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 | 671,505,818 | 13,491 | akuzm | > It shows 10..17% improvement of memory usage, is it real? @akuzm
It's a lie, unfortunately. CDF for memory usage looks like this, meaning no real changes (be careful, the graph is sideways -- no. of queries vs relative change):
![cumulative-memory_usage](https://user-images.githubusercontent.com/36882414/89815376... | 2020-08-10T18:08:32 | 2020-08-10T18:09:02 | {
"+1": 1
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 671,648,461 | 13,360 | mjodeh | > yes. Probably it's for different partitions. Check partition_id column in system.merges.
I checked it and it's for the same partition id! | 2020-08-11T00:01:29 | 2020-08-11T00:01:29 | {} | NONE |
ClickHouse | ClickHouse | 671,879,486 | 13,589 | akuzm | Looks like there are
no significant performance changes https://clickhouse-test-reports.s3.yandex.net/13589/d42dbfce987a7510bb69139eb6e8e099a3f3c0df/performance_comparison/report.html
The memory usage change CDF has some interesting negative tail now.The median is still almost 0, probably because we don't have man... | 2020-08-11T11:02:26 | 2020-08-11T11:02:26 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 671,880,051 | 13,589 | akuzm | For the future reference, another CDF I've built yesterday for a case that doesn't have any memory usage changes: https://github.com/ClickHouse/ClickHouse/pull/13491#issuecomment-671505818 | 2020-08-11T11:03:44 | 2020-08-11T11:03:44 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 672,004,539 | 13,619 | den-crane | >I have two replicas and we upgraded replica A while B stayed on lower version for a few minutes. Meanwhile re
> plication_queue filled up with invalid partitions.
This is an incorrect behavior. Which versions did you use?
> Is there easier way to do this?
https://github.com/ClickHouse/ClickHouse/issues/1036... | 2020-08-11T15:09:52 | 2020-08-11T15:10:07 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 671,190,255 | 13,545 | filimonov | Duplicate of #9339 | 2020-08-10T06:44:25 | 2020-08-10T06:44:25 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 671,351,281 | 11,364 | akuzm | ```
SELECT
arrayJoin(arrayMap(x -> reinterpretAsUInt8(substring(randomString(range(randomString(1048577), NULL), arrayJoin(arrayMap(x -> reinterpretAsUInt8(substring(randomString(range(NULL), 65537), 255)), range(1))), substring(randomString(NULL), x + 7), '257'), 1025)), range(7))) AS byte,
count()... | 2020-08-10T13:23:03 | 2020-08-10T13:23:03 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 671,359,998 | 13,570 | den-crane | You've missed to define a `version` column
https://clickhouse.tech/docs/en/engines/table-engines/mergetree-family/replacingmergetree/
ver — column with version. Type UInt*, Date or DateTime. Optional parameter.
When merging, ReplacingMergeTree from all the rows with the same sorting key leaves only o... | 2020-08-10T13:39:21 | 2020-08-10T13:39:39 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 671,365,698 | 13,563 | den-crane | The first should work. The second is against SQL (in general ANSI/SQL) . | 2020-08-10T13:49:47 | 2020-08-10T13:49:47 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 671,573,546 | 13,459 | akuzm | Can't reproduce the AST fuzzer error Logical error: 'Extremes transform was already added to pipeline.'.
| 2020-08-10T20:28:51 | 2020-08-10T20:28:51 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 671,603,148 | 13,572 | alexey-milovidov | It will not work properly in Arcadia because it has unpatched version of Poco.
