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 | 778,836,942 | 20,366 | alexey-milovidov | https://clickhouse-test-reports.s3.yandex.net/20475/55c17ac93f83746e1ddc92172d45ceeb5973de9f/fuzzer_debug/server.log | 2021-02-14T20:28:00 | 2021-02-14T20:28:00 | {} | MEMBER |
ClickHouse | ClickHouse | 778,841,648 | 20,465 | alexey-milovidov | > AST fuzzer (TSan) — Lost connection to server. See the logs.
OOM. | 2021-02-14T21:05:00 | 2021-02-14T21:05:00 | {} | MEMBER |
ClickHouse | ClickHouse | 778,886,047 | 6,972 | alexey-milovidov | @filimonov Could you please check if this issue exists in version 21.2? | 2021-02-15T01:33:38 | 2021-02-15T01:33:38 | {} | MEMBER |
ClickHouse | ClickHouse | 778,548,768 | 20,452 | den-crane | also parseDateTimeBestEffortUS https://clickhouse.tech/docs/en/sql-reference/functions/type-conversion-functions/ | 2021-02-13T02:33:12 | 2021-02-13T02:33:12 | {
"+1": 1
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 778,668,006 | 20,467 | alexey-milovidov | The desired behaviour is available under `aggregate_functions_null_for_empty` setting. | 2021-02-13T19:41:31 | 2021-02-13T19:41:31 | {} | MEMBER |
ClickHouse | ClickHouse | 778,842,747 | 4,854 | alexey-milovidov | Still relevant in 21.3. | 2021-02-14T21:13:08 | 2021-02-14T21:13:08 | {} | MEMBER |
ClickHouse | ClickHouse | 778,879,996 | 20,472 | alexey-milovidov | Ok: 1.5..2 times performance improvement.
0.6% across all the tests. | 2021-02-15T01:08:28 | 2021-02-15T01:08:28 | {} | MEMBER |
ClickHouse | ClickHouse | 778,509,021 | 20,267 | alexey-milovidov | It's only for 21.2. | 2021-02-12T23:20:32 | 2021-02-12T23:20:32 | {} | MEMBER |
ClickHouse | ClickHouse | 778,542,047 | 19,580 | sundy-li | > test_materialize_mysql_database
cc @zhang2014 | 2021-02-13T01:45:04 | 2021-02-13T01:45:04 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 778,546,679 | 20,454 | alexey-milovidov | I've made this change while looking into code. It's not necessarily better and most likely does not change anything. | 2021-02-13T02:16:23 | 2021-02-13T02:16:23 | {} | MEMBER |
ClickHouse | ClickHouse | 778,640,953 | 18,655 | toktarev | Not sure if it is correct to write here, but curiosity and this is Open Source :).
I have a question to ClickHouse team regarding usage of RocksDB.
Guys, have you faced with RocksDB's memory problems ?
I mean unlimited memory growth during long period of insertion ?
I receive more and more complains durin... | 2021-02-13T16:25:23 | 2021-02-13T16:25:23 | {} | NONE |
ClickHouse | ClickHouse | 778,644,346 | 20,462 | alexey-milovidov | I see the motivation here: https://github.com/ClickHouse/ClickHouse/pull/19580#issuecomment-778333022, let's try... | 2021-02-13T16:51:10 | 2021-02-13T16:51:10 | {} | MEMBER |
ClickHouse | ClickHouse | 778,689,547 | 20,471 | den-crane | ```
SELECT parseDateTimeBestEffortUSOrNull('30/01/2021')
Query id: fcbb75e7-8b59-47c1-b6c8-3db067ad65cb
┌─parseDateTimeBestEffortUSOrNull('30/01/2021')─┐
│ 2021-01-30 00:00:00 │
└───────────────────────────────────────────────┘
``` | 2021-02-13T22:52:55 | 2021-02-13T22:52:55 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 778,713,799 | 18,559 | alexey-milovidov | This can be fixed by a sequence of
ALTER TABLE ... DETACH PARTITION ...
