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 | 830,833,462 | 21,309 | amosbird | It seems we should utilize row policy filter for this purpose. However, currently row filters don't take advantage of any indices. Maybe we should extend it. Or is there any consideration? | 2021-05-02T16:15:16 | 2021-05-02T16:15:16 | {} | COLLABORATOR |
ClickHouse | ClickHouse | 831,132,312 | 23,775 | tritas | Hi. I still hit the authorization header error with the CI build linked (#23844).
Looking at the PRs, it seems the region fix is in #23846 - would it be possible to run CI on that? | 2021-05-03T09:13:43 | 2021-05-03T09:13:43 | {} | NONE |
ClickHouse | ClickHouse | 831,202,717 | 23,860 | alexey-milovidov | That's a good feature. It can be beneficial to enable streaming lz4 compression and disable gz compression of rotated files.
Just to confirm, is it true that:
- lz4 framing format allows streaming compression with appends and it can decompress the data at any point of time?
- the integration in OS (at least Ubuntu... | 2021-05-03T11:34:09 | 2021-05-03T11:34:17 | {} | MEMBER |
ClickHouse | ClickHouse | 830,550,050 | 19,569 | filimonov | > > What is needed in that task - is the ability to stop writing err log
>
> We already have this option.
Yes. And since we can disable writing err log at all - why do you think it is better than giving the user a choice - not write at all, to write errors only, or fatal only? | 2021-05-01T05:38:41 | 2021-05-01T05:39:32 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 830,353,941 | 54 | olgarev | @alexey-milovidov The documentation was updated. | 2021-04-30T20:11:19 | 2021-04-30T20:11:19 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 830,412,162 | 19,627 | filimonov | > I afraid that those files will be almost always empty
Because of that applying that same way as prewhere may have sense (i doubt that we will be able to skip whole granulas often). And if we will use it as skipping index - we anyway will need to process it as a regular filter, similarly to prewhere. So it seems li... | 2021-04-30T21:33:06 | 2021-04-30T21:36:56 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 830,568,895 | 19,569 | alexey-milovidov | > And since we can disable writing err log at all - why do you think it is better than giving the user a choice - not write at all, to write errors only, or fatal only?
It is questionable feature. We can just drop and forget about it and concentrate on other features. | 2021-05-01T07:09:43 | 2021-05-01T07:09:43 | {} | MEMBER |
ClickHouse | ClickHouse | 830,668,642 | 23,775 | excitoon | @tritas Build from this PR #23844 shall solve your issue :man_shrugging: | 2021-05-01T17:49:13 | 2021-05-01T17:49:13 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 830,780,082 | 23,821 | kitaisreal | @den-crane we does not change range dictionary much during dictionaries refactoring. Are there any performance issues with 21.1, 21.2, 21.3, 21.4 versions ?
Just need to be sure that it is bad performance, not a regression. | 2021-05-02T09:41:20 | 2021-05-02T09:41:20 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 831,188,759 | 23,656 | alexey-milovidov | It is possible to check for terminal and obtain its properties with `/dev/tty` file. | 2021-05-03T11:03:58 | 2021-05-03T11:03:58 | {} | MEMBER |
ClickHouse | ClickHouse | 830,652,471 | 23,790 | flyfoxCI | in fa
> ClickHouse may use memory for lot of different (valid) scenarios dictionaries, memory tables, some kind of long running queries. What make you think it's a leak?
>
> Before continue - please update clickhouse at least to the latest 20.8
> ClickHouse may use memory for lot of different (valid) scen... | 2021-05-01T15:47:17 | 2021-05-01T15:47:38 | {} | NONE |
ClickHouse | ClickHouse | 831,211,110 | 23,851 | excitoon | Okay
On Mon, May 3, 2021, 2:00 PM alexey-milovidov ***@***.***>
wrote:
> @filimonov <https://github.com/filimonov>
>
> Lack of memory may produce std::exception.
>
> Not an issue.
>
> Please open another issue with correct title and description.
>
> —
> You are receiving this because you were mentioned.
