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 | 772,711,452 | 20,014 | den-crane | 1. yes. these all cases ( Network issue / ZK xid overflow - ZK restart / No meatadata in Zookeeper)
2. correct ` from system.replicas where is_readonly`
3. attach table with non-existent ZK path
```sql
-- this a new non-existent table with random zk path and name, you don't need to create folders
-- or do any... | 2021-02-03T18:09:35 | 2021-02-03T18:11:59 | {
"heart": 4
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 773,150,441 | 19,911 | alexey-milovidov | 20.10 is not supported. | 2021-02-04T09:07:38 | 2021-02-04T09:07:38 | {} | MEMBER |
ClickHouse | ClickHouse | 772,453,363 | 10,280 | wanghao111 | **same issue!!!**
version:20.10.3.30
select bitmapToArray(bitmapBuild(arrayConcat(range(32),[10003035727])))
result:
[0,1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,1413101135]
10003035727 is transformed to 1413101135.
if change range(32) to range(31),everything is ok~
... | 2021-02-03T11:54:31 | 2021-02-03T11:54:31 | {} | NONE |
ClickHouse | ClickHouse | 587,500,824 | 9,195 | den-crane | It could be atomic if this insert creates only one part.
* only if it inserts into only one partition
* only if number of rows is less than max_insert_block_size
* only if a table does not have MV (there is no atomicity Table <-> MV) | 2020-02-18T14:57:23 | 2021-02-03T12:33:37 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 772,699,911 | 20,029 | alexey-milovidov | > Stress test (thread) — Sanitizer assert (in stderr.log)
#20045 | 2021-02-03T17:52:04 | 2021-02-03T17:52:04 | {} | MEMBER |
ClickHouse | ClickHouse | 772,709,529 | 20,029 | alexey-milovidov | There are no Yandex Synchronization checks for unknown reason. | 2021-02-03T18:06:36 | 2021-02-03T18:06:36 | {} | MEMBER |
ClickHouse | ClickHouse | 772,770,123 | 19,950 | den-crane | it's a copy/paste issue
run client in multiline/multiquery more `clickhouse-client -mn` | 2021-02-03T19:38:05 | 2021-02-03T19:38:05 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 773,040,605 | 20,045 | alexey-milovidov | > AST fuzzer (MSan) — SUMMARY: MemorySanitizer: use-of-uninitialized-value obj-x86_64-linux-gnu/../src/Columns/ColumnUnique.h:524:18 in COW<DB::IColumn>::mutable_...
#19877 | 2021-02-04T05:28:53 | 2021-02-04T05:28:53 | {} | MEMBER |
ClickHouse | ClickHouse | 773,065,394 | 20,055 | zhangjmruc | From the backtrace, we can see there is a table defined in ch with engine MongoDB(StorageMongoDB) with host_port as 47.106.193.101:7017.
To resolve this problem:
1. Check the status of MongoDB to make sure it can be connected.
2. If you have no idea of the problem table, you can choose to delete the metadata for it.... | 2021-02-04T06:27:24 | 2021-02-04T06:27:24 | {
"+1": 1
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 773,069,116 | 20,053 | alexey-milovidov | There are no known bugs with this setting in the release 21.1.
`insert_distributed_sync = 1` has the following advantages:
- data is available to SELECT from replica immediately;
- if all replicas are unavailable, you get the error instead of buffering data on local filesystem;
- better resource usage due to less... | 2021-02-04T06:35:16 | 2021-02-04T06:37:15 | {
"+1": 7
} | MEMBER |
ClickHouse | ClickHouse | 773,073,679 | 20,001 | alexey-milovidov | Most likely no - windowFunnel algorithm is too complex to be auto-vectorized by compiler.
There is also no much chance that it can be vectorized manually.
Looks like it can be optimized by careful tweaks in the algorithm and by looking at the most frequent special cases.
If you want to optimize it, you need to pre... | 2021-02-04T06:44:48 | 2021-02-04T06:44:48 | {} | MEMBER |
ClickHouse | ClickHouse | 773,104,424 | 19,950 | alexey-milovidov | @den-crane I will correct documentation just in case. | 2021-02-04T07:46:20 | 2021-02-04T07:46:20 | {} | MEMBER |
ClickHouse | ClickHouse | 772,646,814 | 20,042 | UnamedRus | Probably it would make sense to have something like RawBinaryWithCharSet(size, '0-9A-Za-z') too.
