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 | 639,380,325 | 11,415 | KochetovNicolai | Failed tests are not connected with this pr. | 2020-06-05T09:58:11 | 2020-06-05T09:58:11 | {} | MEMBER |
ClickHouse | ClickHouse | 639,649,358 | 11,391 | abyss7 | It doesn't look good in some cases.
- What to do if the propagated user is not defined on the whole cluster?
- How to handle situations when incoming connection to the initiator is secured, but outcoming is not?
- How to provide backward compatibility for current production clusters where all the settings and contra... | 2020-06-05T17:25:22 | 2020-06-05T17:25:22 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 639,728,290 | 11,401 | alexey-milovidov | So, what behaviour to expect in production on servers with old Linux kernel?
Maybe it's ready for merge? | 2020-06-05T19:13:06 | 2020-06-05T19:13:06 | {} | MEMBER |
ClickHouse | ClickHouse | 639,858,636 | 11,458 | alexey-milovidov | PS. I have tested performance of big paste manually. It's Ok. | 2020-06-05T22:11:21 | 2020-06-05T22:11:21 | {} | MEMBER |
ClickHouse | ClickHouse | 640,021,791 | 9,113 | alexey-milovidov | Performance tests Ok. | 2020-06-06T09:38:39 | 2020-06-06T09:38:39 | {} | MEMBER |
ClickHouse | ClickHouse | 640,086,336 | 11,470 | den-crane | STR:
```
create table foo(A Int64, V Int64, S String) Engine= Null;
create materialized view fooM Engine = MergeTree order by A as select A, V, S from foo;
insert into foo select number, sipHash64(number), toString(arrayMap(x->sipHash64(number*x), range(10))) from numbers(10000000);
set optimize_move_to_p... | 2020-06-06T16:34:09 | 2020-06-06T16:34:09 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 639,032,365 | 9,477 | alexey-milovidov | When you write a query in the form of two joined subqueries:
```
SELECT * FROM (SELECT ... FROM distributed) JOIN (SELECT ... FROM distributed)
```
data is transferred to initiator node and joined there. So, the JOIN is processed not in distributed fashion and the setting `distributed_product_mode` does not hav... | 2020-06-04T18:29:54 | 2020-06-04T18:29:54 | {
"+1": 1
} | MEMBER |
ClickHouse | ClickHouse | 639,082,692 | 11,451 | jakesylvestre | Figured this out as I was writing the question.
`*ifState` is a function (that's the meaning of any in the documentation) so `minIfState` and `maxIfState` will work ehre. | 2020-06-04T19:56:53 | 2020-06-04T19:56:53 | {} | NONE |
ClickHouse | ClickHouse | 639,464,183 | 11,459 | vzakaznikov | I will look into adding proper support for `PREWHERE` in LiveView tables and should have a PR ready by the end of next week if there are no major road blocks. | 2020-06-05T12:58:01 | 2020-06-05T12:58:01 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 640,079,628 | 11,334 | nikitamikhaylov | I think the issue was to import that library to our `/contrib`, but I don't see it in this PR. In addition, I don't see any test.
According to [this](https://github.com/frankmcsherry/blog/blob/master/posts/2016-02-06.md) and [this](https://arxiv.org/pdf/1909.01917.pdf) papers, we can implement generating of Laplace di... | 2020-06-06T15:40:53 | 2020-06-06T15:40:53 | {} | MEMBER |
ClickHouse | ClickHouse | 639,335,490 | 11,391 | filimonov | Related:
https://github.com/ClickHouse/ClickHouse/issues/9751
https://github.com/ClickHouse/ClickHouse/pull/8926 | 2020-06-05T08:25:01 | 2020-06-05T08:25:01 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 639,406,532 | 11,228 | filimonov | > ```
> SELECT
> dictGet('countryId', 'country', toUInt64(number)) AS country
> FROM numbers(2)
> GROUP BY country
> ```
>
> ` DB::Exception: Function dictGet... takes 3 or 4 arguments.`
This is different: https://github.com/ClickHouse/ClickHouse/issues/11469 | 2020-06-05T10:49:49 | 2020-06-05T10:49:49 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 639,208,172 | 11,395 | donghy-coredumped | Thanks, you are right, my own reason
| 2020-06-05T01:42:59 | 2020-06-05T01:42:59 | {} | NONE |
ClickHouse | ClickHouse | 639,722,804 | 4,796 | jakesylvestre | I assume for this same reason `FixedString` doesn't improve `LowCardinality` performance? | 2020-06-05T19:03:50 | 2020-06-05T19:03:50 | {} | NONE |
ClickHouse | ClickHouse | 639,749,937 | 11,401 | azat | >So, what behaviour to expect in production on servers with old Linux kernel?