> Reply to th... | 2021-05-03T11:54:04 | 2021-05-03T11:54:04 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 831,230,791 | 23,862 | dgloeckner | Note that this issue is different from #15022 as no exception is thrown in the INSERT here. | 2021-05-03T12:35:26 | 2021-05-03T12:35:26 | {} | NONE |
ClickHouse | ClickHouse | 830,348,166 | 23,821 | den-crane | zero lifetime
```sql
CREATE DICTIONARY curs_dict
(
`a` Unt64,
`b` UInt64,
`e` UInt64,
`rate` UInt16
)
PRIMARY KEY a
SOURCE(CLICKHOUSE(TABLE curs DB 'default' USER 'default'))
LIFETIME(0)
LAYOUT(RANGE_HASHED)
RANGE(MIN b MAX e);
0 rows in set. Elapsed: 125.655 sec.
SELECT sum(dictG... | 2021-04-30T19:59:42 | 2021-04-30T20:06:45 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 830,407,563 | 19,627 | filimonov | > but there will be just 1-2 files per part, not bigger then INSERT does, yes there are users who creates lots of parts but does the complexity worth optimizing DELETE for this case?
Yes. I afraid that those files will be almost always empty, while writing them will require to allocate lot of buffers (see https://gi... | 2021-04-30T21:24:24 | 2021-04-30T21:35:08 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 830,512,381 | 23,790 | den-crane | Check what query is eating memory `grep Peak /var/log/clickhouse-server/clickhouse-server.log` | 2021-05-01T04:01:40 | 2021-05-01T04:01:40 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 830,786,330 | 23,844 | excitoon | @Jokser I see, thank you, this is a typo. | 2021-05-02T10:27:47 | 2021-05-02T10:27:47 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 830,857,783 | 23,854 | blanchonvincent | @itzikiusa thank you for pointing this mistake out, that works now!
I'm using now `uniqExact`, `uniqExactState` and `uniqExactMerge`. However, the query on the view is now barely faster than the query on the table `0.827s` vs `0.746s`. It looks like there was other cases related to [slow query with `uniqExact`](https:... | 2021-05-02T19:17:56 | 2021-05-02T19:22:58 | {} | NONE |
ClickHouse | ClickHouse | 831,175,967 | 23,859 | alexey-milovidov | `coalesce(x, 0)` or `ifNull(x, 0)` should be used instead.
`assumeNotNull` is for the case when it's assumed that the value is not NULL. The user is responsible for this assumption.
If assumption does not hold, the function returns implementation specific result.
In other words: `assumeNotNull` is an optimized f... | 2021-05-03T10:37:22 | 2021-05-03T10:38:19 | {} | MEMBER |
ClickHouse | ClickHouse | 831,177,072 | 23,859 | UnamedRus | Thank you
| 2021-05-03T10:39:40 | 2021-05-03T10:39:40 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 831,220,157 | 23,860 | filimonov | But there is a question - how well it will be able to compress in streaming mode (i guess if we will compress each line individually the compression will be poor), but i think it should have some modes for that. I.e. start appending only when clickhouse restart, otherwise write it a single stream.
UPD: it seems we ... | 2021-05-03T12:13:25 | 2021-05-03T13:08:52 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 830,350,524 | 23,800 | SaltTan | Another test
```
drop table if exists table_for_update_field_dictionary;
create table table_for_update_field_dictionary
(
flat_key UInt64,
complex_key Int32 default flat_key,
value String,
last_insert_time DateTime default now()
) engine=Log;
insert into table_for_update_field_dictionary (flat_key)... | 2021-04-30T20:04:32 | 2021-04-30T20:04:32 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 830,357,518 | 23,555 | genzgd | I just checked the logs on a couple of the other affected servers and they all show "Memory limit exceeded" exceptions right about the time that the GET_PART tasks got stuck. I'm sorry I didn't notice that before. | 2021-04-30T20:18:02 | 2021-04-30T20:18:02 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 830,546,341 | 23,778 | filimonov | Hm. Can you also try to enable odbc bridge logs like here:
https://github.com/ClickHouse/ClickHouse/blob/9f5cd35a6963cc556a51218b46b0754dcac7306a/tests/integration/test_odbc_interaction/configs/odbc_logging.xml
| 2021-05-01T05:29:51 | 2021-05-01T05:29:51 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 830,549,127 | 19,569 | filimonov | Actually it's one line fix needed... | 2021-05-01T05:36:13 | 2021-05-01T05:36:13 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 830,574,843 | 19,627 | azat | >Yes. I afraid that those files will be almost always empty, while writing them will require to allocate lot of buffers (see #6943 ). Imagine you removed 1 row from the table with 1000 parts. That will require creating 1998 empty files, and 2 files with some data...