It would help when we have bigger possible charset than just hex. (base64) | 2021-02-03T16:38:03 | 2021-02-03T16:38:03 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 772,708,952 | 20,029 | alexey-milovidov | > test_zookeeper_config/test.py::test_secure_connection
Flaky test. Maybe it's related to TestKeeperServer? @alesapin | 2021-02-03T18:05:46 | 2021-02-03T18:05:46 | {} | MEMBER |
ClickHouse | ClickHouse | 773,083,913 | 20,046 | hexiaoting | I found the sql of create and select that make segfault:
```
CREATE TABLE nt (`x` Nullable(String)) ENGINE = Memory;
SELECT '1048575', NULL FROM nt AS t1 RIGHT JOIN nt AS t2 USING (x) ORDER BY 1048576 + (10 AS x) ASC, '0.0000000010' ASC NULLS LAST, t1.x DESC NULLS FIRST
```
But I still cannot reproduce this se... | 2021-02-04T07:05:24 | 2021-02-04T07:05:24 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 773,093,012 | 20,048 | alexey-milovidov | I think these are just too low-level to be logged. Maybe better to simply remove `ZooKeeperUserExceptions` from your monitoring? It's like some technical low-level event and there are no much insights to get from it.
Also we are using `Trace` level by default and on our production servers. We don't want to make it t... | 2021-02-04T07:22:42 | 2021-02-04T07:23:04 | {
"+1": 2
} | MEMBER |
ClickHouse | ClickHouse | 772,634,399 | 19,950 | akolybelnikov | Yes, I am referring to this tutorial. I run the client in the Docker container [yandex/clickhouse-server](https://hub.docker.com/r/yandex/clickhouse-server/).
There are 2 initial queries, the first one `CREATE TABLE tutorial.hits_v1` doesn't work with the above described error. The second one does work, so I don't thi... | 2021-02-03T16:22:36 | 2021-02-03T16:22:36 | {} | NONE |
ClickHouse | ClickHouse | 772,897,816 | 20,048 | ben-efiz | @alexey-milovidov thx for the follow up. While your example shows a `LOG_TRACE` quite a few of the others ZooKeeper exceptions don't have **any** log. So in my case i have ~20k `ZooKeeperUserExceptions` being reported by Prometheus but no idea what's the reason behind. Even not with logger level set to `Trace`.
Look... | 2021-02-03T23:27:21 | 2021-02-03T23:32:55 | {} | NONE |
ClickHouse | ClickHouse | 772,984,521 | 20,054 | den-crane | JDBC driver does not support SQL queries with Format.
In case of CSV JDBC takes the first line of the result as NamesAndTypes row.
https://github.com/ClickHouse/clickhouse-jdbc/issues/555 | 2021-02-04T02:44:49 | 2021-02-04T03:59:45 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 773,117,781 | 19,896 | depressed-pho | Fixed several build errors found by the build check. | 2021-02-04T08:11:59 | 2021-02-04T08:11:59 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 773,076,721 | 20,052 | hexiaoting | ```
dell123 :) select toInt32(441746) % 150 as a , toTypeName(a), 441746 % 150 as b, toTypeName(b)
Query id: e0111eab-044c-47da-9684-084bf8d07a01
┌────a─┬─toTypeName(modulo(toInt32(441746), 150))─┬───b─┬─toTypeName(modulo(441746, 150))─┐
│ -110 │ Int8 │ 146 │ UInt8 ... | 2021-02-04T06:50:56 | 2021-02-04T08:59:55 | {
"+1": 1
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 772,557,483 | 19,982 | den-crane | `sync` before (2 - Take disk snapshot. ) is probably enough.
>BTW, here is the tutorial that I have been following to do this exercise.
No. I don't see it.
When you read a file from the inside OS you read it from OS cache (+disk).
When you did (2) from the outside you lost OS cache. | 2021-02-03T14:38:17 | 2021-02-03T14:38:17 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 773,001,303 | 20,016 | JackyWoo | @tavplubix Error in this issue is in distributed DDL worker thread but not main thread. So maybe the 3th resolutions is not for this issue. | 2021-02-04T03:31:09 | 2021-02-04T03:31:09 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 773,080,023 | 20,017 | alexey-milovidov | Yes, this query triggered nullptr dereference.
The issue does not reproduce easily - I also checked it.