Wrong timeouts or deadlocks are expected. | 2020-08-10T21:38:28 | 2020-08-10T21:38:28 | {} | MEMBER |
ClickHouse | ClickHouse | 671,879,637 | 12,771 | acmnu | @traceon If we have a number of OS to support, do we have a build and post build test on CI for such OS? I am wondering why we need to check it by hands? | 2020-08-11T11:02:48 | 2020-08-11T11:02:48 | {} | NONE |
ClickHouse | ClickHouse | 671,915,099 | 7,782 | amosbird | It's based on the foreach iterator of hash tables. It might be implemented differently though. I'll take another look some time later... | 2020-08-11T12:25:51 | 2020-08-11T12:25:51 | {} | COLLABORATOR |
ClickHouse | ClickHouse | 671,176,550 | 13,542 | chu1070y | thanks. :) | 2020-08-10T06:02:15 | 2020-08-10T06:02:15 | {} | NONE |
ClickHouse | ClickHouse | 671,236,009 | 13,566 | amosbird | It's already covered in https://github.com/ClickHouse/ClickHouse/pull/12455 | 2020-08-10T08:49:58 | 2020-08-10T08:49:58 | {
"+1": 1
} | COLLABORATOR |
ClickHouse | ClickHouse | 671,389,926 | 13,577 | den-crane | > Thanks a lot. In your opinion what version should I upgrade to?
Any stable version. Try the latest -- 20.6.3.28. **BUT you should test your application for compatibility.**
>And for safety should I upgrade from 18 to 19 and then to 20 step by step?
No need for 99.99% cases. It could be needed for complicated... | 2020-08-10T14:29:21 | 2020-08-10T14:29:21 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 671,432,308 | 13,570 | peaksnail | thanks
```
ver — column with version. Type UInt*, Date or DateTime. Optional parameter.
When merging, ReplacingMergeTree from all the rows with the same sorting key leaves only one:
Last in the selection, if ver not set.
With the maximum version, if ver specified.
```
It has said that if I don't define a ... | 2020-08-10T15:42:30 | 2020-08-10T15:42:30 | {} | NONE |
ClickHouse | ClickHouse | 671,517,390 | 13,579 | vitlibar | No need to backport. | 2020-08-10T18:31:34 | 2020-08-10T18:31:34 | {} | MEMBER |
ClickHouse | ClickHouse | 671,519,094 | 2,914 | alexey-milovidov | @CurtizJ is on vacation, he will clarify after 20th August. | 2020-08-10T18:35:23 | 2020-08-10T18:35:23 | {} | MEMBER |
ClickHouse | ClickHouse | 671,581,010 | 13,402 | Akazz | @excitoon This PR lacks tests to say the least. You should definitely add some stateless tests | 2020-08-10T20:45:41 | 2020-08-10T20:45:41 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 671,678,535 | 13,528 | alexey-milovidov | 00933_test_fix_extra_seek_on_compressed_cache - known bad test. | 2020-08-11T01:55:53 | 2020-08-11T01:55:53 | {} | MEMBER |
ClickHouse | ClickHouse | 671,951,800 | 10,389 | adegtyarev | Using server version 20.6.3 revision 54436 the crash no longer occurs.
Thank you for your work on this issue. | 2020-08-11T13:37:27 | 2020-08-11T13:37:27 | {} | NONE |
ClickHouse | ClickHouse | 671,209,364 | 12,508 | qoega | test_mysql_protocol/test.py::test_mysql_federated
https://clickhouse-test-reports.s3.yandex.net/13530/b633b49e072bdca7419f04fb93469ddfaacc5b39/integration_tests_(release).html
| 2020-08-10T07:43:07 | 2020-08-10T07:43:07 | {} | MEMBER |
ClickHouse | ClickHouse | 671,476,038 | 13,360 | den-crane | >is it ok?
yes. Probably it's for different partitions. Check `partition_id` column in system.merges.
>Is there any formula or estimation to be used?
No. It's a very system specific. NVME for example could perform 10 times better than HDD raid.
