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 | 612,536,900 | 10,196 | azat | And I will revert all changes to `interpretSubquery` then | 2020-04-11T23:45:03 | 2020-04-11T23:45:03 | {} | MEMBER |
ClickHouse | ClickHouse | 612,558,447 | 9,985 | alexey-milovidov | Is that Ok? | 2020-04-12T03:32:57 | 2020-04-12T03:32:57 | {} | MEMBER |
ClickHouse | ClickHouse | 612,617,381 | 10,047 | kamalov-ruslan | so, I have merged latest version of ClickHouse, but 3 tests failed, I checked It locally without my Optimization and it also failing | 2020-04-12T13:48:48 | 2020-04-12T13:48:48 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 612,622,397 | 10,202 | alexey-milovidov | Decimal64 has maximum 18 digits, because it's represented by Int64, and the maximum power of ten, represented within Int64 is 1000000000000000000, so the maximum decimal value with 12 digits after decimal point will be 999999.999999999999. | 2020-04-12T14:18:44 | 2020-04-12T14:18:44 | {} | MEMBER |
ClickHouse | ClickHouse | 612,659,950 | 9,869 | alexey-milovidov | @ekimekim Sampling expression is meant to be uniformly distributed across its domain.
But if your sampling key is UInt32 and it's uniformly distributed and you use it as partition key, you will end up with billions of partitions. This is not practical at all. Even for UInt16 it's not practical. | 2020-04-12T18:50:42 | 2020-04-12T18:50:42 | {} | MEMBER |
ClickHouse | ClickHouse | 612,517,401 | 2,610 | alexey-milovidov | I don't remember whether this specific issue was fixed, but there were multiple issues with subqueries and PREWHERE and the code was rewritten multiple times. | 2020-04-11T21:30:35 | 2020-04-11T21:30:35 | {} | MEMBER |
ClickHouse | ClickHouse | 612,517,504 | 2,610 | alexey-milovidov | @azat Maybe it's worth to commit this test? | 2020-04-11T21:31:15 | 2020-04-11T21:31:15 | {} | MEMBER |
ClickHouse | ClickHouse | 612,518,229 | 2,610 | azat | Ok (Actually I came to this issues while looking into this [issue](https://github.com/ClickHouse/ClickHouse/blob/f472535f54f286d65690572f7d9a0fa63a4b887b/tests/queries/bugs/01224_dist_on_dist_global_in.sql) with distributed and global in, so let's see if I can fix this and then I will add the test from this issue to th... | 2020-04-11T21:35:54 | 2020-04-11T21:36:23 | {} | MEMBER |
ClickHouse | ClickHouse | 612,532,812 | 9,642 | alexey-milovidov | Also we decided to rewrite it completely:
```
Alexey Milovidov, [09.04.20 21:16]
[In reply to nonexistence]
Да, у меня до этого были вопросы про порядок инициализации... сейчас посмотрю ещё раз.
Alexey Milovidov, [09.04.20 21:18]
Очень не нравится битовая маска enum с системными таблицами. Я не успел сформули... | 2020-04-11T23:15:06 | 2020-04-11T23:15:06 | {
"+1": 1
} | MEMBER |
ClickHouse | ClickHouse | 612,548,133 | 9,810 | alexey-milovidov | The most illustrating example for the root cause of this issue:
```
SELECT
t1.*,
t2.*,
'world',
isConstant('world')
FROM
(
SELECT
arrayJoin([1, 2]) AS k,
'hello'
) AS t1
LEFT JOIN
(
SELECT
arrayJoin([1, 3]) AS k,
'world'
) AS t2 ON t1... | 2020-04-12T01:30:33 | 2020-04-12T01:30:33 | {} | MEMBER |
ClickHouse | ClickHouse | 612,625,828 | 10,143 | alexey-milovidov | It has happened because you have created the file with wrong ownership or access rights and clickhouse-server cannot remove it.
