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 | 578,457,342 | 6,358 | alexey-milovidov | Let's encode it as
```
shard1
```
to send data to any replica of specific shard
or
```
shard1_replica1
```
to send data to specific replica of this shard. | 2020-01-26T00:40:48 | 2020-01-26T00:41:11 | {} | MEMBER |
ClickHouse | ClickHouse | 578,457,457 | 8,448 | alexey-milovidov | #8770 | 2020-01-26T00:42:40 | 2020-01-26T00:42:40 | {} | MEMBER |
ClickHouse | ClickHouse | 578,731,703 | 8,841 | filimonov | Why tar? Gives no special benefits, will write all the data once again to the local disk. Again unpacking will require one more data iteration, you need to maintain multiple tars, metadata etc.
As for now the option to rsync / rclone (https://rclone.org/) shadow dir(s) directly to s3 looks the simplest option. Act... | 2020-01-27T12:49:08 | 2020-01-27T13:24:03 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 578,875,732 | 6,168 | barrkel | I erroneously wrote a message here a few minutes ago indicating that `MOVE PARTITION` didn't work with the latest drop of ClickHouse from Docker Hub (20.1.4), but of course it doesn't contain this code yet.
The documentation needs tagging by version at some point, it can be a bit misleading otherwise. | 2020-01-27T18:04:18 | 2020-01-27T18:04:18 | {} | NONE |
ClickHouse | ClickHouse | 579,036,101 | 8,842 | simPod | @zhang2014 do you have a link/reference? Is it this? https://github.com/ClickHouse/ClickHouse/pull/7572
Thanks! | 2020-01-28T01:25:50 | 2020-01-28T01:25:50 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 579,074,074 | 8,864 | filimonov | Special case here is system tables (which do exists but have no metadata). | 2020-01-28T04:29:38 | 2020-01-28T04:29:50 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 579,127,371 | 8,060 | alexmgit | На данный момент есть решение с созданием представления таблицы, где колонка binary(1) конвертируется в CHAR(1):
CREATE OR REPLACE VIEW _test_binary_view (id, status) AS SELECT `id`, CONVERT(`status`, CHAR(1)) FROM _test_binary;
Но это колхоз, конечно же | 2020-01-28T08:07:23 | 2020-01-28T08:07:23 | {} | NONE |
ClickHouse | ClickHouse | 578,451,667 | 8,834 | alexey-milovidov | > it may be overridden during linkage by rewriting extern const bool DB::NULLS_LAST weak symbol
If it is unused right now, better not introduce this tweak at all. | 2020-01-25T23:08:37 | 2020-01-25T23:08:37 | {} | MEMBER |
ClickHouse | ClickHouse | 578,453,588 | 7,336 | alexey-milovidov | We can return 0 for decimals, but I don't think it's a good choice.
Also we can introduce a setting to make Nullable result for all aggregate functions. | 2020-01-25T23:37:44 | 2020-01-25T23:38:03 | {} | MEMBER |
ClickHouse | ClickHouse | 578,454,900 | 5,329 | alexey-milovidov | Similar optimization is possible for just ORDER BY:
```
SELECT a1, a2, ..., b1, b2, ...
FROM table
ORDER BY a1, a2
LIMIT 10
```
Can be rewritten into:
```
SELECT a1, a2, ..., b1, b2, ...
FROM table
PREWHERE (a1, a2, ...) IN
(
SELECT a1, a2, ...
FROM table
ORDER BY a1, a2
LIMIT 10
... | 2020-01-26T00:00:54 | 2020-01-26T00:00:54 | {
"+1": 3
} | MEMBER |
ClickHouse | ClickHouse | 578,543,654 | 7,187 | alexey-milovidov | Yes, I have loaded the data but it shows strange numbers.
And the dataset doesn't contain our repository probably because it is from 2015. | 2020-01-26T21:12:18 | 2020-01-26T21:12:18 | {} | MEMBER |
ClickHouse | ClickHouse | 578,570,100 | 8,842 | zhang2014 | It has been implemented in customhttp patch | 2020-01-27T02:08:22 | 2020-01-27T02:29:05 | {
"heart": 1
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 578,886,986 | 8,834 | zlobober | @alexey-milovidov, would you mind checking again? | 2020-01-27T18:29:35 | 2020-01-27T18:29:35 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 578,899,981 | 8,859 | amosbird | Interesting.
a reduced test:
`select CounterID NOT IN (2) a FROM testmt WHERE CounterID IN (2) group by a;`
however these two variants work:
```
SELECT CounterID NOT IN (3) a FROM testmt WHERE CounterID IN (2) group by a;
SELECT CounterID NOT IN (2) a FROM testmt WHERE CounterID IN (2);
```
Is it so... | 2020-01-27T19:00:26 | 2020-01-27T19:00:26 | {} | COLLABORATOR |
ClickHouse | ClickHouse | 578,452,031 | 8,830 | alexey-milovidov | > Maybe we should introduce special value of 2 (or null) for this flag indicating that certain feature may be supported or not depending on the particular table?