Agree this might be a problem but I think that sol... | 2021-05-01T07:53:52 | 2021-05-01T07:53:52 | {} | MEMBER |
ClickHouse | ClickHouse | 830,675,364 | 23,789 | kitaisreal | Manual backport https://github.com/ClickHouse/ClickHouse/pull/23835. | 2021-05-01T18:31:57 | 2021-05-01T18:31:57 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 830,927,588 | 23,854 | den-crane | There is no any sense to use uniqExact & AggregatingMergeTree.
Try the opposite : hll12State / hll12Merge (or uniqCombine). hyperloglog structures bring % of errors, because they probabilities, but they use 1000 times less space and 1000 times faster. | 2021-05-02T23:44:34 | 2021-05-02T23:44:34 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 830,928,874 | 23,793 | fuwhu | > Let's simply obtain the number of rows/bytes always without `select_sequential_consistency` for system.tables.
>
> How to do it:
>
> * you can copy the settings object;
> * turn off `select_sequential_consistency`;
> * use these settings in getTotalRows/Bytes.
@alexey-milovidov
Thanks Alexey for the sugg... | 2021-05-02T23:54:17 | 2021-05-02T23:54:17 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 831,142,266 | 23,775 | excitoon | Can you provide trace log again or may be you can share some link with some
credentials with me privately, ***@***.*** so I could
reproduce it exactly? Second solution shall definitely help you but this
issue needs to be investigated anyway.
On Mon, May 3, 2021, 12:14 PM tritas ***@***.***> wrote:
> Hi. I still hit t... | 2021-05-03T09:33:10 | 2021-05-03T09:33:10 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 831,174,584 | 23,859 | alexey-milovidov | But it is how it supposed to work. | 2021-05-03T10:34:42 | 2021-05-03T10:34:42 | {} | MEMBER |
ClickHouse | ClickHouse | 830,668,982 | 23,844 | excitoon | @Jokser Can you please look at my changes? | 2021-05-01T17:51:24 | 2021-05-01T17:51:24 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 830,763,961 | 23,790 | filimonov | > is the timeout exceeded query not release memory
It should. Please show how the query look like.
Also please update to the latest 20.8. | 2021-05-02T07:20:05 | 2021-05-02T07:20:05 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 830,796,716 | 22,398 | alexey-milovidov | Tensorbase (an experimental prototype) brief benchmark:
https://tensorbase.io/2021/04/20/base_reload.html | 2021-05-02T11:46:25 | 2021-05-02T11:46:25 | {} | MEMBER |
ClickHouse | ClickHouse | 831,057,420 | 23,719 | filimonov | For sure the reason is the same as described here: https://github.com/ClickHouse/ClickHouse/issues/23778#issuecomment-831055587
As a workaround - you can try to slice the data by some simple WHERE condition to smaller chunks - it may help.
I.e. instead of
```sql
INSERT INTO clickhouse_table SELECT * FROM odbc(.... | 2021-05-03T06:45:15 | 2021-05-03T06:45:15 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 831,238,955 | 23,702 | untitaker | See also https://github.com/ClickHouse/ClickHouse/issues/8684 | 2021-05-03T12:52:16 | 2021-05-03T12:52:16 | {} | NONE |
ClickHouse | ClickHouse | 830,156,397 | 23,778 | jhettler | Cool, I will try it later today and let you know! Really looking forward if it works properly, after spending some time asking our DevOps guys for more and more resources :)
Thanks so far! | 2021-04-30T15:02:38 | 2021-04-30T22:12:42 | {} | NONE |
ClickHouse | ClickHouse | 830,433,738 | 23,778 | jhettler | The memory consumption looks much better, unfortunately I am back with the same issues where I came from.
