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 | 629,734,860 | 10,866 | axolm | @alexey-milovidov since the function is needed, let's merge it! | 2020-05-17T02:52:21 | 2020-05-17T02:52:21 | {} | NONE |
ClickHouse | ClickHouse | 629,765,178 | 10,946 | alexey-milovidov | You did not include the author's commits - we usually do it to respect contributors. | 2020-05-17T09:00:29 | 2020-05-17T09:00:29 | {} | MEMBER |
ClickHouse | ClickHouse | 629,776,728 | 10,858 | azat | >But if there is only one block in INSERT, no squashing is performed.
But while pushing to materialized views (`PushingToViewsBlockOutputStream`/`IBlockOutputStream`) number of blocks is unknown | 2020-05-17T10:40:01 | 2020-05-17T10:40:01 | {} | MEMBER |
ClickHouse | ClickHouse | 629,810,055 | 10,963 | filimonov |
> What I expect at first launch of clickhouse-server is like
> $gosu /usr/bin/clickhouse-server --config-file=$CLICKHOUSE_CONFIG -- listen_host=127.0.0.1 &
where -- listen_host=127.0.0.1 gives the behaviour I search for
That sounds more universal. Can you test it and send a PR? | 2020-05-17T14:47:26 | 2020-05-17T14:47:26 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 629,849,121 | 10,970 | alexey-milovidov | Bad modification, revert. | 2020-05-17T19:36:25 | 2020-05-17T19:36:25 | {} | MEMBER |
ClickHouse | ClickHouse | 629,810,513 | 10,740 | filimonov | Fixed in #10910 | 2020-05-17T14:50:41 | 2020-05-17T14:50:41 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 629,848,727 | 10,858 | azat | >Integration tests (asan) — failed:15, passed:559, error:4
Failed in upstream/master too (7cf35388407ee57f4e75c7040202274fea793ca2)
```
==8==ERROR: AddressSanitizer: heap-use-after-free on address 0x6060000e3148 at pc 0x0000193c70de bp 0x7f29b4d52030 sp 0x7f29b4d52028
READ of size 8 at 0x6060000e3148 thread T48... | 2020-05-17T19:33:32 | 2020-05-17T19:38:05 | {} | MEMBER |
ClickHouse | ClickHouse | 629,968,240 | 4,277 | alexey-milovidov | @qoega Why closed? We have to fix the issue and integrate this test with author's commit.
I will reopen and merge when the issue will be fixed. | 2020-05-18T06:13:17 | 2020-05-18T06:13:17 | {} | MEMBER |
ClickHouse | ClickHouse | 630,053,637 | 10,268 | KochetovNicolai | LGTM, but why not to let me review it before merge? | 2020-05-18T09:13:42 | 2020-05-18T09:13:42 | {} | MEMBER |
ClickHouse | ClickHouse | 630,300,514 | 8,995 | azat | Correct me if I'm wrong, but I guess this is not a leak (i.e. RSS/VIRT is fine) but the memory tracking issue (i.e. `MemoryTracking` from the `asynchronous_metrics`, that also uses for memory limit checks).
If my assumption is correct then I've already have a fix for it (came to it while I was on #10818).
P.S. actu... | 2020-05-18T16:36:47 | 2020-05-18T16:36:47 | {} | MEMBER |
ClickHouse | ClickHouse | 629,748,538 | 10,829 | alexey-milovidov | 01099_parallel_distributed_insert_select - reported. | 2020-05-17T06:04:23 | 2020-05-17T06:04:23 | {} | MEMBER |
ClickHouse | ClickHouse | 630,003,749 | 10,963 | KorneiDontsov | I can, but I'm busy until weekend. Does it suit you? | 2020-05-18T07:41:09 | 2020-05-18T07:41:09 | {} | NONE |
ClickHouse | ClickHouse | 630,244,975 | 11,018 | akuzm | If it's ever merged, please squash, so that we can detect the backport from the commit message. | 2020-05-18T15:06:22 | 2020-05-18T15:06:22 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 630,298,342 | 10,992 | alexey-milovidov | > Performance — 13 faster
- it's a clear win. | 2020-05-18T16:32:34 | 2020-05-18T16:32:34 | {} | MEMBER |
ClickHouse | ClickHouse | 630,308,859 | 10,591 | alexey-milovidov | Ok.
