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 | 769,521,433 | 19,769 | alexey-milovidov | It contains a bug. | 2021-01-29T01:58:52 | 2021-01-29T01:58:52 | {} | MEMBER |
ClickHouse | ClickHouse | 768,929,954 | 19,745 | stigsb | I think a good solution would be to mirror the table replication filters in mysql as MaterializeMySQL database settings, for example:
```
SETTINGS
replicate_do_table = 'tbl list',
replicate_ignore_table = 'tbl list',
replicate_wild_do_table = 'wild tbl list',
replicate_wild_ignore_table = 'wild ... | 2021-01-28T09:43:29 | 2021-01-28T09:43:29 | {
"+1": 2
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 769,011,261 | 18,247 | ghuname | SELECT count()
FROM system.zookeeper
WHERE path = '/clickhouse/task_queue/ddl/'
Node 1 and 2 (shard1) 124, node 3 and 4 (shard2) 1
SELECT
ctime,
name,
value
FROM system.zookeeper
WHERE path = '/clickhouse/task_queue/ddl/'
ORDER BY ctime ASC
LIMIT 3
ctime |name |valu... | 2021-01-28T12:09:50 | 2021-01-28T12:09:50 | {} | NONE |
ClickHouse | ClickHouse | 769,026,207 | 19,684 | yiguolei | @tavplubix You are right. I have another commit. ^-^
But I find it is hard to test it, because the table has to be dropped during the ddl task's running stage. It is very difficult to make such a test. Could you please provide some idea? | 2021-01-28T12:40:02 | 2021-01-28T12:40:02 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 769,172,790 | 19,749 | den-crane | MATERIALIZED VIEW is an insert trigger.
But CH does not have transactions / atomicity in Table -> Materialized View.
1. client sends `INSERT INTO test_1 VALUES('id2', '1A', 2);`
2. INSERT handler gets an inserted buffer
3. INSERT handler writes the inserted buffer into table test_1
4. **INSERT handler** trigger... | 2021-01-28T15:41:13 | 2021-01-28T15:46:55 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 769,502,218 | 19,767 | alexey-milovidov | > AST fuzzer (UBSan) — ../src/Functions/FunctionsRound.h:127:23: runtime error: signed integer overflow: -9223372036854775808 - 100 cannot be represented in type '
#19786 | 2021-01-29T01:06:23 | 2021-01-29T01:06:23 | {} | MEMBER |
ClickHouse | ClickHouse | 769,605,504 | 8,159 | filimonov | You need to use clickhouse 21.1 or newer
Check
https://clickhouse.tech/docs/en/engines/table-engines/integrations/hdfs/
https://github.com/clickhouse/ClickHouse/blob/master/tests/integration/test_storage_kerberized_hdfs/configs/hdfs.xml
| 2021-01-29T06:24:06 | 2021-01-29T06:27:49 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 768,862,926 | 19,703 | alexey-milovidov | You should type `thread apply all backtrace` in gdb after you receive the `Aborted.` message. | 2021-01-28T07:38:16 | 2021-01-28T07:38:16 | {} | MEMBER |
ClickHouse | ClickHouse | 768,883,881 | 19,473 | alexey-milovidov | Help wanted from @KochetovNicolai | 2021-01-28T08:23:11 | 2021-01-28T08:23:11 | {} | MEMBER |
ClickHouse | ClickHouse | 768,903,476 | 19,584 | azat | >Yandex synchronization check (only for Yandex employees)
Lack of autocompletion, will disable the test in arcadia | 2021-01-28T08:58:23 | 2021-01-28T08:58:31 | {} | MEMBER |
ClickHouse | ClickHouse | 768,908,434 | 19,751 | alesapin | :D https://github.com/ClickHouse/ClickHouse/pull/19760 | 2021-01-28T09:06:04 | 2021-01-28T09:06:04 | {} | MEMBER |
ClickHouse | ClickHouse | 768,926,807 | 19,746 | socketpair | FYI, I build CH on ARM64 machine. (i.e. without qemu and not cross-compiling) | 2021-01-28T09:37:45 | 2021-01-28T09:45:09 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 769,019,669 | 19,708 | alesapin | Failures are known. | 2021-01-28T12:26:51 | 2021-01-28T12:26:51 | {} | MEMBER |
ClickHouse | ClickHouse | 769,402,228 | 19,716 | akuzm | ODBC integration test failures and testflows failures are known. | 2021-01-28T21:17:00 | 2021-01-28T21:17:00 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 769,581,248 | 10,998 | excitoon | It was not a fix actually, shall be fixed in #19750 | 2021-01-29T05:18:06 | 2021-01-29T05:18:06 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 769,590,421 | 19,487 | ucasfl | > @ucasFL
>
> ```
> SELECT * FROM system.asynchronous_metrics ORDER BY name
>
> maybe you meant: '['metric']' '['value']'.
