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 | 523,718,694 | 6,597 | alexey-milovidov | Please write a description for changelog that is appropriate for users (rather than developers). | 2019-08-22T02:17:13 | 2019-08-22T02:17:13 | {} | MEMBER |
ClickHouse | ClickHouse | 523,857,443 | 6,454 | akonyaev90 | common part for parquetBlockInput and orcBlockInput was migrated to ArrowColumnToCHColumn class | 2019-08-22T11:03:21 | 2019-08-22T11:03:21 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 524,018,646 | 5,417 | amosbird | > This brings me to a question about the dispatch() method. The main thing it does is switching based on the key length. Then, each branch of the switch transforms the key and calculates the hash, independently from other branches.
I used to implement it in that way, but the key dispatching shows up in perf hot spo... | 2019-08-22T18:16:07 | 2019-08-22T18:16:30 | {} | COLLABORATOR |
ClickHouse | ClickHouse | 524,139,923 | 6,422 | alexey-milovidov | @proller Please merge with master - because no tests have run. | 2019-08-23T01:44:14 | 2019-08-23T01:44:14 | {} | MEMBER |
ClickHouse | ClickHouse | 524,189,855 | 6,608 | filimonov | Most probably relates to #5828 and #5922
Now clickhouse resolves symlinks during default timezone detection.
/cc @alexey-milovidov
| 2019-08-23T06:27:59 | 2019-08-23T06:27:59 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 524,278,216 | 6,626 | 4ertus2 | It's implementations specific. Datatypes that have problems with fixed string representation are hached. | 2019-08-23T11:20:38 | 2019-08-23T11:20:38 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 524,347,280 | 6,625 | alexey-milovidov | You have pasted article inside symlink path. | 2019-08-23T14:57:11 | 2019-08-23T14:57:11 | {} | MEMBER |
ClickHouse | ClickHouse | 523,493,618 | 6,595 | gasabr | There are 2 `group_id`s and something about 8 distinct `idxid`s, totaling in 23M records. In the real world scenario there would be 10k+ groups with, each containing thousands of unique `idxid`s resulting in 40M+ records.
Query:
```
SELECT
count(*) total,
source_id,
countIf(liveness, equals... | 2019-08-21T14:53:53 | 2019-08-21T14:53:53 | {} | NONE |
ClickHouse | ClickHouse | 523,744,194 | 6,603 | alexey-milovidov | This fixes #6563 | 2019-08-22T04:37:10 | 2019-08-22T04:37:10 | {} | MEMBER |
ClickHouse | ClickHouse | 524,048,290 | 6,611 | alexey-milovidov | That's the expected behaviour.
Empty string doesn't read as NULL. To put NULL into TSV file, you have to write `\N`:
```
echo -en "1\ta\t0.1\n2\tb\t\\N" | ch -q 'insert into xx format TSV'
``` | 2019-08-22T19:42:07 | 2019-08-22T19:42:07 | {} | MEMBER |
ClickHouse | ClickHouse | 524,281,547 | 6,594 | WestDragon | `SELECT
DISTINCT general_number
FROM
invoice.item_test
WHERE like(lcase(name),lcase('%каша%'))
LIMIT 100000
`
Code: 246, e.displayText() = DB::Exception: Bad size of marks file '/var/lib/clickhouse/data/invoice/item_test/201607_1_1821_3/skp_idx_name_search.mrk2': 3936, must be: 4824 (v... | 2019-08-23T11:31:42 | 2019-08-23T11:33:33 | {} | NONE |
ClickHouse | ClickHouse | 524,299,275 | 6,627 | alexey-milovidov | Does it reproduce in official build? | 2019-08-23T12:37:17 | 2019-08-23T12:37:17 | {} | MEMBER |
ClickHouse | ClickHouse | 524,468,344 | 6,462 | simPod | The delete works for me. | 2019-08-23T21:29:19 | 2019-08-23T21:29:19 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 524,472,984 | 6,462 | alexey-milovidov | The mechanics of ALTER DELETE and TTLs is different. | 2019-08-23T21:48:39 | 2019-08-23T21:48:39 | {} | MEMBER |
ClickHouse | ClickHouse | 523,508,482 | 6,595 | den-crane | it's another query.
