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 | 529,814,536 | 6,881 | yuzhichang | @alesapin Which Ubuntu release are you using? What arguments do you pass to release script? | 2019-09-10T07:43:08 | 2019-09-10T07:44:38 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 529,887,560 | 6,878 | alexey-milovidov | @vitlibar To check integration test. | 2019-09-10T11:12:32 | 2019-09-10T11:12:32 | {} | MEMBER |
ClickHouse | ClickHouse | 529,899,897 | 6,807 | proller | fixed here - https://github.com/yandex/ClickHouse/commit/54a5b801b708701b1ddbda95887465b9f7ae5740#diff-4f3f04aa07e7cdba3c4a995ef7e5c85dR22 | 2019-09-10T11:53:10 | 2019-09-10T11:53:10 | {
"+1": 1
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 530,174,699 | 6,894 | alexey-milovidov | The idea is very reasonable, I also had some thoughts about it.
You can also check [F14](https://github.com/facebook/folly/blob/master/folly/container/F14.md) as a memory efficient hash table. | 2019-09-11T00:54:57 | 2019-09-11T00:54:57 | {} | MEMBER |
ClickHouse | ClickHouse | 530,176,048 | 6,880 | yingfeng | @filimonov
It does not help when adding partitioning, the performance is even a little worse.
We have found the serialization/deserialization for `RoaringBitmapWithSmallSet` is not the same with the original one from `CRoaring`, we have change it into the [original implementation from `CRoaring`](https://github.co... | 2019-09-11T01:02:14 | 2019-09-11T01:04:21 | {} | NONE |
ClickHouse | ClickHouse | 530,304,451 | 6,902 | filimonov | Sad news, and thanks for the report. :\
Can you describe your setup? (We're trying to cover more cases with tests...) | 2019-09-11T09:39:27 | 2019-09-11T09:39:27 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 530,699,605 | 6,898 | filimonov | > UPD: disappeared without restarting. Any ideas?
It would be really helpful if you will describe exact scenario & used settings. In that case we can try to create test case and try to fix that.
We can also try to update libressl blindly, hoping that new version have that issue fixed.
The issue above says th... | 2019-09-12T07:20:02 | 2019-09-12T07:20:02 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 530,704,262 | 6,917 | filimonov | Related https://github.com/yandex/ClickHouse/pull/3251 | 2019-09-12T07:34:54 | 2019-09-12T07:34:54 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 530,765,501 | 6,921 | fessmage | It hangs with that situation in processes:
```
clickho+ 363548 1.0 5.5 847156 431460 ? Sl 10:15 0:08 /usr/bin/clickhouse-odbc-bridge odbc-bridge --http-port 9018 --listen-host localhost --http-timeout 1800000000
root 363913 0.0 0.0 115304 1568 ? Ss 10:28 0:00 /bin/sh /etc/rc.d/init.d/... | 2019-09-12T10:29:47 | 2019-09-12T10:29:47 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 529,779,819 | 6,879 | volfco | ```
SHOW CREATE TABLE default.metrics
┌─statement──────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────┐
│ CREATE TABLE default... | 2019-09-10T05:44:27 | 2019-09-10T05:50:50 | {} | NONE |
ClickHouse | ClickHouse | 530,111,508 | 6,879 | volfco | @den-crane That's new to me. The documentation makes it sound like LZ4 is used by default unless NONE is used | 2019-09-10T20:44:25 | 2019-09-10T20:44:25 | {} | NONE |
ClickHouse | ClickHouse | 530,615,056 | 6,905 | yingfeng | Hi, @amosbird
I update all submodules recursively. | 2019-09-12T00:23:10 | 2019-09-12T00:23:10 | {} | NONE |
ClickHouse | ClickHouse | 530,695,671 | 6,913 | filimonov | > BTW, @kshvakov used it for SummingMergeTree to do data aggregation of old data and it's considered as a nice feature.
Here are some details: https://github.com/kshvakov/ClickHouse-CPP-Meetup
> We could allow clear only last column of primary key.