ALTER TABLE ... ATTACH PARTITION ...
commands for the affected table. | 2021-02-14T02:57:26 | 2021-02-14T02:57:26 | {
"+1": 1
} | MEMBER |
ClickHouse | ClickHouse | 778,514,507 | 19,802 | alexey-milovidov | @hexiaoting One test failure is related, need to take a look. | 2021-02-12T23:39:56 | 2021-02-12T23:39:56 | {} | MEMBER |
ClickHouse | ClickHouse | 778,841,405 | 20,465 | alexey-milovidov | :heavy_plus_sign:
We should not make destructors noexcept(false).
std::terminate in destructors is Ok. | 2021-02-14T21:03:03 | 2021-02-14T21:03:30 | {} | MEMBER |
ClickHouse | ClickHouse | 778,844,804 | 20,469 | mzealey | Maybe, however my request was for a function that effectively doesn't care if the source is ipv4 or ipv6 (within an ipv6 domain) but returns an ipv4 string if it is ipv4 address. That way and with the new toipv6 improvements in becomes trivial to dual stack stuff (with "legacy" clients expecting an ipv4 string) just us... | 2021-02-14T21:29:16 | 2021-02-14T21:47:39 | {} | NONE |
ClickHouse | ClickHouse | 778,880,722 | 20,168 | alexey-milovidov | > 01396_inactive_replica_cleanup_nodes_zookeeper
Broken after merging NuRaft. | 2021-02-15T01:11:55 | 2021-02-15T01:11:55 | {} | MEMBER |
ClickHouse | ClickHouse | 778,500,575 | 20,346 | den-crane | try simple query
```sql
SELECT x.casino_id, x.game_id, x.round_id, x.trans_id, x.type
FROM transactions t
ANTI RIGHT JOIN transactions_temp_3 x
on t.row_hash = cityHash64(x.casino_id, x.game_id, x.round_id, x.trans_id, x.type)
SETTINGS max_threads = 16
```
example:
```sql
drop ... | 2021-02-12T22:55:22 | 2021-02-12T23:21:18 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 778,569,684 | 20,452 | rlazimi-dev | thank you for the help, this solved my problem! | 2021-02-13T06:05:24 | 2021-02-13T06:05:24 | {} | NONE |
ClickHouse | ClickHouse | 778,623,506 | 20,465 | amosbird | > > Perhaps we should just ignore all errors in dtor as the same way the stdlib does it. If the user cares about errors during close, they must call close() themselves. Otherwise, the dtor calls close and ignores all errors from the call.
>
> That was previous behavior, but it does not looks safe, since "user" (deve... | 2021-02-13T13:58:53 | 2021-02-13T13:58:53 | {} | COLLABORATOR |
ClickHouse | ClickHouse | 778,642,445 | 20,454 | alexey-milovidov | > test_merge_tree_s3/test.py::test_insert_same_partition_and_merge[False]
Flaky test? | 2021-02-13T16:36:25 | 2021-02-13T16:36:25 | {} | MEMBER |
ClickHouse | ClickHouse | 778,710,329 | 19,294 | alexey-milovidov | You should create an issue in the https://github.com/ClickHouse/clickhouse-jdbc repository. | 2021-02-14T02:22:53 | 2021-02-14T02:22:53 | {} | MEMBER |
ClickHouse | ClickHouse | 778,712,838 | 17,759 | alexey-milovidov | Recently we have found and fixed multiple issues with h3 library: https://github.com/ClickHouse/ClickHouse/pull/19383
that can cause the errors you observe. | 2021-02-14T02:48:20 | 2021-02-14T02:48:20 | {
"+1": 1
} | MEMBER |
ClickHouse | ClickHouse | 778,740,941 | 17,759 | hombit | It doesn't fail on 21.2 any more, thank you! I'm going to test if the result is consistent over query runs | 2021-02-14T07:40:01 | 2021-02-14T07:40:01 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 778,797,998 | 20,472 | alexey-milovidov | > test_materialize_mysql_database/test.py::test_select_without_columns_8_0[clickhouse_node1]
Flaky test, @zhang2014 @BohuTANG | 2021-02-14T16:01:14 | 2021-02-14T16:01:14 | {} | MEMBER |
ClickHouse | ClickHouse | 778,841,412 | 19,252 | bharatnc | > No difference:
>
> `clickhouse-benchmark --query "SELECT count() FROM test.hits WHERE NOT ignore(toIPv6(IPv6NumToString(ClientIP6)))"`
>
> 21.1:
>
> ```
> localhost:9000, queries 253, QPS: 10.733, RPS: 95244262.006, MiB/s: 1453.325, result RPS: 10.733, result MiB/s: 0.000.
>
> 0.000% 0.080 sec.