BTW, why delete ttl is Array? If we can only have one delete TTL for table, let's make it simple.
| 2020-05-18T16:51:57 | 2020-05-18T16:51:57 | {} | MEMBER |
ClickHouse | ClickHouse | 630,337,906 | 10,950 | alexey-milovidov | > Calling reset(). All MemoryChunks are unmapped, shared_ptr's becoming invalid.
`reset` should leave currently in-use elements in cache.
Note: this may lead to trouble, because cache won't be 100% invalidated when it was requested by `reset` method. But that's another story, let's postpone it later. | 2020-05-18T17:46:30 | 2020-05-18T17:46:30 | {} | MEMBER |
ClickHouse | ClickHouse | 629,849,181 | 10,970 | alexey-milovidov | It breaks integration tests of parts move. | 2020-05-17T19:37:00 | 2020-05-17T19:37:00 | {} | MEMBER |
ClickHouse | ClickHouse | 629,872,892 | 9,545 | alexey-milovidov | It's ok to leave ALL_CAPS event names for simplicity or to invent our own convenient names for them. | 2020-05-17T22:49:00 | 2020-05-17T22:49:00 | {} | MEMBER |
ClickHouse | ClickHouse | 629,899,791 | 10,972 | alexey-milovidov | It works well and really stresses my terminal (it lags for about a second before displaying it):
| 2020-05-18T01:36:08 | 2020-05-18T01:36:08 | {} | MEMBER |
ClickHouse | ClickHouse | 630,222,395 | 10,982 | akuzm | Oh, it's just a giant mess, there are several python bugs and our bugs conspiring to make proper termination impossible. I use Ctrl+Z + kill %, works kinda better. Tried to fix it once, but broke something else: 573983d4079b280149699c5fe2b522f4ae4c9fb4 Also this fix was for python 3, and now this script uses python 2 f... | 2020-05-18T14:32:16 | 2020-05-18T14:33:44 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 630,309,158 | 10,591 | alexey-milovidov | I see that this is just a mistake.
Fixed in 8bde836. | 2020-05-18T16:52:31 | 2020-05-18T16:53:36 | {
"+1": 1
} | MEMBER |
ClickHouse | ClickHouse | 630,343,893 | 10,950 | myrrc | > > Calling reset(). All MemoryChunks are unmapped, shared_ptr's becoming invalid.
>
> `reset` should leave currently in-use elements in cache.
>
> Note: this may lead to trouble, because cache won't be 100% invalidated when it was requested by `reset` method. But that's another story, let's postpone it later.
... | 2020-05-18T17:58:22 | 2020-05-18T17:58:22 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 630,383,311 | 8,995 | azat | >Correct me if I'm wrong, but I guess this is not a leak
Ignore this message, didn't look carefully, sorry | 2020-05-18T19:13:08 | 2020-05-18T19:13:08 | {} | MEMBER |
ClickHouse | ClickHouse | 630,401,687 | 10,818 | azat | >@azat It will correct tracked memory amount but I doubt that RSS will be lower.