It will write a message
`Can't remove force file to enable table or partition drop`
and continue.
| 2020-04-12T14:39:00 | 2020-04-12T14:39:00 | {
"+1": 3
} | MEMBER |
ClickHouse | ClickHouse | 612,519,827 | 10,092 | alexey-milovidov | FYI. In my case it reproduces on `hits` table, not `visits`. | 2020-04-11T21:46:17 | 2020-04-11T21:46:17 | {} | MEMBER |
ClickHouse | ClickHouse | 612,538,917 | 2,641 | alexey-milovidov | Yes, it's doable with ASOF JOIN:
```
CREATE TEMPORARY TABLE sessions (date DateTime, visitorId String, sessionId String);
CREATE TEMPORARY TABLE orders (date DateTime, visitorId String, orderId String);
INSERT INTO sessions VALUES
('2018-01-01 00:00:00', 'v1', 's1'),
('2018-01-02 00:00:00', 'v1', 's2'),
('20... | 2020-04-12T00:00:12 | 2020-04-12T00:00:12 | {} | MEMBER |
ClickHouse | ClickHouse | 612,660,071 | 9,869 | alexey-milovidov | So, the recommendation is to just give up this idea. | 2020-04-12T18:51:50 | 2020-04-12T18:51:50 | {} | MEMBER |
ClickHouse | ClickHouse | 612,668,709 | 10,211 | youssefqarri | The timezone is available in the OS
[root@xxxx ~]# **ls /usr/share/zoneinfo/Africa/**
Abidjan Asmara Banjul Bujumbura Conakry Douala Harare Khartoum Libreville Lusaka Mbabane Ndjamena Porto-Novo Tunis
Accra Asmera Bissau Cairo Dakar El_Aaiun J... | 2020-04-12T20:04:12 | 2020-04-12T20:06:32 | {} | NONE |
ClickHouse | ClickHouse | 612,690,412 | 10,210 | alexey-milovidov | You don't have to do ALTER UPDATE.
Instead, you can simply create a column with DEFAULT expression. | 2020-04-12T23:14:38 | 2020-04-12T23:14:38 | {} | MEMBER |
ClickHouse | ClickHouse | 612,722,061 | 10,170 | weiyongyuan | > you can `select hostName(), .... from distribute_query_log `
> or `select FQDN(), .... from distribute_query_log`
thank you a lot! | 2020-04-13T02:35:08 | 2020-04-13T02:35:08 | {} | NONE |
ClickHouse | ClickHouse | 612,535,834 | 3,656 | alexey-milovidov | See also https://github.com/ClickHouse/ClickHouse/issues/9715 | 2020-04-11T23:36:55 | 2020-04-11T23:36:55 | {} | MEMBER |
ClickHouse | ClickHouse | 612,669,291 | 10,027 | alexey-milovidov | Thank you! I've applied the patch. | 2020-04-12T20:08:58 | 2020-04-12T20:08:58 | {} | MEMBER |
ClickHouse | ClickHouse | 612,530,216 | 10,085 | alexey-milovidov | Ok.
The name of the setting and implementation functions (`nullIn`, etc.) is slightly non-intuitive but reasonable. | 2020-04-11T22:56:38 | 2020-04-11T22:56:38 | {} | MEMBER |
ClickHouse | ClickHouse | 612,548,256 | 9,810 | alexey-milovidov | https://github.com/ClickHouse/ClickHouse/issues/9810#issuecomment-604995430
https://github.com/ClickHouse/ClickHouse/issues/9810#issuecomment-605214756
that two examples are not very relevant because the user intentionally hacked the query analysis with unusual aliases. They not necessary to be fixed. | 2020-04-12T01:32:29 | 2020-04-12T01:32:48 | {} | MEMBER |
ClickHouse | ClickHouse | 612,667,490 | 10,074 | alexey-milovidov | Yes, it's a great feature request. We need to allow creating replicas without specifying the schema. | 2020-04-12T19:53:00 | 2020-04-12T19:53:00 | {
"+1": 10
} | MEMBER |
ClickHouse | ClickHouse | 612,738,247 | 10,224 | org0000h | why dose the current maximum is 0.00B?
