMBER |
ClickHouse | ClickHouse | 533,771,471 | 6,998 | nauu | > Could you reproduce it on 19.13 or 19.14? There was a fix in this place that is hard to backport. Most probably it's a duplicate for #5859
I have been tested on 19.14.6.12 , the bug has been fixed, thx . | 2019-09-21T06:06:16 | 2019-09-21T06:06:16 | {} | MEMBER |
ClickHouse | ClickHouse | 533,821,453 | 7,037 | silviucpp | @alexey-milovidov do you think somebody more familiar with the CMake files can help with fixing the linking process ?. I'll keep looking myself as well in the meantime. | 2019-09-21T18:46:58 | 2019-09-21T18:46:58 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 533,829,973 | 7,037 | silviucpp | Ohh I didn't saw that you are already working on fixing this. You can get only the things that are not already covered in #7028 and close this one. | 2019-09-21T21:03:03 | 2019-09-21T21:03:03 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 533,857,922 | 6,894 | azat | @alexey-milovidov objections on squashing the following commits?
```
a12cffee91 Update HashedDictionary.h
ab9a7be45a Comments are the must.
9afea6f0b3 Update HashedDictionary.cpp
``` | 2019-09-22T07:43:38 | 2019-09-22T07:43:38 | {} | MEMBER |
ClickHouse | ClickHouse | 533,868,030 | 7,007 | nicelulu | This is the normal result:
```
SELECT count()
FROM
(
SELECT *
FROM system.numbers
LIMIT 1000
)
WHERE 1 IN
(
SELECT 0
WHERE 0
)
┌─count()─┐
│ 0 │
└─────────┘
1 rows in set. Elapsed: 0.008 sec. Processed 1.00 thousand rows, 8.00 KB (117.63 thousand rows/s., 941.06 KB/s.)
``... | 2019-09-22T10:16:20 | 2019-09-22T10:40:17 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 533,879,380 | 7,044 | den-crane | not a bug, but a documentation issue:
> Denny Crane (I don't work at Yandex (never did)), [Sep 20, 2019 at 9:38:35 AM]:
> decimal-ы и float-ы округляются (round) по разному, первый математически, второй по-банковски
> отразить в доке или создать bug на decimal ?
>
> Alexey Milovidov, [Sep 20, 2019 at 10:11:17 A... | 2019-09-22T12:51:06 | 2019-09-22T12:51:06 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 533,118,242 | 6,993 | filimonov | Please try to find the reason in logs why it was ignored. It may be covered by other parts and you don't need to do anything.
Otherwise - create the table with the same structure as original, put that part into the `detached` folder for that new table (removing `ignored_` prefix), and do `alter table ... attach part... | 2019-09-19T12:56:28 | 2019-09-19T12:56:28 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 533,609,249 | 7,018 | AntonSaykovsky | > типа collapsingVersionMT ?
>
> SELECT argMax(tuple(*), (ver,-sign)) FROM test GROUP BY id
>
> select * from test order by id,ver desc, sign asc limit 1 by id
Типо collapsingVersionMT, но не совсем, 'схопывание' же там по праймари идет в любом случае и схему с sign надо соблюдать; у нас же 30+ полей, все под ... | 2019-09-20T15:48:07 | 2019-09-20T15:48:07 | {} | NONE |
ClickHouse | ClickHouse | 533,665,042 | 7,010 | filimonov | Is there some mutation (update/delete) is executed? Can you check if there are some failed mutation in `system.mutations`? | 2019-09-20T18:35:15 | 2019-09-20T18:35:15 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 533,843,181 | 6,957 | alexey-milovidov | That's reasonable. Let's keep your initial naming. | 2019-09-22T01:55:04 | 2019-09-22T01:55:04 | {} | MEMBER |