> ```
>
> There are three wrong things:
>
> 1. Double whitespace.
> 2. Lack of comma between items.
> 3. Bogus format - the items are wrapped into array of a single value for unkno... | 2021-01-29T05:45:41 | 2021-01-29T05:45:41 | {} | COLLABORATOR |
ClickHouse | ClickHouse | 769,622,128 | 19,770 | stigsb | Fixed integration test (failed if system_parts_test database already existed, counted all parts and not just active). | 2021-01-29T07:05:39 | 2021-01-29T07:05:39 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 768,864,073 | 19,671 | alexey-milovidov | @mayankkandari
Ok. Could you please recreate the snapshot table with `SETTINGS index_granularity_bytes = '1M'`?
It may help by reducing the data amount that is read for point queries. | 2021-01-28T07:41:06 | 2021-01-28T07:41:13 | {} | MEMBER |
ClickHouse | ClickHouse | 768,990,657 | 19,690 | ilejn | run "disable pretty", run "enable pretty", provide output. | 2021-01-28T11:30:03 | 2021-01-28T11:30:03 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 768,994,138 | 19,690 | ilejn | Do this when active debugging, e.g. a breakpoint fired. | 2021-01-28T11:36:30 | 2021-01-28T11:36:30 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 769,242,813 | 19,584 | alexey-milovidov | Ok but the test does not look good enough... If you cannot fix it let's proceed without the test (give up). | 2021-01-28T17:20:09 | 2021-01-28T17:20:09 | {} | MEMBER |
ClickHouse | ClickHouse | 769,253,845 | 19,762 | alexey-milovidov | > AST fuzzer (ASan) — 0x000007d2233e is located 2 bytes to the left of global variable '_pcre_ucd_stage2'
https://github.com/ClickHouse/ClickHouse/pull/19720 | 2021-01-28T17:37:22 | 2021-01-28T17:37:22 | {} | MEMBER |
ClickHouse | ClickHouse | 769,254,396 | 19,762 | alexey-milovidov | ANTLR parser does not support query parameters yet?
Need to add to skip_list.json. | 2021-01-28T17:38:17 | 2021-01-28T17:38:17 | {} | MEMBER |
ClickHouse | ClickHouse | 769,290,839 | 19,770 | stigsb | Fixed the build failure (Yandex CI builds without -DENABLE_MYSQL, apparently) | 2021-01-28T18:37:39 | 2021-01-28T18:37:39 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 769,019,252 | 19,690 | ilejn | Well, "1" looks strange and probably does not mean anything.
What looks suspicious is that you are referring the file as "gdbinit", while it should be ".gdbinit" in your home directory.
Obviously all these commands can be issued manually.
It seems a good idea to run
```
python
print(sys.path)
end
```
to be sur... | 2021-01-28T12:25:54 | 2021-01-28T12:26:44 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 769,054,566 | 19,702 | alesapin | Another bug in part checker, need investigation:
```
2021.01.28 13:08:37.187183 [ 16157 ] {} <Information> test_nj731q.r3 (1c125c4e-9a4e-413c-bdf9-6cf859870a91): DB::Exception: No active replica has part all_114_114_0 or covering part
2021.01.28 13:08:37.187275 [ 189 ] {} <Warning> test_nj731q.r3 (ReplicatedMergeTre... | 2021-01-28T13:32:33 | 2021-01-28T13:32:33 | {} | MEMBER |
ClickHouse | ClickHouse | 769,061,991 | 19,770 | stigsb | This is somewhat of a kludge, but so far I have not found a cleaner way to fix this without more refactoring.