my clickhouse ver is 19.16.14.65 | 2020-04-13T03:54:46 | 2020-04-13T03:54:46 | {} | NONE |
ClickHouse | ClickHouse | 612,522,482 | 10,092 | alexey-milovidov | According to the code, it was broken long time ago. | 2020-04-11T22:02:57 | 2020-04-11T22:02:57 | {} | MEMBER |
ClickHouse | ClickHouse | 612,534,463 | 10,196 | alexey-milovidov | Let's implement the following logic:
If the right hand side of GLOBAL IN is not a subquery/function/table, either:
- replace GLOBAL IN to IN;
- or do not create external table but pass and then the GLOBAL IN operator will be executed as regular IN (let's check).
> although that code is used for join, so after t... | 2020-04-11T23:26:50 | 2020-04-11T23:26:50 | {} | MEMBER |
ClickHouse | ClickHouse | 612,548,527 | 9,810 | alexey-milovidov | It may be simply done by assigning qualified names. | 2020-04-12T01:36:33 | 2020-04-12T01:36:33 | {} | MEMBER |
ClickHouse | ClickHouse | 612,523,537 | 10,092 | alexey-milovidov | Minimal reproducing example:
`SELECT arrayJoin([SearchEngineID]) AS search_engine, URL FROM test.hits WHERE SearchEngineID != 0 AND search_engine != 0` | 2020-04-11T22:10:26 | 2020-04-11T22:10:26 | {} | MEMBER |
ClickHouse | ClickHouse | 612,544,391 | 4,767 | alexey-milovidov | arrayMin and arrayMax also can be worth doing. | 2020-04-12T00:42:57 | 2020-04-12T00:42:57 | {} | MEMBER |
ClickHouse | ClickHouse | 612,686,419 | 10,024 | FedericoCeratto | @filimonov it works but it's not done by default. | 2020-04-12T22:34:07 | 2020-04-12T22:34:07 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 612,692,646 | 9,940 | alexey-milovidov | It is reproducing every time on server shutdown if metric_log is enabled:
```
2020.04.13 02:35:36.370276 [ 25812 ] {} <Error> system.metric_log: Cannot quickly remove directory ./data/system/metric_log/delete_tmp_202004_1_4970_1338 by removing files; fallback to recursive removal. Reason: Poco::Exception. Code: 100... | 2020-04-12T23:36:10 | 2020-04-12T23:36:10 | {} | MEMBER |
ClickHouse | ClickHouse | 612,705,067 | 9,336 | bohutang | Comments:
There are already 3 patches in place to address this compatibility issue:
[0] https://github.com/ClickHouse/ClickHouse/pull/10021
[1] https://github.com/ClickHouse/ClickHouse/pull/10140
[2] https://github.com/ClickHouse/ClickHouse/pull/10199
They addressed the **SET commands**:
* `SET NAMES ...`
*... | 2020-04-13T01:08:55 | 2020-04-13T01:08:55 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 612,741,566 | 10,210 | Arthurjal | **Alexey,** this create table below is the only way that I could created the table, but with errors during the importation. ( ERROR: Date must be in YYYY-MM-DD format ). In the other ways below I can not created the table. Remember I can not change the file csv and there are a lot of files.
CREATE TABLE my_table (
... | 2020-04-13T04:13:17 | 2020-04-13T04:13:17 | {} | NONE |
ClickHouse | ClickHouse | 612,518,896 | 9,996 | geldot | Alternative approach:
On the server side, expose clickhouse-client via `Poco.Net.WebSocket` as a protocol upgrade just after [authentication](https://github.com/ClickHouse/ClickHouse/blob/master/programs/server/HTTPHandler.cpp#L544).