ClickHouse | ClickHouse | 533,880,234 | 7,044 | SergeyMirvoda | Отлично! | 2019-09-22T13:03:12 | 2019-09-22T13:03:12 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 533,125,343 | 6,981 | filimonov | Yep. I think @proller was just expecting to see that user-friendly warning, instead of cryptic ' Cannot find source file'
https://github.com/yandex/ClickHouse/blob/06ea75f9fd5581fdeb10ab264ed73f159ab3797e/cmake/find_capnp.cmake#L20 | 2019-09-19T13:15:51 | 2019-09-19T13:16:46 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 533,158,098 | 6,986 | KochetovNicolai | Yes, this setting is only for aggregation phase. Setting `max_bytes_before_external_sort` is only for sorting at `ORDER BY`. | 2019-09-19T14:31:27 | 2019-09-19T14:31:27 | {} | MEMBER |
ClickHouse | ClickHouse | 533,294,048 | 6,615 | alexey-milovidov | Assigned to @millb | 2019-09-19T20:21:25 | 2019-09-19T20:21:25 | {} | MEMBER |
ClickHouse | ClickHouse | 533,377,611 | 5,109 | geldot | The COLUMNS() matcher from #6038 is merged. However it can only select, not deselect, columns due to re2 regex syntax limitations. So a way to negate is still needed. COLUMNS_EXCEPT()? | 2019-09-20T02:27:02 | 2019-09-20T02:27:02 | {
"+1": 1
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 533,389,674 | 6,808 | geldot | Until there are better builtins this sort of works:
```SQL
WITH
['192.168.1.5', '192.168.2.5'] AS ips,
['192.168.1.1/24'] AS nets,
arrayFirst(x -> IPv4StringToNum(ip) BETWEEN x.1 AND x.2, arrayMap(x -> IPv4CIDRToRange(IPv4StringToNum(splitByChar('/',x)[1]), toUInt8(splitByChar('/',x)[2])), nets)) AS net
SELECT ... | 2019-09-20T03:27:14 | 2019-09-20T03:27:14 | {
"+1": 1
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 533,602,432 | 7,018 | den-crane | типа collapsingVersionMT ?
SELECT argMax(tuple(*), (ver,-sign)) FROM test GROUP BY id
select * from test order by id,ver desc, sign asc limit 1 by id | 2019-09-20T15:31:56 | 2019-09-20T15:33:52 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 533,614,110 | 6,028 | alexey-milovidov | @shangshujie365 True.
(you mean `_part` virtual column) | 2019-09-20T16:01:28 | 2019-09-20T16:01:28 | {} | MEMBER |
ClickHouse | ClickHouse | 533,702,298 | 6,969 | alexey-milovidov | Let's add `<events_absolute>` that can be enabled and disabled independent with `<events>`.
The user will even be able to enable both.
Absolute values should go to `ProfileEventsAbsolute` (or `ProfileEventsAbs`) path. | 2019-09-20T20:43:10 | 2019-09-20T20:43:48 | {} | MEMBER |
ClickHouse | ClickHouse | 533,746,029 | 6,969 | azat | >Let's add <events_absolute> that can be enabled and disabled independent with <events>.
Done | 2019-09-20T23:47:43 | 2019-09-20T23:47:43 | {} | MEMBER |
ClickHouse | ClickHouse | 533,775,865 | 7,010 | viputh6 | Yes. you are right. it is fixed after killing those failed mutation.
Thanks a lot filimonov | 2019-09-21T07:26:48 | 2019-09-21T07:26:48 | {} | NONE |
ClickHouse | ClickHouse | 533,802,867 | 6,914 | maqroll | Working in the integration test. | 2019-09-21T14:37:29 | 2019-09-21T14:37:29 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 533,132,935 | 6,972 | 4ertus2 | OK. Thanks.