AFAIR (when I looked into it) this PR does not changes behavior for older linux, since:
- dirty->muzzy **always** uses `MADV_FREE` if defined
- muzzy->clean uses mmap instead of `MADV_DONTNEED` **after this PR**
And actually even upgr... | 2020-06-05T19:38:37 | 2020-06-05T19:38:37 | {} | MEMBER |
ClickHouse | ClickHouse | 639,762,575 | 11,391 | azat | > How to handle situations when incoming connection to the initiator is secured, but outcoming is not?
This PR does not changes this, am I missing something?
> How to provide backward compatibility for current production clusters where all the settings and contraints are properly defined for 'default' user only? ... | 2020-06-05T19:58:23 | 2020-06-05T19:58:23 | {} | MEMBER |
ClickHouse | ClickHouse | 639,831,524 | 11,401 | azat | >Sorry, I'm out of context... do you mean that it's using munmap to return memory?
No, muzzy->clean makes pages "clean", i.e. they are zero-filled and can be used again after this operation (if I understand correctly), hence it uses mmap with MAP_FIXED to clean the arena.
From http://jemalloc.net/jemalloc.3.html
... | 2020-06-05T21:30:53 | 2020-06-05T21:30:53 | {} | MEMBER |
ClickHouse | ClickHouse | 640,081,923 | 11,485 | akuzm | Builds failed because of CI problems. | 2020-06-06T15:59:20 | 2020-06-06T15:59:20 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 639,260,265 | 7,513 | SergeyLossev | btw, here is right link
[https://github.com/ClickHouse/ClickHouse/blob/master/tests/queries/0_stateless/00427_alter_primary_key.sh](https://github.com/ClickHouse/ClickHouse/blob/master/tests/queries/0_stateless/00427_alter_primary_key.sh) | 2020-06-05T05:03:19 | 2020-06-05T05:03:47 | {} | NONE |
ClickHouse | ClickHouse | 639,492,808 | 9,253 | den-crane | When argument is `Null`
```
select if(b_num = 0, -1, b_num),
if(empty(b_str), 'Others', b_str)
from (select Null b_num, Null b_str)
┌─if(equals(b_num, 0), -1, b_num)─┬─if(empty(b_str), 'Others', b_str)─┐
│ ᴺᵁᴸᴸ │ ᴺᵁᴸᴸ │
└─────────────────... | 2020-06-05T13:52:01 | 2020-06-05T14:07:09 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 639,527,249 | 9,253 | 4ertus2 | @den-crane What's the difference with my sample above that does not contain join and returns `other` instead of `Null`?
Your sample with subselect is not relevant cause `b_num` is `Nullable(Nothing)`
```
SELECT
NULL AS b_num,
toTypeName(b_num)
┌─b_num─┬─toTypeName(NULL)──┐
│ ᴺᵁᴸᴸ │ Nullable(Nothin... | 2020-06-05T14:21:47 | 2020-06-05T14:43:42 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 639,560,882 | 11,472 | den-crane | https://github.com/ClickHouse/ClickHouse/issues/10212#issuecomment-612676750 | 2020-06-05T15:20:37 | 2020-06-05T15:20:37 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 640,025,483 | 11,453 | alexey-milovidov | TODO: Replace `strerror` to `strerror_r`. | 2020-06-06T09:54:38 | 2020-06-06T09:54:38 | {} | MEMBER |
ClickHouse | ClickHouse | 640,025,689 | 11,453 | alexey-milovidov | TODO: Add a test with `expect` and two concurrent clients. | 2020-06-06T09:55:38 | 2020-06-06T09:55:38 | {} | MEMBER |
ClickHouse | ClickHouse | 639,225,955 | 11,425 | songenjie | hello? | 2020-06-05T02:48:13 | 2020-06-05T02:48:13 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 639,483,469 | 11,291 | KochetovNicolai | Other tests are mainly ok: https://clickhouse-test-reports.s3.yandex.net/11452/5b08d06846b888c20409df24e24a083a104e66af/performance_comparison/report.html#fail1
Some of queries with sorting and small limit just a little but slower. | 2020-06-05T13:34:31 | 2020-06-05T13:34:31 | {} | MEMBER |
ClickHouse | ClickHouse | 639,743,791 | 11,478 | Turum | Many thanks, it's better, but another error.