Not necessary. Just move only those features that are constantly true or false... | 2020-01-25T23:13:56 | 2020-01-25T23:13:56 | {} | MEMBER |
ClickHouse | ClickHouse | 578,530,810 | 8,830 | alexey-milovidov | ```
2020.01.26 01:50:02.420571 [ 77 ] {dd0f13ee-adb3-4350-85f1-4b510eb3af55} <Debug> executeQuery: (from [::1]:34264) CREATE VIEW test.view AS SELECT CounterID, count() AS c FROM test.hits GROUP BY CounterID
2020.01.26 01:50:02.423212 [ 247 ] {} <Fatal> BaseDaemon: ########################################
2020.01.26... | 2020-01-26T18:47:15 | 2020-01-26T18:47:15 | {} | MEMBER |
ClickHouse | ClickHouse | 578,541,487 | 6,833 | ghost | @alexey-milovidov
https://www.altinity.com/blog/2018/5/10/circular-replication-cluster-topology-in-clickhouse
Following this article, I setup circular replication with 3 shards across a 3 node cluster. Node 1 had shards 1,2, Node 2 had shards 2,3, and Node 3 had shards 3,1. These were all merge tables, and one d... | 2020-01-26T20:46:29 | 2020-01-26T20:46:29 | {} | NONE |
ClickHouse | ClickHouse | 578,112,154 | 8,813 | CurtizJ | Currently have significant performance degradation.
Benchamark on query:
`SELECT count() FROM numbers(10000000) WHERE number IN (... set of 1000 numbers ...)`
20.1:
```
localhost:9000, queries 172, QPS: 15.291, RPS: 153322131.180, MiB/s: 1169.755, result RPS: 15.291, result MiB/s: 0.000.
0.000% 0.062... | 2020-01-24T12:30:49 | 2020-01-27T11:49:28 | {} | MEMBER |
ClickHouse | ClickHouse | 578,767,772 | 8,573 | dstruck | I have now tested the timeout parameter **connect_timeout_with_failover_ms** on the latest version **20.1.2.4**, even set it to 10000ms. Still the same **Timeout exceeded** error (well before the 10 sec. timeout). Please be aware that we do not use replica for this cluster.
Queries run fine on the cluster if TLS is ... | 2020-01-27T14:17:51 | 2020-01-27T14:17:51 | {} | NONE |
ClickHouse | ClickHouse | 578,916,293 | 8,573 | den-crane | check `cat /proc/sys/kernel/random/entropy_avail `
Try rng-tools or haveged at test cluster https://www.cyberciti.biz/open-source/debian-ubuntu-centos-linux-setup-additional-entropy-for-server-using-aveged-rng-tools-utils/ to check that an entropy starvation is the reason.
| 2020-01-27T19:38:47 | 2020-01-27T19:41:17 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 578,927,388 | 8,851 | excitoon | @kmatt This pull request does not consider delimiters at all (in S3 one can choose delimiters), the only requirement is the same/exact bucket for different files. `*` and `?` can not include slashes like in implementations for other storages.
I.e. that example will list everything as expected. | 2020-01-27T20:04:35 | 2020-01-27T22:05:20 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 579,040,665 | 8,568 | dingxiangfei2009 | @huang--yijian Sorry, I type the wrong digit for my issue number. | 2020-01-28T01:46:53 | 2020-01-28T01:46:53 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 578,457,214 | 6,358 | alexey-milovidov | It is wrong that we have credentials in directory name.
Better to put shard number and (optional) replica number instead.
It can be automatically unambiguated from current format.
This task is not in progress right now. | 2020-01-26T00:38:04 | 2020-01-26T00:39:54 | {} | MEMBER |
ClickHouse | ClickHouse | 578,457,596 | 4,385 | alexey-milovidov | We still have this issue. | 2020-01-26T00:45:40 | 2020-01-26T00:45:40 | {} | MEMBER |
ClickHouse | ClickHouse | 578,458,364 | 6,833 | alexey-milovidov | @markcorwin-iex There is no way to configure this limit right now. | 2020-01-26T01:00:25 | 2020-01-26T01:00:25 | {} | MEMBER |
ClickHouse | ClickHouse | 578,507,849 | 8,842 | alexey-milovidov | Yes, we need to implement it. | 2020-01-26T14:40:18 | 2020-01-26T14:40:18 | {
"+1": 4
} | MEMBER |
ClickHouse | ClickHouse | 578,739,092 | 8,830 | alexey-milovidov | @filimonov Separate columns is more ClickHouse style. | 2020-01-27T13:10:31 | 2020-01-27T13:10:31 | {} | MEMBER |
ClickHouse | ClickHouse | 578,809,406 | 8,692 | mrxotey | @LittleMaverick , are you sure?