On the browser side, use the existing [xtermjs](https://github.com/xtermjs/xterm... | 2020-04-11T21:40:20 | 2020-04-11T21:40:20 | {
"+1": 1
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 612,541,010 | 7,994 | alexey-milovidov | @zhang2014 Maybe you can take a look? | 2020-04-12T00:13:14 | 2020-04-12T00:13:14 | {} | MEMBER |
ClickHouse | ClickHouse | 612,669,640 | 10,024 | alexey-milovidov | @filimonov That's perfect! Didn't even remember that we support that. | 2020-04-12T20:11:50 | 2020-04-12T20:11:50 | {} | MEMBER |
ClickHouse | ClickHouse | 612,515,846 | 2,610 | azat | I cannot reproduce this using usptream/master (98c603eaae):
```sql
create table data (key Int) Engine=MergeTree() order by key
insert into data select * from numbers(10)
select * from remote('127.1', currentDatabase(), data) prewhere key global in (select key from data prewhere key = 2)
select * from cluster('te... | 2020-04-11T21:20:26 | 2020-04-11T21:20:26 | {
"+1": 1
} | MEMBER |
ClickHouse | ClickHouse | 612,541,174 | 4,767 | alexey-milovidov | First make it case insensitive... | 2020-04-12T00:14:34 | 2020-04-12T00:14:34 | {} | MEMBER |
ClickHouse | ClickHouse | 612,589,677 | 8,313 | svladykin | @blinkov
Yes. The problem was not solved.
Thus it makes sense to treat this issue as a feature request but not a question.
| 2020-04-12T09:47:13 | 2020-04-12T09:48:12 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 612,627,230 | 10,097 | alexey-milovidov | Let's name this column `query_id` for consistency. | 2020-04-12T14:46:13 | 2020-04-12T14:46:13 | {} | MEMBER |
ClickHouse | ClickHouse | 612,628,313 | 10,083 | alexey-milovidov | @alesapin Need to check if it's fixed in recent versions. | 2020-04-12T14:52:24 | 2020-04-12T14:52:24 | {} | MEMBER |
ClickHouse | ClickHouse | 612,666,263 | 10,211 | alexey-milovidov | It depends on time zone info in your OS:
```
$ clickhouse-local --query "SELECT toTimeZone(now(), 'Africa/Casablanca')"
2020-04-12 20:42:32
``` | 2020-04-12T19:42:57 | 2020-04-12T19:42:57 | {} | MEMBER |
ClickHouse | ClickHouse | 612,666,315 | 10,211 | alexey-milovidov | In Linux, time zone info is located in `/usr/share/zoneinfo/` | 2020-04-12T19:43:26 | 2020-04-12T19:43:26 | {} | MEMBER |
ClickHouse | ClickHouse | 612,676,446 | 10,216 | alexey-milovidov | (test pending) | 2020-04-12T21:10:45 | 2020-04-12T21:10:45 | {} | MEMBER |
ClickHouse | ClickHouse | 612,676,750 | 10,212 | alexey-milovidov | Was intentionally removed in https://github.com/ClickHouse/ClickHouse/pull/8835 | 2020-04-12T21:13:29 | 2020-04-12T21:14:02 | {} | MEMBER |
ClickHouse | ClickHouse | 612,523,389 | 10,092 | alexey-milovidov | But it works on 19.4. | 2020-04-11T22:09:33 | 2020-04-11T22:09:33 | {} | MEMBER |
ClickHouse | ClickHouse | 612,533,823 | 10,196 | alexey-milovidov | But why we are trying to interpret literal as subquery? | 2020-04-11T23:21:58 | 2020-04-11T23:21:58 | {} | MEMBER |
ClickHouse | ClickHouse | 612,536,205 | 10,006 | alexey-milovidov | Ok. The question is how to name this property of functions?
`isDependentOnCatalog`?