```
1. DB::Exception::Exception(std::__1::basic_string<char, std::__1::char_traits<char>, std::__1::allocator<char> > const&, int) @ 0x4ff4bb9 in /usr/bin/clickhouse
2. ? @ 0x49d29c5 in /usr/bin/clickhouse
3. ? @ 0x9157cfd in /usr/bin/clickhouse
4. DB::CSVRowInputFormat... | 2020-06-05T19:28:52 | 2020-06-05T19:29:55 | {} | NONE |
ClickHouse | ClickHouse | 639,852,732 | 11,471 | alexey-milovidov | UndefinedBehaviorSanitizer: undefined-behavior ../src/Compression/tests/gtest_compressionCodec.cpp:750:37 | 2020-06-05T22:05:06 | 2020-06-05T22:05:06 | {} | MEMBER |
ClickHouse | ClickHouse | 640,017,841 | 11,190 | alexey-milovidov | Continued in #11484
@azat All added to the list. | 2020-06-06T09:15:42 | 2020-06-06T09:15:42 | {} | MEMBER |
ClickHouse | ClickHouse | 638,999,149 | 11,443 | alexey-milovidov | The new name means that this setting is intended for all arithmetic-related optimizations?
But as of now it has very specific use - move arithmetic operations out of aggregate functions if they are commutative with aggregation. | 2020-06-04T17:33:18 | 2020-06-04T17:58:13 | {} | MEMBER |
ClickHouse | ClickHouse | 639,084,829 | 11,441 | den-crane | Work in progress https://github.com/ClickHouse/ClickHouse/pull/9404
Duplicate https://github.com/ClickHouse/ClickHouse/issues/7316 | 2020-06-04T20:00:44 | 2020-06-04T20:00:44 | {
"+1": 1
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 639,099,255 | 11,443 | alexey-milovidov | The name "optimize arithmetic"... suggests something more simple.
For example, the user may think that it will perform basic peephole optimizations on associativity, for example, transform:
`x + 1 + 2`
to
`x + (1 + 2)`
But it does something more sophisticated.
That's why I think that this setting should h... | 2020-06-04T20:29:15 | 2020-06-04T20:29:15 | {} | MEMBER |
ClickHouse | ClickHouse | 639,214,647 | 11,459 | den-crane | If you need something like 'select ... order by balance desc' with millisecond realtime response in streaming manner the only way is to recalculate it regularly (e.g. using cron) and store a ready result in another table. There is no Engine which can solve it and I don't see how it can be implemented with MergeTree arc... | 2020-06-05T02:06:27 | 2020-06-05T02:06:27 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 639,306,580 | 9,117 | jakesylvestre | Use case is using `memory` ENGINE in the way described in #11459 (constant recomputes on insert) | 2020-06-05T07:23:24 | 2020-06-05T07:23:24 | {} | NONE |
ClickHouse | ClickHouse | 639,534,216 | 11,458 | alexey-milovidov | Before merge:
- check on white background;
- provide command line argument to disable highlighting. | 2020-06-05T14:34:04 | 2020-06-05T14:34:04 | {} | MEMBER |
ClickHouse | ClickHouse | 639,897,119 | 5,541 | jakesylvestre | By the way, I don't think I have `optimize_move_to_prewhere` on, will check this out now
Yeah - doesn't look like it | 2020-06-05T23:37:54 | 2020-06-05T23:55:34 | {} | NONE |
ClickHouse | ClickHouse | 640,020,659 | 9,751 | azat | Refs:
- https://dev.mysql.com/doc/refman/8.0/en/proxy-users.html
- https://www.postgresql.org/docs/10/sql-set-session-authorization.html | 2020-06-06T09:32:26 | 2020-06-06T09:32:26 | {
"+1": 1
} | MEMBER |
ClickHouse | ClickHouse | 639,151,840 | 11,456 | alexey-milovidov | СС @blinkov | 2020-06-04T22:30:52 | 2020-06-04T22:30:52 | {} | MEMBER |
ClickHouse | ClickHouse | 639,356,017 | 11,291 | KochetovNicolai | BTW, it's pretty cool that all tests have passed and `15 faster, 2 slower` in Performance | 2020-06-05T09:06:07 | 2020-06-05T09:06:07 | {} | MEMBER |
ClickHouse | ClickHouse | 639,486,671 | 11,444 | blinkov | > @blinkov could you look this PR again? I add all missed SYSTEM statements descriptions
It looks like there's an issue with an incorrect link on single-page version:
```
INFO:root:Tag <a href="#query_language-system-restart-replicas">RESTART REPLICAS</a>
INFO:root:Link to nowhere: #query_language-system-restart-re... | 2020-06-05T13:40:33 | 2020-06-05T13:40:33 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 639,744,563 | 11,401 | akuzm | > So, what behaviour to expect in production on servers with old Linux kernel?