| 2021-01-28T13:44:52 | 2021-01-28T13:44:52 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 768,876,277 | 19,588 | alexey-milovidov | #19755 | 2021-01-28T08:07:40 | 2021-01-28T08:07:40 | {} | MEMBER |
ClickHouse | ClickHouse | 768,882,442 | 19,584 | alexey-milovidov | It is Ok. | 2021-01-28T08:20:29 | 2021-01-28T08:20:29 | {} | MEMBER |
ClickHouse | ClickHouse | 769,158,728 | 19,715 | alesapin | > An unpleasant side effect of this approach is that now we have to mark all the tests that explicitly create a database as sequential-only... This might be technically correct, but we could probably run them concurrently with other tests.
I think it's OK | 2021-01-28T15:21:03 | 2021-01-28T15:21:03 | {} | MEMBER |
ClickHouse | ClickHouse | 769,172,693 | 19,746 | socketpair | This is my build log. We build CH using RPM .spec file. Again, compiling for ARM64 linux on ARM64 linux.
[ch_ideco.log](https://github.com/ClickHouse/ClickHouse/files/5888349/ch_ideco.log)
it contains:
`-- Not using FastMemcpy` | 2021-01-28T15:41:05 | 2021-01-28T15:43:29 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 769,177,297 | 19,746 | filimonov | Duplicate of https://github.com/ClickHouse/ClickHouse/issues/18951
Should be fixed in https://github.com/ClickHouse/ClickHouse/pull/19369/files#diff-28dfb239b7a9ef0686c4c80be08bc28984acfd6e0376454b9f5767dcb4dd0067
| 2021-01-28T15:47:30 | 2021-01-28T15:49:05 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 769,181,047 | 19,692 | den-crane | @jjtjiang show the banner:
```
ClickHouse client version 21.2.1.5790 (official build).
Connecting to localhost:9000 as user default.
Connected to ClickHouse server version 21.2.1 revision 54447.
```
It seems you have outdated ClickHouse client | 2021-01-28T15:52:55 | 2021-01-28T15:52:55 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 768,849,318 | 19,673 | azat | >Yandex synchronization check (only for Yandex employees) failed
Don't have any ideas (all new tests are for distributed and disabled for arcadia)
Any clue? Some private fields from StorageDistributed had been used in arcadia?
(HEAD was 2c4b9ea2705c37c05e27c4dcb7634da5fd24cc0d) | 2021-01-28T07:10:45 | 2021-01-28T20:22:01 | {} | MEMBER |
ClickHouse | ClickHouse | 769,468,707 | 19,709 | den-crane | >ENGINE = AggregatingMergeTree()
>ORDER BY (ts, zid1, zid2)
>SAMPLE BY (ts, zid1, zid2);
1. ENGINE = AggregatingMergeTree() -- AggregatingMergeTree must used only with AggregateFunctions or SimpleAggregateFunctions.
2. ORDER BY (ts, zid1, zid2) -- mostly you should never place high-cardinality column (ts) as ... | 2021-01-28T23:34:00 | 2021-01-28T23:34:44 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 768,878,103 | 19,702 | alexey-milovidov | > AST fuzzer (debug) — Received signal 11
https://github.com/ClickHouse/ClickHouse/pull/19736/ | 2021-01-28T08:11:24 | 2021-01-28T08:11:24 | {} | MEMBER |
ClickHouse | ClickHouse | 768,905,724 | 18,377 | vdimir | > I'll put here some details about implementation, correct me if I understand something wrong ...
Could you also comment this from my previous message? | 2021-01-28T09:01:48 | 2021-01-28T09:02:23 | {} | MEMBER |
ClickHouse | ClickHouse | 768,928,661 | 19,690 | ilejn | @FavonianKong I suggest you should issue disable pretty/enable pretty.