```
localhost :) create table q (`id` Nullable(String), `status` Nullable(Enum8('NEW' = 0, 'CANCEL' = 1)), `nested.nestedType` Array(Nullable(String)), `partition` Date) ENGINE = MergeTree() PARTITION BY partition ORDER BY partition SETTINGS index_granularity = 8192
CREATE TABLE q... | 2020-01-27T15:43:51 | 2020-01-27T15:43:51 | {} | NONE |
ClickHouse | ClickHouse | 578,934,350 | 8,854 | alexey-milovidov | We should allow to select from anything returning a table like DESCRIBE, SHOW TABLES, ... | 2020-01-27T20:21:13 | 2020-01-27T20:21:13 | {
"hooray": 1
} | MEMBER |
ClickHouse | ClickHouse | 578,446,811 | 8,833 | excitoon | @alexey-milovidov what about 8.3? | 2020-01-25T22:01:34 | 2020-01-25T22:01:34 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 578,460,926 | 7,187 | alexey-milovidov | @lorenz Could you please describe this dataset?
And why I don't see ClickHouse inside? | 2020-01-26T01:48:39 | 2020-01-26T01:48:39 | {} | MEMBER |
ClickHouse | ClickHouse | 578,481,144 | 8,826 | MeteHanC | Thanks for the suggestions. No, there is no table with Kafka Engine in our cluster. I got the following after executing the command ;
```
Could not attach to process. If your uid matches the uid of the target
process, check the setting of /proc/sys/kernel/yama/ptrace_scope, or try
again as the root user. For mo... | 2020-01-26T08:37:08 | 2020-01-26T08:37:08 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 578,490,875 | 8,826 | alexey-milovidov | In some cases you can kill zombie process by `kill -9` again. | 2020-01-26T11:02:08 | 2020-01-26T11:02:08 | {} | MEMBER |
ClickHouse | ClickHouse | 578,526,156 | 8,695 | alexey-milovidov | @alex-krash Don't you mind if we close? | 2020-01-26T17:52:51 | 2020-01-26T17:52:51 | {} | MEMBER |
ClickHouse | ClickHouse | 578,838,280 | 8,856 | mvcalder-xbk | I tried starting the server in interactive mode and it complains about a permission error:
```
root@914727f99de4:/# sudo -u clickhouse /usr/bin/clickhouse-server --config-file /etc/clickhouse-server/config.xml
Include not found: clickhouse_remote_servers
Include not found: clickhouse_compression
Couldn't save pr... | 2020-01-27T16:43:32 | 2020-01-27T16:43:32 | {} | NONE |
ClickHouse | ClickHouse | 578,893,317 | 8,861 | alexey-milovidov | Now it works perfectly:
```
~/ClickHouse/build_gcc9/dbms/programs/clickhouse local --query "SELECT data FROM file('*.json.gz', TSV, 'data String') WHERE JSONExtractString(data, 'actor', 'login') = 'akuzm' LIMIT 10" | jq
``` | 2020-01-27T18:44:59 | 2020-01-27T18:44:59 | {} | MEMBER |
ClickHouse | ClickHouse | 579,073,623 | 8,573 | filimonov | Also:
1) do you face that issue only for Distributed queries?
2) clickhouse-client connected from the same node to secure port works properly?
3) Do you see any errors in logs of the nodes initiating query, or in logs or shards, when the exception happen?
4) you say you're using https too, did you configure ports... | 2020-01-28T04:27:12 | 2020-01-28T04:27:12 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 578,446,021 | 8,833 | excitoon | Progress on #8760 | 2020-01-25T21:49:20 | 2020-01-25T21:49:30 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 578,454,307 | 7,187 | alexey-milovidov | @lorenz It is wonderful that we have this dataset publicly available!
Maybe we can use it in our performance tests and examples...
Is it Ok if we will copy and redistribute this dataset from our servers? | 2020-01-25T23:50:40 | 2020-01-25T23:50:48 | {} | MEMBER |
ClickHouse | ClickHouse | 578,463,510 | 7,187 | lorenz | @alexey-milovidov The dataset is for directly attaching to ClickHouse. You should be able to extract it to `/var/lib/clickhouse/data/$some_db/` and then just use the attach commands in the first post.