It was found by fuzzer (that runs enormous amounts of randomized queries) and in this case - the server was run under Memory Sanitizer. The error may depend on the settings that were changed (fuzzer also changes ... | 2021-02-04T06:57:47 | 2021-02-04T06:57:47 | {} | MEMBER |
ClickHouse | ClickHouse | 773,103,326 | 20,047 | alexey-milovidov | > AST fuzzer (UBSan) — ../src/Common/assert_cast.h:50:12: runtime error: downcast of address 0x7f63f41107d0 which does not point to an object of type 'const DB::Co
https://github.com/ClickHouse/ClickHouse/issues/19754 | 2021-02-04T07:44:03 | 2021-02-04T07:44:03 | {} | MEMBER |
ClickHouse | ClickHouse | 772,556,241 | 32 | crakjie | Is stored expression a possible feature ?
```
CREATE TABLE Toto
(
`id` String,
`fn` Int32 => Int32
)
ENGINE = Memory()
```
The idea is to in one query both data preparation and selecton. Data preparation depending on the row and some dynamic value. | 2021-02-03T14:36:34 | 2021-02-03T14:36:53 | {} | NONE |
ClickHouse | ClickHouse | 772,571,199 | 20,029 | alexey-milovidov | > AST fuzzer (MSan) — SUMMARY: MemorySanitizer: use-of-uninitialized-value (/workspace/clickhouse+0x8d6f8de) in memcmp
https://github.com/ClickHouse/ClickHouse/pull/19972 | 2021-02-03T14:57:42 | 2021-02-03T14:57:42 | {} | MEMBER |
ClickHouse | ClickHouse | 772,708,498 | 20,029 | alexey-milovidov | > AST fuzzer (TSan) — Received signal 11
https://github.com/ClickHouse/ClickHouse/issues/20046 | 2021-02-03T18:05:08 | 2021-02-03T18:05:08 | {} | MEMBER |
ClickHouse | ClickHouse | 772,728,451 | 19,673 | azat | >AST fuzzer (UBSan) — ../src/Functions/formatReadableTimeDelta.cpp:180:28: runtime error: -nan is outside the range of representable values of type 'unsigned long
Already fixed in #19791 | 2021-02-03T18:36:26 | 2021-02-03T18:36:26 | {} | MEMBER |
ClickHouse | ClickHouse | 773,058,494 | 20,017 | hexiaoting | Is this query leading to nullptr dereference?
```
SELECT intDiv(t2.y, 3), t2.* FROM t1 FULL OUTER JOIN t2 ON and(toUInt32(intDiv(intDiv(t1.y, 3), 1048576)) = t2.x) ORDER BY (t2.y AND (NULL AND (t2.y AND (t1.x = t2.x)) AND (t1.x = t2.x)) AND (t1.x = t2.x)) AND (intDiv(t1.y, NULL) AND t2.y AND (t1.x = t2.x)) AND (t1.x ... | 2021-02-04T06:12:09 | 2021-02-04T06:19:30 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 773,103,636 | 20,001 | fuwhu | > Performance optimizations require quick iterations on realistic datasets.
> Maybe you can create obfuscated dataset?
>
> Run `clickhouse-obfuscator --help` and follow the instruction.
>
> You can have 100 bn records but to reproduce performance maybe some subset of a few million records can be enough?
sure,... | 2021-02-04T07:44:40 | 2021-02-04T07:44:40 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 773,105,058 | 19,950 | alexey-milovidov | But it's already alright. | 2021-02-04T07:47:32 | 2021-02-04T07:47:32 | {} | MEMBER |
ClickHouse | ClickHouse | 772,719,519 | 19,768 | den-crane | New feature has been implemented
```
SELECT timezoneOffset(toDateTime('2018-08-21 22:20:00', 'Australia/Lord_Howe'));
```
afaik will be available in CH 21.3+ | 2021-02-03T18:21:56 | 2021-02-03T18:21:56 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 773,055,624 | 19,896 | depressed-pho | Marked the test as being skipped in fasttest, where SSL is disabled. | 2021-02-04T06:05:46 | 2021-02-04T06:05:46 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 773,094,706 | 20,011 | alexey-milovidov | Ok. | 2021-02-04T07:25:58 | 2021-02-04T07:25:58 | {} | MEMBER |
ClickHouse | ClickHouse | 772,443,374 | 20,032 | alexey-milovidov | @tavplubix It means that the server exited early. | 2021-02-03T11:35:35 | 2021-02-03T11:35:35 | {} | MEMBER |
ClickHouse | ClickHouse | 772,454,119 | 20,030 | tavplubix | It happens because database name is not specified in `Distributed` engine arguments of `default.graphite`, so `default` database is used and `default.graphite` recursively reads from itself. You should either specify database name in engine arguments like this: `ENGINE = Distributed('graphite', 'shard_01', 'graphite', ... | 2021-02-03T11:55:57 | 2021-02-03T11:55:57 | {} | MEMBER |
ClickHouse | ClickHouse | 772,634,989 | 20,040 | den-crane | >But when I try to dump data from mysql to clickhouse it shows me following error :
Dump how?