You are right (I'm writing long reply in the PR) | 2020-05-18T19:53:22 | 2020-05-18T19:53:22 | {} | MEMBER |
ClickHouse | ClickHouse | 629,757,596 | 10,512 | alexey-milovidov | For some reason, performance test does not show the comparison. | 2020-05-17T07:45:59 | 2020-05-17T07:45:59 | {} | MEMBER |
ClickHouse | ClickHouse | 629,890,309 | 9,516 | alexey-milovidov | Bad test has been found: #10989 | 2020-05-18T00:50:02 | 2020-05-18T00:50:02 | {} | MEMBER |
ClickHouse | ClickHouse | 629,971,317 | 4,277 | alexey-milovidov | I have merged it manually: https://github.com/ClickHouse/ClickHouse/commit/efab707fbc3fb72dfe2caae54323ea6288054b08 | 2020-05-18T06:22:41 | 2020-05-18T06:22:41 | {} | MEMBER |
ClickHouse | ClickHouse | 630,021,358 | 10,930 | alexey-milovidov | Discussed one issue in TG chat. | 2020-05-18T08:15:13 | 2020-05-18T08:15:13 | {} | MEMBER |
ClickHouse | ClickHouse | 630,250,505 | 11,015 | akuzm | If it's merged, please squash, so that the backport can be detected from the commit message. | 2020-05-18T15:14:56 | 2020-05-18T15:14:56 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 630,402,908 | 11,022 | azat | >memory leak in the test for uniqState (or is there any memory leak?);
Hm, I didn't look carefully at RSS/VIRT, indeed, this is a different issue (will drop fixes tag for the #8995)
>memory usage that is higher than usual and sustained in about 6x the memory usage of query but with no leak.
This is about #1081... | 2020-05-18T19:55:53 | 2020-05-18T19:55:53 | {} | MEMBER |
ClickHouse | ClickHouse | 629,740,634 | 10,971 | zhbdesign | With 1.5 billion data, the execution takes 633 seconds to complete | 2020-05-17T04:16:29 | 2020-05-17T04:16:29 | {} | NONE |
ClickHouse | ClickHouse | 629,748,162 | 10,937 | alexey-milovidov | Some of the CI checks get stuck. | 2020-05-17T06:00:01 | 2020-05-17T06:00:01 | {} | MEMBER |
ClickHouse | ClickHouse | 629,846,376 | 10,856 | amosbird | I'm confused that `long` a string type | 2020-05-17T19:15:34 | 2020-05-17T19:15:34 | {} | COLLABORATOR |
ClickHouse | ClickHouse | 629,846,575 | 10,981 | champtar | I saw this PR is touching QuantileTDigest.h and RadixSort.h so I remembered my old ticket, but my intuition was wrong, thanks for the clarification. | 2020-05-17T19:16:58 | 2020-05-17T19:16:58 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 630,005,159 | 10,842 | Zer0Divis0r | @excitoon , any news? I am eager to implement data life cycles, but there is this blank point about replication. Thanks. | 2020-05-18T07:44:03 | 2020-05-18T07:44:03 | {} | NONE |
ClickHouse | ClickHouse | 630,195,509 | 9,545 | akuzm | Having a third set of names that is neither API constants nor perf stat names is the worst option.
Don't worry about this now, I'll rename them myself before merging.
| 2020-05-18T13:49:21 | 2020-05-18T13:49:21 | {
"+1": 1
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 630,236,992 | 11,005 | den-crane | try
```
sudo yum install -y clickhouse-server-20.3.8.53 clickhouse-client-20.3.8.53 clickhouse-common-static-20.3.8.53
```
>Is 20.4 stable to install?
yes. https://repo.yandex.ru/clickhouse/rpm/stable/x86_64/ | 2020-05-18T14:54:26 | 2020-05-18T14:54:26 | {
"+1": 1
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 630,245,038 | 10,158 | wutchzone | I have noticed this error occurs as well on every materialzed view
```
2020.05.18 15:01:34.701077 [ 625 ] {} <Error> default.`.inner.bytessent_50p_graph_day`: Cannot quickly remove directory /var/lib/clickhouse/data/default/%2Einner%2Ebytessent_50p_graph_day/delete_tmp_20200408_253156_810463_10470 by removing files; ... | 2020-05-18T15:06:28 | 2020-05-18T15:06:28 | {} | NONE |
ClickHouse | ClickHouse | 630,361,294 | 10,983 | azat | >people will be disappointed if we will make performance (with is not great) worse
I have the same thoughts, but I've verified that tests passes on the internal CI
But if this is indeed performance related, then people will not see this, since there are debug logs enabled in this tests, which people don't do on pro... | 2020-05-18T18:31:23 | 2020-05-18T18:31:23 | {} | MEMBER |
ClickHouse | ClickHouse | 629,724,361 | 10,818 | azat | >Threads in ThreadPool should take jobs in preferred order, not randomly.