... | 2021-02-14T21:03:07 | 2021-02-14T21:03:07 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 778,508,695 | 20,303 | alexey-milovidov | > AST fuzzer (TSan) — Lost connection to server. See the logs.
> Program terminated with signal SIGKILL, Killed.
Yes, it's possible because:
- memory tracking is not enabled for new/delete in sanitizers builds;
- some aggregate functions can allocate memory with new/delete in generic case.
Ideas how to avoid t... | 2021-02-12T23:19:42 | 2021-02-12T23:19:59 | {} | MEMBER |
ClickHouse | ClickHouse | 778,590,918 | 19,580 | azat | >2021.02.12 14:22:18.238037 [ 375 ] {} <Fatal> BaseDaemon: (version 21.3.1.5996, build id: 8DBCED54529C989F7AD4D991F51410774D55DE6C) (from thread 17728) Terminate called for uncaught exception:
@alesapin should be fixed in #20464 | 2021-02-13T09:38:08 | 2021-02-13T09:38:08 | {} | MEMBER |
ClickHouse | ClickHouse | 778,593,644 | 19,451 | azat | >My local build fails to start up after this PR.
@amosbird #20465 | 2021-02-13T10:04:13 | 2021-02-13T10:04:13 | {} | MEMBER |
ClickHouse | ClickHouse | 778,706,453 | 19,754 | alexey-milovidov | https://clickhouse-test-reports.s3.yandex.net/20472/5bdc57004682a5e0236ec630546d20ad752c2fde/fuzzer_asan/server.log | 2021-02-14T01:42:05 | 2021-02-14T01:42:05 | {} | MEMBER |
ClickHouse | ClickHouse | 778,709,607 | 19,645 | alexey-milovidov | ```
DROP DATABASE IF EXISTS A1;
CREATE DATABASE A1;
CREATE TABLE A1.B
(
`id1` Nullable(String),
`id2` Nullable(String),
`id3` Nullable(String),
`id4` Nullable(String),
`id5` Nullable(String),
`date` Nullable(Date),
`shuzhi` Nullable(Int32),
`数值2` Nullable(Int32)
)
ENGINE ... | 2021-02-14T02:15:24 | 2021-02-14T02:18:33 | {} | MEMBER |
ClickHouse | ClickHouse | 778,712,180 | 14,568 | alexey-milovidov | In previous versions there was possibility of deadlocks in ALTER ... PARTITION queries.
In recent versions there is no such possibility.
In previous versions DROP TABLE can be locked by other queries to the same table.
After we enabled `Atomic` database by default it's not longer the case. | 2021-02-14T02:41:05 | 2021-02-14T02:41:05 | {} | MEMBER |
ClickHouse | ClickHouse | 778,713,167 | 18,813 | alexey-milovidov | Always worth to check `dmesg` for messages about faulty RAM. | 2021-02-14T02:51:26 | 2021-02-14T02:51:26 | {} | MEMBER |
ClickHouse | ClickHouse | 778,713,601 | 10,745 | alexey-milovidov | This is a diagnostic message about behaviour of the clients and network interaction.
The server correctly informs about connection reset. There are two possibilities:
- if these messages are appeared only occasionally: no action required;
- if these messages appear every time and the system is disrupted: misconfigur... | 2021-02-14T02:55:35 | 2021-02-14T02:55:35 | {} | MEMBER |
ClickHouse | ClickHouse | 778,713,944 | 14,039 | alexey-milovidov | @linlancer The example you provided is not complete. I cannot reproduce the issue by simply copy-pasting it. | 2021-02-14T02:58:39 | 2021-02-14T02:58:39 | {} | MEMBER |
ClickHouse | ClickHouse | 778,765,959 | 17,759 | hombit | Looks like the result is deterministic now! Two runs have shown the same result, it was never the case for 21.1 | 2021-02-14T11:42:38 | 2021-02-14T11:42:38 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 778,585,392 | 19,580 | azat | >Not related to changes
```
Code: 24, e.displayText() = DB::Exception: Cannot write to ostream at offset 262994, Stack trace (when copying this message, always include the lines below):
```
This came from #19451
I have some ideas, but let's improve diagnostics in stress tests first #20462 | 2021-02-13T08:47:20 | 2021-02-13T08:47:20 | {} | MEMBER |
ClickHouse | ClickHouse | 778,615,719 | 20,455 | qoega | @vdimir is it somewhere near joins you are working on? | 2021-02-13T12:58:53 | 2021-02-13T12:58:53 | {} | MEMBER |
ClickHouse | ClickHouse | 778,756,832 | 19,685 | hexiaoting | > It's slightly related to #19856, not I'm trying to realize can it be solved independently or we should make changes common for both cases.