>CRITICAL:ex='datetime.timedelta' object has no attribute 'encode'
It does not look like CH error. | 2021-02-03T16:23:21 | 2021-02-03T16:23:21 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 772,992,583 | 19,592 | olgarev | @Akazz Hi, can you review these docs, pls? | 2021-02-04T03:06:52 | 2021-02-04T03:06:52 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 773,062,916 | 20,053 | hexiaoting | It will be slower to return. | 2021-02-04T06:22:16 | 2021-02-04T06:22:25 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 773,076,620 | 20,053 | alexey-milovidov | In case of any error you will get exception in your client - that means - you need to retry the INSERT query again.
If INSERT query succeeded - then data was successfully written to replica.
Only if you do something really wrong, e.g. run INSERT query and don't look at its result - you will lose the data. | 2021-02-04T06:50:43 | 2021-02-04T06:50:43 | {} | MEMBER |
ClickHouse | ClickHouse | 773,098,419 | 20,001 | fuwhu | > Most likely no - windowFunnel algorithm is too complex to be auto-vectorized by compiler.
> There is also no much chance that it can be vectorized manually.
>
> Looks like it can be optimized by careful tweaks in the algorithm and by looking at the most frequent special cases.
> If you want to optimize it, you n... | 2021-02-04T07:33:44 | 2021-02-04T07:38:42 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 772,589,099 | 19,995 | leosunli | the error message as follow when I call ninja.
[10/3427] Linking CXX executable contrib/protobuf/protoc-3.13.0.0
FAILED: contrib/protobuf/protoc-3.13.0.0
: && /usr/local/bin/g++ -fdiagnostics-color=always -std=c++2a -fsized-deallocation -pipe -msse4.1 -msse4.2 -mpopcnt -Wall -Werror -w -O2 -g -DNDEBUG -O3 -fno-... | 2021-02-03T15:21:32 | 2021-02-03T15:21:32 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 772,894,312 | 20,011 | azat | @alexey-milovidov @vdimir DateTime64 will be fixed separately, since this is not only about UBsan reports, but there is just no specializations for DataTypeDateTime64 from numeric, and it simply uses raw cast https://github.com/ClickHouse/ClickHouse/blob/9d8033d8d67c42e098f70e41c8b2fd425c052f0e/src/Functions/FunctionsC... | 2021-02-03T23:19:21 | 2021-02-04T05:32:27 | {} | MEMBER |
ClickHouse | ClickHouse | 773,123,270 | 19,529 | fuwhu | i tried this change, it caused segment fault when starting clickhouse server.
not looking into it yet. | 2021-02-04T08:21:57 | 2021-02-04T08:22:22 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 772,424,983 | 20,016 | JackyWoo | Mabye there are two resolutions:
**1. add a distributed lock to distributed DDL processing**
The lock should take 'DDLWorker::runCleanupThread' procedure in consideration. The lock can still employ Zookeeper just like 'ZooKeeperLock', but a read-write-lock is better.