The most proper way: is to allow to clear one or more column ... | 2019-09-12T07:06:38 | 2019-09-12T07:08:43 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 529,711,872 | 6,879 | volfco | ```
SELECT *
FROM system.merges
┌─database─┬─table───┬────elapsed─┬───────────progress─┬─num_parts─┬─source_part_names───────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────┬─result_part_name───────┬─partition_id─┬─is_mutation─┬─total_... | 2019-09-09T23:58:57 | 2019-09-09T23:59:14 | {} | NONE |
ClickHouse | ClickHouse | 529,777,783 | 6,879 | filimonov | Can you share the results of
```
SHOW CREATE TABLE default.metrics
```
Normally merges are IO, not CPU intensive operation, maybe there is something special in your table. | 2019-09-10T05:35:33 | 2019-09-10T05:35:33 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 530,413,084 | 6,905 | yingfeng | @alexey-milovidov
The cmake build is still broken after I update master to
```
commit 6b1352f65caf818d306518de602662678f806a80
Merge: f1f177c 62f85c0
Author: alexey-milovidov <milovidov@yandex-team.ru>
Date: Wed Sep 11 16:02:35 2019 +0300
Merge pull request #6893 from malkfilipp/client_older_than_serve... | 2019-09-11T14:42:58 | 2019-09-11T14:42:58 | {} | NONE |
ClickHouse | ClickHouse | 529,712,977 | 6,879 | volfco | ```
SELECT *
FROM system.merges
┌─database─┬─table───┬───────elapsed─┬────────────progress─┬─num_parts─┬─source_part_names─────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────... | 2019-09-10T00:04:16 | 2019-09-10T00:04:16 | {} | NONE |
ClickHouse | ClickHouse | 530,034,521 | 6,889 | alexey-milovidov | CC @nikvas0 | 2019-09-10T17:14:33 | 2019-09-10T17:14:33 | {} | MEMBER |
ClickHouse | ClickHouse | 530,114,308 | 6,879 | den-crane | Thought it's something wrong with a formatting in en doc.
It has `If a codec is specified, the default codec doesn't apply. ` | 2019-09-10T20:52:14 | 2019-09-10T20:52:14 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 530,138,796 | 6,879 | den-crane | >PARTITION BY toStartOfHour(Timestamp) | 2019-09-10T22:05:23 | 2019-09-10T22:05:23 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 530,152,955 | 6,890 | alexey-milovidov | But it's already merged in https://github.com/yandex/ClickHouse/pull/6602 | 2019-09-10T23:02:22 | 2019-09-10T23:02:22 | {} | MEMBER |
ClickHouse | ClickHouse | 530,280,906 | 6,881 | yuzhichang | I changed to `env DIST=bionic EXTRAPACKAGES="gcc-9 g++-9" DEB_CC=gcc-9 DEB_CXX=g++-9 CMAKE_FLAGS=" -DNO_WERROR=1 " ./release --no-pbuilder --rpm`. The rpm built out performs as well as the official one. | 2019-09-11T08:35:54 | 2019-09-11T08:35:54 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 530,324,637 | 3,140 | ilyas-pro | The same situation with "alias" columns:
```sql
drop table if exists test1;
drop table if exists test2;
create table test1(col UInt64, testCol UInt64 alias col)
Engine=MergeTree partition by tuple() order by tuple();
insert into test1 (col) values (1),(2),(3),(4),(5);
create table test2(testCol UInt64)
... | 2019-09-11T10:39:28 | 2019-09-11T10:39:28 | {} | NONE |
ClickHouse | ClickHouse | 530,331,877 | 6,892 | alesapin | Clone error (in CI):
```
No submodule mapping found in .gitmodules for path 'gradle'
Failed to recurse into submodule path 'contrib/poco'
``` | 2019-09-11T11:03:34 | 2019-09-11T11:03:51 | {} | MEMBER |
ClickHouse | ClickHouse | 530,360,484 | 6,906 | amosbird | well, MySQL is already disabled in nonssl build now. This pr just fixes build. I guess nonssl MySQL deserves another pr | 2019-09-11T12:36:15 | 2019-09-11T12:36:15 | {} | COLLABORATOR |
ClickHouse | ClickHouse | 530,396,027 | 6,862 | KochetovNicolai | I think I have fixed this issue in another way. @nicelulu can you check if #6911 fixes your case? | 2019-09-11T14:05:14 | 2019-09-11T14:05:14 | {} | MEMBER |
ClickHouse | ClickHouse | 530,615,585 | 6,905 | amosbird | Please post the exact cmake command you used. | 2019-09-12T00:26:29 | 2019-09-12T00:26:29 | {} | COLLABORATOR |
ClickHouse | ClickHouse | 530,038,289 | 6,565 | den-crane | @fengchangfight do you have reproducible example ? (like this https://gist.github.com/den-crane/32cac4f069f4227b35ab86553bbc649b ) ? | 2019-09-10T17:23:35 | 2019-09-10T17:23:35 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 530,107,021 | 6,879 | volfco | Using ClickHouse client version 19.14.3.3 (official build) from docker. I pulled a sample set of data from QA and duplicated it for this test.