> Maybe it's ready for merge?
I want to try another approach -- disable DONTNEED_ZEROS so that MADV_FREE is used + increase muzzy decay time.
By the way, this PR didn't significantly affect the variability of perf test results -- I'm se... | 2020-06-05T19:30:08 | 2020-06-05T19:30:08 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 639,860,554 | 11,458 | alexey-milovidov | Also it's tempting to add more pleasant, naturally distributed 24bit colors as we have in logs.
But only if we can detect whether it's supported by terminal (for example, it does not work in `screen` by default). | 2020-06-05T22:14:56 | 2020-06-05T22:14:56 | {} | MEMBER |
ClickHouse | ClickHouse | 640,084,866 | 11,483 | alexey-milovidov | Unit tests and builds fixed in a73b360
Then CI was temporarily broken. | 2020-06-06T16:22:52 | 2020-06-06T16:22:52 | {} | MEMBER |
ClickHouse | ClickHouse | 640,092,138 | 11,482 | den-crane | STR
```
drop table if exists jaeger_index_v2;
CREATE TABLE jaeger_index_v2 (
timestamp DateTime CODEC(Delta, ZSTD(1)),
traceID String CODEC(ZSTD(1)),
service LowCardinality(String) CODEC(ZSTD(1))
) ENGINE MergeTree()
PARTITION BY toDate(timestamp)
ORDER BY (service, timestamp)
SETTINGS index_granul... | 2020-06-06T17:18:42 | 2020-06-06T17:35:09 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 631,068,181 | 10,790 | akuzm | A couple of issues introduced in https://github.com/ClickHouse/ClickHouse/pull/10773: -- all fixed now
* [x] red-green color contrast is not accessible for people with protanopia
* [x] measurement units are not abbreviations, periods are not used after them | 2020-05-19T20:36:22 | 2020-06-04T17:41:48 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 639,802,679 | 11,401 | akuzm | > > disable DONTNEED_ZEROS so that MADV_FREE is used
>
> Interesting, can you point at the code? (I can't find such bits there)
This was my misinterpretation of your above comment. Looks like DONTNEED without DONTNEED_ZEROS is a no-op, and decommit is used instead if one of them is not defined.
https://github.... | 2020-06-05T20:38:53 | 2020-06-05T20:38:53 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 640,017,047 | 11,326 | weiyongyuan | Today,i found that in the /proc/ckpid/fd,there ara so many socket created but not use.