**2. Adjust distributed DDL queue clean pol... | 2021-02-03T11:04:25 | 2021-02-03T12:35:16 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 772,479,552 | 20,032 | tavplubix | But how does tsan prints something to stderr after `_exit`? It should terminate process immediately, but maybe tsan intercepts it. Does it mean, that tsan will always produce false positive "thread leak" report when server is shutdown forcefully? | 2021-02-03T12:42:29 | 2021-02-03T12:42:29 | {} | MEMBER |
ClickHouse | ClickHouse | 772,492,069 | 20,032 | alexey-milovidov | https://github.com/llvm/llvm-project/blob/7afdc89c2054d6fa3a6c2f2f24411becfc764676/compiler-rt/lib/tsan/rtl/tsan_interceptors_posix.cpp#L2814 | 2021-02-03T13:04:22 | 2021-02-03T13:04:22 | {} | MEMBER |
ClickHouse | ClickHouse | 772,498,426 | 20,034 | alexey-milovidov | The latest release is 21.1. Does it help to upgrade? | 2021-02-03T13:15:47 | 2021-02-03T13:15:47 | {} | MEMBER |
ClickHouse | ClickHouse | 772,586,157 | 19,315 | OmarBazaraa | I even tried to concatenate the fields instead of comparing tuples, but I get the exact same error:
```sql
ALTER TABLE clarity.page_data DELETE
WHERE concat(toString(ProjectId), toString(UserId), toString(SessionId)) NOT IN (
SELECT concat(toString(ProjectId), toString(UserId), toString(SessionId))
... | 2021-02-03T15:17:43 | 2021-02-03T15:17:43 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 772,700,677 | 19,646 | den-crane |
Then you can execute fetch partition for all partitions ( copy them from old shard to another into detached folder of table ).
https://clickhouse.tech/docs/en/sql-reference/statements/alter/partition/#alter_fetch-partition
```
ALTER TABLE users FETCH PARTITION 201902 FROM '/clickhouse/tables/01-01/visits';
``... | 2021-02-03T17:53:13 | 2021-02-03T17:53:13 | {
"+1": 1
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 772,776,255 | 19,950 | alexey-milovidov | @den-crane There should not be any copy-paste issues in clickhouse-client. | 2021-02-03T19:48:29 | 2021-02-03T19:48:29 | {} | MEMBER |
ClickHouse | ClickHouse | 772,490,010 | 20,032 | alexey-milovidov | > but maybe tsan intercepts it
Yes, I think.
> Does it mean, that tsan will always produce false positive "thread leak" report when server is shutdown forcefully?
Yes. | 2021-02-03T13:01:04 | 2021-02-03T13:01:04 | {} | MEMBER |
ClickHouse | ClickHouse | 772,582,586 | 19,982 | ibradwan | Makes sense... Anyway I think `sync` will only reduce the probability of this happening but won't eliminate it. My only concern here is the large and old parts, because small & recent parts are acceptable to be lost in a disastrous recovery. But if this approach corrupts large parts that were backed up while they were ... | 2021-02-03T15:13:09 | 2021-02-03T16:05:38 | {} | NONE |
ClickHouse | ClickHouse | 772,717,416 | 19,602 | den-crane | very confusing feature , AFAIK they are going to remove it from Mysql.
I don't like it. | 2021-02-03T18:18:32 | 2021-02-03T18:18:32 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 772,768,628 | 19,673 | azat | Could someone please take a look at this? (This contains some tiny refactoring, so conflicts may pops up over time) | 2021-02-03T19:36:20 | 2021-02-03T19:36:20 | {} | MEMBER |
ClickHouse | ClickHouse | 773,068,156 | 20,001 | fuwhu | > @alesapin maybe you can help?
>
> The build has finished successfully but it cannot be found in s3.
Hi Alexey,
Have you tested the performance difference of windowFunnel after this change ?
Per my observation, the performance bottle neck of the windowFunnel function seems be the memory store and read of dat... | 2021-02-04T06:33:14 | 2021-02-04T06:33:14 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 773,149,640 | 19,990 | alexey-milovidov | No such function in this release. | 2021-02-04T09:06:25 | 2021-02-04T09:06:25 | {} | MEMBER |
ClickHouse | ClickHouse | 772,496,729 | 20,036 | alexey-milovidov | Did you check my PR? https://github.com/ClickHouse/ClickHouse/pull/11305/files | 2021-02-03T13:12:34 | 2021-02-03T13:12:34 | {} | MEMBER |
ClickHouse | ClickHouse | 772,513,867 | 20,038 | nvartolomei | We do that successfully for small and large workloads. There are some tradeoffs, mostly specific to how big the batches are, normal mergetree tables vs. aggregation* merge tree tables and so on. | 2021-02-03T13:40:59 | 2021-02-03T13:40:59 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 772,690,265 | 20,001 | alexey-milovidov | @alesapin maybe you can help?