```
SELECT count(*)
FROM metrics
┌──count()─┐
│ 80000000 │
└──────────┘
```
**Stock Table Schema**
```CREATE TABLE default.metrics (`Timestamp` DateTime CODEC(Dou... | 2019-09-10T20:31:35 | 2019-09-10T20:31:35 | {} | NONE |
ClickHouse | ClickHouse | 530,257,487 | 6,898 | ilyakarol | Full error log:
```
2019.09.11 07:22:56.075740 [ 53 ] {} <Error> ServerErrorHandler: Code: 210, e.displayText() = DB::NetException: SSL Exception: error:04FFF06A:rsa routines:CRYPTO_internal:block type is not 01, while reading from socket (yyy.yyy.yyy.yyy:47072), Stack trace:
0. /usr/bin/clickhouse-server(StackTra... | 2019-09-11T07:25:54 | 2019-09-11T07:25:54 | {} | NONE |
ClickHouse | ClickHouse | 530,298,970 | 6,898 | ilyakarol | UPD: disappeared without restarting. Any ideas? | 2019-09-11T09:25:05 | 2019-09-11T09:25:05 | {} | NONE |
ClickHouse | ClickHouse | 530,313,522 | 6,902 | millin | Looks like Kafka engine is broken at all.
1. I just cleaned all data/metadata on my test machine and started CH server. It started up.
2. Then I created a simple table with the Kafka engine
```
CREATE TABLE test
(
`data` String
)
ENGINE = Kafka
SETTINGS kafka_broker_list = 'xxx', kafka_topic_list = 'xxx',... | 2019-09-11T10:05:12 | 2019-09-11T10:05:39 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 530,617,747 | 6,905 | yingfeng | @amosbird
It's my fault.
I use `git pull --recurse-submodules` previously, but it has not cloned `contribu/orc` untill `git submodule update --init --recursive` was used once more. | 2019-09-12T00:38:52 | 2019-09-12T00:38:52 | {} | NONE |
ClickHouse | ClickHouse | 530,726,611 | 6,871 | VaBezruchko | We cannot upgrade version to latest stable soon. Anyway we suppose that this issue appears only once in six months, so we need a long time to reproduce issue on new version (For now we have restarted all nodes). | 2019-09-12T08:39:44 | 2019-09-12T08:39:44 | {
"laugh": 1
} | NONE |
ClickHouse | ClickHouse | 530,729,820 | 6,862 | KochetovNicolai | @nicelulu thank you for the test case. | 2019-09-12T08:48:37 | 2019-09-12T08:48:37 | {} | MEMBER |
ClickHouse | ClickHouse | 530,743,492 | 6,862 | nicelulu | > @nicelulu thank you for the test case.
@alexey-milovidov @KochetovNicolai thank you. | 2019-09-12T09:25:33 | 2019-09-12T09:25:33 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 530,745,858 | 6,880 | yingfeng | @andyyzh
Could you please review PR [6908](https://github.com/yandex/ClickHouse/pull/6908) ? Why did you choose the original serialization/deserialization approaches different from the official corresponding parts of `croaring`?