```
clickhouse1 :) INSERT INTO clickhouse.DETAIL_VIEWS select DATE_VALID, COMPANY_G2ID, VACANCY_G2ID, VACANCY_EXTERNAL_NAME, VACANCY_INTERNAL_NAME, RECRUITMENT_PROCESS_G2ID, RECRUITER_G2ID, FIRST_PUBLISHED_AT, ... | 2021-04-30T22:27:34 | 2021-04-30T22:27:34 | {} | NONE |
ClickHouse | ClickHouse | 830,568,748 | 19,569 | alexey-milovidov | > Actually it's one line fix needed...
Ok, do it. | 2021-05-01T07:08:36 | 2021-05-01T07:08:36 | {} | MEMBER |
ClickHouse | ClickHouse | 830,677,889 | 23,826 | den-crane | use setting
--optimize_skip_merged_partitions arg Skip partitions with one part with level > 0 in optimize final | 2021-05-01T18:52:47 | 2021-05-01T18:52:47 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 830,361,815 | 23,822 | excitoon | This definitely fixes behavior of `+` but I was not able to reproduce a bug with `=`. | 2021-04-30T20:26:57 | 2021-04-30T20:26:57 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 830,584,477 | 21,770 | taylor12805 | > > I am very keen to see when it will get merged?
>
> let's give the author a week to respond, and then we'll see
Sorry, I am working on the implementation. Have only weekend time to do it, so it is not going quite smoothly. Have a vacation from 5.1 to 5.5, I try to finish it druing this time | 2021-05-01T09:01:09 | 2021-05-01T09:01:09 | {
"+1": 1
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 830,810,365 | 23,851 | alexey-milovidov | @excitoon What is the point of this task?
Having exception in this case is perfectly correct. | 2021-05-02T13:32:01 | 2021-05-02T13:32:01 | {} | MEMBER |
ClickHouse | ClickHouse | 830,833,171 | 4,419 | amosbird | It's likely to be superceded by aggregate projections in https://github.com/ClickHouse/ClickHouse/pull/20202 | 2021-05-02T16:12:55 | 2021-05-02T16:12:55 | {} | COLLABORATOR |
ClickHouse | ClickHouse | 828,845,704 | 23,656 | kssenii | btw, if there is `<` (`clickhouse-client --progress --query "INSERT ..." < file`) in non-interactive mode in clickhouse-client, it results in `isatty(STDIN_FILENO) == false` (in `getTerminalWidth()`) => it cannot get terminal width and progress will be shown only with progress percentage and the amount of how much is a... | 2021-04-28T23:31:16 | 2021-05-03T09:58:46 | {} | MEMBER |
ClickHouse | ClickHouse | 831,176,693 | 23,859 | UnamedRus | Ok.
In that case docs looks a bit misleading:
```
Returned values
The original value from the non-Nullable type, if it is not NULL.
The default value for the non-Nullable type if the original value was NULL.
``` | 2021-05-03T10:38:55 | 2021-05-03T10:38:55 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 831,176,998 | 23,859 | alexey-milovidov | I have corrected the docs: https://github.com/ClickHouse/ClickHouse/commit/41a6fd677e633124a2341b22d518307ed9e596e2 | 2021-05-03T10:39:31 | 2021-05-03T10:39:31 | {} | MEMBER |
ClickHouse | ClickHouse | 831,186,756 | 23,851 | alexey-milovidov | @filimonov
> Lack of memory may produce std::exception.
Not an issue.
Please open another issue with correct title and description.
| 2021-05-03T10:59:57 | 2021-05-03T10:59:57 | {} | MEMBER |
ClickHouse | ClickHouse | 831,209,961 | 23,782 | gyuton | Internal documentation ticket: DOCSUP-9168. | 2021-05-03T11:51:15 | 2021-05-03T11:51:15 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 830,392,186 | 23,196 | kssenii | BTW, executing a query like `select number from numbers(1000000000) format Null;` in clickhouse-client on your branch does not render any progress bar, but it does so on master, looks like there is a bug.. | 2021-04-30T21:05:44 | 2021-04-30T21:07:04 | {} | MEMBER |
ClickHouse | ClickHouse | 830,536,971 | 19,569 | alexey-milovidov | > What is needed in that task - is the ability to stop writing err log
We already have this option. | 2021-05-01T04:46:51 | 2021-05-01T04:46:51 | {} | MEMBER |
ClickHouse | ClickHouse | 830,622,699 | 21,107 | petuhovskiy | Everything works now, but performance in some cases should be improved. There are also some places in the code that I think can be done better.