Try to reduce insert rate somehow: engine=Buffer / clickhouse-bul... | 2020-08-10T17:06:41 | 2020-08-10T17:07:14 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 671,731,955 | 13,577 | sliontc | ClickHouse Update
If ClickHouse was installed from deb packages, execute the following commands on the server:
sudo apt-get update
sudo apt-get install clickhouse-client clickhouse-server
sudo service clickhouse-server restart
So I just go with this guide to upgrade? | 2020-08-11T05:21:34 | 2020-08-11T05:21:34 | {} | NONE |
ClickHouse | ClickHouse | 671,841,978 | 13,597 | zhang2014 | Maybe help you:
``` brew install binutils``` | 2020-08-11T09:37:33 | 2020-08-11T09:37:33 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 671,877,932 | 13,544 | abyss7 | Stress-test has found a heap-after-free and stacktrace is correlated with your changes. Looks like there is an issue somewhere in your code. Will take a closer look.
https://gist.github.com/abyss7/e8dcc7f3ce7e1cc335f5f04507b9661a | 2020-08-11T10:58:27 | 2020-08-11T10:58:27 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 672,016,957 | 13,612 | vzakaznikov | @alexey-milovidov, I think this PR is ready for merge. | 2020-08-11T15:28:26 | 2020-08-11T15:28:37 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 672,059,872 | 13,584 | alexey-milovidov | Duplicate #12562 | 2020-08-11T16:13:30 | 2020-08-11T16:13:30 | {} | MEMBER |
ClickHouse | ClickHouse | 671,175,243 | 13,543 | chu1070y | Great. It worked :) | 2020-08-10T05:57:26 | 2020-08-10T05:57:26 | {} | NONE |
ClickHouse | ClickHouse | 671,320,934 | 13,357 | akuzm | 01428_hash_set_nan_key broken in 20.6 | 2020-08-10T12:18:21 | 2020-08-10T12:18:21 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 671,826,542 | 12,661 | jmalicevic | Just to follow up, we have temporarly by-passed the issue by using a proxy as suggested by @oandrew and replacing the '-' with '_' .
| 2020-08-11T09:05:42 | 2020-08-11T09:05:42 | {} | NONE |
ClickHouse | ClickHouse | 671,925,442 | 13,598 | filimonov | Hm, it looks like the behavior has changed just in the last releases (most probably https://github.com/ClickHouse/ClickHouse/pull/12376 ) Check 20.5.4 | 2020-08-11T12:47:01 | 2020-08-11T12:49:28 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 671,955,097 | 13,563 | filimonov | Sometimes i need something like Engine=Merge but picking the first matching table only (and empty row if no matching tables found), i think that issue is about smth similar.
@mbtolou am I right, can you describe the use case a bit? | 2020-08-11T13:43:23 | 2020-08-11T13:43:23 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 672,120,515 | 13,613 | Slach | @alesapin
also related to https://github.com/ClickHouse/ClickHouse/issues/10006
look like if() function try to load dictionary even when I use dictGetUInt32 and dictGetUInt8 and I don't understand why | 2020-08-11T17:31:50 | 2020-08-11T17:31:50 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 671,394,682 | 10,851 | zhang2014 | Update:
Change _sign and _version from virtual column to materialize column.
This will help implement the following case:
```sql
CREATE DATABASE test_database ENGINE = MaterializeMySQL('localhost:3306', test, 'test', 'test')
CREATE MATERIALIZED VIEW test_materialized_view ENGINE = MergeTree() PARTITION BY tu... | 2020-08-10T14:37:09 | 2020-08-10T14:40:41 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 671,479,904 | 13,099 | akuzm | > @akuzm It shows 22..24% performance improvement, but for unknown reason, performance test does not show it as confident result.
Let's remove all unknown reasons. Which test do you mean, `synthetic_hardware_benchmark`? We have set `max_ignored_relative_change = 0.2` for it, because it had some unstable queries. Thi... | 2020-08-10T17:14:31 | 2020-08-10T17:14:31 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 671,494,500 | 13,099 | alexey-milovidov | Yes, I meant this test. Ok. | 2020-08-10T17:44:46 | 2020-08-10T17:44:46 | {} | MEMBER |
ClickHouse | ClickHouse | 671,913,719 | 13,544 | amosbird | Ah, I forgot to shutdown `tmp_volume`. Now it should be fine. | 2020-08-11T12:23:02 | 2020-08-11T12:23:02 | {} | COLLABORATOR |
ClickHouse | ClickHouse | 671,951,636 | 13,613 | filimonov | But in dict definition you have `value UInt32` can you double check everything?