There are thirty something libcxx pretty printers. If you see more, you probably use wrong pretty printers. | 2021-01-28T09:41:08 | 2021-01-28T09:41:08 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 769,356,264 | 19,584 | azat | >but the test does not look good enough
Let's try one more time... (I've tweaked timeouts) | 2021-01-28T20:17:24 | 2021-01-28T20:17:57 | {} | MEMBER |
ClickHouse | ClickHouse | 769,544,509 | 13,644 | xuelinchao | how to use LocalDateTime? | 2021-01-29T03:11:21 | 2021-01-29T03:12:36 | {} | NONE |
ClickHouse | ClickHouse | 769,581,525 | 19,750 | excitoon | It needs manual test on AWS. | 2021-01-29T05:18:57 | 2021-01-29T05:18:57 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 768,873,197 | 19,588 | alexey-milovidov | #19754 | 2021-01-28T08:01:15 | 2021-01-28T08:01:15 | {} | MEMBER |
ClickHouse | ClickHouse | 769,007,037 | 19,690 | FavonianKong | > Do this when active debugging, e.g. a breakpoint fired.
@ilejn You mean when the debugger hit the breakpoint？ it show like this：
```
$ gdb clickhouse
GNU gdb (GDB) 10.1
Copyright (C) 2020 Free Software Foundation, Inc.
License GPLv3+: GNU GPL version 3 or later <http://gnu.org/licenses/gpl.html>
This is free... | 2021-01-28T12:01:21 | 2021-01-28T12:01:21 | {} | NONE |
ClickHouse | ClickHouse | 769,370,568 | 19,756 | CurtizJ | Minimal repro:
```sql
SELECT
number
FROM numbers(256)
ORDER BY
arrayFilter(x -> materialize(255), ['257']) ASC
LIMIT 10
``` | 2021-01-28T20:35:37 | 2021-01-28T20:35:37 | {} | MEMBER |
ClickHouse | ClickHouse | 769,460,066 | 19,688 | SaltTan | clickhouse-copier creates intermediate tables in order to handle exceptions better.
Just wait a little longer.
You can also read copier's logs to find out what it is doing.
Or at least check system.processes.
There is some information in this discussion: https://github.com/ClickHouse/ClickHouse/pull/9075 | 2021-01-28T23:11:27 | 2021-01-28T23:11:27 | {
"+1": 1
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 768,879,337 | 19,679 | kitaisreal | Fixed in https://github.com/ClickHouse/ClickHouse/pull/19712. | 2021-01-28T08:14:02 | 2021-01-28T08:14:02 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 768,905,440 | 19,487 | alexey-milovidov | Ok. Thank you! | 2021-01-28T09:01:21 | 2021-01-28T09:01:37 | {} | MEMBER |
ClickHouse | ClickHouse | 768,907,899 | 19,584 | azat | >there was also one failure of the test for roaring bitmaps that I also disabled for Arcadia in master.
I've rebased against upstream too, so the check should pass | 2021-01-28T09:05:08 | 2021-01-28T09:05:08 | {} | MEMBER |
ClickHouse | ClickHouse | 768,943,115 | 19,573 | stigsb | See also #14735 for BIT support.
As for primary keys, you really want to have one. MaterializeMySQL could perhaps treat a unique index for a non-null column (or concatenated non-null columns) as a primary key, if there's only one such index, but IMHO that would be putting even more hardcoded assumptions into Materi... | 2021-01-28T10:06:01 | 2021-01-28T10:06:01 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 769,036,654 | 6,402 | genzgd | This seems broken still for shards with many replicas? Maybe as a result of the new storage symlinks? Also this particular cluster uses the new cluster <secret> authentication
```Bad path syntax: /opt/data/clickhouse/store/216/216bb68e-7864-42f9-8192-cb2553df0366/default@cdn%2Dmaple%2Dch2%2D059%2Eprod%2Ecomcast%2E... | 2021-01-28T13:00:59 | 2021-01-28T13:00:59 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 769,541,078 | 15,753 | bharatnc | > @bharatnc It depends on the browser window width. With width about 1200px or lower, the link becomes not clickable at all.
>
> This still considered a major bug.
Right, I see it now. | 2021-01-29T02:59:39 | 2021-01-29T02:59:39 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 769,556,093 | 13,644 | den-crane | > how to use LocalDateTime?
https://github.com/ClickHouse/clickhouse-jdbc/issues | 2021-01-29T03:49:24 | 2021-01-29T03:49:24 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 769,610,254 | 19,794 | alexey-milovidov | Looks like it does not reproduce on 20.11 but reproduces on 20.12 and master.