>PREWHERE group_id = 4 GROUP BY source_id;
VS
>WHERE idxid == '...' GROUP BY idxid;
statistics is useless
| 2019-08-21T15:26:51 | 2019-08-21T15:26:51 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 523,718,556 | 6,597 | alexey-milovidov | Is this the reason for duplicates after a normal restart? | 2019-08-22T02:16:27 | 2019-08-22T02:16:27 | {} | MEMBER |
ClickHouse | ClickHouse | 523,734,006 | 6,563 | weiqxu | @alexey-milovidov We also remove the DT_HASH branch as the work round.In my env(centos 7.3) , I found the GCC will use DT_HASH, and clang use DT_GNU_HASH as default. Which compiler was used by the offical build?
And I'm not sure is the DT_HASH symbol is generatored by the JIT or the origin clickhouse executable pac... | 2019-08-22T03:40:24 | 2019-08-22T03:40:24 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 523,479,280 | 6,575 | yingfeng | More information:
Table `dist_trips2` has two shards with following schema:
```
DESCRIBE TABLE dist_trips2
┌─name────────────┬─type─────┬─default_type─┬─default_expression─┬─comment─┬─codec_expression─┬─ttl_expression─┐
│ id │ UInt64 │ │ │ │ ... | 2019-08-21T14:22:03 | 2019-08-22T05:52:53 | {} | NONE |
ClickHouse | ClickHouse | 523,830,064 | 6,597 | abyss7 | > Is this the reason for duplicates after a normal restart?
It's a very probable reason, but I'm still trying to reproduce it. | 2019-08-22T09:38:15 | 2019-08-22T09:38:15 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 524,029,832 | 6,589 | 4ertus2 | > BTW distributed_product_mode affects only on subqueries.
OK. Any case, there's a hask that rewrites table into subselect for push down into join. It's possible to remove this restriction if really needed. | 2019-08-22T18:48:02 | 2019-08-22T18:48:02 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 524,047,298 | 5,417 | alexey-milovidov | > probably because crc32 has more collisions than a proper hash function
Collisions don't lead to more resizes. | 2019-08-22T19:39:33 | 2019-08-22T19:39:33 | {} | MEMBER |
ClickHouse | ClickHouse | 524,170,049 | 5,925 | alexey-milovidov | Fixed in master. | 2019-08-23T04:42:36 | 2019-08-23T04:42:36 | {} | MEMBER |
ClickHouse | ClickHouse | 524,307,268 | 6,627 | den-crane | 19.13.3.26
Linux xxxx 4.9.0-4-amd64 #1 SMP Debian 4.9.65-3 (2017-12-03) x86_64 GNU/Linux
Also crashes
---------------------------
And 19.14.1.1144 (official build) (Debian 3.16.0-4-amd64) crashes too | 2019-08-23T13:03:44 | 2019-08-23T13:11:37 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 524,499,454 | 6,645 | alexey-milovidov | Three tests have failed instead. | 2019-08-24T00:31:13 | 2019-08-24T00:31:13 | {} | MEMBER |
ClickHouse | ClickHouse | 523,992,385 | 6,598 | BayoNet | @blinkov Look at the updated version, please. RU translation I'll update in the future pull about T64 codec. | 2019-08-22T17:05:44 | 2019-08-22T17:05:44 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 524,140,155 | 6,343 | alexey-milovidov | > UPD: Not sure if i should fix style of a huge table (Style check — Found 101 errors)
Sure, you have to. | 2019-08-23T01:45:29 | 2019-08-23T01:45:29 | {} | MEMBER |
ClickHouse | ClickHouse | 524,218,114 | 469 | filimonov | https://github.com/yandex/ClickHouse/pull/6055 | 2019-08-23T08:08:45 | 2019-08-23T08:08:45 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 524,428,600 | 6,641 | den-crane | Yes, I think these results should be consistent (the same) a1=a2, b1=b2
null:
```
a1)
select count(), count(a), max(a), min(a), avg(a), sum(a)
from (select cast(Null,'Nullable(Float64)') a)
┌─count()─┬─count(a)─┬─max(a)─┬─min(a)─┬─avg(a)─┬─sum(a)─┐
│ 1 │ 0 │ ᴺᵁᴸᴸ │ ᴺᵁᴸᴸ │ ᴺᵁᴸᴸ │ ᴺᵁ... | 2019-08-23T19:06:22 | 2019-08-23T19:27:08 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 524,442,139 | 5,097 | Akazz | @zhang2014
> ClickHouse does not provide "snapshot isolation" for queries
This above is too strong a statement and of course we have important exceptions to it: CH provides snapshot isolation for *MergeTree as well as for Memory, Log/StripeLog engines (due to the fact that new data is only being appended).