Loosely related to
https://github.com/ClickHouse/ClickHouse/issues/8004
https://github.com/ClickHouse/ClickHouse/issues/10397
https://github.com/ClickHouse/ClickHouse/issues/10994
and
https://github.com/ClickHouse/ClickHouse/... | 2020-08-11T13:37:11 | 2020-08-11T13:37:43 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 671,351,152 | 13,546 | akuzm | https://github.com/ClickHouse/ClickHouse/issues/11364 | 2020-08-10T13:22:47 | 2020-08-10T13:22:47 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 671,425,317 | 13,584 | den-crane | 20.3.12.112
```
SELECT 100 data FORMAT JSON;
...
{
"data": 100
}
```
20.7.1
```
SELECT 100 data FORMAT JSON;
...
{
"data": 100
``` | 2020-08-10T15:30:07 | 2020-08-10T15:30:07 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 671,905,048 | 13,595 | filimonov | The exception is correct. ClickHouse has written the data to a target table already, but can't commit an offset to Kafka broker. (so you may get duplicate at that point).
"Broker: Unknown member"
Most probably Kafka broker kicks you off from the consumer group due to some timeouts. | 2020-08-11T12:05:21 | 2020-08-11T12:06:37 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 671,930,287 | 4,194 | filimonov | @pikhovkin я закрываю тикет (потому что очень старый). Если вы нашли причину или смогли устранить проблему - пожалуйста напишите пару слов о том как вы это сделали, если проблема всё ещё актуальна и проявляется на новых версиях кликхауса - создайте новый тикет с описанием шагов для воспроизведения проблемы (пример схем... | 2020-08-11T12:56:20 | 2020-08-11T12:56:20 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 671,969,883 | 13,613 | Slach | I checked twice
and change the dictionary to `value String`
run
```
docker-compose down
docker-compose run clickhouse
```
and got the same behavior with same error
```sql
CREATE DICTIONARY IF NOT EXISTS default.dict_prod_mb2_params (
partner_id UInt32,
`order` UInt8,
`name` String,
display_name String,
is_forwarded UInt8
... | 2020-08-11T16:59:53 | 2020-08-11T17:33:56 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 671,207,861 | 13,513 | qoega | >It requires pip3 install codespell (or maybe apt-get install is also ok?).
There in no container for style check https://nda.ya.ru/t/uQY4n48E3WUUDb
Probably we can create one. | 2020-08-10T07:38:52 | 2020-08-10T07:39:05 | {} | MEMBER |
ClickHouse | ClickHouse | 671,354,926 | 13,571 | alexey-milovidov | Changelog is published here: https://github.com/ClickHouse/ClickHouse/blob/master/CHANGELOG.md
Website https://clickhouse.tech/docs release is lagged behind in a few hours. | 2020-08-10T13:29:59 | 2020-08-10T13:29:59 | {} | MEMBER |
ClickHouse | ClickHouse | 671,478,157 | 13,570 | den-crane | @peaksnail it will It will choose the row from the last **insert**.
1592768014809 1st insert
1593961745804 2nd insert <-- **the last insert**
MERGE process uses parts' numbers to determine which insert was the last.
EXAMPLE:
```
create table XX (A String) Engine=ReplacingMergeTree order by tuple();
i... | 2020-08-10T17:11:06 | 2020-08-10T17:22:54 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 671,892,509 | 7,782 | abyss7 | @amosbird what's with this PR? Description says to merge after #7662 which is closed. | 2020-08-11T11:34:03 | 2020-08-11T11:34:03 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 671,274,040 | 1,463 | vmihailenco | `quantilesTDigestMerge` is probably fast, but `quantilesTDigest` is not - https://github.com/ClickHouse/ClickHouse/issues/2668
`quantiles` is fast, but `quantilesMerge` is slow - 2 seconds to merge 30k rows.
`quantilesTiming` is the only the working solution, but only for values in range [0, 30000].
:( | 2020-08-10T10:17:39 | 2020-08-10T10:17:39 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 671,354,822 | 13,574 | filimonov | We discussed that idea some time ago, and hopefully, it will be taken / delivered. But there is a very problematic moment: commit into 3 places (filesystem on source shard, the filesystem on target shard, and zookeeper) which can't be executed safe enough in the distributed system.