(this property will be also applicable to `joinGet` and `hasColumnInTable`) | 2020-04-11T23:39:40 | 2020-04-11T23:39:40 | {
"+1": 1
} | MEMBER |
ClickHouse | ClickHouse | 612,544,132 | 4,767 | alexey-milovidov | dec3e0f986 | 2020-04-12T00:39:18 | 2020-04-12T00:39:18 | {} | MEMBER |
ClickHouse | ClickHouse | 612,545,172 | 9,889 | alexey-milovidov | It does not work at all:
```
SELECT * FROM test.hits FORMAT MsgPack
```
```
Program received signal SIGSEGV, Segmentation fault.
memcpy (destination=0x7ffff5fb49e3, source=0x10, size=size@entry=1048541) at /usr/lib/gcc/x86_64-linux-gnu/9/include/emmintrin.h:727
727 *__P = __B;
(gdb) bt
#0 memcpy (de... | 2020-04-12T00:52:21 | 2020-04-12T00:52:21 | {} | MEMBER |
ClickHouse | ClickHouse | 612,558,249 | 10,053 | alexey-milovidov | Ok.
Let's make it as simple as possible.
Let's name the setting `log_queries_min_type` and it will take numeric values. We will not validate these values. And simply apply this setting by comparison.
Reason: this will make the implementation of this setting obvious (that it compares numeric values of `type`). | 2020-04-12T03:30:45 | 2020-04-12T03:31:24 | {} | MEMBER |
ClickHouse | ClickHouse | 612,623,173 | 10,202 | alexey-milovidov | So, it's not a bug.
But we have to improve diagnostics. | 2020-04-12T14:23:20 | 2020-04-12T14:23:20 | {} | MEMBER |
ClickHouse | ClickHouse | 612,626,672 | 10,128 | alexey-milovidov | First need to get done with small issues in the list: https://github.com/ClickHouse/ClickHouse/pulls/excitoon | 2020-04-12T14:43:22 | 2020-04-12T14:43:22 | {} | MEMBER |
ClickHouse | ClickHouse | 612,661,217 | 10,207 | alexey-milovidov | Surprisingly it didn't break the tests. | 2020-04-12T19:01:38 | 2020-04-12T19:01:38 | {} | MEMBER |
ClickHouse | ClickHouse | 612,668,025 | 10,041 | alexey-milovidov | @ssky1 Yes, it's expected that ClickHouse-JDBC driver is slower than ClickHouse-Native-JDBC and it's consequently slower than clickhouse-client. | 2020-04-12T19:58:08 | 2020-04-12T19:58:08 | {} | MEMBER |
ClickHouse | ClickHouse | 612,668,178 | 10,041 | alexey-milovidov | No space left on device - is the error we get from the OS. Multiple reasons possible... | 2020-04-12T19:59:35 | 2020-04-12T19:59:35 | {} | MEMBER |
ClickHouse | ClickHouse | 612,670,354 | 10,010 | alexey-milovidov | - HAProxy;
- nginx;
- clickhouse-server with Distributed table on localhost;
- application side load balancing;
- LVS;
- load balancer inside your cloud provider, both L3/L7 will do;
- ... | 2020-04-12T20:17:52 | 2020-04-12T20:17:52 | {} | MEMBER |
ClickHouse | ClickHouse | 612,519,002 | 9,908 | alexey-milovidov | @filimonov The task appeared to be more difficult.
We will fix the issue in multiple steps instead.
1. Add timeouts to locks: #10073
2. Remove deadlock checks at all: deadlocks will remain in very rare cases but they will timeout.
3. Remove the number of cases when table RWLock is used: #7512
4. Simpify the imp... | 2020-04-11T21:41:02 | 2020-04-11T21:41:02 | {} | MEMBER |
ClickHouse | ClickHouse | 612,530,258 | 10,085 | alexey-milovidov | The query `SELECT NULL IN (1, NULL, 2)` does not work. But it's still Ok. | 2020-04-11T22:57:02 | 2020-04-11T22:57:02 | {} | MEMBER |
ClickHouse | ClickHouse | 612,533,991 | 10,196 | alexey-milovidov | `select * from remote('localhost', system.one) where dummy global in (1);`
Should not it work in the same way is regular IN? | 2020-04-11T23:23:14 | 2020-04-11T23:23:14 | {} | MEMBER |
ClickHouse | ClickHouse | 612,622,961 | 10,202 | alexey-milovidov | And the value of `locations.accuracy`: -2147467.79 looks suspicious.