We have met some strange phenomenon that some kinds of bitmap data could not be persi... | 2019-09-12T09:32:05 | 2019-09-12T09:32:05 | {} | NONE |
ClickHouse | ClickHouse | 530,761,095 | 6,902 | millin | Solved in [v19.15.1.1310-testing](https://github.com/yandex/ClickHouse/releases/tag/v19.15.1.1310-testing) | 2019-09-12T10:16:02 | 2019-09-12T10:16:02 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 529,854,297 | 6,881 | alesapin | No additional options passed to release script https://github.com/yandex/ClickHouse/blob/master/docker/packager/deb/build.sh (we call it from docker image). Ubuntu version doesn't matter, because almost all clickhouse libraries linked statically from contrib. | 2019-09-10T09:31:00 | 2019-09-10T09:31:00 | {
"+1": 1
} | MEMBER |
ClickHouse | ClickHouse | 530,060,381 | 6,602 | amosbird | now cmake 3.12 requirement is dropped | 2019-09-10T18:21:43 | 2019-09-10T18:21:43 | {} | COLLABORATOR |
ClickHouse | ClickHouse | 530,175,730 | 6,894 | alexey-milovidov | It's in question - should we provide the choice of the data structure via attribute or via different `layout`. I think, different layout is better.
Another point: current data structure is larger than needed, because it uses separate hash tables for each dictionary attribute. It's possible to use single data structu... | 2019-09-11T01:00:49 | 2019-09-11T01:00:49 | {} | MEMBER |
ClickHouse | ClickHouse | 529,824,441 | 6,880 | yingfeng | The output of `callgrind` is attached.
[callgrind.out.2454.zip](https://github.com/yandex/ClickHouse/files/3594444/callgrind.out.2454.zip)
It has contained the overhead for bootstrapping, together with several queries.
We could see that the serialization/deserialization have occupied dominant percentage:
<img w... | 2019-09-10T08:11:36 | 2019-09-10T08:11:36 | {} | NONE |
ClickHouse | ClickHouse | 530,176,634 | 6,895 | alexey-milovidov | > since in some cases 30 seconds can be too large
But after you removed the limit, it will go arbitrary large: if the server was down, when it become alive, the waiting time will be proportional to elapsed downtime (for example - if the server was down for one day, recovery time will also be about one day - that's t... | 2019-09-11T01:05:01 | 2019-09-11T01:05:55 | {} | MEMBER |
ClickHouse | ClickHouse | 530,354,738 | 6,905 | yuzhichang | I tried build each commit of the master branch, and found that the bad commit is `254102c5bd42189d9ced3547eb928db563f69c33`(introduced by #6602) which cause cmake error. | 2019-09-11T12:19:23 | 2019-09-11T12:29:31 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 530,449,617 | 6,900 | kreuzerkrieg | I totally forgot to mention when the previous solution wouldnt work - when switching from bundled cmake to any other cmake the previous string matching (which looks for "clion" token) will stop working thats why I resorted to more robust way to check if it is a CLion | 2019-09-11T16:05:10 | 2019-09-11T16:05:10 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 530,472,942 | 5,315 | MakarovYaroslav | +1
I have the same problem.
Query with LIMIT 1 quickly (<1s) returns first string and then it takes a very long time to complete query (about 20s) | 2019-09-11T17:04:52 | 2019-09-11T17:04:52 | {} | NONE |
ClickHouse | ClickHouse | 529,783,167 | 6,879 | filimonov | What if you will try without DoubleDelta? | 2019-09-10T05:59:16 | 2019-09-10T05:59:16 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 530,110,856 | 6,879 | den-crane | @volfco Are you aware CODEC(DoubleDelta) <-- It means NO compression?
It's supposed to be CODEC(DoubleDelta, LZ4) or CODEC(DoubleDelta, ZSTD)
| 2019-09-10T20:42:25 | 2019-09-10T20:42:25 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 530,142,123 | 6,879 | den-crane | ```
CREATE TABLE test_dd (A Int64, T DateTime CODEC(DoubleDelta, ZSTD))
Engine=MergeTree order by A;
insert into test_dd select number, number%2 from numbers(10000000);
OPTIMIZE TABLE test_dd FINAL
0 rows in set. Elapsed: 0.491 sec.
0 rows in set. Elapsed: 0.493 sec.
0 rows in set. Elapsed: 0.535 sec.
0 rows... | 2019-09-10T22:17:39 | 2019-09-10T22:19:31 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 530,229,682 | 6,894 | azat | > You can also check F14 as a memory efficient hash table.
I will take a look, but seems that sparsehash is more memory efficient anyway.