So it seems that we need to allow ... | 2020-08-10T13:29:48 | 2020-08-10T13:29:48 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 671,372,832 | 12,550 | myrrc | Merging this PR would also resolve #13392 | 2020-08-10T14:01:41 | 2020-08-10T14:01:41 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 671,738,860 | 13,360 | mjodeh | I stopped the insertion and `tail -f /var/log/clickhouse-server/clickhouse-server.log` I see the following:
`2020.08.11 10:08:56.670500 [ 2603 ] {} <Error> void DB::SystemLog<LogElement>::flushImpl(const std::__1::vector<_Tp>&, uint64_t) [with LogElement = DB::TraceLogElement; uint64_t = long unsigned int]: Code: 25... | 2020-08-11T05:43:37 | 2020-08-11T05:43:37 | {} | NONE |
ClickHouse | ClickHouse | 672,081,342 | 9,855 | 4ertus2 | Works in master regardless of `multiple_joins_rewriter_version`. It's hard to find now where it has been fixed.
Issue needs a test before close. | 2020-08-11T16:43:26 | 2020-08-11T16:43:26 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 672,024,352 | 13,613 | Slach | look like same with https://github.com/ClickHouse/ClickHouse/issues/13059
`dictGetString` instead of `dictGet` works
| 2020-08-11T15:40:45 | 2020-08-11T17:13:45 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 671,351,731 | 13,516 | alexey-milovidov | > Some downside - bash scripts which rely on $(pidof -s clickhouse-server) may misbehave. Not sure if possible to avoid (use another processname for watchdog?)
Yes, and this is already done in this PR. | 2020-08-10T13:23:57 | 2020-08-10T13:23:57 | {} | MEMBER |
ClickHouse | ClickHouse | 671,444,473 | 13,585 | alexey-milovidov | We don't support this grammar: https://github.com/ClickHouse/ClickHouse/blob/master/utils/grammar/README.md
The new experimental ANTLR grammar is being prepared in #11298 | 2020-08-10T16:04:45 | 2020-08-10T16:04:45 | {} | MEMBER |
ClickHouse | ClickHouse | 671,518,673 | 2,914 | alexey-milovidov | This optimization is implemented by @dimarub2000 and prepared for production quality by @CurtizJ.
There are a few more things to consider:
- it is disabled by default, because some queries may slow down
(and @CurtizJ is working on further optimization opportunities);
- AFAIK it does not work when we wrap keys into ... | 2020-08-10T18:34:28 | 2020-08-10T18:34:38 | {} | MEMBER |
ClickHouse | ClickHouse | 671,629,859 | 12,741 | azat | - the row-level security filter is in WHERE also because of https://github.com/ClickHouse/ClickHouse/blob/18ef0c0c964530866f85fec3adb8ee8ac23b0b10/src/Interpreters/InterpreterSelectQuery.cpp#L426
- expression from `WHERE` hasn't been pushed to `PREWHERE` because of `optimize_move_to_prewhere` does not works if `PREWHE... | 2020-08-10T22:56:24 | 2020-08-10T22:56:24 | {} | MEMBER |
ClickHouse | ClickHouse | 671,933,909 | 13,360 | den-crane | >Keeps complaining about Merges are processing significantly
>slower than inserts even though the insertion is stopped 15 min before!
