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 | 760,890,923 | 18,373 | alexey-milovidov | The lack of comments delayed understanding of this PR. | 2021-01-15T11:36:46 | 2021-01-15T11:36:46 | {
"+1": 1
} | MEMBER |
ClickHouse | ClickHouse | 760,921,587 | 19,028 | KochetovNicolai | Ok. ILLEGAL INSTRUCTION somewhere inside openssl file generated by perl script
https://github.com/ClickHouse-Extras/openssl/blob/master/crypto/arm64cpuid.pl#L73
I think I cannot debug this.
You can try to build without ssl support, like `cmake -DENABLE_SSL=OFF` | 2021-01-15T12:45:21 | 2021-01-15T12:45:21 | {} | MEMBER |
ClickHouse | ClickHouse | 760,235,380 | 11,871 | alexey-milovidov | > we have several coordinate systems (2d plane, sphere or WGS84) And we have some policy how to calculate distance, for example
That is the definition of metric, exactly. | 2021-01-14T14:35:31 | 2021-01-14T14:35:31 | {} | MEMBER |
ClickHouse | ClickHouse | 760,249,565 | 19,066 | alexey-milovidov | The issue should not exist on newer versions, as ALTER query has been made completely lock-free. | 2021-01-14T14:57:59 | 2021-01-14T14:57:59 | {} | MEMBER |
ClickHouse | ClickHouse | 760,284,875 | 18,689 | alex-zaitsev | > > Why don't measure elapsed time on client (clickhouse-client does it to calculate speed)?
>
> Re clickhouse-client - yes, and I questioned/mentioned that above.
> Re why we don't do it at client side - @alex-zaitsev could you please comment on this?
We need it for HTTP queries in order to see the query time w... | 2021-01-14T15:52:32 | 2021-01-14T15:52:32 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 760,251,774 | 18,689 | traceon | Also, please note the following somewhat relevant places, where I decided not to update the code to read/write elapsed time:
1. This uses its own stopwatch to measure things, it covers a bit more than what would be accumulated in progress.elapsed time, so not sure if I should change this:
https://github.com/ClickHo... | 2021-01-14T15:01:14 | 2021-01-14T15:06:23 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 760,650,871 | 19,099 | l1t1 | and the rows read are very different when read one table and read joined two tables
```sql
mysql> select count(*) from catalog_sales;
+----------+
| count() |
+----------+
| 14401261 |
+----------+
1 row in set (0.00 sec)
Read 1 rows, 4.01 KiB in 0.00058857 sec., 1699 rows/sec., 6.65 MiB/sec.
mysql> select... | 2021-01-15T05:14:30 | 2021-01-15T05:14:30 | {} | NONE |
ClickHouse | ClickHouse | 760,690,642 | 18,853 | azat | >AST fuzzer — Logical error: 'Block structure mismatch in PipelineExecuting stream: different columns:
#19007 | 2021-01-15T06:46:53 | 2021-01-15T06:46:53 | {} | MEMBER |
ClickHouse | ClickHouse | 760,783,181 | 19,102 | alexey-milovidov | Rows written while merge are also counted. | 2021-01-15T09:36:35 | 2021-01-15T09:36:35 | {} | MEMBER |
ClickHouse | ClickHouse | 760,792,940 | 19,077 | alexey-milovidov | Are they published in checks? | 2021-01-15T09:55:28 | 2021-01-15T09:55:28 | {} | MEMBER |
ClickHouse | ClickHouse | 760,834,084 | 18,979 | tavplubix | AST fuzzer - #19108
Functional stateless tests (ANTLR debug) - `Logical error: 'Incomplete granules are not allowed while blocks are granules size...` - known issue | 2021-01-15T10:43:39 | 2021-01-15T10:43:39 | {} | MEMBER |
ClickHouse | ClickHouse | 760,901,200 | 18,955 | alexey-milovidov | There is already `MergeTreeData::getPartsCount()` method. | 2021-01-15T12:00:09 | 2021-01-15T12:00:09 | {} | MEMBER |
ClickHouse | ClickHouse | 760,218,217 | 11,871 | nikitamikhaylov | > LocalWGS84
@alexey-milovidov Could you please explain a little more precise and detailed why do we need "do the following" on every calculation. Why `Euclidean` and `LocalWGS84` are metrics? How do the differ? I googled and watched some presentation, but I don't really understand how it correlates with mathematica... | 2021-01-14T14:06:52 | 2021-01-14T14:06:52 | {} | MEMBER |
ClickHouse | ClickHouse | 760,304,721 | 19,079 | alexey-milovidov | Culprit: `SELECT isConstant(multiIf('dict1' = 'dict1', 'dictName1', 'dictName2'))`
It is the implementation detail of `multiIf`.