lrwx------ 1 root root 64 Jun 6 10:01 994 -> socket:[80756151]
lrwx------ 1 root root 64 Jun 6 10:01 995 -> socket:[80756154]
lrwx------ 1 root root 64 Jun 6 10:01 996 -> socket:[80747433]
lrwx------ 1 root root 64 Jun 6 ... | 2020-06-06T09:11:11 | 2020-06-06T09:11:11 | {} | NONE |
ClickHouse | ClickHouse | 639,162,932 | 11,190 | azat | > test_insert_distributed_load_balancing https://clickhouse-test-reports.s3.yandex.net/11114/aa656f95dd331d00d152e76ff8bca70d1bbd3441/integration_tests_(release).html#fail1
Fixed in #11355 | 2020-06-04T23:07:24 | 2020-06-04T23:07:24 | {
"+1": 1
} | MEMBER |
ClickHouse | ClickHouse | 639,202,594 | 2,765 | jakesylvestre | I believe this is intended behavior | 2020-06-05T01:20:56 | 2020-06-05T01:20:56 | {} | NONE |
ClickHouse | ClickHouse | 639,554,712 | 9,253 | 4ertus2 | OK. But problem is not in JOIN
```
SELECT
toTypeName(b_num),
b_num,
if(b_num = 0, 'true', 'false')
FROM
(
SELECT if(1 = 0, toNullable(toInt8(0)), NULL) AS b_num
)
┌─toTypeName(b_num)─┬─b_num─┬─if(equals(b_num, 0), 'true', 'false')─┐
│ Nullable(Int8) │ ᴺᵁᴸᴸ │ true ... | 2020-06-05T15:09:18 | 2020-06-05T15:09:18 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 639,859,706 | 11,458 | alexey-milovidov | It does not look alright on white background. So, we cannot merge it.
Need to replace support for colors in replxx, so WHITE will actually mean "intense". | 2020-06-05T22:13:22 | 2020-06-05T22:13:22 | {} | MEMBER |
ClickHouse | ClickHouse | 640,080,629 | 11,334 | alexey-milovidov | If I remember correctly, Artemy has found out that the library is not directly applicable. | 2020-06-06T15:48:55 | 2020-06-06T15:48:55 | {} | MEMBER |
ClickHouse | ClickHouse | 638,964,997 | 11,401 | akuzm | Some comments from Telegram (in Russian):
Alexander Kuzmenkov, [03.06.20 22:37]
О, нашёл ишшую где описано, зачем им двухстадийное освобождение страниц. Нам надо вторую стадию как-то отключить, что ли...
https://github.com/jemalloc/jemalloc/issues/521
Alexey Milovidov, [03.06.20 22:56]
[In reply to Alexande... | 2020-06-04T16:27:22 | 2020-06-04T16:27:22 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 639,104,576 | 11,447 | alexey-milovidov | Update: the codec actually has an error in implementation that can be reproduced on very small block sizes. | 2020-06-04T20:40:23 | 2020-06-04T20:40:29 | {} | MEMBER |
ClickHouse | ClickHouse | 639,425,544 | 11,071 | r-zenine | hi @Akazz ,
Thanks for the review, I will take time to fix everything in the up comming days. | 2020-06-05T11:31:02 | 2020-06-05T11:31:02 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 639,549,553 | 11,226 | akuzm | Fuzz test failed for `SELECT __getScalar('\0')` -- reported.
Perf test -- won't run because of the new settings. Will check in master.
Other tests have passed before, merging now. | 2020-06-05T15:00:17 | 2020-06-05T15:00:17 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 639,616,233 | 5,747 | afedotov89 | @akonyaev90 Hello! when can this feature be released? Thanks | 2020-06-05T16:35:24 | 2020-06-05T16:35:24 | {} | NONE |
ClickHouse | ClickHouse | 639,819,379 | 11,478 | Turum | I found the issue, thank you for the clue. | 2020-06-05T21:08:24 | 2020-06-05T21:08:24 | {} | NONE |
ClickHouse | ClickHouse | 640,094,899 | 11,293 | azat | >I think that there is no difference, it can lead to the same deadlock (that will be timed out after 120 seconds by default).
Indeed, so how about restoring this config directive? | 2020-06-06T17:42:58 | 2020-06-06T17:42:58 | {} | MEMBER |
ClickHouse | ClickHouse | 640,096,982 | 11,388 | abyss7 | Yandex check failure is unrelated | 2020-06-06T18:01:53 | 2020-06-06T18:01:53 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 638,985,194 | 11,410 | akuzm | Another fuzz test failure: https://clickhouse-test-reports.s3.yandex.net/11410/029005be7c13e00d2ad41ecd16869e094810de84/functional_stateless_tests_(debug).html#fail1
```
SELECT * FROM jdbc((SELECT [], '\0'), ['\0'])
Logical error: 'Scalar `3696643124553813592_9682321359984966147` doesn't exist (internal bug)'.