Using thread based on the last executed callbacks (queue of few elements, to overcome periodic jobs scheduling) should work, do you have better ideas?
Also for queries like in your example (`SELECT ... GROUP BY ... LIMIT <L>`, i.e. no `ORDE... | 2020-05-17T00:28:39 | 2020-05-17T00:31:32 | {} | MEMBER |
ClickHouse | ClickHouse | 629,731,192 | 10,818 | alexey-milovidov | > Also for queries like in your example (SELECT ... GROUP BY ... LIMIT <L>, i.e. no ORDER BY and GROUP BY specials) group_by_overflow_mode=any + max_rows_to_group_by=<L> can be forced (some kind of optimization), thoughts?
Only if it's executed in a single thread on a single server. | 2020-05-17T01:59:26 | 2020-05-17T01:59:26 | {} | MEMBER |
ClickHouse | ClickHouse | 629,760,553 | 10,920 | alexey-milovidov | Ok.
But the overall design still looks flawed. | 2020-05-17T08:16:39 | 2020-05-17T08:16:39 | {} | MEMBER |
ClickHouse | ClickHouse | 629,850,804 | 10,961 | alexey-milovidov | Does not reproduce.
Need stack traces of all threads at that moment.
(attach with gdb and type `thread apply all backtrace`) | 2020-05-17T19:49:14 | 2020-05-17T19:49:27 | {} | MEMBER |
ClickHouse | ClickHouse | 629,998,796 | 9,814 | alexey-milovidov | logical error in ReplicatedMergeTreeQueue, 01035_concurrent_move_partition_from_table_zookeeper: https://clickhouse-test-reports.s3.yandex.net/9813/da7fa82ee0746fe70906476cd5f474f6a600ebbc/functional_stateless_tests_(debug).html#fail1
Reported in CI fest. | 2020-05-18T07:30:42 | 2020-05-18T07:30:42 | {} | MEMBER |
ClickHouse | ClickHouse | 630,288,106 | 10,950 | myrrc | Help needed due to [server failing](https://clickhouse-test-reports.s3.yandex.net/10950/4dedcdc623db906efd7a833de1bd1c42be5dad5a/stress_test_logs/test_output/stderr.log) in stress test.
1) Allocating new RegionMetadata.create() **<- returning shared_ptr on request**
```
#1 0x196f7b4a in DB::IGrabberAllocator::Re... | 2020-05-18T16:14:32 | 2020-05-18T16:20:39 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 630,312,470 | 10,982 | myrrc | > Oh, it's just a giant mess, there are several python bugs and our bugs conspiring to make proper termination impossible. I use Ctrl+Z + kill %, works kinda better. Tried to fix it once, but broke something else: [573983d](https://github.com/ClickHouse/ClickHouse/commit/573983d4079b280149699c5fe2b522f4ae4c9fb4) Also t... | 2020-05-18T16:58:36 | 2020-05-18T17:03:14 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 629,763,963 | 10,858 | alexey-milovidov | Is the issue relevant only for queries with multiple INSERTed blocks?
| 2020-05-17T08:49:06 | 2020-05-17T08:49:06 | {} | MEMBER |
ClickHouse | ClickHouse | 629,764,906 | 10,751 | alexey-milovidov | @alexandr-orlov ok? | 2020-05-17T08:57:53 | 2020-05-17T08:57:53 | {} | MEMBER |
ClickHouse | ClickHouse | 629,842,111 | 10,935 | saravsars | Yes, when TCP health check is configured in the AWS Target Group this exception is getting printed. When the target group is deleted the exceptions stopped.