(at least if I use 100 concurrent connecting threads) | 2021-01-29T06:37:03 | 2021-01-29T06:43:33 | {} | MEMBER |
ClickHouse | ClickHouse | 768,883,554 | 19,487 | alexey-milovidov | I have to validate this message manually. It's suspected that it can output two consecutive whitespaces. | 2021-01-28T08:22:35 | 2021-01-28T08:22:35 | {} | MEMBER |
ClickHouse | ClickHouse | 768,909,159 | 19,751 | alexey-milovidov | Fast test Ok.
| 2021-01-28T09:07:15 | 2021-01-28T09:07:15 | {} | MEMBER |
ClickHouse | ClickHouse | 769,272,467 | 9,363 | lazjake | i'm running into the same issue, except returning long text fields truncated to 1023 chars. and no luck adding `Max...Size` options to odbc.ini. on clickhouse version 20.2.12.112. any luck figuring this one out?
| 2021-01-28T18:06:39 | 2021-01-28T18:06:39 | {} | NONE |
ClickHouse | ClickHouse | 769,398,057 | 6,402 | filimonov | Check https://github.com/ClickHouse/ClickHouse/pull/16728
On older versions - just enable the setting. | 2021-01-28T21:11:44 | 2021-01-28T21:11:44 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 769,525,640 | 8,159 | ys-zhaoyh | How do I use kerberos authorized HDFS engine in HA mode?@wbigdata | 2021-01-29T02:10:42 | 2021-01-29T02:10:42 | {} | NONE |
ClickHouse | ClickHouse | 769,586,468 | 19,487 | alexey-milovidov | @ucasFL
```
SELECT * FROM system.asynchronous_metrics ORDER BY name
maybe you meant: '['metric']' '['value']'.
```
There are three wrong things:
1. Double whitespace.
2. Lack of comma between items.
3. Bogus format - the items are wrapped into array of a single value for unknown reason.
Ideally it sh... | 2021-01-29T05:34:19 | 2021-01-29T05:34:26 | {} | MEMBER |
ClickHouse | ClickHouse | 769,607,518 | 19,794 | alexey-milovidov | Cannot reproduce on remote server in Yandex.Cloud. | 2021-01-29T06:29:32 | 2021-01-29T06:29:32 | {} | MEMBER |
ClickHouse | ClickHouse | 768,991,851 | 19,690 | FavonianKong | > run "disable pretty", run "enable pretty", provide output.
@ilejn
Here is the output:
```
(gdb) disable pretty
1 printer disabled
0 of 1 printers enabled
----------------------------
(gdb) enable pretty
1 printer enabled
1 of 1 printers enabled
``` | 2021-01-28T11:32:09 | 2021-01-28T11:32:40 | {} | NONE |
ClickHouse | ClickHouse | 769,577,178 | 19,769 | hombit | Thank you for the fast fix! | 2021-01-29T05:05:32 | 2021-01-29T05:05:32 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 768,879,068 | 18,662 | bohutang | @zhang2014 ping for help | 2021-01-28T08:13:32 | 2021-01-28T08:13:32 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 768,880,120 | 19,702 | alexey-milovidov | > AST fuzzer (TSan)
#19704 | 2021-01-28T08:15:45 | 2021-01-28T08:15:45 | {} | MEMBER |
ClickHouse | ClickHouse | 768,966,979 | 19,690 | FavonianKong | > If you see more, you probably use wrong pretty printers.
@ilejn
I use [this one](https://github.com/koutheir/libcxx-pretty-printers)，which is suggested by alexey.
the `gdbinit` file is：
```
set print thread-events off
python ... | 2021-01-28T10:47:22 | 2021-01-28T11:13:45 | {} | NONE |
ClickHouse | ClickHouse | 769,272,035 | 19,234 | qoega | Another one in GRPC.