... | 2020-06-04T17:05:57 | 2020-06-04T17:05:57 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 639,039,436 | 11,434 | den-crane | duplicate https://github.com/ClickHouse/ClickHouse/issues/10212 | 2020-06-04T18:38:15 | 2020-06-04T18:38:15 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 639,065,625 | 11,442 | alesapin | Integration tests OK. | 2020-06-04T19:21:07 | 2020-06-04T19:21:07 | {} | MEMBER |
ClickHouse | ClickHouse | 639,279,610 | 11,444 | Slach | @blinkov could you look this PR again? I add all missed SYSTEM statements descriptions | 2020-06-05T06:10:00 | 2020-06-05T06:10:00 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 639,713,201 | 11,459 | den-crane | @jakesyl you can insert into an intermediate table, drop a partition in the main table, and attach / move from the intermediate. Or you can rename tables (do swap in a single rename). | 2020-06-05T18:53:49 | 2020-06-05T18:53:49 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 639,729,956 | 11,478 | den-crane | ```
file ttrades.csv
ttrades.csv: Little-endian UTF-16 Unicode text, with CRLF line terminators
```
I think CH expects UTF-8 with CR line terminators
try
`iconv -f UTF-16LE -t UTF-8 ttrades.csv > ttrades.csv.uft8`
| 2020-06-05T19:14:29 | 2020-06-05T19:14:44 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 639,853,839 | 11,471 | alexey-milovidov | I have already fixed that: #11345 | 2020-06-05T22:06:01 | 2020-06-05T22:06:01 | {} | MEMBER |
ClickHouse | ClickHouse | 639,899,787 | 11,479 | 4ertus2 | It looks like we have `mysqlxx::checkError(conn->getDriver())` when connection is destroyed earlier then `CreatingSetsBlockInputStream` finished.
@KochetovNicolai, do you remember the case with totals + early query exit. Could it be the same issue here? | 2020-06-05T23:44:38 | 2020-06-05T23:44:56 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 640,013,288 | 11,471 | alexey-milovidov | Still don't work:
```
SUMMARY: UndefinedBehaviorSanitizer: undefined-behavior ../src/Compression/CompressionCodecDoubleDelta.cpp:293:27
``` | 2020-06-06T08:37:22 | 2020-06-06T08:37:22 | {} | MEMBER |
ClickHouse | ClickHouse | 640,091,595 | 11,293 | azat | Got it, but what prevents from deadlock when flusher thread creates table? | 2020-06-06T17:14:33 | 2020-06-06T17:14:33 | {} | MEMBER |
ClickHouse | ClickHouse | 639,049,139 | 11,443 | 4ertus2 | > The new name means that this setting is intended for all arithmetic-related optimizations?
Yes. The idea is we need several cummulative settings for AST optimizer. Not a setting per optimization but a setting per group of optimizations. Generally we have to support optimizers settings forever cause they could make... | 2020-06-04T18:48:27 | 2020-06-04T18:48:27 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 639,241,526 | 11,461 | LiemLin | @proller Please take a look at it，thanks | 2020-06-05T03:46:22 | 2020-06-05T03:46:22 | {} | NONE |
ClickHouse | ClickHouse | 639,215,573 | 11,459 | jakesylvestre | Got it @den-crane is there some sort of way to store the ready result so it's queryable or should I just be dropping/recreating tables? (Asking as I know the fundamental problem here is deletes are expensive in LSM)
By the way - it might be possible to acheive this at the engine level by moving records between parti... | 2020-06-05T02:10:03 | 2020-06-05T04:49:37 | {} | NONE |
ClickHouse | ClickHouse | 639,383,322 | 5,541 | KochetovNicolai | @vzakaznikov I think we just need to disable `prewhere` for LV. Coult you please take a look? | 2020-06-05T10:05:05 | 2020-06-05T10:05:05 | {} | MEMBER |
ClickHouse | ClickHouse | 639,437,955 | 11,464 | blinkov | @zzsmdfj the test error is `WARNING:mkdocs.structure.pages:Documentation file 'sql-reference/functions/type-conversion-functions.md' contains a link to 'sql-reference/data-types/lowcardinality.md' which is not found in the documentation files.`. Could you fix it by adding the lowcardinality.md which is missing in zh ve... | 2020-06-05T12:00:43 | 2020-06-05T12:00:43 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 639,607,025 | 11,450 | alesapin | At least it doesn't become slower. | 2020-06-05T16:20:18 | 2020-06-05T16:20:18 | {} | MEMBER |
ClickHouse | ClickHouse | 639,679,400 | 11,459 | jakesylvestre | @vzakaznikov does the above seem like it would work with a `LiveView` (considering number of rows?) | 2020-06-05T18:14:27 | 2020-06-05T18:14:27 | {} | NONE |
ClickHouse | ClickHouse | 639,747,411 | 11,459 | jakesylvestre | @den-crane appreciate the reccomendation. Trying this right now and having some issues around write performance but will let you know how this goes | 2020-06-05T19:34:32 | 2020-06-05T19:34:32 | {} | NONE |
ClickHouse | ClickHouse | 639,758,178 | 11,401 | alexey-milovidov | > muzzy->clean uses mmap instead of MADV_DONTNEED after this PR
Sorry, I'm out of context... do you mean that it's using `munmap` to return memory?