A f... | 2019-08-23T19:54:57 | 2019-08-23T19:54:57 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 523,621,761 | 5,680 | abyss7 | Should be fixed in 19.14+. Possibly was backported to 19.11 and 19.13 | 2019-08-21T19:49:30 | 2019-08-21T19:49:30 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 523,878,506 | 6,591 | dimarub2000 | Reopened because of github issues | 2019-08-22T12:11:57 | 2019-08-22T12:11:57 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 524,147,476 | 5,925 | alexey-milovidov | It doesn't work:
```
milovidov-Latitude-E7470 :) SELECT number, neighbor(number, number) FROM system.numbers LIMIT 10
SELECT
number,
neighbor(number, number)
FROM system.numbers
LIMIT 10
┌─number─┬─neighbor(number, number)─┐
│ 0 │ 0 │
│ 1 │ ... | 2019-08-23T02:25:28 | 2019-08-23T02:25:28 | {} | MEMBER |
ClickHouse | ClickHouse | 524,275,396 | 6,627 | 4ertus2 | No crash in debug master build. Need some investigation. | 2019-08-23T11:11:00 | 2019-08-23T11:11:00 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 524,440,958 | 6,638 | SaltTan | ```
$ addr2line -afiCe /usr/bin/clickhouse 0x7f6deaa 0x7434b5e 0x73a6aeb 0x734390c
0x0000000007f6deaa
CityHash_v1_0_2::CityHash64(char const*, unsigned long)
/build/obj-x86_64-linux-gnu/../contrib/cityhash102/src/city.cc:205
0x0000000007434b5e
StringRefHash64::operator()(StringRef) const
/build/obj-x86_64-linux-... | 2019-08-23T19:50:28 | 2019-08-23T19:50:28 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 523,493,000 | 6,563 | alexey-milovidov | @amosbird has helped me to check that it does reproduce even in official binary.
```
Alexey Milovidov, [21.08.19 08:06]
About SymbolIndex issue...
Does the issue reproduce if you remove
if (it->d_tag == DT_HASH)
branch and leave only
if (it->d_tag == DT_GNU_HASH)
?
Another question: does it reproduce wit... | 2019-08-21T14:52:29 | 2019-08-21T14:52:29 | {} | MEMBER |
ClickHouse | ClickHouse | 523,593,535 | 6,591 | alexey-milovidov | Please fill the description. | 2019-08-21T18:32:37 | 2019-08-21T18:32:37 | {} | MEMBER |
ClickHouse | ClickHouse | 524,000,847 | 5,891 | dimarub2000 | Can't reproduce the problem
`dimarub2000@virtual:~/ClickHouse/build/dbms/programs$ ./clickhouse-client -q "insert into mtree format CSV" < ~/common/kek.csv `
```
Received exception from server (version 19.14.1):
Code: 252. DB::Exception: Received from localhost:9000. DB::Exception: Too many partitions for single ... | 2019-08-22T17:28:40 | 2019-08-22T17:29:52 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 524,259,762 | 6,621 | bkarlson | Пардон за неочевидность, ошибка при переливке (и экпорте) данных — каждый раз, при выборке — спорадически.
Альтеров не было, просто добавляются новые данные каждые 10 минут, и всё работало в таком режиме на протяжении полугода без малейших проблем. Сегодня утром зашёл в аналитику и увидел пустую страницу, пошёл раз... | 2019-08-23T10:15:51 | 2019-08-23T10:15:51 | {} | NONE |
ClickHouse | ClickHouse | 524,297,537 | 3,490 | alexey-milovidov | Does it reproduce with the same test case in recent ClickHouse versions (master, 19.13. 19.11)?