Could you also attach your schema (only used tables and columns are needed). We'll try to reproduce the issue. | 2019-09-19T13:34:02 | 2019-09-19T13:34:02 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 533,473,401 | 6,948 | imiskolee | can we have any temp solutions for this case? beacuse we're global SaaS service. | 2019-09-20T09:05:54 | 2019-09-20T09:05:54 | {} | NONE |
ClickHouse | ClickHouse | 533,548,282 | 6,956 | yingfeng | This is the screenshot taken from the `callgrind` output of the above SQL:
<img width="308" src="https://user-images.githubusercontent.com/7248/65329060-4b14df00-dbea-11e9-8eff-3be0f30b3e5d.png">
<img width="697" src="https://user-images.githubusercontent.com/7248/65366694-fb6dfc00-dc59-11e9-80b2-dfbbffdf3d4e.pn... | 2019-09-20T13:11:42 | 2019-09-21T02:27:02 | {} | NONE |
ClickHouse | ClickHouse | 533,843,339 | 7,007 | alexey-milovidov | Some tests have failed. | 2019-09-22T01:59:29 | 2019-09-22T01:59:29 | {} | MEMBER |
ClickHouse | ClickHouse | 533,173,187 | 6,991 | amosbird | Interesting interface. At first glance I was thinking that we can have dynamic UDFs storing the code text along with the data. It seems not yet possible though. Since we are already doing code generation by text manipulation, shall we use sized array instead of std::vector for column data? | 2019-09-19T15:02:54 | 2019-09-19T15:03:24 | {} | COLLABORATOR |
ClickHouse | ClickHouse | 533,481,016 | 7,008 | blinkov | #5747 | 2019-09-20T09:28:22 | 2019-09-20T09:28:22 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 533,585,208 | 7,018 | 4ertus2 | Посмотрел по коду, в StorageJoin прилетает попытка использовать Type::keys128 в качестве ключа (т.к. ключ составной). Похоже, engine Join пока не поддерживает составные ключи (позволяет создать, но в части использовать недоимплемент). | 2019-09-20T14:49:51 | 2019-09-20T14:50:12 | {
"+1": 2,
"laugh": 1
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 533,593,481 | 6,615 | nicelulu | > Assigned to @millb
Sorry, I didn't see it when I was doing it. | 2019-09-20T15:10:50 | 2019-09-20T15:10:50 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 533,667,593 | 6,948 | filimonov | @imiskolee can you please descibe your usecase? The best workaround in common case is to store offset in seconds in separate field.
I.e.
```
create table x (
time_utc DateTime,
tz_offset Int32) engine=Log;
Insert into x values (now(), 3600), (now() + 1, -7200):
Select toString(time_utc + tz_offset) from x;
... | 2019-09-20T18:43:11 | 2019-09-20T18:43:34 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 533,796,624 | 7,019 | geldot | The regression has now been introduced to the 19.13 series as of yesterday's point release:
HTTP requests with `cancel_http_readonly_queries_on_client_close=1`:
Failing: 19.13.5.44
Working: 19.13.4.32
| 2019-09-21T13:06:07 | 2019-09-21T13:06:07 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 533,843,163 | 7,037 | alexey-milovidov | #7028 is in progress. I will merge this first. | 2019-09-22T01:54:25 | 2019-09-22T01:54:25 | {} | MEMBER |
ClickHouse | ClickHouse | 533,153,686 | 6,994 | ghuname | Please give to me some time to get the approval from my superior. | 2019-09-19T14:22:12 | 2019-09-19T14:22:12 | {} | NONE |
ClickHouse | ClickHouse | 533,405,817 | 2,542 | alexcohaniuc | Having the same issue. Is this planned to be fixed. Or are we doing something wrong | 2019-09-20T04:58:13 | 2019-09-20T04:58:13 | {} | NONE |
ClickHouse | ClickHouse | 533,510,545 | 7,013 | chaitas90 | My mistake using recent clickhouse-jdbc 2.0 we can achieve it so closing this issue.. | 2019-09-20T11:08:46 | 2019-09-20T11:08:46 | {} | NONE |
ClickHouse | ClickHouse | 533,524,241 | 6,972 | 4ertus2 | Probably related #6767 | 2019-09-20T12:00:33 | 2019-09-20T12:00:33 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 533,549,906 | 1,827 | blitzarx1 | @filimonov @proller @alexey-milovidov Hi there I have the same problem as @631068264 with your official docker build. When I start container with additional config file, I do it like it is said in your docker instruction
```bash
docker run -d --name some-clickhouse-server --ulimit nofile=262144:262144 -p 9000:9000... | 2019-09-20T13:16:41 | 2019-09-20T13:18:44 | {} | NONE |
ClickHouse | ClickHouse | 533,665,597 | 7,018 | den-crane | xxHash64.
Engine=Join это in-memory движок (как впрочем и сам HashJoin). | 2019-09-20T18:36:57 | 2019-09-20T18:36:57 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 533,829,234 | 7,037 | alexey-milovidov | Most of the changes are already prepared in #7028
Your PR also has some meaningful changes that are not in #7028
I can merge it when Linux builds will be finished (I have no Mac to check OS X build).
Only cross-compilation (that is implemented in #7028) will solve this problem.
PS.