The main performance feature is to allow to query single row of the table. For example:
```sql
-- test.mv table declared above
SELECT * FROM test.mv WHERE Region = 'Mos... | 2021-05-01T12:12:51 | 2021-05-01T12:12:51 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 830,626,469 | 23,775 | tritas | Please find the error message below. Some exponentially backed-off retries are omitted and the bucket name is replaced with a ``<bucket>`` placeholder.
```
2021.05.01 12:03:07.093903 [ 2110027 ] {} <Warning> AWSClient: ClientConfiguration: Retry Strategy will use the default max attempts.
2021.05.01 12:03:12.204... | 2021-05-01T12:41:00 | 2021-05-01T12:41:00 | {} | NONE |
ClickHouse | ClickHouse | 830,671,111 | 23,775 | tritas | Amazing. Will craft a build and report back here (if I can win the fight against CMake).
Unless you want to mark it resolved ASAP. | 2021-05-01T18:05:16 | 2021-05-01T18:05:16 | {} | NONE |
ClickHouse | ClickHouse | 830,781,944 | 23,844 | Jokser | So, now any positive status code (200) will be treated as error? | 2021-05-02T09:57:30 | 2021-05-02T09:57:30 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 831,063,667 | 23,778 | filimonov | @jhettler
1) please check - just before the exception how much ram do `clickhouse-odbc-bridge` use? How much free RAM do you have at that moment?
2) There are chances that [this simple workaround](https://github.com/ClickHouse/ClickHouse/issues/23719#issuecomment-831057420) will help you.
| 2021-05-03T06:59:04 | 2021-05-03T07:02:07 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 831,070,781 | 23,778 | jhettler | I will check this around 10. | 2021-05-03T07:13:32 | 2021-05-03T07:13:32 | {
"+1": 1
} | NONE |
ClickHouse | ClickHouse | 830,540,615 | 23,766 | alexey-milovidov | https://cmake.org/cmake/help/latest/command/find_package.html
> Version ranges are only supported with CMake 3.19 or later.
But it's unclear what version of CMake do we have.
> set (LLVM_PATHS "/usr/local/lib/llvm")
But if we install it from packages, it will be in `/usr/lib/...` | 2021-05-01T05:11:36 | 2021-05-01T05:11:36 | {} | MEMBER |
ClickHouse | ClickHouse | 830,548,657 | 19,569 | filimonov | > > ability to control the level of stdout logger
>
> We also have this.
You are wrong. It uses the same logging level as main log. I need different (reread the first post)
https://github.com/ClickHouse/ClickHouse/blob/9f5cd35a6963cc556a51218b46b0754dcac7306a/base/loggers/Loggers.cpp#L146 | 2021-05-01T05:35:08 | 2021-05-01T05:41:56 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 830,581,052 | 23,793 | alexey-milovidov | Let's simply obtain the number of rows/bytes always without `select_sequential_consistency` for system.tables.
How to do it:
- you can copy the settings object;
- turn off `select_sequential_consistency`;
- use these settings in getTotalRows/Bytes. | 2021-05-01T08:34:53 | 2021-05-01T08:35:29 | {} | MEMBER |
ClickHouse | ClickHouse | 830,621,491 | 23,787 | kitaisreal | Manual backport https://github.com/ClickHouse/ClickHouse/pull/23835. | 2021-05-01T12:03:33 | 2021-05-01T12:03:33 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 830,677,690 | 23,826 | den-crane | already implemented
https://github.com/ClickHouse/ClickHouse/issues/12759
https://github.com/ClickHouse/ClickHouse/pull/15939 | 2021-05-01T18:51:14 | 2021-05-01T18:51:14 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 830,757,912 | 23,790 | flyfoxCI | > Check what query is eating memory `grep Peak /var/log/clickhouse-server/clickhouse-server.log
After checking the log with your methods I found that there's a query always excuting time out when memory going up.