How big is performance difference? | 2019-08-23T12:31:21 | 2019-08-23T12:31:21 | {} | MEMBER |
ClickHouse | ClickHouse | 524,398,750 | 6,627 | 4ertus2 | Crash cause of stack corruption here https://github.com/yandex/ClickHouse/blob/master/dbms/src/Common/OptimizedRegularExpression.cpp#L416
MAX_SUBPATTERNS = 5,
limit = 6
No need for visitParamExtractString(). extractAll() crashes itself.
```
SELECT extractAll('Mozilla/5.0 (Windows NT 10.0; WOW64) AppleWebKit/53... | 2019-08-23T17:31:45 | 2019-08-23T17:40:10 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 524,408,032 | 6,381 | dimarub2000 | Reproduced the issue, trying to find the problem.
when `select * from mysql('localhost:3306', 'test', 'kek', 'dsrub', 'kek') where id=123 and 1=1;`
mysql gets ```SELECT `s`, `id` FROM `test`.`kek` WHERE (`id` = 123) AND (1 = 1)```
when `select * from mysql('localhost:3306', 'test', 'kek', 'dsrub', 'kek') where lef... | 2019-08-23T18:01:05 | 2019-08-23T18:05:46 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 524,534,735 | 6,648 | filimonov | Related / similar #5318 | 2019-08-24T09:10:53 | 2019-08-24T09:10:53 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 523,717,980 | 6,422 | alexey-milovidov | @abyss7 ? | 2019-08-22T02:13:27 | 2019-08-22T02:13:27 | {} | MEMBER |
ClickHouse | ClickHouse | 524,178,764 | 6,562 | MetikovVadim | Now i try to count the INSERT ... SELECT ... for both tables command time. | 2019-08-23T05:33:57 | 2019-08-23T05:33:57 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 524,480,559 | 6,636 | nvartolomei | Or, make this part of resource pools feature, which ideally would be hierarchical. One pool for critical queries with say limit of 10 concurrent requests. Another pool for production workload with say limit of 100, with nested pools to split up this production workload between services if necessary. | 2019-08-23T22:23:08 | 2019-08-23T22:23:08 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 524,537,841 | 5,318 | alexey-milovidov | I also have in my stash:
- randomString(length, [seed]) - random binary string of specified length,
- fuzzBits(s, prob, [seed]) - alternate bits in a string with desired probability,
- fuzzBytes(s, prob, [seed]) - alternate bytes in a string with desired probability. | 2019-08-24T09:56:35 | 2019-08-24T09:59:16 | {} | MEMBER |
ClickHouse | ClickHouse | 523,591,071 | 6,585 | alexey-milovidov | @proller How to separate different `adduser` programs in a more sane way?
Maybe we need to use `useradd` instead of `adduser`? | 2019-08-21T18:26:25 | 2019-08-21T18:26:25 | {} | MEMBER |
ClickHouse | ClickHouse | 523,593,445 | 6,591 | alexey-milovidov | Does it help? | 2019-08-21T18:32:25 | 2019-08-21T18:32:25 | {} | MEMBER |
ClickHouse | ClickHouse | 523,830,142 | 6,597 | abyss7 | > Please write a description for changelog that is appropriate for users (rather than developers).
Ok, sure | 2019-08-22T09:38:31 | 2019-08-22T09:38:31 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 524,157,994 | 6,405 | qietingfengling | yes ,the DEFAULT database ,I use my database not this problem | 2019-08-23T03:24:56 | 2019-08-23T03:24:56 | {} | NONE |
ClickHouse | ClickHouse | 523,704,035 | 6,563 | yingfeng | It seems this corruption could be triggered more easily when I pulled codes to latest master:
For example, I have a distributed table `dist_trips2` having two shards with following schema:
```
DESCRIBE TABLE dist_trips2
┌─name────────────┬─type─────┬─default_type─┬─default_expression─┬─comment─┬─codec_expressio... | 2019-08-22T00:58:16 | 2019-08-22T01:32:55 | {} | NONE |
ClickHouse | ClickHouse | 523,922,311 | 6,589 | den-crane | @4ertus2 BTW distributed_product_mode affects only on subqueries.