(it should be significantly slower than MADV_DONTNEED and MADV_FREE) | 2020-06-05T19:51:40 | 2020-06-05T19:51:40 | {} | MEMBER |
ClickHouse | ClickHouse | 640,021,841 | 9,113 | alexey-milovidov | 00746_sql_fuzzy - fixed in master. | 2020-06-06T09:38:53 | 2020-06-06T09:38:53 | {} | MEMBER |
ClickHouse | ClickHouse | 640,093,750 | 11,293 | alexey-milovidov | I think that there is no difference, it can lead to the same deadlock (that will be timed out after 120 seconds by default). | 2020-06-06T17:32:50 | 2020-06-06T17:32:50 | {} | MEMBER |
ClickHouse | ClickHouse | 639,110,679 | 9,051 | alexey-milovidov | We have received second complaint about this usability issue. | 2020-06-04T20:53:48 | 2020-06-04T20:53:54 | {} | MEMBER |
ClickHouse | ClickHouse | 639,336,734 | 6,843 | filimonov | Alternative option #9751 | 2020-06-05T08:27:45 | 2020-06-05T08:27:45 | {
"+1": 1
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 640,083,479 | 11,293 | azat | @alexey-milovidov do you have details on why the deadlock occurred? | 2020-06-06T16:12:15 | 2020-06-06T16:12:15 | {} | MEMBER |
ClickHouse | ClickHouse | 639,194,808 | 5,541 | jakesylvestre | I'm seeing `Code: 182. DB::Exception: Received from localhost:9000. DB::Exception: Storage Blocks doesn't support PREWHERE.` even with SELECT *. Schema is laid out in detail on #11459 | 2020-06-05T00:51:42 | 2020-06-05T00:51:42 | {} | NONE |
ClickHouse | ClickHouse | 639,324,777 | 11,462 | filimonov | Please check your certificates (follow the tutorial: https://www.altinity.com/blog/2019/3/5/clickhouse-networking-part-2 )
Try `server.crt`, `server.key`, `dhparam.pem` from clickhouse tests: see https://github.com/ClickHouse/ClickHouse/tree/master/tests/config | 2020-06-05T08:03:43 | 2020-06-05T08:03:58 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 639,429,420 | 9,253 | 4ertus2 | I've looked at the sample and do not understand what do you want from JOIN here. The same result for `if(empty(), ...)` you could get without join:
```
select if(1 = 0, toNullable('x'), NULL) as b_str, empty(b_str), if(empty(b_str), 'other', b_str);
```
```
┌─b_str─┬─empty(if(equals(1, 0), toNullable('x'), NUL... | 2020-06-05T11:40:26 | 2020-06-05T11:40:26 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 639,547,166 | 9,253 | den-crane | argument (b_num) Null then expected result false
`if null then false`
```
SELECT
CAST(NULL, 'Nullable(UInt8)') AS b_num,
if(b_num, 'true', 'false')
┌─b_num─┬─if(CAST(NULL, 'Nullable(UInt8)'), 'true', 'false')─┐
│ ᴺᵁᴸᴸ │ false │
└───────┴───────────────... | 2020-06-05T14:56:25 | 2020-06-05T15:01:08 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 640,014,063 | 11,471 | alexey-milovidov | cbf71d6 | 2020-06-06T08:44:38 | 2020-06-06T08:44:38 | {} | MEMBER |
ClickHouse | ClickHouse | 640,021,523 | 11,421 | alexey-milovidov | Yandex check - broken by Yandex DevTools.