>maybe it's better to unify it and use single library (remove old library)?
Sure, will take a look
>It's in question - should we provide the choice of the data structure ... | 2019-09-11T05:47:14 | 2019-09-11T05:47:14 | {} | MEMBER |
ClickHouse | ClickHouse | 530,358,410 | 6,906 | proller | Is possible to keep mysql support without ssl ? | 2019-09-11T12:30:13 | 2019-09-11T12:30:13 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 530,364,242 | 6,877 | KochetovNicolai | Hi.
The bug is weird indeed. I haven't reproduced it yet (tried different alters with corner cases).
Will try to crash server after alter, maybe it will help.
It would be very helpful if you make some reproducible example. | 2019-09-11T12:47:05 | 2019-09-11T12:47:05 | {} | MEMBER |
ClickHouse | ClickHouse | 530,614,851 | 6,862 | alexey-milovidov | @KochetovNicolai Could you please also import the test from this PR? | 2019-09-12T00:22:09 | 2019-09-12T00:22:09 | {} | MEMBER |
ClickHouse | ClickHouse | 530,641,312 | 6,862 | nicelulu | > I think I have fixed this issue in another way. @nicelulu can you check if #6911 fixes your case?
ok,it can. | 2019-09-12T02:44:02 | 2019-09-12T02:44:49 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 530,701,837 | 6,898 | ilyakarol | Nothing special, 4 Clickhouse instances as cluster (2 shards, 2 replicas).
No updates/inserts since very first data load, we just test queries and performance.
Problems only with queries in Distributed tables.
Which settings could be useful? | 2019-09-12T07:27:00 | 2019-09-12T07:27:11 | {} | NONE |
ClickHouse | ClickHouse | 529,914,436 | 6,063 | zhang2014 | This is a feature that hasn't been implemented for some reason, and I will implement them as soon as possible : ) | 2019-09-10T12:36:41 | 2019-09-10T12:36:51 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 530,137,250 | 6,879 | den-crane | >Single merge should be execute in one thread.
Yes and we observe 12 merges for 12 different partitions. | 2019-09-10T22:00:26 | 2019-09-10T22:00:26 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 530,185,818 | 6,894 | data-man | Compare with [Tessil](https://github.com/Tessil)'s collections. ([results](https://tessil.github.io/2016/08/29/benchmark-hopscotch-map.html))
Especially with [hat-trie](https://github.com/Tessil/hat-trie) for strings. ([results](https://github.com/Tessil/hat-trie#benchmark)) | 2019-09-11T01:53:01 | 2019-09-11T01:53:01 | {} | NONE |
ClickHouse | ClickHouse | 530,447,421 | 6,905 | amosbird | Looks like your submodules aren't up to date? Recently orc build is fixed by @proller | 2019-09-11T15:59:53 | 2019-09-11T15:59:53 | {} | COLLABORATOR |
ClickHouse | ClickHouse | 530,571,194 | 6,903 | filimonov | Please consider that behaviour non determenistic, and don't rely on that. You've requested SOME 10 row and you get some.
If you need 10 rows by some order, use ORDER BY. | 2019-09-11T21:21:47 | 2019-09-11T21:21:47 | {
"+1": 1
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 530,152,772 | 6,890 | alexey-milovidov | > ../contrib/libcxx/include/list:984: _LIBCPP_ASSERT '!empty()' failed. list::back called on empty list
Probably it need to be merged with master. | 2019-09-10T23:01:40 | 2019-09-10T23:01:40 | {} | MEMBER |
ClickHouse | ClickHouse | 530,358,009 | 6,878 | vitlibar | > @vitlibar To check integration test.
I'll fix the integration test `test_dictionaries_all_layouts_and_sources` soon. | 2019-09-11T12:28:58 | 2019-09-11T12:28:58 | {} | MEMBER |
ClickHouse | ClickHouse | 530,366,586 | 6,871 | alexey-milovidov | Does the issue reproduce on the latest stable release? | 2019-09-11T12:53:48 | 2019-09-11T12:53:48 | {} | MEMBER |
ClickHouse | ClickHouse | 530,473,722 | 6,914 | alexey-milovidov | Security considerations should be noted in setting description. | 2019-09-11T17:06:59 | 2019-09-11T17:06:59 | {} | MEMBER |
ClickHouse | ClickHouse | 530,549,302 | 6,913 | CurtizJ | > Clearing a column from primary key does not break ordering of data. I think, it should be possible.