So In which situations should cross join optimized to inner join?
A cross join with WHERE clause includes at least a condition related to two tables in AND expression?
And ... | 2021-02-14T10:17:28 | 2021-02-14T10:17:28 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 778,764,013 | 19,982 | ibradwan | Thanks Denny for pointing this out.
> No, I do NOT see any sense to stop CH.
Sorry I was confused, but yes given that the parts are immutable, and their swapping is atomic, and also the `FREEZE` operation uses hard-links, syncing the pages after the `FREEZE` operation should be enough.
I had to dig a bit deep... | 2021-02-14T11:24:05 | 2021-02-14T11:30:46 | {} | NONE |
ClickHouse | ClickHouse | 778,816,304 | 20,213 | nikitamikhaylov | @MaxJPowers If 1 minute is Ok for you dictionary source, you can tweak `query_wait_timeout_milliseconds` in dictionary's config. BTW, which version of ClickHouse is being used?
> Also what are the factors that go into determining the loading time of a cache dictionary?
Cache dictionary will load keys and attribu... | 2021-02-14T18:08:32 | 2021-02-14T18:08:32 | {} | MEMBER |
ClickHouse | ClickHouse | 778,841,568 | 20,465 | alexey-milovidov | > 00992_system_parts_race_condition_zookeeper
Broken after merging NuRaft. | 2021-02-14T21:04:25 | 2021-02-14T21:04:25 | {} | MEMBER |
ClickHouse | ClickHouse | 778,880,428 | 20,472 | alexey-milovidov | > Stress test (thread) — Killed by signal (in clickhouse-server.log)
Fixed in #20465 | 2021-02-15T01:10:34 | 2021-02-15T01:10:34 | {} | MEMBER |
ClickHouse | ClickHouse | 778,507,345 | 20,387 | alexey-milovidov | About empty part after merge - should consult with @alesapin or @CurtizJ. | 2021-02-12T23:15:48 | 2021-02-12T23:15:48 | {} | MEMBER |
ClickHouse | ClickHouse | 778,610,289 | 20,465 | azat | >Perhaps we should just ignore all errors in dtor as the same way the stdlib does it. If the user cares about errors during close, they must call close() themselves. Otherwise, the dtor calls close and ignores all errors from the call.
That was previous behavior, but it does not looks safe, since "user" (developer) ... | 2021-02-13T12:15:51 | 2021-02-13T12:15:51 | {} | MEMBER |
ClickHouse | ClickHouse | 778,642,363 | 18,655 | toktarev | @alexey-milovidov , I see , thanks.
| 2021-02-13T16:35:48 | 2021-02-13T16:35:48 | {} | NONE |
ClickHouse | ClickHouse | 778,635,786 | 20,455 | den-crane | В общем-то непонятно как должно быть правильно, потому что синтаксиса такого нигде наверное нет.
То что существует `RIGTH ANTI ` и `ANTI RIGHT ` и это не одно и тоже, стало для меня открытием.
Идея зачем понятна, чтобы можно было поменять таблицы местами положить в хеш-таблицу маленькую таблицу. Поэтому я тупо по... | 2021-02-13T15:43:15 | 2021-02-13T20:58:47 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 778,706,488 | 20,472 | alexey-milovidov | > AST fuzzer (ASan) — 0x6100000436d0 is located 144 bytes inside of 192-byte region [0x610000043640,0x
https://github.com/ClickHouse/ClickHouse/issues/19754 | 2021-02-14T01:42:30 | 2021-02-14T01:42:30 | {} | MEMBER |
ClickHouse | ClickHouse | 778,711,752 | 15,233 | alexey-milovidov | @durchgedreht Yes, it is known behaviour of function transform - it does not support nullable arguments. | 2021-02-14T02:36:53 | 2021-02-14T02:36:53 | {} | MEMBER |
ClickHouse | ClickHouse | 778,837,070 | 20,475 | alexey-milovidov | > AST fuzzer (debug) — Logical error: 'Arguments of 'intDiv' have incorrect data types: '10' of type 'UInt8', '2' of type 'UInt8''.