Yes, it's possible to make it return constant expression in this case. | 2021-01-14T16:23:48 | 2021-01-14T16:23:48 | {} | MEMBER |
ClickHouse | ClickHouse | 760,899,648 | 18,955 | alexey-milovidov | The test does not work even with infinite retries.
Something is wrong with consistency of these metrics?
BTW, I don't like the idea to maintaining the consistency manually.
Maybe better to move them to AsynchronousMetrics? | 2021-01-15T11:56:20 | 2021-01-15T11:56:20 | {} | MEMBER |
ClickHouse | ClickHouse | 760,900,538 | 18,955 | alexey-milovidov | But it will be inefficient for large number of data parts. | 2021-01-15T11:58:33 | 2021-01-15T11:58:33 | {} | MEMBER |
ClickHouse | ClickHouse | 760,901,381 | 18,671 | filimonov | Seems cmake scripts should be adjusted this time. | 2021-01-15T12:00:32 | 2021-01-15T12:00:32 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 760,926,200 | 5,747 | alexey-milovidov | My team is not working on this feature, we hope it will be implemented by community (maybe by Altinity or ArenaData, but I'm not sure). | 2021-01-15T12:55:31 | 2021-01-15T12:55:31 | {} | MEMBER |
ClickHouse | ClickHouse | 760,973,989 | 19,001 | alesapin | Not implemented | 2021-01-15T14:29:10 | 2021-01-15T14:29:10 | {} | MEMBER |
ClickHouse | ClickHouse | 760,977,335 | 19,009 | alesapin | `test_ttl_replicated/test.py::test_ttl_empty_parts` known flaky, will cherry-pick fix manually. | 2021-01-15T14:35:14 | 2021-01-15T14:35:14 | {} | MEMBER |
ClickHouse | ClickHouse | 760,457,732 | 18,446 | azat | >I mean, we can wrap cast in identity, so it will be identity(cast(..... The type will be correct, but it will not be optimized away.
I was trying to avoid two function calls, but I guess it is not significant for this place, OK updated to `identity(cast())` | 2021-01-14T20:30:00 | 2021-01-14T20:30:00 | {} | MEMBER |
ClickHouse | ClickHouse | 760,675,275 | 19,097 | templarzq | @l1t1 ok,the code formatted. | 2021-01-15T06:09:31 | 2021-01-15T06:09:31 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 760,698,392 | 18,959 | filimonov | > Do you have any documents related to atomic database? I can't find them in the official documents.
Check
https://github.com/ClickHouse/ClickHouse/issues/18123
https://github.com/ClickHouse/ClickHouse/issues/17906
https://github.com/ClickHouse/ClickHouse/issues/6787
There are also some materials (Russian):
... | 2021-01-15T07:05:50 | 2021-01-15T07:05:50 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 760,711,902 | 6,503 | wangweiA | @gubinjie
Add these two configuration parameters in the file "/etc/clickhouse-server/users.xml" and restart clickhouse-server :
<http_send_timeout>3600</http_send_timeout>
<http_receive_timeout>3600</http_receive_timeout>
the unit is Seconds. | 2021-01-15T07:38:39 | 2021-01-15T07:38:39 | {} | NONE |
ClickHouse | ClickHouse | 760,779,847 | 18,542 | alexey-milovidov | @nautaa Congratulations for merged PR! 🎉
Now we can add `decodeHTMLComponent` function:
https://en.wikipedia.org/wiki/List_of_XML_and_HTML_character_entity_references | 2021-01-15T09:29:59 | 2021-01-15T09:29:59 | {} | MEMBER |
ClickHouse | ClickHouse | 760,790,003 | 19,100 | alexey-milovidov | Distribution of data by blocks is implementation specific and depends on internal algorithms. | 2021-01-15T09:49:46 | 2021-01-15T09:49:46 | {} | MEMBER |
ClickHouse | ClickHouse | 760,916,638 | 5,747 | glader | @alexey-milovidov do you have any plans to implement this functionality? | 2021-01-15T12:34:13 | 2021-01-15T12:34:13 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 760,244,548 | 18,689 | traceon | Originally, we needed elapsed time in `X-ClickHouse-Progress` HTTP headers for our platform (at Altinity.) It could be probably grafted in an ad-hoc manner somewhere around that code, but that wouldn't be the right way to do it.