Thanks | 2020-05-17T18:40:23 | 2020-05-17T18:40:23 | {} | NONE |
ClickHouse | ClickHouse | 630,195,793 | 10,777 | alesapin | `01196_max_parser_depth` - flap | 2020-05-18T13:49:48 | 2020-05-18T13:49:48 | {} | MEMBER |
ClickHouse | ClickHouse | 630,231,111 | 5,681 | den-crane | Distibuted Auto-increment using ZK.
`create distributed sequence x_id cache 1billion`
ZK x_id = 0
Host books a billion values x_id +=1b and uses this reservation until it exhaustion. Books next. Holes up to cache size is possible but it does not matter. | 2020-05-18T14:45:55 | 2020-05-18T14:46:29 | {
"hooray": 1
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 630,249,571 | 10,998 | den-crane | https://github.com/ClickHouse/ClickHouse/issues/10417
https://github.com/ClickHouse/ClickHouse/issues/9988 | 2020-05-18T15:13:25 | 2020-05-18T15:13:25 | {
"+1": 1
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 624,912,487 | 10,684 | tavplubix | It will be in 20.3.10 | 2020-05-06T21:57:16 | 2020-05-18T15:46:58 | {} | MEMBER |
ClickHouse | ClickHouse | 629,731,584 | 10,953 | alexey-milovidov | Successful. | 2020-05-17T02:05:02 | 2020-05-17T02:05:02 | {} | MEMBER |
ClickHouse | ClickHouse | 629,758,501 | 10,866 | alexey-milovidov | It's not ready for merge because
> Description check — Category section not found
> ClickHouse build check — 16/17 builds are OK
(built for Mac OS is failed)
But don't worry, I will do it for you. | 2020-05-17T07:55:28 | 2020-05-17T07:55:28 | {} | MEMBER |
ClickHouse | ClickHouse | 629,821,911 | 10,981 | g-arslan | Hi @champtar
Actually it shouldn't affect QuantileTDigest, the only case this patch may improve something is when you need permutation of sorted data | 2020-05-17T16:12:19 | 2020-05-17T16:12:19 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 629,895,500 | 10,984 | alexey-milovidov | Performance — 14 faster.
| 2020-05-18T01:18:09 | 2020-05-18T01:18:09 | {} | MEMBER |
ClickHouse | ClickHouse | 629,747,812 | 10,962 | jason-godden | This is great! Is there a plan to add simple SQL UDFs though? Essentially like macros? Would be very useful when there are complex formulas that you want to reuse.
| 2020-05-17T05:55:37 | 2020-05-17T05:55:37 | {} | NONE |
ClickHouse | ClickHouse | 629,774,551 | 10,858 | alexey-milovidov | But if there is only one block in INSERT, no squashing is performed. | 2020-05-17T10:20:22 | 2020-05-17T10:20:22 | {} | MEMBER |
ClickHouse | ClickHouse | 629,891,986 | 9,516 | alexey-milovidov | For unknown reason, it took 12 minutes starting odbc-bridge and it does not response in next second. | 2020-05-18T01:00:49 | 2020-05-18T01:00:49 | {} | MEMBER |
ClickHouse | ClickHouse | 629,909,064 | 10,961 | alexey-milovidov | `/home/kot/clickhouse/build/programs/clickhouse`
\- this is alternative ClickHouse build.