Unit tests - fixed in master. | 2020-06-06T09:37:15 | 2020-06-06T09:37:15 | {} | MEMBER |
ClickHouse | ClickHouse | 640,062,664 | 11,479 | KochetovNicolai | Well, it's probably because connection pool is owned by `StorageMySQL`:
https://github.com/ClickHouse/ClickHouse/blob/e230632645a29d917a934a1b46d2b8a81a5bb9d1/src/TableFunctions/TableFunctionMySQL.cpp#L120-L122
This storage is temporary and could possible be destroyed too early. | 2020-06-06T13:38:40 | 2020-06-06T13:38:40 | {} | MEMBER |
ClickHouse | ClickHouse | 640,093,887 | 11,293 | alexey-milovidov | It will be fixed automatically when we will change the engine of system database to Atomic #7512
but only for new servers. | 2020-06-06T17:33:54 | 2020-06-06T17:34:12 | {} | MEMBER |
ClickHouse | ClickHouse | 640,094,094 | 11,482 | den-crane | @bobrik check your query with `set optimize_read_in_order=0` | 2020-06-06T17:35:41 | 2020-06-06T17:35:41 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 639,599,016 | 6,843 | 243f6a8885a308d313198a2e037 | #9751 sounds enough to me! | 2020-06-05T16:06:49 | 2020-06-05T16:06:49 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 639,840,222 | 11,190 | azat | One more flacky test:
- 01079_parallel_alter_detach_table_zookeeper - https://clickhouse-test-reports.s3.yandex.net/11455/5fea34d0e1e72b6d5aa233e332127336a3dd1237/functional_stateless_tests_(debug).html#fail1 | 2020-06-05T21:47:39 | 2020-06-05T21:47:39 | {} | MEMBER |
ClickHouse | ClickHouse | 639,959,296 | 11,459 | jakesylvestre | So I applied some of the optimizations above (namely [`LowCardinality`](https://github.com/ClickHouse/ClickHouse/issues/4796#issuecomment-477610482), [`CollapsingMergeTree`](https://clickhouse.tech/docs/en/engines/table-engines/mergetree-family/versionedcollapsingmergetree/). (with `tx_count` as version), [skipping ind... | 2020-06-06T02:22:52 | 2020-06-06T02:23:09 | {} | NONE |
ClickHouse | ClickHouse | 640,090,357 | 11,293 | alexey-milovidov | We have loaded system database and started up tables. Some tables started to perform background merge. Then we initialize SystemLogs and it tries to do RENAME TABLE.
It is introducing a sequence of R W R locks:
```
thread 1: R R
thread 2: W
```
that caused deadlock because our RWLocks are fair.
| 2020-06-06T17:04:44 | 2020-06-06T17:05:12 | {} | MEMBER |
ClickHouse | ClickHouse | 640,102,430 | 11,391 | abyss7 | > > How to handle situations when incoming connection to the initiator is secured, but outcoming is not?
>
> This PR does not changes this, am I missing something?
>
Yes, but AFAIK current implementation never re-sends passwords.
> > Also you should make sure, that a raw password isn't stored in RAM on the lon... | 2020-06-06T18:46:58 | 2020-06-06T18:46:58 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 640,280,083 | 11,494 | den-crane | Can you copy/paste an exact error message. It's kinda unclear which DB has threw it. | 2020-06-07T21:07:41 | 2020-06-07T21:08:44 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 640,314,250 | 11,496 | alexey-milovidov | > ClickHouse build check — 20/21 builds are OK
Unbundled build is accidentially broken by @abyss7. Fixed in master. | 2020-06-08T01:25:53 | 2020-06-08T01:25:53 | {} | MEMBER |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.