```
ESC[36mnode_1 |ESC[0m ==7==WARNING: MemorySanitizer: use-of-uninitialized-value
ESC[36mnode_1 |ESC[0m #0 0x3e618d61 in pollable_process_events(grpc_pollset*, pollable*, bool) obj-x86_64-linux-gnu/../contrib/grpc/src/core/lib/iomgr/ev_epollex_linux.cc:880:9
ESC[36mnode_1 |ESC[0m ... | 2021-01-28T18:05:57 | 2021-01-28T18:05:57 | {} | MEMBER |
ClickHouse | ClickHouse | 769,410,085 | 19,709 | kuraga | @alexey-milovidov ,
> Maybe you can provide more details?
```sql
CREATE TABLE InputTable
(
ts DateTime('Europe/Moscow'),
zid1 UInt32,
zid2 UInt32,
i1 UInt32,
i2 UInt32,
...
)
ENGINE = AggregatingMergeTree()
ORDER BY (ts, zid1, zid2)
SAMPLE BY (ts, zid1, zid2);
```
```sql
CREATE TABLE ... | 2021-01-28T21:28:02 | 2021-01-28T21:28:02 | {} | NONE |
ClickHouse | ClickHouse | 769,412,433 | 19,668 | kitaisreal | Fixed in https://github.com/ClickHouse/ClickHouse/issues/19668. | 2021-01-28T21:31:33 | 2021-01-28T21:31:33 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 769,412,784 | 19,517 | kitaisreal | @MaxJPowers can we close this ? | 2021-01-28T21:32:12 | 2021-01-28T21:32:12 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 769,532,444 | 19,749 | ClownfishYang | > MATERIALIZED VIEW is an insert trigger.
> But CH does not have transactions / atomicity in Table -> Materialized View.
>
> 1. client sends `INSERT INTO test_1 VALUES('id2', '1A', 2);`
> 2. INSERT handler gets an inserted buffer
> 3. INSERT handler writes the inserted buffer into table test_1
> 4. **INSERT hand... | 2021-01-29T02:31:40 | 2021-01-29T02:31:40 | {} | NONE |
ClickHouse | ClickHouse | 769,567,309 | 19,791 | alexey-milovidov | > AST fuzzer (ASan) — 0x000007d1f33e is located 2 bytes to the left of global variable '_pcre_ucd_stage
https://github.com/ClickHouse/ClickHouse/pull/19720 | 2021-01-29T04:29:51 | 2021-01-29T04:29:51 | {} | MEMBER |
ClickHouse | ClickHouse | 768,867,754 | 19,407 | templarzq | @alexey-milovidov @qoega @CurtizJ would you please give any suggestion? | 2021-01-28T07:49:32 | 2021-01-28T07:49:32 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 768,904,957 | 19,584 | alexey-milovidov | Ok, there was also one failure of the test for roaring bitmaps that I also disabled for Arcadia in master. | 2021-01-28T09:00:39 | 2021-01-28T09:00:39 | {} | MEMBER |
ClickHouse | ClickHouse | 768,928,182 | 19,719 | inkrement | Indeed, the flag solved the problem! | 2021-01-28T09:40:20 | 2021-01-28T09:40:20 | {} | NONE |
ClickHouse | ClickHouse | 768,959,442 | 18,422 | Enmk | Failed to demonstrate any improvement in performance, closing for now | 2021-01-28T10:34:36 | 2021-01-28T10:34:36 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 769,063,698 | 19,746 | l1t1 | > > FastMemcpy
>
> This library should not be built on AArch64.
need the official non debug info release | 2021-01-28T13:47:21 | 2021-01-28T13:47:21 | {} | NONE |
ClickHouse | ClickHouse | 769,093,670 | 19,702 | alesapin | ```
/// It's possible in rare case:
/// 1) We have tried to execute merge but don't have all source parts
/// 2) So we tried to fetch the already merged part, but this part doesn't exist yet (nobody merged it yet)
/// 3) We asked part checked thread to... | 2021-01-28T14:27:22 | 2021-01-28T14:27:22 | {} | MEMBER |
ClickHouse | ClickHouse | 769,153,749 | 19,709 | kuraga | > Also you can lower the `aggregation_memory_efficient_merge_threads` - it will give a tradeoff between speed and memory usage.