We could allow clear only last column of primary key. Otherwise, ordering of data may become broken in case of complex key.
E.g: with key of two columns:
```
┌─a─┬─b─┐
│ 1 │ 1 │
│ 1 │ 2 │
│ 2 │ 1 │
│ 2 │ 2 │... | 2019-09-11T20:21:06 | 2019-09-11T20:21:06 | {
"+1": 2
} | MEMBER |
ClickHouse | ClickHouse | 530,768,209 | 6,912 | akuzm | The test you added fails under AddressSanitizer. Here's how to enable it: https://github.com/yandex/ClickHouse/blob/master/dbms/tests/instructions/sanitizers.md#how-to-use-address-sanitizer
Other test failures are probably unrelated. To be sure, you can rebase to a latest `testing` tag, e.g. v19.15.1.1310-testing --... | 2019-09-12T10:38:31 | 2019-09-12T10:38:31 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 529,812,529 | 6,881 | alesapin | We use gcc-9 for our releases. | 2019-09-10T07:37:19 | 2019-09-10T07:37:19 | {} | MEMBER |
ClickHouse | ClickHouse | 530,226,814 | 6,895 | azat | >But after you removed the limit, it will go arbitrary large: if the server was down
Didn't thought about that, [previous version](https://github.com/yandex/ClickHouse/compare/master...azat-archive:distributed_directory_monitor_max_sleep_time) of the patch as adds `distributed_directory_monitor_max_sleep_time` confi... | 2019-09-11T05:34:15 | 2019-09-11T05:34:15 | {} | MEMBER |
ClickHouse | ClickHouse | 530,371,052 | 6,905 | alexey-milovidov | @amosbird has fixed this issue in master, can you please check it? | 2019-09-11T13:05:50 | 2019-09-11T13:05:50 | {} | MEMBER |
ClickHouse | ClickHouse | 529,747,231 | 6,795 | slimtom95 | > Hi @xuxudede would you like to share more how you guys resolve this issue? I got this same exception, but the data caused it is not available
Edit:
This error occurs when the precision of data fraction part exceed `S` in field definition.
| 2019-09-10T02:56:26 | 2019-09-11T13:47:47 | {} | NONE |
ClickHouse | ClickHouse | 530,568,519 | 6,489 | alesapin | performance flap | 2019-09-11T21:14:04 | 2019-09-11T21:14:04 | {} | MEMBER |
ClickHouse | ClickHouse | 529,737,777 | 3,981 | xiashuijun | > That is caused by client side disconnect on http protocol, for example due to read timeout on client side. So server can't write anymore to the socket after client close that.
>
> Test case:
>
> ```shell
> wget 'http://localhost:8123/?query=SELECT%201,sleepEachRow(0.01)%20FROM%20numbers(200)' -qO- --read-timeo... | 2019-09-10T02:09:53 | 2019-09-10T02:09:53 | {} | NONE |
ClickHouse | ClickHouse | 530,315,702 | 6,902 | abyss7 | I see where the problem is. Can you provide the full log from start til crash, so I can dig a little bit more details about actual scenario? | 2019-09-11T10:11:38 | 2019-09-11T10:11:38 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 530,368,233 | 6,895 | alexey-milovidov | > Didn't thought about that, previous version of the patch as adds distributed_directory_monitor_max_sleep_time config directive, does this sounds reasonable?
> And thinking about this more, I assume that better way will be to replace distributed_directory_monitor_max_sleep_time with a setting, just like distributed... | 2019-09-11T12:58:17 | 2019-09-11T12:58:17 | {} | MEMBER |
ClickHouse | ClickHouse | 530,533,609 | 6,913 | alexey-milovidov | Clearing a column from primary key does not break ordering of data. I think, it should be possible.
> get wrong result in some queries, because of broken index
Need to check what are these queries and why they work wrong.