@alex-zaitsev may provide more details for when we needed this. | 2021-01-14T14:50:13 | 2021-01-14T14:50:13 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 760,451,370 | 18,952 | azat | @gonzalezjo Am I understand right that `Debug, all off` != upstream, since it will include `-Wl,-O0`? | 2021-01-14T20:17:15 | 2021-01-14T20:17:15 | {} | MEMBER |
ClickHouse | ClickHouse | 760,616,824 | 10,280 | JJWIT | sql : "SELECT bitmapToArray(bitmapBuild([695788478213916450,693475830172484980,693566989590856286,693439459143715197,693457216849577977,693459823102002107,692675502728544393,693813533132981670,693563537879991179,693519689237660505,693459553655720434,693517537362575591,693749165821266823,693584915760939132,6937992492587... | 2021-01-15T03:12:31 | 2021-01-15T03:12:31 | {} | NONE |
ClickHouse | ClickHouse | 760,796,658 | 19,020 | alexey-milovidov | > test_ttl_replicated/test.py::test_ttl_empty_parts
Bad test, fixed in master. | 2021-01-15T10:02:38 | 2021-01-15T10:02:38 | {} | MEMBER |
ClickHouse | ClickHouse | 760,977,190 | 19,009 | alesapin | Race https://github.com/ClickHouse/ClickHouse/issues/19130 reported | 2021-01-15T14:34:56 | 2021-01-15T14:34:56 | {} | MEMBER |
ClickHouse | ClickHouse | 760,333,353 | 18,446 | akuzm | > So it looks like the CAST is required, since only it can represent the correct type for types that does not have it's own DataType, i.e. DateTime and similar.
I mean, we can wrap cast in `identity`, so it will be `identity(cast(....`. The type will be correct, but it will not be optimized away. | 2021-01-14T17:09:18 | 2021-01-14T17:09:18 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 760,524,553 | 18,954 | filimonov | p.2 : https://github.com/ClickHouse/ClickHouse/pull/19096 | 2021-01-14T22:45:47 | 2021-01-14T22:45:47 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 760,699,444 | 10,745 | wangweiA | Is this issue resolved ?? i am getting this issue in 20.9.4.76 version?? | 2021-01-15T07:08:25 | 2021-01-15T07:08:25 | {} | NONE |
ClickHouse | ClickHouse | 760,703,755 | 18,959 | filimonov | @tavplubix it looks like different error code than in https://github.com/ClickHouse/ClickHouse/issues/17661
| 2021-01-15T07:18:44 | 2021-01-15T07:19:18 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 760,773,704 | 18,671 | alexey-milovidov | https://github.com/edenhill/librdkafka/issues/3215 | 2021-01-15T09:17:57 | 2021-01-15T09:17:57 | {
"+1": 1
} | MEMBER |
ClickHouse | ClickHouse | 760,818,096 | 18,542 | nautaa | Thanks for your review and adjust code! I will try it later. @alexey-milovidov | 2021-01-15T10:36:05 | 2021-01-15T10:36:05 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 760,254,490 | 18,689 | traceon | > Why don't measure elapsed time on client (clickhouse-client does it to calculate speed)?
Re clickhouse-client - yes, and I questioned/mentioned that above.