It looks like it indicated an error in your application as it looks like -2^31 / 1000 with some additional error. | 2020-04-12T14:22:00 | 2020-04-12T14:22:00 | {} | MEMBER |
ClickHouse | ClickHouse | 612,674,082 | 10,215 | alexey-milovidov | It did not help. | 2020-04-12T20:50:19 | 2020-04-12T20:50:19 | {} | MEMBER |
ClickHouse | ClickHouse | 612,771,996 | 8,986 | ghost | @alexey-milovidov @amosbird @KochetovNicolai
How are you doing well in this critical environment with COVID-19?
I would appreciate if you could share the status of this PR.
| 2020-04-13T06:30:03 | 2020-04-13T06:30:03 | {} | NONE |
ClickHouse | ClickHouse | 612,536,426 | 10,196 | azat | >replace GLOBAL IN to IN;
Great idea! (I was thinking to do this initially, but by some reason rejected it) | 2020-04-11T23:41:04 | 2020-04-11T23:41:04 | {} | MEMBER |
ClickHouse | ClickHouse | 612,557,857 | 10,094 | alexey-milovidov | I've fixed minor issues. | 2020-04-12T03:26:28 | 2020-04-12T03:26:28 | {} | MEMBER |
ClickHouse | ClickHouse | 612,669,269 | 10,027 | alexey-milovidov | 20dcc4decd | 2020-04-12T20:08:43 | 2020-04-12T20:08:43 | {} | MEMBER |
ClickHouse | ClickHouse | 612,678,581 | 9,973 | alexey-milovidov | ```
CREATE TABLE mt (d Date, x UInt8) ENGINE = MergeTree(d, x, 8192);
INSERT INTO mt VALUES (52392, 1), (62677, 2);
``` | 2020-04-12T21:29:31 | 2020-04-12T21:29:31 | {} | MEMBER |
ClickHouse | ClickHouse | 612,759,112 | 9,336 | meob | :) | 2020-04-13T05:36:46 | 2020-04-13T05:36:46 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 612,512,828 | 10,184 | alexey-milovidov | Minor issue with the list of libraries to link with test util:
https://clickhouse-builds.s3.yandex.net/10184/c9542b66018bc2e0ca124572642af364f627ebcb/build_log_650442673_1586564198.txt | 2020-04-11T21:01:41 | 2020-04-11T21:01:41 | {} | MEMBER |
ClickHouse | ClickHouse | 612,628,008 | 10,091 | alexey-milovidov | @qoega Please validate that everything is correct. | 2020-04-12T14:50:34 | 2020-04-12T14:50:34 | {} | MEMBER |
ClickHouse | ClickHouse | 612,667,295 | 10,075 | alexey-milovidov | Thanks @amosbird! | 2020-04-12T19:51:19 | 2020-04-12T19:51:19 | {} | MEMBER |
ClickHouse | ClickHouse | 612,668,396 | 10,041 | alexey-milovidov | When you insert small amount of data in a short period of time, it may take a few times more space due to write amplification (background merging of data parts). But this write amplification will be cleaned up after 8 minutes (the default lifetime of data parts after background merge). | 2020-04-12T20:01:39 | 2020-04-12T20:01:39 | {} | MEMBER |
ClickHouse | ClickHouse | 612,672,383 | 10,211 | youssefqarri | Yes. It is available in both as the client is hosted in the server as the server | 2020-04-12T20:34:44 | 2020-04-12T20:34:44 | {} | NONE |
ClickHouse | ClickHouse | 612,673,562 | 10,211 | alexey-milovidov | Confirmed. I can reproduce it with the following commands:
```
$ docker run --rm -it --volume /home/milovidov/work/ClickHouse/build:/build centos:8 /build/programs/clickhouse local --query "SELECT toTimeZone(now(), 'Africa/Casablanca')"
Exception: Cannot load time zone Africa/Casablanca
$ docker run --rm -it --vo... | 2020-04-12T20:45:43 | 2020-04-12T20:45:43 | {
"+1": 1
} | MEMBER |
ClickHouse | ClickHouse | 612,687,542 | 10,024 | alexey-milovidov | By the way, how to decrypt the acronym `XDG`, what does it mean? | 2020-04-12T22:45:23 | 2020-04-12T22:45:23 | {} | MEMBER |
ClickHouse | ClickHouse | 612,536,738 | 2,641 | alexey-milovidov | This should be the case for ASOF JOIN.