> Could not find a column of minimum size in MergeTree
It's about clearing all the col... | 2019-09-11T19:38:07 | 2019-09-11T19:38:07 | {} | MEMBER |
ClickHouse | ClickHouse | 529,902,941 | 6,565 | fengchangfight | > > I believe a cause is CREATE MATERIALIZED VIEW.... **SETTINGS index_granularity=8192**
> > Try index_granularity=64 ... 16 ... 8
> > https://gist.github.com/den-crane/32cac4f069f4227b35ab86553bbc649b
> > Or check adaptive index_granularity in the recent releases.
>
> Firstly, thanks for your reply.I have chang... | 2019-09-10T12:02:38 | 2019-09-10T12:02:38 | {} | NONE |
ClickHouse | ClickHouse | 530,113,275 | 6,879 | den-crane | @volfco Hmm, it could be lost in translation.
Russian documentation has this sentence: `If a column has its own sequence of codecs, then the general table codec is not applied (must be specified in the sequence forcibly, if necessary).` (in Russian). | 2019-09-10T20:49:28 | 2019-09-10T20:49:28 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 530,113,691 | 6,879 | volfco | @den-crane I'm doing an A/B right now to test. So I'll have answer in a minute | 2019-09-10T20:50:37 | 2019-09-10T20:50:37 | {} | NONE |
ClickHouse | ClickHouse | 530,153,961 | 6,878 | azat | (Rebased & push forced after #6602) | 2019-09-10T23:06:48 | 2019-09-10T23:06:48 | {} | MEMBER |
ClickHouse | ClickHouse | 530,466,424 | 6,574 | maqroll | #6914 | 2019-09-11T16:48:12 | 2019-09-11T16:48:12 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 530,778,472 | 6,867 | akuzm | > but the new chunk will also be unable to allocate 4 bytes
At Arena.h, 104:
```
if (size_after_grow < min_next_size)
size_after_grow = min_next_size;
```
nextSize() has a protection from this -- it always allocates at least as much as we asked. In your example, it will allocate 103 bytes, s... | 2019-09-12T11:12:51 | 2019-09-12T11:12:51 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 529,929,710 | 6,880 | filimonov | BTW: does it help if you will add partitioning, for example by xxHash32(tag_id) % 512 or similar? | 2019-09-10T13:17:03 | 2019-09-10T13:17:03 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 530,115,036 | 6,879 | volfco | ```f2bd24850ace :) select * from system.columns where name = 'Timestamp';
SELECT *
FROM system.columns
WHERE name = 'Timestamp'
┌─database─┬─table────┬─name──────┬─type─────┬─default_kind─┬─default_expression─┬─data_compressed_bytes─┬─data_uncompressed_bytes─┬─marks_bytes─┬─comment─┬─is_in_partition_key─┬─is_in... | 2019-09-10T20:54:04 | 2019-09-10T20:58:20 | {} | NONE |
ClickHouse | ClickHouse | 530,122,008 | 6,879 | filimonov | > @volfco Are you aware CODEC(DoubleDelta) <-- It means NO compression?
> It's supposed to be CODEC(DoubleDelta, LZ4) or CODEC(DoubleDelta, ZSTD)
That is true, but DoubleDelta itself do a simple bit compression. | 2019-09-10T21:12:07 | 2019-09-10T21:12:07 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 530,320,353 | 6,902 | millin | ```
2019.09.11 13:19:15.014126 [ 1 ] {} <Information> : Starting ClickHouse 19.15.1.1299 with revision 54426
2019.09.11 13:19:15.015923 [ 1 ] {} <Information> Application: starting up
2019.09.11 13:19:15.017866 [ 1 ] {} <Debug> Application: rlimit on number of file descriptors is 500000
2019.09.11 13:19:15.017899 [... | 2019-09-11T10:25:23 | 2019-09-11T10:25:23 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 529,926,878 | 6,880 | filimonov | Scalar subqueries are quite specific as executed on query parsing stage.
But in should still give linear increase:
```
SELECT concat(
(
SELECT toString(min(rand()))
FROM numbers(50000000)
),
(
SELECT toString(min(rand(1)))
FROM numbers(50000000)
),
(
... | 2019-09-10T13:10:15 | 2019-09-10T13:10:15 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 529,928,307 | 6,862 | abyss7 | Can't figure out what caused the problem in the first place. Can you explain it a little bit more? | 2019-09-10T13:13:51 | 2019-09-10T13:13:51 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 530,034,333 | 5,387 | akuzm | https://github.com/yandex/ClickHouse/pull/6892 | 2019-09-10T17:14:03 | 2019-09-10T17:14:03 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 530,206,385 | 6,862 | nicelulu | > Can't figure out what caused the problem in the first place. Can you explain it a little bit more?