Re why we don't do it at client side - @alex-zaitsev could you please comment on this? | 2021-01-14T15:04:56 | 2021-01-14T15:04:56 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 760,694,391 | 19,097 | templarzq | this draft can only optimize bitmap copy and join action during the same block. if there are same bitmap in different blocks,this draft has no optimization. On the other way , search for same bitmap on every row will cost some time ,this will slow the query without duplicated bitmap somehow. | 2021-01-15T06:56:08 | 2021-01-15T06:56:31 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 760,789,215 | 18,671 | filimonov | Yep, find it already too. Will try 1.6 | 2021-01-15T09:48:12 | 2021-01-15T09:48:12 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 760,810,032 | 19,102 | gj-zhang | thanks | 2021-01-15T10:28:34 | 2021-01-15T10:28:34 | {} | NONE |
ClickHouse | ClickHouse | 760,927,260 | 5,747 | alexey-milovidov | Kerberos for HDFS: #16621
Kerberos for Kafka: #12771
Kerberos for ClickHouse own auth: #14995
Kerberos for ZK: not currently in progress. | 2021-01-15T12:57:58 | 2021-01-15T12:57:58 | {
"+1": 2
} | MEMBER |
ClickHouse | ClickHouse | 760,486,866 | 19,055 | alesapin | Ok, the only docker changes work. | 2021-01-14T21:26:38 | 2021-01-14T21:26:38 | {} | MEMBER |
ClickHouse | ClickHouse | 760,516,376 | 19,095 | alexey-milovidov | Could you please share the file to reproduce the issue?
You can obfuscate data or send it privately to clickhouse-feedback@yandex-team.com | 2021-01-14T22:29:35 | 2021-01-14T22:29:35 | {} | MEMBER |
ClickHouse | ClickHouse | 760,625,757 | 19,097 | templarzq | just reverse the join order,the query speed will change very much:
select a.dim,bitmapCardinality(b.ids) from test_bm_join2 b right join test_bm2 a using(dim) | 2021-01-15T03:43:45 | 2021-01-15T03:43:45 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 759,345,124 | 17,041 | vdimir | Could you add test for new feature with flag enabled?
Also it would be nice to have performance comparison. | 2021-01-13T10:07:28 | 2021-01-15T08:23:55 | {} | MEMBER |
ClickHouse | ClickHouse | 760,915,174 | 19,095 | glader | Sent to clickhouse-feedback@yandex-team.com | 2021-01-15T12:31:17 | 2021-01-15T12:31:17 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 760,938,363 | 18,913 | alesapin | Work as expected during manual insert/select from bad part:
```
$ ~/code/cpp/BuildCH/utils/check-marks/check-marks message.mrk2 message.bin
Mark 0, points to 0, 0, has rows after 8192, decompressed size 1048576.
Mark 1, points to 116015, 0, has rows after 8192, decompressed size 791228.
Mark 2, points to 217163, 0... | 2021-01-15T13:20:51 | 2021-01-15T13:21:07 | {} | MEMBER |
ClickHouse | ClickHouse | 760,224,914 | 11,871 | alexey-milovidov | Metric in the sense of "metric tensor". It defines scalar product and allows to calculate lengths of vectors (and, consequently, lengths of curves). We can treat it as additional coefficients that should be applied when we calculate lengths. Example: https://en.wikipedia.org/wiki/Metric_tensor#The_round_metric_on_a_sph... | 2021-01-14T14:18:14 | 2021-01-14T14:21:35 | {} | MEMBER |
ClickHouse | ClickHouse | 760,334,682 | 19,089 | den-crane | Mutations are only supported by *MergeTree and Memory Engines.
use mutations directly against company_db.events `ALTER TABLE company_db.events ON CLUSTER 'company_cluster' DELETE WHERE uid = 100;` | 2021-01-14T17:11:29 | 2021-01-14T17:16:11 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 760,367,186 | 19,024 | den-crane | Поправил.
>PS. А кто сказал, что поддерживаются выражения от столбцов?