Because you inserted data into Distributed table with async mode. Distributed does real insertions later until when it stores data in .bin files in own folder. | 2020-08-11T13:03:39 | 2020-08-11T13:03:39 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 671,971,472 | 13,577 | sliontc | Thank you so much guys. | 2020-08-11T14:12:24 | 2020-08-11T14:12:24 | {} | NONE |
ClickHouse | ClickHouse | 671,988,062 | 13,573 | amosbird | Tests are added but not verified locally (I don't have docker right now). Let's see if I can get it right ... | 2020-08-11T14:41:38 | 2020-08-11T14:41:38 | {} | COLLABORATOR |
ClickHouse | ClickHouse | 671,352,237 | 13,575 | den-crane | ```
20.7.1
:) SELECT * FROM t3 WHERE c1 = '2020-02-05 14:34:12.333'
┌──────────────────────c1─┐
│ 2020-02-05 14:34:12.333 │
└─────────────────────────┘
20.6.3.28
┌──────────────────────c1─┐
│ 2020-02-05 14:34:12.333 │
└─────────────────────────┘
20.5.4.40
┌──────────────────────c1─┐
│ 2020-02-0... | 2020-08-10T13:24:54 | 2020-08-10T13:30:55 | {
"+1": 1
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 671,701,413 | 13,570 | den-crane | Closed for now. Reopen if needed. | 2020-08-11T03:21:46 | 2020-08-11T03:21:54 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 671,847,124 | 13,572 | abyss7 | > It will not work properly in Arcadia because it has unpatched version of Poco.
> Wrong timeouts or deadlocks are expected.
Yes, it's a possible outcome - we had an agreement to turn it on and test. | 2020-08-11T09:48:58 | 2020-08-11T09:48:58 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 672,005,884 | 13,619 | simPod | There was 20.6.3.28 and 20.1.11.73 running alongside.
If there were no exceptions in replication_queue, should `SYSTEM RESTART REPLICAS` do the trick? | 2020-08-11T15:12:08 | 2020-08-11T15:12:08 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 671,299,422 | 13,571 | filimonov | Check / help review: https://github.com/ClickHouse/ClickHouse/pull/13421/files | 2020-08-10T11:24:39 | 2020-08-10T11:24:39 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 671,517,508 | 13,581 | vitlibar | No need to backport. | 2020-08-10T18:31:50 | 2020-08-10T18:31:50 | {} | MEMBER |
ClickHouse | ClickHouse | 671,876,171 | 13,597 | PatricYan | > Maybe help you:
> ` brew install binutils`
No, it doesn't work. The error is same as before | 2020-08-11T10:53:53 | 2020-08-11T10:53:53 | {} | NONE |
ClickHouse | ClickHouse | 671,300,750 | 13,527 | filimonov | How do you calculate it? Can you share some sample data and queries? | 2020-08-10T11:27:57 | 2020-08-11T12:38:24 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 671,922,007 | 13,598 | Anshelen | @filimonov i repeated your requests:
```
CREATE TABLE test (ts DATETIME, id VARCHAR, ver Int8, str Nullable(VARCHAR), num Nullable(DOUBLE)) ENGINE=MergeTree PARTITION BY toStartOfDay(ts) ORDER BY id;
INSERT INTO test VALUES (now(), '1', 1, null, null);
CREATE MATERIALIZED VIEW test_mv2 ENGINE=AggregatingMergeTr... | 2020-08-11T12:40:09 | 2020-08-11T12:40:09 | {} | NONE |
ClickHouse | ClickHouse | 671,931,822 | 4,391 | filimonov | Something like that. ClickHouse is a columnar database, so the responce is send in a form of so called block containing 2 columns.
One with id, another with aggregation function states.
I.e. smth like:
id column: 1,2,3,4,5
count(id) column: 100,100,100,100,100
| 2020-08-11T12:59:38 | 2020-08-11T12:59:38 | {
"+1": 1
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 671,971,955 | 4,061 | filimonov | @georgantasp actually your issue differs. To diagnose that you can try disabling suggestions and try running queries from:
https://github.com/ClickHouse/ClickHouse/blob/5a97f0c19e3990d96805b9460c93064b713a0fe2/programs/client/Suggest.cpp#L88-L120 | 2020-08-11T14:13:17 | 2020-08-11T14:13:17 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 672,123,018 | 13,619 | den-crane | >Clickhouse started processing it but in a very slow pace. There was like 1500 lost parts and it was going 1 part per 5 minutes.