Perform an INSERT SELECT. When there is an asynchronous read data like UnionBlockInputStream in the execution plan, the data production is slow in some cases, and nothing can be read in the copyData. This situation w... | 2019-09-11T03:41:07 | 2019-09-11T03:41:07 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 530,254,413 | 6,368 | StephenSorriaux | Hi,
If somebody wants to do the same thing, suggestion from @millin works pretty well, just need to also add the `kafka_row_delimiter='\n'` parameter:
```
create table kafka_ingest (
jsonString String,
jsonField1 String DEFAULT JSONExtractString(jsonString, 'field1'),
jsonField2 UInt64 DEFAULT JSONExtractUInt(j... | 2019-09-11T07:16:09 | 2019-09-11T07:16:09 | {} | NONE |
ClickHouse | ClickHouse | 530,367,064 | 6,894 | alexey-milovidov | > That my second concern, but this can/should be done separately anyway (objections?)
Ok. | 2019-09-11T12:55:00 | 2019-09-11T12:55:00 | {} | MEMBER |
ClickHouse | ClickHouse | 530,385,557 | 6,880 | yuzhichang | PR is #6908. | 2019-09-11T13:41:19 | 2019-09-11T13:41:19 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 530,469,140 | 6,913 | den-crane | related https://github.com/yandex/ClickHouse/issues/3207 | 2019-09-11T16:55:04 | 2019-09-11T16:55:04 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 530,493,584 | 6,913 | alexey-milovidov | BTW, @kshvakov used it for SummingMergeTree to do data aggregation of old data and it's considered as a nice feature. | 2019-09-11T17:56:23 | 2019-09-11T17:56:23 | {} | MEMBER |
ClickHouse | ClickHouse | 530,626,584 | 6,903 | peaksnail | thanks, I just want to get the data by sorted key, and the create_date is the sorted key.
If using as a time series db,all data is inserted ORDER BY create_date, when I want to get the first 10 row, so should I always use ORDER BY create_date limit 10?
And I find that the performance is bad to ORDER BY with larg... | 2019-09-12T01:26:58 | 2019-09-12T01:27:29 | {} | NONE |
ClickHouse | ClickHouse | 529,938,133 | 5,682 | abyss7 | > @abyss7
>
> Have you failed this task?
This task is just in the line with Kafka and all other assigned issues. | 2019-09-10T13:35:59 | 2019-09-10T13:35:59 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 530,126,313 | 6,879 | filimonov | ```
FROM system.merges
12 rows in set. Elapsed: 0.041 sec.
```
Single merge should be execute in one thread. But I see here 12 concurrent merges. | 2019-09-10T21:23:30 | 2019-09-10T21:23:30 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 530,175,150 | 6,894 | alexey-milovidov | As we already use sparsehash/densehash
https://github.com/yandex/ClickHouse/tree/master/contrib/libsparsehash
maybe it's better to unify it and use single library (remove old library)? | 2019-09-11T00:57:32 | 2019-09-11T00:57:32 | {} | MEMBER |
ClickHouse | ClickHouse | 530,346,425 | 6,898 | filimonov | Maybe related: https://github.com/libressl-portable/portable/issues/445 | 2019-09-11T11:53:21 | 2019-09-11T11:53:21 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 506,341,264 | 5,771 | mfridental | The try to build on a local machine:
cd ClickHouse
mkdir build
cd build
cmake ..
ninja
produces this:
[248/1326] Building CXX object dbms/programs/clang/Compiler-6.0.0/CMakeFiles/clickhouse-compiler-lib.dir/driver.cpp.o
FAILED: /usr/bin/g++-7 -DUNALIGNED_OK -DWITH_GZFILEOP -DX86_64 -DZLIB_COMPAT -isystem ... | 2019-06-27T13:16:15 | 2019-06-27T13:16:15 | {} | CONTRIBUTOR |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.