The build has finished successfully but it cannot be found in s3. | 2021-02-03T17:37:47 | 2021-02-03T17:37:47 | {} | MEMBER |
ClickHouse | ClickHouse | 772,703,708 | 20,034 | filimonov | ```
root@6e51c0edbe0f:/# addr2line -afiCe /usr/bin/clickhouse 0xe755f51 0x15469861 0x1532d8f0 0x15f4c2b2 0x15f4d9c5 0x15f8e262 0xe5ea637 0xe5eadaa 0xe5e9b47 0xe5e8093 0x7f1980820fa3 0x7f19807424cf
0x000000000e755f51
long std::__1::__libcpp_atomic_refcount_decrement<long>(long&)
/build/obj-x86_64-linux-gnu/../contr... | 2021-02-03T17:58:03 | 2021-02-03T17:58:03 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 772,714,297 | 18,383 | alexey-milovidov | If EXPLAIN does not show `ReadFromStorage ... with order` then optimization is not applied.
If the setting is enabled but reading is not in order - then the optimization is not applicable for the query. | 2021-02-03T18:13:53 | 2021-02-03T18:13:53 | {} | MEMBER |
ClickHouse | ClickHouse | 772,714,604 | 19,950 | bharatnc | Interesting, it works fine for me atleast:
Followed your method of using client in docker container (server also runs in same container).
```
docker run -d --name some-clickhouse-server --ulimit nofile=262144:262144 yandex/clickhouse-server
b13db9ff368b425ec030a4e783fc8e9e589318a94e04656d82067b2886c74798
➜ docke... | 2021-02-03T18:14:17 | 2021-02-03T18:14:17 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 772,813,585 | 20,043 | ghost | Same issue (table is named test2, but created exactly from your CREATE/INSERT statements):
```curl -i 'http://localhost:8123/?query=SELECT+count%28%2A%29+FROM+default.test2+WHERE+title+LIKE+%27%25as%27&send_progress_in_http_headers=1'
HTTP/1.1 200 OK
Date: Wed, 03 Feb 2021 20:46:40 GMT
Connection: Keep-Alive
Con... | 2021-02-03T20:49:54 | 2021-02-03T20:49:54 | {} | NONE |
ClickHouse | ClickHouse | 772,491,671 | 20,016 | tavplubix | 3. Reinitialize `DDLWorker` state on unexpected errors (#18285), so it will work like server was just restarted. As for now, `ZNONODE` [is not considered as unexpected error](https://github.com/ClickHouse/ClickHouse/blob/3b52b0400863f0d90a0d01ece3a6492d4dfdaf3b/src/Interpreters/DDLWorker.cpp#L669), but it should be. | 2021-02-03T13:03:42 | 2021-02-03T13:03:42 | {} | MEMBER |
ClickHouse | ClickHouse | 772,688,768 | 9,800 | bharatnc | > @bharatnc @alexey-milovidov is it resolved and can be closed? Or there is smth more to do here?
Last PR is open. We can close this out once that is merged. | 2021-02-03T17:35:28 | 2021-02-03T17:35:28 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 772,942,184 | 19,743 | mdianjun | @nikitamikhaylov PTAL. | 2021-02-04T00:57:12 | 2021-02-04T00:57:12 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 772,395,158 | 19,995 | leosunli | hi @alexey-milovidov @filimonov
there are no code on https://github.com/msgpack/msgpack-c/tree/master as follow. Which branch of msgpack-c is recommended?
Thanks @alexey-milovidov @filimonov | 2021-02-03T10:15:04 | 2021-02-04T01:49:52 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 773,024,717 | 19,896 | depressed-pho | Fixed !USE_SSL again. | 2021-02-04T04:45:21 | 2021-02-04T04:45:21 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 773,096,456 | 19,886 | alexey-milovidov | Ok.
I'm not able to carefully review the parts about Kafka and RabiitMQ but the code LGTM. | 2021-02-04T07:29:35 | 2021-02-04T07:29:43 | {} | MEMBER |
ClickHouse | ClickHouse | 773,103,079 | 20,058 | excitoon | There is a potential problem with this PR, by default configuration file does contain specified tag, empty, so if a user has their custom edited configuration, it may effectively block access to all external URLs for him after an upgrade. | 2021-02-04T07:43:32 | 2021-02-04T07:43:48 | {
"+1": 1
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 773,145,796 | 20,052 | alexey-milovidov | Minimal example: `SELECT toInt32(-199) % 200`. | 2021-02-04T09:00:27 | 2021-02-04T09:00:27 | {} | MEMBER |
ClickHouse | ClickHouse | 772,460,486 | 20,001 | alexey-milovidov | I will trigger the task manually to run on appropriate servers. | 2021-02-03T12:07:09 | 2021-02-03T12:07:09 | {} | MEMBER |
ClickHouse | ClickHouse | 772,496,302 | 20,037 | alexey-milovidov | @abyss7 We can do it intentionally to skip data and it's Ok. | 2021-02-03T13:11:45 | 2021-02-03T13:11:45 | {} | MEMBER |
ClickHouse | ClickHouse | 772,501,194 | 20,037 | abyss7 | > @abyss7 We can do it intentionally to skip data and it's Ok.