<del>SELECT * FROM table1 ANY LEFT JOIN table2 SETTINGS distributed_product_mode=LOCAL<del>
SELECT * FROM table1 ANY LEFT JOIN (select * from table2) SETTINGS distributed_product_mode=LOCAL
And distributed_product_mode makes sense for the case o... | 2019-08-22T14:08:02 | 2019-08-22T14:11:52 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 524,140,468 | 6,514 | alexey-milovidov | https://clickhouse-test-reports.s3.yandex.net/6343/086dd815ac19070e119643929164663aa196c035/functional_stateless_tests_(ubsan).html | 2019-08-23T01:47:10 | 2019-08-23T01:47:10 | {} | MEMBER |
ClickHouse | ClickHouse | 524,259,409 | 6,620 | blinkov | merged in #6625 | 2019-08-23T10:14:39 | 2019-08-23T10:14:39 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 524,280,598 | 6,615 | 4ertus2 | It's not easy in general.
```
SELECT count() FROM system.numbers WHERE number IN (SELECT abs(x) as y WHERE y < 0)
```
We could calculate function result ranges, append them to ASTIdentifiers and rewrite filters with this knowlage. It needs aliases map, const argument detection and ranges for all functions (with res... | 2019-08-23T11:28:20 | 2019-08-23T12:02:01 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 524,321,452 | 6,621 | bkarlson | В общем, из бэкапа и с накатом пропущенных данных всё заработало, если будут проблемы, переоткрою issue. | 2019-08-23T13:46:09 | 2019-08-23T13:46:09 | {} | NONE |
ClickHouse | ClickHouse | 524,328,555 | 6,422 | abyss7 | > https://a.yandex-team.ru/review/912102
There are still 2 build errors in the 8th iteration. | 2019-08-23T14:05:50 | 2019-08-23T14:05:50 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 524,484,073 | 6,636 | alexey-milovidov | > Maybe instead add a option to skip max_concurrent_queries check?
The user will be able to set this limit to zero, that means unlimited. Also the user can set it to some high enough value like 1000 while still limiting other users to 100.
> Or, make this part of resource pools feature, which ideally would be hie... | 2019-08-23T22:42:30 | 2019-08-23T22:43:29 | {} | MEMBER |
ClickHouse | ClickHouse | 523,935,537 | 6,422 | proller | https://a.yandex-team.ru/review/912102 | 2019-08-22T14:40:42 | 2019-08-22T14:40:42 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 524,091,962 | 6,425 | alexey-milovidov | It was too early to merge.
https://clickhouse-test-reports.s3.yandex.net/0/a52c004b5b364d64fa2fd6c5ae521beda60d106d/stress_test_(thread).html
```
2019.08.23 00:00:17.715007 [ 937 ] {} <Fatal> BaseDaemon: ########################################
2019.08.23 00:00:17.716036 [ 937 ] {} <Fatal> BaseDaemon: (version ... | 2019-08-22T21:51:52 | 2019-08-22T21:51:52 | {} | MEMBER |
ClickHouse | ClickHouse | 524,149,228 | 5,925 | alexey-milovidov | ```
milovidov-Latitude-E7470 :) SELECT number, neighbor(number, number, 0) FROM system.numbers LIMIT 10
SELECT
number,
neighbor(number, number, 0)
FROM system.numbers
LIMIT 10
┌─number─┬─neighbor(number, number, 0)─┐
│ 0 │ 0 │
│ 1 │ 2 ... | 2019-08-23T02:35:07 | 2019-08-23T02:35:07 | {} | MEMBER |