You can download the official build in "ClickHouse Build Check" and see what will happen:
https://clickhouse-builds.s3.yandex.net/0/6f1d522b6aa2dd119e452f20c5cd4ca12da6d072/report.html | 2020-05-18T02:15:01 | 2020-05-18T02:15:01 | {} | MEMBER |
ClickHouse | ClickHouse | 630,126,181 | 10,831 | nikitamikhaylov | It is easier for me to fix tests failures by myself. This PR will be merged automatically. | 2020-05-18T11:42:54 | 2020-05-18T11:42:54 | {} | MEMBER |
ClickHouse | ClickHouse | 629,724,721 | 10,964 | alexey-milovidov | Tests did not run under 4 hours. | 2020-05-17T00:33:02 | 2020-05-17T00:33:02 | {} | MEMBER |
ClickHouse | ClickHouse | 629,725,494 | 10,927 | azat | Failures looks unrelated | 2020-05-17T00:41:06 | 2020-05-17T00:41:06 | {} | MEMBER |
ClickHouse | ClickHouse | 629,740,421 | 10,910 | alexey-milovidov | https://github.com/ClickHouse/ClickHouse/pull/10970 | 2020-05-17T04:13:49 | 2020-05-17T04:13:49 | {} | MEMBER |
ClickHouse | ClickHouse | 629,856,320 | 10,983 | azat | @alexey-milovidov FYI I'm observing failure for the test_storage_kafka::test_kafka_virtual_columns2, and it passes if I run only it, but this PR should not affect it, maybe only due to some timing issues, not sure, will take a look | 2020-05-17T20:31:09 | 2020-05-17T20:31:09 | {} | MEMBER |
ClickHouse | ClickHouse | 629,881,507 | 10,961 | myrrc | [trace.txt](https://github.com/ClickHouse/ClickHouse/files/4641341/trace.txt)
| 2020-05-17T23:55:48 | 2020-05-17T23:55:48 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 629,901,725 | 10,972 | alexey-milovidov | The more sophisticated version of this function would work on the list of the ranges of assigned code points. And also it may work directly on the list of UTF8 sequences. | 2020-05-18T01:44:33 | 2020-05-18T01:44:33 | {} | MEMBER |
ClickHouse | ClickHouse | 629,972,025 | 4,277 | alexey-milovidov | And we appreciate adding more tests to `bugs` directory! | 2020-05-18T06:24:39 | 2020-05-18T06:24:39 | {} | MEMBER |
ClickHouse | ClickHouse | 630,327,293 | 11,002 | alesapin | Wow! It works, just need to make these options opt-in or fix failed. | 2020-05-18T17:26:32 | 2020-05-18T17:26:51 | {} | MEMBER |
ClickHouse | ClickHouse | 630,397,462 | 10,818 | alexey-milovidov | @azat It will correct tracked memory amount but I doubt that RSS will be lower. | 2020-05-18T19:43:59 | 2020-05-18T19:43:59 | {} | MEMBER |
ClickHouse | ClickHouse | 629,753,820 | 10,940 | azat | >So, the only difference is that file is closed earlier?
Not exactly, there is write after close that was ignored, see #10953 | 2020-05-17T07:07:58 | 2020-05-17T07:07:58 | {} | MEMBER |
ClickHouse | ClickHouse | 629,765,201 | 10,600 | alexey-milovidov | Merged in #10946 | 2020-05-17T09:00:42 | 2020-05-17T09:00:42 | {} | MEMBER |
ClickHouse | ClickHouse | 629,771,487 | 10,925 | Enmk | Confirmed, fixing | 2020-05-17T09:51:33 | 2020-05-17T09:51:33 | {
"hooray": 1
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 629,787,330 | 10,962 | jason-godden | Another question here - would it be easy enough to extend this to allow for anonymous functions? So something like:
```python
SELECT
id,
pythonUDF('UInt8', 'parameter1', $$
import somemodule
return somemodule.somethingwithparams(parameter1)$$)(someclickhousecolumn)
FROM table
```
Wh... | 2020-05-17T12:13:05 | 2020-05-17T12:13:05 | {} | NONE |
ClickHouse | ClickHouse | 630,209,930 | 11,000 | den-crane | duplicate https://github.com/ClickHouse/ClickHouse/issues/10894 | 2020-05-18T14:13:53 | 2020-05-18T14:13:53 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 630,326,742 | 9,624 | CurtizJ | Failures under memory-sanitizer - ok. They were finally fixed and MSan tests were enabled after 20.3.