Is there a point on trying to refactor this? It will potentially simplify the usage and debugging for multi-layer buffers' composition.
At least I'd like to add `ignore(available())` where possible. | 2021-02-03T13:20:27 | 2021-02-03T13:21:31 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 772,648,674 | 19,315 | den-crane | I think mutations are not designed to handle such `where subqueries`
as a WA I would create a table Engine=Join and inserted into this Join table IDs which should be deleted using `insert select`
then run delete like this
```
ALTER TABLE clarity.page_data DELETE
WHERE joinHas(, , (ProjectId, serId, SessionId) ... | 2021-02-03T16:40:32 | 2021-02-03T16:40:32 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 773,095,105 | 20,011 | alexey-milovidov | > Stress test (thread) — Sanitizer assert (in stderr.log)
#20045 | 2021-02-04T07:26:53 | 2021-02-04T07:26:53 | {} | MEMBER |
ClickHouse | ClickHouse | 773,102,314 | 20,001 | alexey-milovidov | Performance optimizations require quick iterations on realistic datasets.
Maybe you can create obfuscated dataset?
Run `clickhouse-obfuscator --help` and follow the instruction.
You can have 100 bn records but to reproduce performance maybe some subset of a few million records can be enough? | 2021-02-04T07:41:54 | 2021-02-04T07:42:11 | {} | MEMBER |
ClickHouse | ClickHouse | 773,144,210 | 20,052 | alexey-milovidov | You are right, we need to use larger signed type to accomodate the result. | 2021-02-04T08:58:00 | 2021-02-04T08:58:00 | {} | MEMBER |
ClickHouse | ClickHouse | 773,150,384 | 19,899 | alexey-milovidov | 20.10 is not supported. | 2021-02-04T09:07:33 | 2021-02-04T09:07:33 | {} | MEMBER |
ClickHouse | ClickHouse | 772,446,846 | 20,032 | alexey-milovidov | This:
```
2021.02.02 11:42:35.017967 [ 311 ] {} <Debug> Application: Received termination signal.
2021.02.02 11:42:35.020440 [ 311 ] {} <Debug> Application: Waiting for current connections to close.
2021.02.02 11:42:35.754593 [ 311 ] {} <Information> Application: Closed all listening sockets. Waiting for 11 outst... | 2021-02-03T11:42:03 | 2021-02-03T11:42:03 | {} | MEMBER |
ClickHouse | ClickHouse | 772,676,339 | 20,042 | alexey-milovidov | I would not like if it will be implemented before user defined types. | 2021-02-03T17:17:32 | 2021-02-03T17:17:32 | {} | MEMBER |
ClickHouse | ClickHouse | 772,694,696 | 20,042 | filimonov | #18568 #4420 | 2021-02-03T17:44:34 | 2021-02-03T17:44:34 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 772,712,312 | 18,383 | alexey-milovidov | EXPLAIN shows the difference of pipelines if aggregation is processed in order:
```
$ clickhouse-client
ClickHouse client version 21.3.1.1.
Connecting to localhost:9000 as user default.
Connected to ClickHouse server version 21.3.1 revision 54447.
milovidov-desktop :) SELECT CounterID, count() FROM test.hits... | 2021-02-03T18:10:54 | 2021-02-03T18:10:54 | {} | MEMBER |
ClickHouse | ClickHouse | 772,717,962 | 19,692 | den-crane | No repro / proof. Closing. | 2021-02-03T18:19:24 | 2021-02-03T18:19:24 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 772,773,799 | 20,048 | alexey-milovidov | `ZooKeeperUserExceptions` don't indicate errors. They are part of the logic of distributed coordination.
They can appear due to multiple various reasons.