Then I don't really understand the issue. What was in system.replication_queue? Postpone reason? | 2020-08-11T17:35:35 | 2020-08-11T17:35:35 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 671,371,864 | 13,577 | sliontc | Thanks a lot. In your opinion what version should I upgrade to? | 2020-08-10T14:00:13 | 2020-08-10T14:00:13 | {} | NONE |
ClickHouse | ClickHouse | 671,402,609 | 13,577 | alexey-milovidov | We guarantee that new server will work with the data from old server. | 2020-08-10T14:50:49 | 2020-08-10T14:50:49 | {} | MEMBER |
ClickHouse | ClickHouse | 671,903,519 | 13,592 | filimonov | Non constant DEFAULT can require reordering of existing data, that's why it's forbidden.
Constant DEFAULTs actually may be allowed. | 2020-08-11T12:02:02 | 2020-08-11T12:02:02 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 671,362,708 | 13,567 | den-crane | WA expression (cast - toDateTime(d) ):
```
CREATE TABLE example_table
(
d DateTime64(3),
a Int
)
ENGINE = MergeTree
PARTITION BY toYYYYMM(d)
ORDER BY d
TTL toDateTime(d) + INTERVAL 1 MONTH;
```
| 2020-08-10T13:44:19 | 2020-08-10T13:44:19 | {
"+1": 2
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 671,391,061 | 13,574 | xjewer | >So it seems that we need to allow duplicate parts to exists for some time on both shards, and we need to remember the mapping of parts that were moved, and the Distributed table should be aware of that.
That's exactly what the proposal covers (see 3rd step) by adding metadata to the moving part.
Hence the executor... | 2020-08-10T14:31:08 | 2020-08-10T14:31:08 | {
"+1": 1
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 671,796,326 | 12,261 | yinjiakang | > 19.5.2.6 is too old.
> The serialization of groupBitmapState has been adjusted, additionally, previous aggregation data format requires to be escaped with Base64 encoding, which leads to remarkable overhead.
so , from which version is this feature available ? thx | 2020-08-11T08:06:44 | 2020-08-11T08:06:44 | {} | NONE |
ClickHouse | ClickHouse | 671,952,143 | 12,278 | akuzm | Merged all parts of this except ALWAYS_INLINE which is also in variant 2 PR. | 2020-08-11T13:38:05 | 2020-08-11T13:38:05 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 671,374,730 | 13,577 | sliontc | And for safety should I upgrade from 18 to 19 and then to 20 step by step? | 2020-08-10T14:04:29 | 2020-08-10T14:04:29 | {} | NONE |
ClickHouse | ClickHouse | 671,603,519 | 13,589 | alexey-milovidov | AFAIR it gives huge performance degradation. | 2020-08-10T21:39:28 | 2020-08-10T21:39:28 | {} | MEMBER |
ClickHouse | ClickHouse | 671,631,198 | 13,343 | azat | >the problem is that the header wasn't written, while it should.
@akuzm fixed | 2020-08-10T23:00:51 | 2020-08-10T23:00:51 | {} | MEMBER |
ClickHouse | ClickHouse | 671,920,222 | 13,599 | filimonov | The style is described here: https://clickhouse.tech/docs/en/development/style/
And there is a style check script: see https://github.com/ClickHouse/ClickHouse/tree/master/utils/check-style (fired automatically for every PR in CI)
AFAIK most programmers just do formatting manually, if you see the issues in formatt... | 2020-08-11T12:36:26 | 2020-08-11T12:36:26 | {
"+1": 1
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 671,955,924 | 13,577 | den-crane | Yes, but I would use
sudo apt-get install **clickhouse-common-static** clickhouse-client clickhouse-server
or specified the desired version
apt-get install clickhouse-server=20.6.2.15 clickhouse-client=20.6.2.15 clickhouse-common-static=20.6.2.15
And test it at your testing environment first. Your appli... | 2020-08-11T13:44:53 | 2020-08-11T13:44:53 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 672,045,852 | 13,612 | alexey-milovidov | #11364 | 2020-08-11T16:00:33 | 2020-08-11T16:00:33 | {} | MEMBER |
ClickHouse | ClickHouse | 672,068,428 | 13,594 | qoega | It is probably clickhouse-odbc issue https://github.com/ClickHouse/clickhouse-odbc/issues
You can provide more information about `CESHI1` table? Schema etc.