```
executeQuery: Code: 209, e.displayText() = DB::NetException: Timeout exceeded while writing to ... | 2021-05-02T06:16:22 | 2021-05-02T07:18:04 | {} | NONE |
ClickHouse | ClickHouse | 830,668,120 | 23,778 | jhettler | Hi,
clikhouse-odbc-bridge.log
```
2021.05.01 17:32:14.367364 [ 102 ] {} <Information> : Starting ClickHouse 21.5.2.25 with revision 54450, , PID 102
2021.05.01 17:32:14.380166 [ 102 ] {} <Information> SentryWriter: Sending crash reports is disabled
2021.05.01 17:32:14.380324 [ 102 ] {} <Trace> Pipe: Pipe capaci... | 2021-05-01T17:45:21 | 2021-05-03T04:22:10 | {} | NONE |
ClickHouse | ClickHouse | 831,055,587 | 23,778 | filimonov | @jhettler, your help is very useful, thanks! :+1:
It seems like we have 2 issues here:
1) exception inside clickhouse-odbc-bridge during sending the data. @kssenii can you try to understand what was the reason of that exception? here is the stack trace:
```
root@a8db08d64b81:/# addr2line -afiCe /usr/bin/click... | 2021-05-03T06:40:37 | 2021-05-03T07:03:35 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 831,067,392 | 23,719 | jhettler | This will probably work. When I tried to find the cause of this trouble, I tried to load smaller datasets and it worked without any troubles. Usually the trouble starts around higher tens of milions. I have never had this issues under ten milions of rows. | 2021-05-03T07:07:13 | 2021-05-03T07:07:13 | {} | NONE |
ClickHouse | ClickHouse | 831,162,788 | 23,775 | tritas | Confirmed #23844 works, previous report was a user error. Sending you an e-mail. | 2021-05-03T10:11:46 | 2021-05-03T10:11:46 | {} | NONE |
ClickHouse | ClickHouse | 830,347,210 | 23,283 | excitoon | Problem here is that we use AWS C++ SDK and Poco together, but these products encode/decode URI's in different manner. | 2021-04-30T19:57:41 | 2021-04-30T19:57:41 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 830,424,409 | 23,785 | filimonov | > just different results.
> query: select * from AIS.dynamics where mmsi = 235089333 and lon > -45 and lat > -3
What exactly do you mean by different results? Number or returned rows differ? Their order? Something else?
Can you also check if that query
```
select * from AIS.dynamics where mmsi = 235089333 and... | 2021-04-30T22:03:19 | 2021-04-30T22:08:35 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 830,537,283 | 19,569 | alexey-milovidov | > ability to control the level of stdout logger
We also have this. | 2021-05-01T04:49:38 | 2021-05-01T04:49:38 | {} | MEMBER |
ClickHouse | ClickHouse | 830,544,496 | 23,766 | alexey-milovidov | > When a version range is specified but the package is only designed to expect a single version, the package will ignore the upper end point of the range and only take the single version at the lower end of the range into account.
It's unclear how the package is "designed". | 2021-05-01T05:25:45 | 2021-05-01T05:25:45 | {} | MEMBER |
ClickHouse | ClickHouse | 831,031,321 | 23,851 | filimonov | @excitoon please research it / try make it better if possible. | 2021-05-03T05:36:51 | 2021-05-03T05:37:02 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 831,218,397 | 23,860 | filimonov | ```bash
apt-get update
apt-get install lz4
# yes, it is appendable:
$ echo 1 | lz4c >> log.lz4
$ echo 2 | lz4c >> log.lz4
$ echo 3 | lz4c >> log.lz4
$ lz4cat log.lz4
1
2
3
# yes, there are nice command line wrappers:
$ lz4cat log.lz4 | grep 2
2
$ lz4cat log.lz4 | less
# ripgrep su... | 2021-05-03T12:09:34 | 2021-05-03T12:15:58 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 830,355,187 | 23,555 | genzgd | Yes, we did have "Memory limit exceeded" exceptions at about the same time. | 2021-04-30T20:13:33 | 2021-04-30T20:13:33 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 830,385,916 | 19,569 | filimonov | > > Also I get once the request to set up error log to collect only logs with ERROR level (no WARNINGS).