Упс, не обратил внимания, там этот бред видимо давно. | 2021-01-14T18:06:47 | 2021-01-14T18:06:47 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 760,652,041 | 19,097 | l1t1 | please use ``` instead of ... to display code in good format.
| 2021-01-15T05:17:55 | 2021-01-15T05:17:55 | {} | NONE |
ClickHouse | ClickHouse | 760,710,498 | 15,889 | qoega | Recent PR
```
2021.01.14 17:46:57.656157 [ 66 ] {a3ab16bf-a2ef-4bb9-9dba-d4e5f5d5163a} <Fatal> : Logical error: 'Block structure mismatch in PipelineExecuting stream: different columns:
ignore(-9223372036854775808, toDateTime(addDays(addDays(ignore(addDays(toDateTime(256), 1)), addDays(NULL, toDateTime(addDays(toDat... | 2021-01-15T07:35:05 | 2021-01-15T07:35:05 | {} | MEMBER |
ClickHouse | ClickHouse | 760,762,924 | 18,671 | alexey-milovidov | ```
WARNING: ThreadSanitizer: data race (pid=7)
Write of size 4 at 0x7b6000182e10 by thread T65:
#0 rd_kafka_toppar_op_serve obj-x86_64-linux-gnu/../contrib/librdkafka/src/rdkafka_partition.c:2133:39 (clickhouse+0x15b2a54d)
#1 rd_kafka_op_handle obj-x86_64-linux-gnu/../contrib/librdkafka/src/rdkafka_op.c:... | 2021-01-15T08:56:46 | 2021-01-15T08:56:46 | {} | MEMBER |
ClickHouse | ClickHouse | 760,781,194 | 19,099 | alexey-milovidov | @l1t1 ClickHouse has optimization for trivial count query from MergeTree table. These queries processed from table metadata without touching the data. | 2021-01-15T09:32:38 | 2021-01-15T09:32:38 | {} | MEMBER |
ClickHouse | ClickHouse | 760,878,028 | 19,064 | alexey-milovidov | > test_ttl_replicated/test.py::test_ttl_empty_parts
Fixed in master. | 2021-01-15T11:09:42 | 2021-01-15T11:09:42 | {} | MEMBER |
ClickHouse | ClickHouse | 760,224,461 | 18,689 | vdimir | Also could you tell why do you need elapsed_time in progress? | 2021-01-14T14:17:31 | 2021-01-14T14:17:31 | {} | MEMBER |
ClickHouse | ClickHouse | 760,419,635 | 18,979 | alexey-milovidov | > test_ttl_replicated/test.py::test_ttl_empty_parts
Fixed in master. | 2021-01-14T19:28:58 | 2021-01-14T19:28:58 | {} | MEMBER |
ClickHouse | ClickHouse | 760,745,639 | 19,089 | etc-crontab | Thanks a lot @den-crane for the fast answer and the provided query. | 2021-01-15T08:19:14 | 2021-01-15T08:19:14 | {} | NONE |
ClickHouse | ClickHouse | 760,798,356 | 18,189 | alexey-milovidov | Not enough info. Reopen when it will be reproduced next time. | 2021-01-15T10:05:43 | 2021-01-15T10:05:43 | {} | MEMBER |
ClickHouse | ClickHouse | 760,923,883 | 18,913 | alesapin | Unfortunately, the problem is deeper. We get marks bigger than fixed granularity during the horizontal merge. Vertical merge error is a consequence.
The additional check will fail before this bad horizontal merge, but I hope to figure out the error earlier.
https://github.com/ClickHouse/ClickHouse/pull/19123 | 2021-01-15T12:50:23 | 2021-01-15T12:50:45 | {} | MEMBER |
ClickHouse | ClickHouse | 760,974,671 | 19,008 | alesapin | Stress -- rocks db, known flaky test. Integration fixed in master (I'll cherry-pick manually) | 2021-01-15T14:30:25 | 2021-01-15T14:30:25 | {} | MEMBER |
ClickHouse | ClickHouse | 760,247,096 | 18,689 | alexey-milovidov | Why don't measure elapsed time on client (clickhouse-client does it to calculate speed)?