Have you tried to reproduce it using another DB? | 2020-08-11T16:24:54 | 2020-08-11T16:24:54 | {} | MEMBER |
ClickHouse | ClickHouse | 671,217,014 | 13,343 | azat | >The reason the performance test fails is that now we can't read an empty file as TSVWithNamesAndTypes:
Not really, the problem is that the header wasn't written, while it should.
P.S. and reading empty files is not supported anyway, this is the result w/o this PR:
```sql
SELECT * FROM file('/dev/null', TSVWi... | 2020-08-10T08:03:23 | 2020-08-10T08:03:23 | {} | MEMBER |
ClickHouse | ClickHouse | 671,240,132 | 13,516 | filimonov | I like the idea, in future the responsibilities of watchdog can be slightly extended.
For example, many webservers (nginx / httpd) start parent process that's owned by root (to listen privilegued ports) which then forks copies of worker nodes up using a less credentialed user.
A lot of initialization (and also p... | 2020-08-10T08:59:29 | 2020-08-10T09:07:04 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 671,253,230 | 13,570 | peaksnail | I execute sql
```
optimize table goods_sale final
```
the latest record disappear and the older record is saved, it's so weird | 2020-08-10T09:28:37 | 2020-08-10T09:28:37 | {} | NONE |
ClickHouse | ClickHouse | 671,369,067 | 13,577 | den-crane | No. Nothing will be lost. CH supports upgrades.
But you should test a new version first because there are backward incompatible changes https://github.com/ClickHouse/ClickHouse/blob/master/CHANGELOG.md
And should test an upgrade procedure as well. | 2020-08-10T13:55:56 | 2020-08-10T13:55:56 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 671,517,448 | 13,580 | vitlibar | No need to backport. | 2020-08-10T18:31:41 | 2020-08-10T18:31:41 | {} | MEMBER |
ClickHouse | ClickHouse | 671,657,022 | 13,590 | alexey-milovidov | The difference is that unescaped carriage return will break value.
I'm afraid that I said somewhere that the only characters has to be escaped in TSV are backslash, tab and line feed.
For example, if you formatting binary data, you don't have to escape zero byte, vertical tab, form feed, etc... | 2020-08-11T00:32:50 | 2020-08-11T00:33:37 | {} | MEMBER |
ClickHouse | ClickHouse | 671,669,746 | 13,570 | peaksnail | thanks, I resync the data and the result is right , I will keep to observe | 2020-08-11T01:21:24 | 2020-08-11T01:21:24 | {} | NONE |
ClickHouse | ClickHouse | 671,914,469 | 13,598 | filimonov | W/o OrNull it works :)
```
CREATE MATERIALIZED VIEW test_mv2 ENGINE=AggregatingMergeTree PARTITION BY toStartOfDay(ts) ORDER BY id POPULATE AS SELECT ts,id, argMaxState(str, ver) str, argMaxState(num, ver) num FROM test GROUP BY ts, id;
SELECT ts, id, argMaxMerge(str), argMaxMerge(num) FROM test_mv2 GROUP BY ts,... | 2020-08-11T12:24:33 | 2020-08-11T12:24:33 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 671,915,000 | 13,598 | filimonov | @hczhcz can you check that? | 2020-08-11T12:25:36 | 2020-08-11T12:25:36 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 670,417,721 | 13,418 | qoega | In your automaton you can use something like `{CLUSTER_PLACEHOLDER_shard}` and replace `CLUSTER_PLACEHOLDER` in query to cluster name before query execution. | 2020-08-07T09:11:01 | 2020-08-07T09:11:01 | {} | MEMBER |
ClickHouse | ClickHouse | 670,703,232 | 13,501 | alexey-milovidov | See also https://clickhouse-test-reports.s3.yandex.net/0/c6ac3398977e76b3b2123759c29c509f4587ef80/coverage_report/test_output/Functions/index-sort-l.html | 2020-08-07T20:37:32 | 2020-08-07T20:37:32 | {} | MEMBER |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.