ClickHouse | ClickHouse | 524,329,093 | 6,422 | proller | because macos build broken and one internal problem not solved | 2019-08-23T14:07:18 | 2019-08-23T14:07:30 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 524,421,725 | 6,644 | 4ertus2 | fixes #6627 | 2019-08-23T18:43:52 | 2019-08-23T18:43:52 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 523,500,217 | 5,937 | vitlibar | > Is it fixed by patching simdjson? [lemire/simdjson#276](https://github.com/lemire/simdjson/pull/276)
Yes. For example, https://clickhouse-test-reports.s3.yandex.net/0/0e896fb341f5474b0e1c6a2a3dfeea83ab61655e/performance_test/comparison_min_time_gcc_9.html
test | rapidjson (s) | simdjson - avx2 (s)
----- | ----... | 2019-08-21T15:09:04 | 2019-08-21T15:10:55 | {
"+1": 1
} | MEMBER |
ClickHouse | ClickHouse | 523,621,327 | 5,545 | abyss7 | Should be fixed in 19.14+. Possibly was backported to 19.13 and 19.11 | 2019-08-21T19:48:15 | 2019-08-21T19:48:15 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 523,717,489 | 6,596 | alexey-milovidov | Your own test has failed. | 2019-08-22T02:10:57 | 2019-08-22T02:10:57 | {} | MEMBER |
ClickHouse | ClickHouse | 523,763,501 | 6,589 | filimonov | Sharding schema is not need to be global. You can use different sharding (or replication) schema from each table.
So for example you can create `table2_local` as replicated across all shards (just don't add `{shard}` macro in zookeeper path for ReplicatedMergeTree).
Additionally you can also configure one more c... | 2019-08-22T06:11:08 | 2019-08-22T06:11:37 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 523,511,914 | 5,029 | dengwanc | seems there is no such way parse something like this
`[{"a": 1, "b": 2}, {"a":1, "b":3}]`
@alexey-milovidov shoule we have some function like this
```
select JSONExtractArrayRaw('[{"a": 1, "b": 2}, {"a":1, "b":3}]')
=>
Array of Raw String
['{"a": 1, "b": 2}', '{"a":1, "b":3}']
```
| 2019-08-21T15:33:54 | 2019-08-22T14:42:41 | {} | NONE |
ClickHouse | ClickHouse | 524,217,500 | 6,611 | filimonov | Related / duplicate: https://github.com/yandex/ClickHouse/issues/469 | 2019-08-23T08:06:53 | 2019-08-23T08:06:53 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 524,261,523 | 6,621 | 4ertus2 | > Запрос, хотя мне кажется дело просто в повреждении данных
Скорее всего. Стоит проверить размеры и чексуммы файлов c данными и фалов индексов руками. | 2019-08-23T10:21:56 | 2019-08-23T10:21:56 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 524,267,946 | 3,490 | lapkofear | Unfortunately this bug is still reproduced. Not as critical as when the bug was opened, but still seriously affects performance | 2019-08-23T10:45:41 | 2019-08-23T11:56:12 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 524,382,590 | 6,627 | 4ertus2 | gcc build crashes | 2019-08-23T16:39:34 | 2019-08-23T16:39:34 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 523,505,712 | 6,597 | abyss7 | Should fix #6414 | 2019-08-21T15:20:59 | 2019-08-21T15:20:59 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 523,744,696 | 6,563 | alexey-milovidov | @weiqxu Thank you for investigation!
> I found the GCC will use DT_HASH, and clang use DT_GNU_HASH as default. Which compiler was used by the offical build?