Example:
```
else if (code == Coordination::Error::ZBADVERSION)
{
ProfileEvents::increment(ProfileEvents::NotCreatedLogEntryForMerge);
... | 2021-02-03T19:44:25 | 2021-02-03T19:46:28 | {} | MEMBER |
ClickHouse | ClickHouse | 773,040,443 | 19,877 | alexey-milovidov | ```
SELECT NULL * 3, (val + 1024) = (rval * 1), NULL FROM (SELECT (val + NULL) = (NULL * 256), (NULL + val) = (NULL * '3'), (val + 3) = (1023 * 1025), toLowCardinality(dummy) AS val FROM system.one) AS s1 ANY LEFT JOIN (SELECT toLowCardinality(dummy) AS rval FROM system.one) AS s2 ON (val + 1) = (rval * 100)
```
h... | 2021-02-04T05:28:30 | 2021-02-04T05:28:30 | {} | MEMBER |
ClickHouse | ClickHouse | 773,064,049 | 20,053 | gubinjie | Will there be some unexpected bugs | 2021-02-04T06:24:40 | 2021-02-04T06:24:40 | {} | NONE |
ClickHouse | ClickHouse | 773,071,102 | 20,053 | gubinjie | @alexey-milovidov
Does this sentence "have more transient errors" cause data loss | 2021-02-04T06:39:26 | 2021-02-04T06:39:26 | {} | NONE |
ClickHouse | ClickHouse | 773,089,632 | 20,056 | alexey-milovidov | ClickHouse is using implementation specific behaviour when hour/minute/second is out of range.
It may overflow to the next time period.
This is made for performance reasons - to avoid extra checks, and we have no plans to change it. | 2021-02-04T07:16:29 | 2021-02-04T07:16:29 | {} | MEMBER |
ClickHouse | ClickHouse | 773,122,147 | 20,052 | zhangjmruc | A simple analysis:
The incorrect result may happen when the type A is signed and type B is usigned, because the result type of modulo is signed B. In this case, type A is signed Int32, type B is unsigned UInt8, hence the result type is Int8. The value of UInt8 from 128 to 255 is considered as negative for type Int8.
... | 2021-02-04T08:19:59 | 2021-02-04T08:19:59 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 773,133,651 | 19,580 | alesapin | One of 100 runs of `test_testkeeper_multinode` with thread sanitizer had timed out. Strange in logs:
```
[11:05:00]alesap@click:~/.../ClickHouse/tests/integration$ grep '5ae61917-0a22-48de-8cdd-3b341c83ae69' test_testkeeper_multinode/_instances/node2/logs/clickhouse-server.log
2021.02.04 00:39:48.171294 [ 11 ] {5ae6... | 2021-02-04T08:40:14 | 2021-02-04T08:40:14 | {} | MEMBER |
ClickHouse | ClickHouse | 773,149,574 | 19,989 | alexey-milovidov | No such function in this release. | 2021-02-04T09:06:18 | 2021-02-04T09:06:18 | {} | MEMBER |
ClickHouse | ClickHouse | 772,553,146 | 17,865 | amosbird | It's already fixed in https://github.com/ClickHouse/ClickHouse/pull/13793 . | 2021-02-03T14:32:10 | 2021-02-03T14:32:10 | {} | COLLABORATOR |
ClickHouse | ClickHouse | 772,691,702 | 20,043 | den-crane | cannot reproduce with 21.2.1.5834
Check my test with your environment
```
CREATE TABLE default.test
(
`id` UInt64,
`title` String,
`time` DateTime
)
ENGINE = MergeTree
PARTITION BY toStartOfMonth(time)
ORDER BY (id, title, time)
SETTINGS index_granularity = 8192
insert into default.test s... | 2021-02-03T17:40:02 | 2021-02-03T17:41:34 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 772,848,773 | 18,247 | SaltTan | Can you reproduce the issue? | 2021-02-03T21:52:11 | 2021-02-03T21:52:11 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 776,589,181 | 18,242 | k0st1an | Upgrade to version 21.2.2.8 not helped. | 2021-02-10T09:54:58 | 2021-02-10T09:54:58 | {} | NONE |
ClickHouse | ClickHouse | 776,904,864 | 19,317 | abyss7 | ASAN Fuzzer: #20017 | 2021-02-10T18:07:07 | 2021-02-10T18:07:07 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 777,082,459 | 20,229 | alexey-milovidov | Let's add `distributed_depth` to `ClientInfo`, pass it from client to server, increment it on every query, and throw exception if it is larger than `max_distributed_depth` setting (default = 5). | 2021-02-10T22:37:47 | 2021-02-10T22:37:47 | {} | MEMBER |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.