>
> You should convince the client that this feature is unneeded.
Our err logs are often very noisy, and polluted by some silly things like https://github.com/ClickHouse/ClickHouse/issues/10935
That forces ... | 2021-04-30T20:57:42 | 2021-04-30T21:01:55 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 830,430,602 | 23,790 | filimonov | ClickHouse may use memory for lot of different (valid) scenarios dictionaries, memory tables, some kind of long running queries. What make you think it's a leak?
Before continue - please update clickhouse at least to the latest 20.8 | 2021-04-30T22:19:50 | 2021-04-30T22:22:02 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 830,537,542 | 10,935 | alexey-milovidov | @filimonov Let's create another task with a feature request to decrease the logging level in this place. | 2021-05-01T04:51:14 | 2021-05-01T04:51:14 | {} | MEMBER |
ClickHouse | ClickHouse | 830,662,743 | 14,735 | tlightsky | also have similar issue | 2021-05-01T17:05:24 | 2021-05-01T17:05:24 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 830,668,952 | 23,701 | sevirov | @kochetovnicolai Hi. Please, review this PR. | 2021-05-01T17:51:13 | 2021-05-01T17:51:13 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 830,784,278 | 23,844 | excitoon | > So, now any positive status code (200) will be treated as error?
@Jokser No, thing is AWS client handles HTTP errors itself (by response code), so we shall not report `Client` type of error in these cases. I need to fix few tests though, mostly because allowed responses are either XML or JSON. | 2021-05-02T10:13:38 | 2021-05-02T10:14:19 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 830,843,190 | 23,854 | itzikiusa | I think it should be uniqExact and not uniq
This is what distinct does | 2021-05-02T17:30:52 | 2021-05-02T17:30:52 | {} | NONE |
ClickHouse | ClickHouse | 830,349,969 | 23,821 | den-crane | ```sql
SELECT sum(dictGetUInt16('curs_dict', 'rate', toUInt64(1),
toYYYYMMDDhhmmss(toDateTime('2020-01-01 00:00:00') + (rand() % 54645654)))) AS s
FROM numbers(100)
1 rows in set. Elapsed: 0.445 sec.
SELECT sum(dictGetUInt16('curs_dict', 'rate', toUInt64(1),
toYYYYMMDDhhmmss(toDateTime('2020-01-01... | 2021-04-30T20:03:22 | 2021-04-30T20:03:55 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 830,358,628 | 23,555 | genzgd | Example:
```
│ auto DB::StorageReplicatedMergeTree::processQueueEntry(ReplicatedMergeTreeQueue::SelectedEntryPtr)::(anonymous class)::operator()(DB::StorageReplicatedMergeTree::LogEntryPtr &) const: Code: 241, e.displayText() = DB::Exception: Memory limit (total) exceeded: would use 496.64 GiB (attempt to allocate c... | 2021-04-30T20:20:16 | 2021-04-30T20:20:16 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 830,643,364 | 23,775 | excitoon | I wonder does it expect gra to be written exactly in uppercase...