Let's check if it's applicable... | 2020-04-11T23:43:42 | 2020-04-11T23:43:42 | {} | MEMBER |
ClickHouse | ClickHouse | 612,540,896 | 6,222 | alexey-milovidov | How exactly we should modify the example command? | 2020-04-12T00:12:12 | 2020-04-12T00:12:12 | {} | MEMBER |
ClickHouse | ClickHouse | 612,546,272 | 9,950 | alexey-milovidov | I'm mostly Ok with this solution.
BTW, What will happen in query:
`select cast(7 as String), * from (select number "'String'" FROM numbers(2));`
? | 2020-04-12T01:05:41 | 2020-04-12T01:05:41 | {} | MEMBER |
ClickHouse | ClickHouse | 612,547,887 | 9,810 | alexey-milovidov | @akuzm @4ertus2 To solve this issue, we have to assign unique names only to constant expressions that will be materialized and altered during JOINs.
e.g. LEFT JOINed constants that may become NULL.
| 2020-04-12T01:26:50 | 2020-04-12T01:26:50 | {} | MEMBER |
ClickHouse | ClickHouse | 612,565,233 | 7,994 | zhang2014 | Ok, I will add it to my todo list. | 2020-04-12T05:07:08 | 2020-04-12T05:07:08 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 612,660,513 | 9,869 | alexey-milovidov | Why the `max_parallel_replicas` setting is implemented via sampling?
It is needed to provide consistent split of data, regardless to how the data was merged on different replicas.
But reading sample of data is not always N times faster than reading the whole data. It depends on the sampling key and if data is not o... | 2020-04-12T18:55:37 | 2020-04-12T18:55:37 | {} | MEMBER |
ClickHouse | ClickHouse | 612,667,039 | 10,078 | alexey-milovidov | Names of identifiers (columns, tables, etc.) in ClickHouse SQL dialect are case sensitive.
(and it was not changed in any release) | 2020-04-12T19:49:19 | 2020-04-12T19:49:55 | {} | MEMBER |
ClickHouse | ClickHouse | 612,534,333 | 10,196 | azat | >Should not it work in the same way is regular IN?
Agree (although that code is used for join, so after the following will be allowed I guess `select 1 union all 2`) | 2020-04-11T23:25:52 | 2020-04-11T23:25:52 | {} | MEMBER |
ClickHouse | ClickHouse | 612,538,027 | 10,196 | azat | Description had been update (don't forget to update labels) | 2020-04-11T23:53:35 | 2020-04-11T23:53:35 | {} | MEMBER |
ClickHouse | ClickHouse | 612,539,976 | 7,115 | alexey-milovidov | The issue is no longer relevant.
Our CI images are using Ubuntu 19.10. | 2020-04-12T00:06:20 | 2020-04-12T00:06:20 | {} | MEMBER |
ClickHouse | ClickHouse | 612,540,215 | 6,673 | alexey-milovidov | Now it is checked in our CI on Ubuntu 19.10:
https://clickhouse-builds.s3.yandex.net/0/121bf7b8c2bbfc2f09379afa2232a8b1b8754844/build_log_650970136_1586648038.txt | 2020-04-12T00:07:49 | 2020-04-12T00:07:49 | {} | MEMBER |
ClickHouse | ClickHouse | 612,671,679 | 10,211 | alexey-milovidov | Is it available both on client and server side?
(client may need it for data formatting) | 2020-04-12T20:28:57 | 2020-04-12T20:28:57 | {} | MEMBER |
ClickHouse | ClickHouse | 612,680,753 | 6,443 | alexey-milovidov | #9973 | 2020-04-12T21:46:04 | 2020-04-12T21:46:04 | {} | MEMBER |
ClickHouse | ClickHouse | 612,680,857 | 6,443 | alexey-milovidov | It is caused by wrong data improted into table.