Failures with DatabaseAtomic - ok. It isn't implemented in 20.3. | 2020-05-18T17:25:26 | 2020-05-18T17:26:43 | {} | MEMBER |
ClickHouse | ClickHouse | 629,738,981 | 10,910 | alexey-milovidov | Stress test shows use-after-free, need to investigate. | 2020-05-17T03:55:12 | 2020-05-17T03:56:29 | {} | MEMBER |
ClickHouse | ClickHouse | 629,760,146 | 10,867 | alexey-milovidov | Ok.
It's rather big change to be backported (backport to stable versions will be risky) and it doesn't look like a bugfix, more like an implementation of missing feature. | 2020-05-17T08:12:31 | 2020-05-17T08:12:31 | {} | MEMBER |
ClickHouse | ClickHouse | 629,871,464 | 10,856 | alexey-milovidov | Yes, it's string in MySQL and integer in MS Access. I have removed this synonim. | 2020-05-17T22:37:11 | 2020-05-17T22:37:11 | {} | MEMBER |
ClickHouse | ClickHouse | 630,387,564 | 10,983 | filimonov | Thread names during the execution of `test_kafka_rebalance`
```
ps H -o 'tid comm' $(pidof clickhouse)
TID COMMAND
8692 clickhouse
8808 clickhouse
8809 clickhouse
8815 clickhouse
8818 ExterLdrReload
8819 SystemLogFlush
8820 SystemLogFlush
8821 clickhouse
8822 SystemLogFlush
8823 SystemLogFlu... | 2020-05-18T19:21:58 | 2020-05-18T19:21:58 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 629,730,139 | 7,512 | tavplubix | It's `SIGABRT` caused by `std::terminate()` call when `ALTER` tries to rewrite metadata after database was dropped.
#10968 | 2020-05-17T01:44:20 | 2020-05-17T01:44:20 | {} | MEMBER |
ClickHouse | ClickHouse | 629,748,375 | 10,969 | alexey-milovidov | Build check Ok. | 2020-05-17T06:02:15 | 2020-05-17T06:02:15 | {} | MEMBER |
ClickHouse | ClickHouse | 629,766,956 | 10,858 | azat | >Is the issue relevant only for queries with multiple INSERTed blocks?
Not only, since each block will be pushed to MV, each MV will have separate squashing, so until the minimal number of rows will be accumulated it will be kept in memory, so if you have let's say 10 MVs then you need 10x more memory (depends on sq... | 2020-05-17T09:13:32 | 2020-05-17T09:13:32 | {} | MEMBER |
ClickHouse | ClickHouse | 629,969,381 | 10,993 | blinkov | @haarchri I think you'd better post this to the repository where you got that helm chart. | 2020-05-18T06:16:48 | 2020-05-18T06:16:48 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 630,081,689 | 10,751 | alexandr-orlov | > @alexandr-orlov ok?
Ok! | 2020-05-18T10:04:32 | 2020-05-18T10:04:32 | {} | NONE |
ClickHouse | ClickHouse | 630,244,002 | 11,008 | den-crane | duplicate https://github.com/ClickHouse/ClickHouse/issues/10471 | 2020-05-18T15:04:51 | 2020-05-18T15:04:51 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 630,299,582 | 10,992 | alexey-milovidov | Next steps (in subsequent PRs):
1. Check that min/max are already vectorized for all data types.
2. Implement the same method for avg.
3. Implement specialization for Nullable for min/max/avg. | 2020-05-18T16:34:57 | 2020-05-18T16:35:10 | {} | MEMBER |
ClickHouse | ClickHouse | 630,359,049 | 10,983 | filimonov | 1) Logging in clickhouse logs looks great, interesing if it's able to log (for example) during destroy.