Another question - how elapsed times are merged from multiple servers in distributed query execution? | 2021-01-14T14:54:03 | 2021-01-14T14:54:03 | {} | MEMBER |
ClickHouse | ClickHouse | 760,296,668 | 18,937 | mathfool | TreeRewriter:568

TableJoin.cpp:97
Chinese characters considered not a qualified identifier and a table prefix added. This check should also apply to the source columns to make the logic aligned.

ENGINE = MergeTree()
ORDER BY( dim )
SETTINGS index_granularity = 8192;
create table test_bm_join2(
dim UInt64,
ids AggregateFunction(groupBitmap, UInt64) )
ENGINE = M... | 2021-01-15T03:42:49 | 2021-01-15T03:42:49 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 760,652,665 | 18,761 | xinge-1 | > ClickHouse使用的版本是?
找到原因了 是MaterializeMySQL同步不支持V1版本的binlog | 2021-01-15T05:19:44 | 2021-01-15T05:19:44 | {} | NONE |
ClickHouse | ClickHouse | 760,616,548 | 18,824 | tangxueming | > Check my advice. Reduce granularity in index to hour.` Order by (toStartOfHour(timestamp), hash(traceId))`
I failed with timestamp. I need list record order by timestamp desc, so I set primary key with timestamp.
Now I think it is may satisfy in my case.
When query one day data, I use query without sample.
W... | 2021-01-15T03:11:34 | 2021-01-15T06:06:15 | {} | NONE |
ClickHouse | ClickHouse | 760,768,483 | 18,671 | alexey-milovidov | @filimonov They have 1.6.0 RC1, try to update first. | 2021-01-15T09:07:57 | 2021-01-15T09:07:57 | {} | MEMBER |
ClickHouse | ClickHouse | 760,882,603 | 18,185 | alexey-milovidov | Could you please merge with master? | 2021-01-15T11:18:40 | 2021-01-15T11:18:40 | {} | MEMBER |
ClickHouse | ClickHouse | 760,884,076 | 14,700 | alexey-milovidov | https://clickhouse-test-reports.s3.yandex.net/17310/d7200ee2ede7b739a682c8f5c06e47c7e624a121/fuzzer/report.html#fail1 | 2021-01-15T11:21:31 | 2021-01-15T11:21:31 | {} | MEMBER |
ClickHouse | ClickHouse | 760,923,898 | 19,028 | alexey-milovidov | @zhangjingwei-gwi Could you please also try the latest (master) version instead of 20.9? | 2021-01-15T12:50:24 | 2021-01-15T12:50:24 | {} | MEMBER |
ClickHouse | ClickHouse | 760,943,477 | 16,610 | alexey-milovidov | Looks like #18806. | 2021-01-15T13:31:16 | 2021-01-15T13:31:16 | {} | MEMBER |
ClickHouse | ClickHouse | 760,960,554 | 19,077 | qoega | They will be published. Currently check status will be green, but individual test results will be written to CI database correctly.
And it is currently not committed and I run them manually for debug purposes. | 2021-01-15T14:05:07 | 2021-01-15T14:05:07 | {} | MEMBER |
ClickHouse | ClickHouse | 760,975,418 | 19,009 | alesapin | Functional stateless tests (ANTLR debug) — unrelated race. | 2021-01-15T14:31:44 | 2021-01-15T14:31:44 | {} | MEMBER |
ClickHouse | ClickHouse | 760,236,024 | 19,028 | KochetovNicolai | Can you please show gdb output?
```
gdb ./clickhouse
r server config.xml
```
then wait for crash and show result for `bt full` | 2021-01-14T14:36:34 | 2021-01-14T14:36:34 | {} | MEMBER |
ClickHouse | ClickHouse | 760,415,164 | 19,058 | alexey-milovidov | > test_ttl_replicated/test.py::test_ttl_empty_parts
Fixed in master. | 2021-01-14T19:23:39 | 2021-01-14T19:23:39 | {} | MEMBER |
ClickHouse | ClickHouse | 760,471,389 | 18,671 | filimonov | Have no idea what was the reason of the failure, it looks like some network / CI issue (may be smth happened during iptables manipulations).I've did rebase to fresh master, let's see what will happen this time. | 2021-01-14T20:57:25 | 2021-01-14T20:57:25 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 760,713,443 | 6,503 | gubinjie | @wangweiA Thank you, it has been solved long ago. It is better to use socket_timeout. This configuration item contains http_send_timeout and http_receive_timeout. | 2021-01-15T07:42:18 | 2021-01-15T07:42:18 | {} | NONE |
ClickHouse | ClickHouse | 760,938,019 | 16,610 | alexey-milovidov | @KochetovNicolai The fix was not backported to 20.8:
```
$ clickhouse-client
ClickHouse client version 21.1.1.5646.