#20461 | 2021-02-14T20:29:08 | 2021-02-14T20:29:08 | {} | MEMBER |
ClickHouse | ClickHouse | 778,838,143 | 20,462 | alexey-milovidov | Note that it should not prevent receiving fatal messages with stack trace in client. | 2021-02-14T20:37:50 | 2021-02-14T20:37:50 | {} | MEMBER |
ClickHouse | ClickHouse | 778,514,606 | 19,802 | alexey-milovidov | Also need to add a test for a new feature. | 2021-02-12T23:40:19 | 2021-02-12T23:40:19 | {} | MEMBER |
ClickHouse | ClickHouse | 778,708,708 | 20,343 | alexey-milovidov | @vitlibar SETTINGS clause should be either after INSERT SELECT or before FORMAT. | 2021-02-14T02:05:20 | 2021-02-14T02:05:20 | {} | MEMBER |
ClickHouse | ClickHouse | 778,712,903 | 17,759 | alexey-milovidov | I recommend to test on the most recent version: 21.2. | 2021-02-14T02:48:46 | 2021-02-14T02:48:46 | {} | MEMBER |
ClickHouse | ClickHouse | 778,822,857 | 20,472 | alexey-milovidov | > Functional stateless tests (ubsan) — Some queries hung, fail: 1, passed: 2768, skipped: 10
Broken after merging of NuRaft. | 2021-02-14T18:48:48 | 2021-02-14T18:48:48 | {} | MEMBER |
ClickHouse | ClickHouse | 777,012,006 | 20,253 | traceon | After discussing this in chat, the missing part was identified to be a `{user_dn}` substitution in `<search_filter>` section specifically, which would store the real DN of the user. By default, it could be equal to `bind_dn` itself, but for cases like this issue, a separate LDAP search should be performed to retrieve t... | 2021-02-10T20:35:15 | 2021-02-15T00:25:43 | {
"+1": 2
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 778,509,913 | 20,206 | alexey-milovidov | @tavplubix Let's add a test and merge... | 2021-02-12T23:23:12 | 2021-02-12T23:23:12 | {} | MEMBER |
ClickHouse | ClickHouse | 778,690,016 | 20,471 | stas-sl | Cool! Sorry for bothering | 2021-02-13T22:57:12 | 2021-02-13T23:00:44 | {} | NONE |
ClickHouse | ClickHouse | 778,708,402 | 20,343 | alexey-milovidov | @filimonov We should not allow SETTINGS clause after FORMAT.
Everything after FORMAT is parsed as data.
Do you remember a pull request where such a wrong behaviour was introduced? | 2021-02-14T02:02:43 | 2021-02-14T02:02:43 | {} | MEMBER |
ClickHouse | ClickHouse | 778,512,942 | 20,114 | alexey-milovidov | So, table will be shut down but not dropped, because ZooKeeper is unavailable. Will it be possible to drop table later? | 2021-02-12T23:33:52 | 2021-02-12T23:33:52 | {} | MEMBER |
ClickHouse | ClickHouse | 778,543,520 | 20,455 | den-crane | mind blowing
```sql
SELECT c,d FROM B1 RIGTH ANTI JOIN T1 USING (c);
┌─c─┬─d─┐
│ 4 │ 4 │
│ 5 │ 5 │
│ 6 │ 6 │
└───┴───┘
SELECT c,d FROM B1 RIGTH ANTI JOIN T1 on T1.c = B1.c;
┌─c─┬─d─┐
│ 4 │ 4 │
│ 5 │ 5 │
│ 6 │ 6 │
└───┴───┘
SELECT c,d FROM B1 ANTI RIGHT JOIN T1 USING (c);
┌─c─┬─d─┐
│ 3 │... | 2021-02-13T01:55:36 | 2021-02-13T01:55:36 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 778,699,403 | 20,206 | alexey-milovidov | > Functional stateless tests (ubsan) — fail: 0, passed: 2770, skipped: 9
Ok. | 2021-02-14T00:32:12 | 2021-02-14T00:32:12 | {} | MEMBER |
ClickHouse | ClickHouse | 778,734,050 | 20,465 | amosbird | Destructors are noexcept by default. We can annotate it with `noexcept(false)` and check if there is any `std::uncaught_exception` on the fly. If yes, we log the exception. If no, we throw. | 2021-02-14T06:17:48 | 2021-02-14T06:18:04 | {} | COLLABORATOR |
ClickHouse | ClickHouse | 778,758,654 | 20,477 | azat | I want to run all tests for this patches, but right now I'm expecting at least one test failure, can someone add a `force tests` label please? | 2021-02-14T10:34:38 | 2021-02-14T10:35:37 | {} | MEMBER |
ClickHouse | ClickHouse | 778,832,511 | 20,480 | den-crane | 21.3.1.5998.