2) Thread names are a bit 'out of convention' (other threads use CamelCasing with shortcuts) but ok.
3) I also have 2 failures locally: test_kafka_rebalance & test_kafka_virtual_columns2 I didn't check but may... | 2020-05-18T18:27:16 | 2020-05-18T18:29:49 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 630,373,482 | 10,818 | azat | >Is there a current workaround?
@genzgd is your original query has the same pattern? i.e.
- has GROUP BY
- has LIMIT
- no ORDER BY
If so, then as a workaround you can use one of the following:
- ORDER BY (any column, the most lightwegiht)
- WITH TOTALS (will also work) | 2020-05-18T18:53:13 | 2020-05-18T18:53:13 | {} | MEMBER |
ClickHouse | ClickHouse | 629,722,806 | 10,963 | KorneiDontsov | > If I understand CLI correctly I guess I could get this behaviour if I specify <listen_host>127.0.0.1</listen_host> in config.xml and execute docker run with -- listen_host=0.0.0.0.
I checked this workaround. It works. | 2020-05-17T00:11:29 | 2020-05-17T00:11:29 | {} | NONE |
ClickHouse | ClickHouse | 630,054,974 | 10,019 | filimonov | > rules configured for user A affect user B.
That is definitely a bug, which exists in 20.3. | 2020-05-18T09:16:07 | 2020-05-18T09:16:07 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 630,158,389 | 1,287 | unflag | Same issue. We have to truncate some tables periodically and it would be much more convenient if our apps wouldn't know anything about cluster shards and their tables - only cluster name. | 2020-05-18T12:45:18 | 2020-05-18T12:45:18 | {
"+1": 1
} | NONE |
ClickHouse | ClickHouse | 629,725,120 | 7,512 | azat | Looks like this can lead to SIGSEGV on parallel `ALTER`? See https://clickhouse-test-reports.s3.yandex.net/10927/c059ee008519cff43b42022fc6f6a9d8ee86c56c/functional_stateless_tests_(release,_databaseatomic).html#fail1 | 2020-05-17T00:36:52 | 2020-05-17T00:36:52 | {} | MEMBER |
ClickHouse | ClickHouse | 629,766,245 | 10,520 | alexey-milovidov | @qoega ping. | 2020-05-17T09:08:16 | 2020-05-17T09:08:16 | {} | MEMBER |
ClickHouse | ClickHouse | 629,767,153 | 10,858 | azat | >I'm Ok with these settings and the code LGTM.
Great! Will rebase against master and address the comments | 2020-05-17T09:15:07 | 2020-05-17T09:15:07 | {} | MEMBER |
ClickHouse | ClickHouse | 629,812,353 | 10,981 | champtar | Hi @g-arslan
I'm curious what numbers you get before/after your patch for this test case: https://github.com/ClickHouse/ClickHouse/issues/2668#issuecomment-598313960 | 2020-05-17T15:03:34 | 2020-05-17T15:03:46 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 629,846,082 | 10,981 | alexey-milovidov | @champtar Yes, it's completely independent - QuantileTDigest has some simple flaw in implementation (independent of particular choice of sort algorithm), we should dig into it. | 2020-05-17T19:13:13 | 2020-05-17T19:13:13 | {} | MEMBER |
ClickHouse | ClickHouse | 615,888,767 | 10,301 | alexey-milovidov | It is possible to distinguish if the first column cannot contain binary data. | 2020-04-18T15:21:11 | 2020-04-18T15:21:11 | {} | MEMBER |
ClickHouse | ClickHouse | 616,232,968 | 10,354 | oandrew | @alexey-milovidov
review pls when you have a chance.
| 2020-04-19T22:08:10 | 2020-04-19T22:08:10 | {} | CONTRIBUTOR |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.