We use gcc for official builds. But happy enough, it (actually - the linker that is used) creates both types of hash tables:
```
$ readelf --sections ~/C... | 2019-08-22T04:40:01 | 2019-08-22T04:40:01 | {} | MEMBER |
ClickHouse | ClickHouse | 523,455,920 | 6,589 | 4ertus2 | You need distributed subqueries: https://clickhouse.yandex/docs/en/query_language/select/#select-distributed-subqueries
In example you may try:
```sql
SELECT * FROM table1 GLOBAL ANY LEFT JOIN table2
```
or place table2 on each shard and run
```sql
SELECT * FROM table1 ANY LEFT JOIN table2 SETTINGS distributed... | 2019-08-21T13:26:32 | 2019-08-22T06:02:39 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 524,260,354 | 215 | philodavies | @alexey-milovidov you say that it should be accepted, any clue why it was failing in my sample? | 2019-08-23T10:17:45 | 2019-08-23T10:17:45 | {} | NONE |
ClickHouse | ClickHouse | 524,445,415 | 6,422 | abyss7 | Are all those fixes in the master already? Otherwise we'll have to make them again in the next port to arcadia. | 2019-08-23T20:07:16 | 2019-08-23T20:07:50 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 524,468,297 | 6,647 | alexey-milovidov | CC @blinkov | 2019-08-23T21:29:09 | 2019-08-23T21:29:09 | {} | MEMBER |
ClickHouse | ClickHouse | 523,848,837 | 6,595 | gasabr | Sorry, misunderstood you the first time, here are stats:
```
PARTITION BY toYYYYMM(created) ORDER BY (group_id, idxid);
"statistics":
{
"elapsed": 0.3618612,
"rows_read": 5603328,
"bytes_read": 111997196
}
```
your suggested changes:
```
PARTITION BY (toYYYYMM(created)) ORDER BY (idxid, group_id)
"... | 2019-08-22T10:36:46 | 2019-08-22T10:36:46 | {} | NONE |
ClickHouse | ClickHouse | 523,878,468 | 6,422 | abyss7 | > @abyss7 Waiting for your review.
I have viewed the changes, but can't reliably check their purpose - there are no any build logs from `yamake`: link to the internal review would be appreciated. At least the CI builds should be restarted to make sure that release-branch isn't broken. | 2019-08-22T12:11:48 | 2019-08-22T12:11:48 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 524,199,360 | 6,562 | MetikovVadim | I created a reduced table with only two columns(Date Date , Path String). After merging i will try to insert into new tables and optimize them and record consumed time. | 2019-08-23T07:05:05 | 2019-08-23T07:05:05 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 524,298,692 | 6,615 | alexey-milovidov | You can start executing a query and return quickly from CreatingSetsBlockInputStream after realizing that the set is empty if it's IN or INNER/RIGHT JOIN. This is one-line change. | 2019-08-23T12:35:30 | 2019-08-23T12:35:30 | {} | MEMBER |
ClickHouse | ClickHouse | 524,464,093 | 6,646 | alexey-milovidov | It became obsolete with the following PR: #6435
It can be returned back whenever needed. | 2019-08-23T21:12:56 | 2019-08-23T21:12:56 | {} | MEMBER |
ClickHouse | ClickHouse | 524,468,064 | 6,462 | volfco | EDIT. Yeah. I'll make my own issue and the devs can link | 2019-08-23T21:28:13 | 2019-08-23T21:29:09 | {} | NONE |
ClickHouse | ClickHouse | 524,070,673 | 6,587 | alexey-milovidov | > select * from table where date >'2019-07-26'
It's already supported. | 2019-08-22T20:46:26 | 2019-08-22T20:46:26 | {} | MEMBER |
ClickHouse | ClickHouse | 524,190,645 | 208 | filimonov | One more possible approach: #6294 | 2019-08-23T06:31:14 | 2019-08-23T06:31:14 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 524,479,491 | 6,636 | nvartolomei | Maybe instead add a option to skip max_concurrent_queries check? Use cases include not just dba, but also metrics exporters (prometheus exporter) that need to query system tables.
At the moment if the limit is hit you lose visibility into ClickHouse metrics which is not ideal. | 2019-08-23T22:17:53 | 2019-08-23T22:17:53 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 524,488,645 | 6,625 | alexey-milovidov | @zhang2014 Could you please make another PR? | 2019-08-23T23:08:32 | 2019-08-23T23:08:32 | {} | MEMBER |
ClickHouse | ClickHouse | 524,012,342 | 5,417 | akuzm | Some news on the progress:
The PR is gradually taking the shape I had in mind. It is still more of a
prototype and has some rough edges, but the general outline looks plausible. I
split it into several commits. First, I changed the hash table interfaces to make it
simpler to support compound hash tables, and then... | 2019-08-22T17:58:45 | 2019-08-22T17:58:45 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 524,048,901 | 6,611 | alexey-milovidov | @tavplubix FYI that's about possible implementation of `input_format_tsv_empty_as_default` setting. | 2019-08-22T19:43:43 | 2019-08-22T19:43:43 | {} | MEMBER |
ClickHouse | ClickHouse | 524,071,696 | 6,588 | alexey-milovidov | @thyn There is a way to move table to a separate device:
1. Stop `clickhouse-server`.
2. Move table data directory to desired location.
3. Create symlink in a place of previous directory that points to new place. Make sure it is accessible to `clickhouse` user.