Connecting to localhost:9000 as user default.
Connected to ClickHouse server version 20.8.12 revision 54438.
ClickHouse server version is older than ClickHouse client. It may indicate that the se... | 2021-01-15T13:20:17 | 2021-01-15T13:20:17 | {} | MEMBER |
ClickHouse | ClickHouse | 760,957,356 | 18,772 | alexey-milovidov | @azat I think we should make all such blockers to account memory for server and not account for query/user. | 2021-01-15T13:59:05 | 2021-01-15T13:59:05 | {} | MEMBER |
ClickHouse | ClickHouse | 760,972,679 | 18,369 | filimonov | Fixed in #18853 | 2021-01-15T14:26:42 | 2021-01-15T14:26:42 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 760,239,990 | 11,871 | alexey-milovidov | More details:
Suppose you want to calculate scalar product of two vectors.
You can treat first vector as row and second vector as column and do matrix multiplication:
`vec1_row * vec2_column`
You can also calculate the length of the vector: `sqrt(scalar_product(vec, vec))`
But in some coordinate systems you ... | 2021-01-14T14:42:51 | 2021-01-14T14:42:51 | {} | MEMBER |
ClickHouse | ClickHouse | 760,486,825 | 18,542 | alexey-milovidov | I have simplified code and added comments: https://github.com/ClickHouse/ClickHouse/pull/18542/commits/9db1299f4135831c6cbc5cf9dddd88596bb9d91b | 2021-01-14T21:26:35 | 2021-01-14T21:26:35 | {} | MEMBER |
ClickHouse | ClickHouse | 760,753,811 | 18,671 | alexey-milovidov | @filimonov The same test has failed. | 2021-01-15T08:36:28 | 2021-01-15T08:36:28 | {} | MEMBER |
ClickHouse | ClickHouse | 756,394,052 | 18,849 | igorbb | Thanks, @den-crane
I think the example I posted, might have been too simple.
I know it will fail, would be great to have an alias to be computed only if included on the query. Like in the sample query.
In our case, we depend on other conditions, so I think SimpleAggregateFunction would not be enough for us.
Un... | 2021-01-07T21:24:13 | 2021-01-15T09:51:20 | {} | NONE |
ClickHouse | ClickHouse | 760,965,302 | 18,913 | alesapin | Another example:
```
$ ~/code/cpp/BuildCH/utils/check-marks/check-marks memory_usage.mrk2 memory_usage.bin
Mark 0, points to 0, 0, has rows after 7452, decompressed size 122216.
Mark 1, points to 0, 59616, has rows after 7825, decompressed size 62600.