> The authorization header is malformed; the region 'us-east-1' is wrong; expecting 'GRA' | 2021-05-01T14:42:14 | 2021-05-01T14:42:14 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 830,675,490 | 23,775 | excitoon | You can gather a build from CI btw —
<s> https://clickhouse-builds.s3.yandex.net/23844/844b2992c5e99e94d163269d4cc7af9d912fa30c/clickhouse_build_check/report.html </s>
On Sat, May 1, 2021, 9:05 PM tritas ***@***.***> wrote:
> Amazing. Will craft a build and report back here (if I can win the fight
> against CMa... | 2021-05-01T18:33:05 | 2021-05-02T10:49:39 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 831,139,289 | 21,414 | amosbird | ya.make failure is not related to this PR | 2021-05-03T09:27:24 | 2021-05-03T09:27:24 | {} | COLLABORATOR |
ClickHouse | ClickHouse | 831,158,916 | 23,656 | kssenii | Will be finished after https://github.com/ClickHouse/ClickHouse/pull/23196. | 2021-05-03T10:04:02 | 2021-05-03T10:04:02 | {} | MEMBER |
ClickHouse | ClickHouse | 830,581,204 | 23,793 | alexey-milovidov | Also changed the category to "Improvement" as there is no bug. | 2021-05-01T08:36:12 | 2021-05-01T08:36:12 | {} | MEMBER |
ClickHouse | ClickHouse | 830,734,725 | 18,275 | amosbird | I think we can introduce a dedicated combinator for this purpose: `-ExplicitMerge` . This combinator accepts an additional aggregate function as argument and used for states finalization. For example, if we use `uniqExplicitMerge(user_id, sum)` as an aggregate, it will have state storing the finalized data of uniq, and... | 2021-05-02T03:46:52 | 2021-05-02T03:46:52 | {} | COLLABORATOR |
ClickHouse | ClickHouse | 830,814,441 | 23,851 | excitoon | @alexey-milovidov 1. The message is totally unclear. 2. With 2 writing threads 512 MB buffer each, it somehow uses more than 32GB of RAM. | 2021-05-02T14:02:22 | 2021-05-02T14:02:22 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 830,814,631 | 23,851 | excitoon | 3. It may help someone who search this text | 2021-05-02T14:03:29 | 2021-05-02T14:03:29 | {
"+1": 2
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 830,421,448 | 23,283 | excitoon | @Zhiqisc which version of ClickHouse have you used? | 2021-04-30T21:54:52 | 2021-04-30T21:54:52 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 830,651,330 | 23,775 | excitoon | This error can be easily reproduced by:
```
clickhouse local --logger.level=trace -q "select count() from s3('https://storage.gra.cloud.ovh.net/AUTH_10e1a9235c63431c95e5b84a247830db/prod/artwork/511850_1_m.jpg', '123', '234', 'TSV', 's String')"
```
```
<Error><Code>AuthorizationHeaderMalformed</Code><Message>Th... | 2021-05-01T15:38:32 | 2021-05-01T15:38:43 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 830,677,165 | 23,834 | den-crane | duplicate https://github.com/ClickHouse/ClickHouse/issues/11 https://github.com/ClickHouse/ClickHouse/issues/32 | 2021-05-01T18:46:58 | 2021-05-01T18:46:58 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 830,784,897 | 23,844 | Jokser | @excitoon I mean metrics increment `ProfileEvents::increment(select_metric(S3MetricType::Errors));` | 2021-05-02T10:17:50 | 2021-05-02T10:17:50 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 830,793,210 | 22,647 | excitoon | @azat @alexey-milovidov This PR breaks `--disable-net-host` option on any test:
```
Stderr:
Pulling proxy2 ... error
Pulling proxy1 ... error
Pulling minio1 ... error
Pulling s3_max_redirects ... error
Pulling restricted_dummy ... error
Pulling dummy ... error
Pulling... | 2021-05-02T11:19:16 | 2021-05-02T11:20:35 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 834,893,650 | 23,941 | den-crane | >Is it ok like following?
No. For a query.
```
create table x (a Int64, b Int64, s String) Engine=MergeTree partition by a order by b;
insert into x select number%5, number%999, toString(number) from numbers(100000);
show estimates for select uniq(s), a from x where a=1 and b > 42 group by a;
┌─parts─┬─mark... | 2021-05-08T00:20:49 | 2021-05-08T00:20:49 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 835,230,915 | 23,941 | fastio | > Please also take distributed queries into account.
Sure. | 2021-05-08T09:09:00 | 2021-05-08T09:09:00 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 835,240,597 | 23,912 | azat | >Need to return some headers back...
Done | 2021-05-08T09:25:43 | 2021-05-08T09:25:43 | {} | MEMBER |
ClickHouse | ClickHouse | 835,343,685 | 23,959 | alexey-milovidov | @divanorama ClickHouse will not work on 32bit machines (either x86 or ARM) for many reasons. | 2021-05-08T12:39:04 | 2021-05-08T12:39:04 | {
"+1": 1
} | MEMBER |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.