4. Start `clickhouse-server`. | 2019-08-22T20:49:12 | 2019-08-22T20:49:12 | {} | MEMBER |
ClickHouse | ClickHouse | 524,262,479 | 6,575 | yingfeng | Why `'\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0'` appeared in the parsed ASTSelectQuery ?
```
2019.08.23 10:08:42.826201 [ 47 ] {8b21794c-1a90-4c69-865b-a80e2a2f11e8} <Trace> InterpreterSelectQuery: FetchColumns -> Complete
2019.08.23 10:08:42.826266 [ 47 ] {8b21794c-1a90-4c69-865b-a80e2a2f11e8} <Trace> Aggreg... | 2019-08-23T10:25:07 | 2019-08-23T10:25:07 | {} | NONE |
ClickHouse | ClickHouse | 524,467,740 | 6,462 | simPod | I don't think that's related at all. | 2019-08-23T21:26:58 | 2019-08-23T21:26:58 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 523,708,573 | 6,240 | geldot | This patch introduces a lot of complexity. 34 files patched, many prototypes changes.
Perhaps it's time to come up with a clean way to store hostname/secrets/credentials instead? Either in flat file, a table or similar. The logic is already there for dictionaries and could be reused.
We had a security alert weeks... | 2019-08-22T01:24:17 | 2019-08-22T01:24:17 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 523,720,550 | 6,601 | alexey-milovidov | @nikvas0 | 2019-08-22T02:27:00 | 2019-08-22T02:27:00 | {} | MEMBER |
ClickHouse | ClickHouse | 523,910,418 | 6,608 | den-crane | It's not a bug. The reason of that tzdata2019.1.
```
SELECT timezone()
┌─timezone()─┐
│ Etc/UTC │
└────────────┘
vi /etc/default/clickhouse
export TZ=UTC
/etc/init.d/clickhouse-server restart
SELECT timezone()
┌─timezone()─┐
│ UTC │
└────────────┘
```
https://github.com/yandex/Cli... | 2019-08-22T13:38:52 | 2019-08-22T13:40:25 | {
"+1": 1
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 524,216,288 | 6,621 | filimonov | Вторая проблема (при переливке данных) тоже воспроизводится не каждый раз?
Можно ли подбирая WHERE условия для переливки найти фрагмент данных который "поломан"?
Выполнялись ли какие-то альтеры в последнее время на этой таблице?
Есть ли какие-то другие сообщения об ошибках (например при мержах на этой таблице) в ло... | 2019-08-23T08:03:04 | 2019-08-23T08:03:04 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 524,304,146 | 6,627 | den-crane | official build 19.13.1.1066 (Debian LXC) Debian 3.16.43-2+deb8u2
Exception on client:
Code: 32. DB::Exception: Attempt to read after eof: while receiving packet from localhost:9000
Connecting to localhost:9000 as user default.
Code: 210. DB::NetException: Connection refused (localhost:9000)
ClickHouse client v... | 2019-08-23T12:53:53 | 2019-08-23T12:54:13 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 524,422,831 | 5,097 | Akazz | - ClickHouse does not provide "snapshot isolation" for queries - only a simple "multiple readers/single writer" model (reads and writes cannot run on the same data concurrently)
- In the absence of such isolation the semantics of `INSERT INTO $self SELECT * from $self` is unclear
- Technically `INSERT INTO $self ... | 2019-08-23T18:47:13 | 2019-08-23T19:57:18 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 517,225,589 | 6,252 | kaijianding | maybe inode exhausted? you can check it with df -i | 2019-08-01T10:27:11 | 2019-08-01T10:27:11 | {} | NONE |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.