```
CREATE TABLE d (oid UInt64 CODEC(Delta(8), LZ4) ) ENGINE = MergeTree() ORDER BY oid ;
insert into d select cityHash64(number) oid from numbers(100000000) ;
for i in `seq 20`; do clickhouse-client -q 'SELECT medianDeterministic(oid, oid) FROM d'; done
9167152341473662000
9163761196733... | 2021-02-14T19:55:33 | 2021-02-14T19:55:33 | {
"+1": 1
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 778,641,887 | 18,655 | alexey-milovidov | @toktarev It's not actively used yet and we have no reports about this issue.
There are no tests for long period of insertion - maximum are fuzz and stress tests that last only one hour (for every commit).
And that tests are not specific - they run queries with RocksDB only occasionally (by random choice among othe... | 2021-02-13T16:32:29 | 2021-02-13T16:32:29 | {} | MEMBER |
ClickHouse | ClickHouse | 778,711,565 | 13,360 | alexey-milovidov | Most likely you have very slow or overloaded IO subsystem on the server.
Running `dstat` and `iostat` on the server will help to quantify it. | 2021-02-14T02:34:51 | 2021-02-14T02:34:51 | {} | MEMBER |
ClickHouse | ClickHouse | 778,712,605 | 17,759 | alexey-milovidov | The issue looks important, we will try to download data and reproduce it. | 2021-02-14T02:45:52 | 2021-02-14T02:45:52 | {} | MEMBER |
ClickHouse | ClickHouse | 778,839,068 | 20,462 | azat | >Note that it should not prevent receiving fatal messages with stack trace in client.
Right. Actually I don't think that this should break this, since the server will continue after gdb will print stacktraces | 2021-02-14T20:44:36 | 2021-02-14T20:44:36 | {} | MEMBER |
ClickHouse | ClickHouse | 778,844,119 | 20,469 | bharatnc | Yes we will need `toIPv4`, `FunctionIPv4StringToNum` mirror the current behavior `toIPv6` & `FunctionIPv6StringToNum`.
I can work on this if someone can /assign this to me (related to my previous work on #19518).
Currently toIPv4 does:
```sql
SELECT toIPv4('::ffff:192.168.1.1')
┌─toIPv4('::ffff:192.168.... | 2021-02-14T21:23:31 | 2021-02-14T21:25:22 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 778,642,393 | 20,454 | alexey-milovidov | The test of GROUP BY two fields has been speed up more than 1.5 times. | 2021-02-13T16:36:06 | 2021-02-13T16:36:06 | {} | MEMBER |
ClickHouse | ClickHouse | 778,689,643 | 20,471 | den-crane | ClickHouse release v21.2.2.8-stable, 2021-02-07
Added functions parseDateTimeBestEffortUSOrZero, parseDateTimeBestEffortUSOrNull. #19712 (Maksim Kita). | 2021-02-13T22:53:46 | 2021-02-13T22:53:46 | {
"+1": 1
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 778,710,344 | 19,294 | alexey-milovidov | CC @zhicwu | 2021-02-14T02:23:05 | 2021-02-14T02:23:05 | {} | MEMBER |
ClickHouse | ClickHouse | 778,800,110 | 20,479 | ucasfl | Current fix may lead to another problem:
```
:) SELECT transform(number / 2, [1.0, 2.0], ['Hello', 'World'], '-') FROM numbers(10)
SELECT transform(number / 2, [1., 2.], ['Hello', 'World'], '-')
FROM numbers(10)
Query id: 8469547a-653a-4937-a2f2-0aa898618b07
┌─transform(divide(number, 2), [1., 2.], ['Hello'... | 2021-02-14T16:16:32 | 2021-02-14T16:16:32 | {} | COLLABORATOR |
ClickHouse | ClickHouse | 778,508,040 | 20,456 | den-crane | ```sql
SELECT
temp.casino,
temp.game,
temp.round,
temp.trans,
temp.type
FROM temp
ANTI LEFT JOIN t ON t.row_hash = cityHash64(temp.casino, temp.game, temp.round, temp.trans, temp.type)
┌─casino─┬─game─────────────────┬─round─┬─trans─┬─type─┐
│ 0 │ 4761183170873013810 │ 0 │... | 2021-02-12T23:17:45 | 2021-02-12T23:17:45 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 778,519,500 | 6,402 | genzgd | That setting appears to have worked, thanks! | 2021-02-12T23:58:34 | 2021-02-12T23:58:34 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 778,593,893 | 19,673 | azat | @alexey-milovidov can you please take a look? | 2021-02-13T10:06:33 | 2021-02-13T10:06:33 | {} | MEMBER |
ClickHouse | ClickHouse | 778,706,320 | 20,472 | alexey-milovidov | Also we can try trie. | 2021-02-14T01:40:42 | 2021-02-14T01:40:42 | {} | MEMBER |
ClickHouse | ClickHouse | 778,711,021 | 13,875 | alexey-milovidov | 1. The issue may happen if you have huge number of data parts and low amount of memory.