Mark 2, points to 2795, 0, has rows after 8192, decompressed si... | 2021-01-15T14:13:38 | 2021-01-15T14:13:38 | {} | MEMBER |
ClickHouse | ClickHouse | 760,524,809 | 18,927 | filimonov | > Also uid/gid was changed:
That part fixed in #19096 | 2021-01-14T22:46:25 | 2021-01-14T22:46:41 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 760,917,594 | 17,838 | alexey-milovidov | @weeds085490 See https://github.com/ClickHouse/ClickHouse/pull/19122. | 2021-01-15T12:36:22 | 2021-01-15T12:36:22 | {} | MEMBER |
ClickHouse | ClickHouse | 760,972,759 | 19,002 | alesapin | Drop is not implemented in this branch. | 2021-01-15T14:26:53 | 2021-01-15T14:26:53 | {} | MEMBER |
ClickHouse | ClickHouse | 760,330,197 | 18,924 | snar | OOps, sorry, missed that :( Downloaded, all tests passed. Thanks! | 2021-01-14T17:04:20 | 2021-01-14T17:04:20 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 760,402,136 | 18,952 | gonzalezjo | # A Look at the Data on Realistic Hardware
### Challenges
Returning to this after a brutal past day and a half of attempting to collect data. Attempting to benchmark this on slow storage was a bad idea: I only had cheap USB drives, and they slow down a lot after a couple of builds because of the cheap NAND a... | 2021-01-14T18:58:05 | 2021-01-14T18:59:08 | {
"+1": 2
} | NONE |
ClickHouse | ClickHouse | 760,587,752 | 19,028 | ElapsedSoul | @KochetovNicolai
Got this:
```
[root@10-20-103-5 programs]# gdb ./clickhouse
GNU gdb (GDB) EulerOS 8.3.1-11.ky10
Copyright (C) 2019 Free Software Foundation, Inc.
License GPLv3+: GNU GPL version 3 or later <http://gnu.org/licenses/gpl.html>
This is free software: you are free to change and redistribute it.
The... | 2021-01-15T01:40:11 | 2021-01-15T03:16:19 | {} | NONE |
ClickHouse | ClickHouse | 760,756,481 | 18,898 | taylor12805 | > The implementation in #18791 is to cast the second arg column to be `UInt64` column, then we can support UInt8 -> UInt64 types.
Yes, I did't see it. I will close this PR | 2021-01-15T08:42:46 | 2021-01-15T08:42:46 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 760,775,912 | 18,671 | alexey-milovidov | @filimonov
How to read the results of integration tests:
- download and unpack the tar;
- read logs and stderr. | 2021-01-15T09:22:06 | 2021-01-15T09:22:06 | {} | MEMBER |
ClickHouse | ClickHouse | 760,796,018 | 19,007 | alexey-milovidov | @KochetovNicolai Arcadia does not support distributed queries:
```
Code: 279. DB::Exception: Received from localhost:35650. DB::Exception: All connection tries failed. Log:
E
E Code: 210, e.displayText() = DB::NetException: Connection refused (localhost:59000) (version 21.1.1.5646-arcadia)
E Code: 210, e... | 2021-01-15T10:01:32 | 2021-01-15T10:01:32 | {} | MEMBER |
ClickHouse | ClickHouse | 760,863,262 | 18,959 | tavplubix | > @tavplubix it looks like different error code than in #17661
Yes, it's `EINVAL`, fixed in 20.8.4: https://github.com/ClickHouse/ClickHouse/blob/fb6d1dc18e1e5641e4753c69f28ad73d6c1db3bb/src/Common/renameat2.cpp#L68-L69
However, WSL may also return `ENOSYS`, it's also fixed, but not backported to 20.8. | 2021-01-15T10:57:14 | 2021-01-15T10:57:14 | {} | MEMBER |
ClickHouse | ClickHouse | 735,996,563 | 17,512 | azat | >The conflict doesn't seem to be easy to resolve.
This issue looks pretty serious, and actually backport is not that hard, there was just some refactoring that make it hard, here is a backport - #17644 | 2020-11-30T19:36:03 | 2020-11-30T19:36:03 | {} | MEMBER |
ClickHouse | ClickHouse | 736,115,282 | 16,947 | azat | >Is this problem related to the fact that we match remote columns by name? This is not the first one, we probably should match them by index, like for UNION.
@akuzm Not really, let's consider the following example:
Some node got:
```sql
SELECT _shard_num, count() FROM remote('127.0.0.{1,2}', system.one) GROUP... | 2020-11-30T23:23:31 | 2020-11-30T23:23:31 | {} | MEMBER |
ClickHouse | ClickHouse | 736,129,521 | 17,453 | azat | @alexey-milovidov should it be rebased to restart the tests?
I found one more thing that will conflict with this, BlockerInThread (and LockExceptionInThread after this PR) is not nested, and this may be a problem in some cases. | 2020-12-01T00:07:11 | 2020-12-01T00:07:11 | {} | MEMBER |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.