> From what I understand tha... | 2019-09-21T20:49:58 | 2019-09-21T20:49:58 | {} | MEMBER |
ClickHouse | ClickHouse | 533,387,835 | 6,991 | hczhcz | > shall we use sized array instead of std::vector for column data?
The data is passed per-block. If the array size is given in compile-time, it will require re-compiling for different blocks.
As an alternative, we may pass the array size and a pointer to the data. The performance will be much better than std::vecto... | 2019-09-20T03:16:57 | 2019-09-20T03:28:52 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 533,445,076 | 3,750 | fancyqlx | thanks, we did not see it again. | 2019-09-20T07:37:20 | 2019-09-20T07:37:20 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 533,575,411 | 7,018 | 4ertus2 | Попробуйте toUInt64(). С UInt8 должно тоже работать - видимо баг. | 2019-09-20T14:24:47 | 2019-09-20T14:24:47 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 533,694,829 | 7,019 | geldot | OK, it looks like missing coverage: The existing test is only checking for a timeout, which also happens when the HTTP request hangs, so it doesn't fail). A more basic `SELECT 1` test as above will catch this. | 2019-09-20T20:16:29 | 2019-09-20T20:16:29 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 533,821,123 | 7,037 | silviucpp | On OSX `si_overrun` from `siginfo_t` is not available and I don't see similar functionality . Not sure if the fix I did in QueryProfiler is the proper one (just make sure that part is not executed on OSX). From what I understand that part is to make sure that in case the signals are called too frequently the performanc... | 2019-09-21T18:42:20 | 2019-09-21T18:42:20 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 533,122,743 | 6,994 | filimonov | Can you provide create table statements from `table_a` and `table_b`? If it is distributed tables - underlying tables too.
If possible - try to minimize the schema only to the minimum number of columns which allow reproducing the problem.
So start with
`select customerId, A_Dn_Sem, A_Up_Sem FROM database.table... | 2019-09-19T13:08:54 | 2019-09-19T13:08:54 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 533,141,059 | 5,417 | akuzm | Some news about what I'm working on now: before, I changed emplace() to return a pointer to Mapped type, not an iterator. I realized that this makes it impossible to support a consecutive keys cache, because the cache must store the persistent key, and there is no way to get it from the table. The code that is now in #... | 2019-09-19T13:53:22 | 2019-09-19T13:53:22 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 533,479,736 | 5,618 | kainever7 | @den-crane
should not use the disk? Why still reporting memory limits?
Because final aggregation result should fit into memory.
So for group by SOME_HIGH_CARDINALITY_SET_OF_COLUMNS, ch will do intermediate partial aggregations on a disk, but in the end CH will gather the result into one part in memory.
---- ... | 2019-09-20T09:24:27 | 2019-09-20T09:24:27 | {
"+1": 1
} | NONE |
ClickHouse | ClickHouse | 533,563,372 | 7,018 | 4ertus2 | Ваша проблема в функции max() она вернет Float64, который не поддерживается в движке Join (в качестве ключей) | 2019-09-20T13:53:56 | 2019-09-20T13:54:51 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 533,599,075 | 5,635 | alexey-milovidov | 1. `clickhouse-performance-test filename.xml`
Does it assume that "filename.xml" is a garbage and completely ignores it instead of throwing an exception about unsupported command line parameter? This is bug.
2. Compare, how easy is to run `clickhouse-test` for specific test case. I expect that `clickhouse-perform... | 2019-09-20T15:24:48 | 2019-09-20T15:25:14 | {} | MEMBER |
ClickHouse | ClickHouse | 533,609,375 | 6,615 | alexey-milovidov | But it isn't quite done. | 2019-09-20T15:48:26 | 2019-09-20T15:48:26 | {} | MEMBER |
ClickHouse | ClickHouse | 533,879,774 | 6,894 | alexey-milovidov | Yes, you can always squash my commits in your PR.
I merged this PR because I don't care too much about the number of commits. | 2019-09-22T12:56:47 | 2019-09-22T12:56:47 | {} | MEMBER |
ClickHouse | ClickHouse | 539,180,607 | 6,448 | alexey-milovidov | BTW, I found these changes very useful. Especially: forcing functions to have internal linkage, forcing `override`... | 2019-10-07T20:04:12 | 2019-10-07T20:04:12 | {} | MEMBER |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.