2. The performance of directory removal was fixed in version 21.2 here https://github.com/ClickHouse/ClickHouse/pull/19086.
3. With daily partitions you will have more data parts and this is bad for server startup - monthly partit... | 2021-02-14T02:29:25 | 2021-02-14T02:29:25 | {} | MEMBER |
ClickHouse | ClickHouse | 778,711,233 | 14,764 | alexey-milovidov | > 9- What about using Buffer tables on cluster? is it possible to do so?
Yes, you can use Buffer tables on every server of your cluster.
> 10- Also I tested the Compact format on the staging server and rows=90000 and the insertion rate 2k/sec I noticed that the size it takes is far more the Wide format, is it cor... | 2021-02-14T02:31:29 | 2021-02-14T02:31:29 | {} | MEMBER |
ClickHouse | ClickHouse | 778,715,374 | 20,472 | alexey-milovidov | > AST fuzzer (UBSan) — ../src/Functions/DivisionUtils.h:103:40: runtime error: 9.22429e+18 is outside the rang
https://github.com/ClickHouse/ClickHouse/pull/20475 | 2021-02-14T03:12:13 | 2021-02-14T03:12:13 | {} | MEMBER |
ClickHouse | ClickHouse | 778,833,021 | 20,480 | den-crane | even without cityHash64 (faster insert)
```
CREATE TABLE d (oid UInt64 CODEC(Delta(8), LZ4) ) ENGINE = MergeTree() ORDER BY oid ;
insert into d select (number) oid from numbers(100000000) ;
for i in `seq 20`; do clickhouse-client -q 'SELECT medianDeterministic(oid, oid) FROM d'; done
49497039
4949690... | 2021-02-14T19:59:01 | 2021-02-14T19:59:01 | {
"+1": 2
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 778,842,810 | 4,854 | alexey-milovidov | ```
milovidov-desktop :) SELECT toTypeName(now())
SELECT toTypeName(now())
Query id: 7b95112a-c7bb-456b-83ea-6780c799d813
┌─toTypeName(now())─┐
│ DateTime │
└───────────────────┘
1 rows in set. Elapsed: 0.002 sec.
milovidov-desktop :) SELECT toTypeName(now() - 1)
SELECT toTypeName(now() -... | 2021-02-14T21:13:46 | 2021-02-14T21:13:46 | {} | MEMBER |
ClickHouse | ClickHouse | 778,536,025 | 19,529 | alexey-milovidov | This change did not give performance benefits because our arrays appeared to already have the "perfect" size for allocator. | 2021-02-13T01:08:10 | 2021-02-13T01:08:10 | {} | MEMBER |
ClickHouse | ClickHouse | 778,592,809 | 19,451 | amosbird | My local build fails to start up after this PR.