Kafka engine is trying to import it again and again, it cannot skip this data.
This is the expected behaviour.
To solve the issue, you have to remove broken data from Kafka topic or manually update the offset. | 2020-04-12T21:46:55 | 2020-04-12T21:47:52 | {} | MEMBER |
ClickHouse | ClickHouse | 612,688,041 | 9,940 | alexey-milovidov | @excitoon
Most likely Ilya manually inserted data with incorrect Sign via `generateRandom` function. This is expected. | 2020-04-12T22:50:09 | 2020-04-12T22:50:09 | {} | MEMBER |
ClickHouse | ClickHouse | 612,702,289 | 10,221 | alexey-milovidov | No test possible. | 2020-04-13T00:50:45 | 2020-04-13T00:50:45 | {} | MEMBER |
ClickHouse | ClickHouse | 612,522,640 | 10,092 | alexey-milovidov | More than one year ago, starting from 19.4. | 2020-04-11T22:04:00 | 2020-04-11T22:04:00 | {} | MEMBER |
ClickHouse | ClickHouse | 612,548,357 | 9,810 | alexey-milovidov | What about the following logic:
If it's a JOIN action and we have a column with constant expression and without an alias, then assign a unique name. | 2020-04-12T01:33:55 | 2020-04-12T01:34:02 | {} | MEMBER |
ClickHouse | ClickHouse | 612,599,688 | 10,157 | CurtizJ | Fixed in master. | 2020-04-12T11:22:49 | 2020-04-12T11:22:49 | {} | MEMBER |
ClickHouse | ClickHouse | 612,625,076 | 10,183 | alexey-milovidov | @filimonov Let it fail completely and not continue. | 2020-04-12T14:34:47 | 2020-04-12T14:34:47 | {} | MEMBER |
ClickHouse | ClickHouse | 612,627,755 | 10,091 | alexey-milovidov | You are using old compiler from OS.
Instead, you should install fresh `clang` version. The minimum version is 9*.
This manual should work: https://clickhouse.tech/docs/en/development/build_osx/
\* - please note that AppleClang compiler from the OS is using different version numbering and is actually older. | 2020-04-12T14:48:56 | 2020-04-12T14:49:55 | {} | MEMBER |
ClickHouse | ClickHouse | 612,660,684 | 9,869 | alexey-milovidov | We need to implement different way for efficient query processing across multiple replicas. | 2020-04-12T18:57:00 | 2020-04-12T18:57:00 | {} | MEMBER |
ClickHouse | ClickHouse | 608,496,378 | 10,010 | filimonov | AFAIK in JDBC client, you can provide addresses of a few servers in connection settings. | 2020-04-03T15:09:47 | 2020-04-12T20:15:20 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 612,731,429 | 10,200 | AlexanderJLiu | It seems that not only the ` join query` would cause `Segment fault`, some `normal queries` will also lead to this problem. For examples:
```sql
select sum(click) as c
from db.distributed_table
where Date>='2020-04-11' and Date<='2020-04-11' and id='xxx'
``` | 2020-04-13T03:22:33 | 2020-04-13T03:22:33 | {} | NONE |
ClickHouse | ClickHouse | 613,942,461 | 10,277 | azat | > test_settings_constraints_distributed/test.py::test_insert_clamps_settings https://clickhouse-test-reports.s3.yandex.net/10049/824255f603abbf1363c6bfe2e9a078011f9131d8/integration_tests_(asan).html @vitlibar
Can be fixed by adding `SYSTEM FLUSH DISTRIBUTED` (b7dde1a86f0a6724e0f04f9854d3b4f411417844) | 2020-04-15T09:58:11 | 2020-04-15T09:58:11 | {} | MEMBER |
ClickHouse | ClickHouse | 614,043,330 | 3,621 | alexey-milovidov | Workaround possible. | 2020-04-15T13:33:08 | 2020-04-15T13:33:08 | {} | MEMBER |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.