Yes, it works! Thanks. | 2021-01-28T15:14:49 | 2021-01-28T15:14:49 | {} | NONE |
ClickHouse | ClickHouse | 769,197,446 | 19,715 | akuzm | The test failures are from master. | 2021-01-28T16:16:37 | 2021-01-28T16:16:37 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 768,880,229 | 19,702 | alexey-milovidov | > AST fuzzer (MSan)
https://github.com/ClickHouse/ClickHouse/issues/19756 | 2021-01-28T08:15:59 | 2021-01-28T08:15:59 | {} | MEMBER |
ClickHouse | ClickHouse | 768,941,670 | 19,108 | KochetovNicolai | I suppose it is ok have stack overflow in coroutine if it's depth ~600. Now I use 128KB, so it's about 200 byte per frame.
Also, last page of stack is protected:
https://github.com/ClickHouse/ClickHouse/blob/cf085b0687d72ec9e9f3a49be0f4084e83b5d420/src/Common/FiberStack.h#L46-L46
Probably, we can handle this so so... | 2021-01-28T10:03:42 | 2021-01-28T10:03:42 | {} | MEMBER |
ClickHouse | ClickHouse | 769,241,066 | 19,703 | alexey-milovidov | The info you posted indicates that our ARM64 build is broken somehow (exception handling does not work in some cases), similar to https://github.com/ClickHouse/ClickHouse/issues/15638 | 2021-01-28T17:17:35 | 2021-01-28T17:17:35 | {} | MEMBER |
ClickHouse | ClickHouse | 769,519,957 | 5,747 | ys-zhaoyh | How do I use kerberos authorized HDFS engine in HA mode?@alexey-milovidov | 2021-01-29T01:55:06 | 2021-01-29T01:55:06 | {} | NONE |
ClickHouse | ClickHouse | 768,909,119 | 19,703 | lurenjia528 | > You should type `thread apply all backtrace` in gdb after you receive the `Aborted.` message.
sorry , I don't use gdb often.
```bash
root@node-4:/nas/ygt/clickhouse# gdb clickhouse
GNU gdb (Ubuntu 7.11.1-0kord1~16.5) 7.11.1
Copyright (C) 2016 Free Software Foundation, Inc.
License GPLv3+: GNU GPL version 3... | 2021-01-28T09:07:11 | 2021-01-28T09:07:11 | {} | NONE |
ClickHouse | ClickHouse | 768,939,134 | 19,746 | socketpair | First difference I found is:
```
my: Default libraries: -nodefaultlibs /usr/lib64/clang/11.0.0/lib/linux/libclang_rt.builtins-aarch64.a -lc -lm -lrt -lpthread -ldl
official: Default libraries: -nodefaultlibs -lgcc -lc -lm -lrt -lpthread -ldl
```
gcc ? what ? | 2021-01-28T09:59:21 | 2021-01-28T09:59:32 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 769,061,528 | 19,684 | tavplubix | Take a look at `*.sh` tests (such as `01150_ddl_guard_rwr` and `00993_system_parts_race_condition_drop_zookeeper`). You can run two threads, one thread will execute `CREATE` and `DROP` and other thread will execute `ALTER ... ON CLUSTER`. | 2021-01-28T13:44:06 | 2021-01-28T13:44:06 | {} | MEMBER |
ClickHouse | ClickHouse | 769,511,203 | 19,703 | lurenjia528 | > The info you posted indicates that our ARM64 build is broken somehow (exception handling does not work in some cases), similar to #15638
thanks | 2021-01-29T01:29:04 | 2021-01-29T01:29:04 | {} | NONE |
ClickHouse | ClickHouse | 769,508,477 | 19,407 | templarzq | risk:if there are insertMergeFrom operations after copy, this pr will cause wrong query result. | 2021-01-29T01:22:16 | 2021-01-29T01:51:58 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 768,864,612 | 19,671 | alexey-milovidov | Also, if the first table works fast enough for all types of queries you may not need to duplicate data in the second table. | 2021-01-28T07:42:15 | 2021-01-28T07:42:15 | {} | MEMBER |
ClickHouse | ClickHouse | 768,907,795 | 19,690 | alexey-milovidov | The test program most likely was compiled with gcc's libstdc++ (it is the default on Linux) while ClickHouse is using libc++.