```
2021.02.13 12:52:22.392174 [ 1499211 ] {} <Fatal> BaseDaemon: (version 21.3.1.1, build id: 4E2F3CF7C0210B16AB70B7FB87CFA04F66570550) (from thread 1499154) Terminate called for uncaught exception:
Code: 75, e.displayText() = DB::ErrnoException: Cannot write to fil... | 2021-02-13T09:56:13 | 2021-02-13T09:56:13 | {} | COLLABORATOR |
ClickHouse | ClickHouse | 778,604,032 | 20,457 | CurtizJ | Introduced in #19007. | 2021-02-13T11:34:36 | 2021-02-13T11:34:36 | {} | MEMBER |
ClickHouse | ClickHouse | 778,667,741 | 20,466 | alexey-milovidov | LGTM | 2021-02-13T19:39:17 | 2021-02-13T19:39:17 | {} | MEMBER |
ClickHouse | ClickHouse | 778,714,885 | 20,468 | alexey-milovidov | 10..20% improvement on perf test and about 1.5 times on real query. | 2021-02-14T03:06:52 | 2021-02-14T03:06:52 | {} | MEMBER |
ClickHouse | ClickHouse | 778,824,198 | 20,462 | azat | This is also required in other places (like fast-test - https://clickhouse-test-reports.s3.yandex.net/20477/8ad20fcee5aaa642c2a2dd873d02103692d554f4/fast_test.html#fail1) | 2021-02-14T18:58:20 | 2021-02-14T18:58:20 | {} | MEMBER |
ClickHouse | ClickHouse | 778,838,702 | 20,465 | azat | > Destructors are noexcept by default.
Right
> We can annotate it with noexcept(false) and check if there is any std::uncaught_exception on the fly. If yes, we log the exception. If no, we throw.
I tried to do this, but the problem is - there is too much classes that should be changed (including poco).
And ... | 2021-02-14T20:41:55 | 2021-02-14T20:41:55 | {} | MEMBER |
ClickHouse | ClickHouse | 778,879,243 | 20,475 | alexey-milovidov | > 01396_inactive_replica_cleanup_nodes_zookeeper
Broken after merging NuRaft. | 2021-02-15T01:04:44 | 2021-02-15T01:04:44 | {} | MEMBER |
ClickHouse | ClickHouse | 778,546,523 | 20,453 | alexey-milovidov | SIGTRAP is triggered when code reaches `int3` instruction that can be inserted by compiler in code branches explicitly marked as unreachable. | 2021-02-13T02:15:24 | 2021-02-13T02:15:24 | {
"eyes": 1
} | MEMBER |
ClickHouse | ClickHouse | 778,564,833 | 20,350 | amosbird | > Maybe we should forbid subqueries as function arguments on parser level? We can allow it for table functions only by passing special flag (something like `is_table_function`) to `ParserFunction` constructor from `ParserTableExpression::parseImpl(...)`, `ParserCreateTableQuery::parseImpl(...)` and similar places.
G... | 2021-02-13T05:08:36 | 2021-02-13T05:08:36 | {} | COLLABORATOR |
ClickHouse | ClickHouse | 778,710,172 | 19,351 | alexey-milovidov | @kssenii RabbitMQ storage engine should support FormatFactory settings similar to Kafka. | 2021-02-14T02:21:25 | 2021-02-14T02:21:25 | {} | MEMBER |
ClickHouse | ClickHouse | 778,806,860 | 20,479 | alexey-milovidov | Yes, I also started patching and faced the same issue.
You can check my patch just in case:
[patch.txt](https://github.com/ClickHouse/ClickHouse/files/5978104/patch.txt)
| 2021-02-14T17:06:38 | 2021-02-14T17:06:38 | {
"+1": 1
} | MEMBER |
ClickHouse | ClickHouse | 778,885,118 | 7,997 | alexey-milovidov | Since version 20.5 there is no notion of "leader" in ClickHouse and this issue is no longer relevant. | 2021-02-15T01:29:38 | 2021-02-15T01:29:38 | {} | MEMBER |
ClickHouse | ClickHouse | 770,852,156 | 15,607 | alexey-milovidov | I've reopened #16616. | 2021-02-01T13:21:03 | 2021-02-01T13:21:03 | {
"+1": 1
} | MEMBER |
ClickHouse | ClickHouse | 770,895,390 | 19,873 | filimonov | Can you also check memory usage? (see #10461)
| 2021-02-01T14:25:32 | 2021-02-01T14:25:32 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 771,028,611 | 19,945 | den-crane | You have to use aliases in this case. `ARRAY JOIN t.score as array_join_alias`
Original array column is also available in select section by original `t.score` name
https://clickhouse.tech/docs/ru/sql-reference/statements/select/array-join/ | 2021-02-01T17:37:38 | 2021-02-01T17:38:04 | {} | CONTRIBUTOR |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.