I don't know why pretty-printers don't work in your case - actually I don't use it by myself but some of my collegues managed to have success with it. | 2021-01-28T09:05:00 | 2021-01-28T09:05:00 | {} | MEMBER |
ClickHouse | ClickHouse | 768,970,602 | 19,745 | zhang2014 | Maybe consider the following DDL(I'm not sure if it's better):
```
ALTER DATABASE `db_name` STOP REPLICATE;
ALTER DATABASE `db_name` START REPLICATE;
ALTER DATABASE `db_name` RESTART REPLICATE;
ALTER DATABASE `db_name` ADD IGNORE TABLES `table_name` | (`table_name_1`, `table_name_2`);
``` | 2021-01-28T10:54:17 | 2021-01-28T10:54:55 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 769,149,708 | 19,706 | qoega | Incorporates changes from #19697 | 2021-01-28T15:09:12 | 2021-01-28T15:09:12 | {} | MEMBER |
ClickHouse | ClickHouse | 769,152,220 | 19,768 | den-crane | yes, the only way now:
```sql
SELECT (toDateTime('2020-01-01 00:00:00', 'America/New_York') AS x) - toDateTime(toString(x), 'UTC') AS y
┌─────y─┐
│ 18000 │
└───────┘
``` | 2021-01-28T15:12:47 | 2021-01-28T15:12:47 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 769,164,489 | 19,753 | den-crane | All `*MergeTree` engines (except plain the MergeTree) work this way.
They have not fully aggregated data **ALWAYS**.
Every insert adds rows, every merge **can** reduce number of rows.
Select COUNT(*) does not have practical sense against these engines.
Example:
```sql
create table aggtest(
user_id UI... | 2021-01-28T15:29:18 | 2021-01-28T15:30:19 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 769,240,295 | 19,703 | alexey-milovidov | Does it help to disable `mysql_port` in config? | 2021-01-28T17:16:30 | 2021-01-28T17:16:30 | {} | MEMBER |
ClickHouse | ClickHouse | 769,260,088 | 19,724 | alexey-milovidov | Should not it be printed only in interactive mode? | 2021-01-28T17:47:28 | 2021-01-28T17:47:28 | {} | MEMBER |
ClickHouse | ClickHouse | 769,608,954 | 8,159 | ys-zhaoyh | I have these two documents.
I'm using the latest stable version(21.1.2.15).
Maybe I need to look at the configuration file again Or I can show you my configuration file.
At 2021-01-29 14:24:20, "filimonov" <notifications@github.com> wrote:
You need to use 21.1 or newer
Check
https://clickhouse.tech/docs/en/engines... | 2021-01-29T06:33:25 | 2021-01-29T06:33:25 | {} | NONE |
ClickHouse | ClickHouse | 723,693,826 | 12,297 | alexey-milovidov | Does not reproduce in version 20.12. | 2020-11-09T00:39:55 | 2020-11-09T00:39:55 | {} | MEMBER |
ClickHouse | ClickHouse | 723,695,348 | 10,505 | alexey-milovidov | It is interesting that we don't use `tcmalloc` in ClickHouse.
The message `src/tcmalloc.cc:283] Attempt to free invalid pointer 0x7f3ef801f000` is unrelated to ClickHouse.
Even if we'd use `tcmalloc` you would not get this message in release builds.
Conclusion: it is the bug of Google Colab. | 2020-11-09T00:46:33 | 2020-11-09T00:47:31 | {} | MEMBER |
ClickHouse | ClickHouse | 723,710,168 | 7,963 | zhang2014 | looks like the bloom filter does not support the `has` function on version 19.16.1.1
for more information:
https://github.com/ClickHouse/ClickHouse/pull/6984/commits/b25151b61beded046494cc262ef621d3df666d4e
commit tags:
v19.16.19.85-stable v19.16.17.80-stable v19.16.16.72-stable v19.16.15.69-stable v19.16.14.65-s... | 2020-11-09T01:56:00 | 2020-11-09T01:59:30 | {} | CONTRIBUTOR |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.