The data was not collected by me, it comes from [here](https://boyter.org/posts/an-informal-survey-of-10-million-git... | 2020-01-26T02:48:14 | 2020-01-26T02:57:30 | {} | NONE |
ClickHouse | ClickHouse | 578,486,927 | 8,826 | MeteHanC | No, I have already stopped ClickHouse but yes there is still a zombie process which prevents ClickHouse from restarting it. As I said, there seems to be no other zombie process. And parent process is 1 which is 'init'. However after executing the 'pstree -p -s 10926' command, this came up;
```
systemd(1)───clickhou... | 2020-01-26T10:01:48 | 2020-01-26T10:13:14 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 578,674,485 | 8,692 | mrxotey | Same bug for 19.17.5.18
```sql
CREATE TABLE test (
a Date,
b UInt32,
c UInt64,
p Nested (
at1 String,
at2 String
)
) ENGINE = MergeTree()
PARTITION BY a
ORDER BY b
SETTINGS index_granularity = 8192
```
```sql
INSERT INTO test (a, b, c, p.at1, p.at2)
VALUES (now()... | 2020-01-27T10:01:54 | 2020-01-27T12:09:56 | {} | NONE |
ClickHouse | ClickHouse | 579,077,412 | 7,851 | filimonov | That looks like that commit https://github.com/ClickHouse/ClickHouse/pull/7683/commits/cd45d962caaa368305c3f044379b2a3fe499faa5 exists in 19.17 branch only? Do you expect to do rolling update to 20 only with a transition through 19.17 (stuff like that make support of updates painful) | 2020-01-28T04:42:51 | 2020-01-28T04:42:51 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 579,080,762 | 6,358 | filimonov | > to send data to any replica of specific shard
Corner case here is when new shards added - shard numbering can be shifted (not very important) | 2020-01-28T05:00:16 | 2020-01-28T05:00:39 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 578,490,335 | 8,826 | alexey-milovidov | @filimonov
> Can you collect thread stacktraces at the moment of hung
Zombie process is already exited, it cannot have stack. | 2020-01-26T10:55:00 | 2020-01-26T11:01:17 | {} | MEMBER |
ClickHouse | ClickHouse | 578,129,275 | 8,813 | CurtizJ | Build from the original PR shows the same performance as latest releases. But when I rebased it on master, performance has been degraded even without commit 0a8080e. I suppose that function `convertFieldToType` or `evaluateConstantExpression` has been slowed down since `May 19, 2019` | 2020-01-24T13:27:32 | 2020-01-27T11:49:10 | {} | MEMBER |
ClickHouse | ClickHouse | 578,789,093 | 8,854 | filimonov | > `SELECT name FROM (DESCRIBE TABLE ABC) where type='Int32'`
didn't check that, but most probably allowing that will be simpler/more logical (comparing to introducing table function which is able to accept another tables / table functions). | 2020-01-27T15:00:08 | 2020-01-27T15:01:23 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 578,846,309 | 8,859 | alesapin | @amosbird Maybe you have any ideas about how to fix it? | 2020-01-27T17:00:18 | 2020-01-27T17:00:18 | {} | MEMBER |
ClickHouse | ClickHouse | 578,474,025 | 8,826 | MeteHanC | No, it does not get fixed automatically, I have waited for a day before opening the issue with same thoughts. And yes it happens when I run CH server from terminal too. Raising the same error;
```
StatusFile: Status file /var/lib/clickhouse/status already exists - unclean restart. Contents:
PID: 10926
Started at:... | 2020-01-26T06:59:36 | 2020-01-26T07:00:45 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 578,479,196 | 8,826 | filimonov | Do you use Engine=Kafka? There is one known issue with librdkafka leading to similar behaviour (something like deadlock when client is stopped in unfortunate moment, after application hungs forever waiting for thread joining - I have it in my Kafka todo list).
Can you collect thread stacktraces at the moment of hun... | 2020-01-26T08:17:47 | 2020-01-26T08:35:43 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 578,512,724 | 8,843 | alexey-milovidov | @KochetovNicolai why we had them? | 2020-01-26T15:27:07 | 2020-01-26T15:27:07 | {} | MEMBER |
ClickHouse | ClickHouse | 578,621,042 | 8,830 | filimonov | Maybe just array?
```
features: ['settings' , 'skipping_indices' , 'ttl', 'sort_order']
```
It's easier to extend and horizontal output is not so wide. Checking with `select * from where has(features, 'ttl')` looks very natural | 2020-01-27T07:19:06 | 2020-01-27T07:19:30 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 578,758,604 | 8,738 | hhell | > I.e. you can have something like
Yes, that would be good.
Not sure how it would work with some of the formats (e.g. CSV, which can have any amount of newlines in data), but for many others (TSV, JSONCompactEachRow) it would be perfect. | 2020-01-27T13:59:03 | 2020-01-27T13:59:03 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 578,846,510 | 8,856 | mvcalder-xbk | I fixed the problem by changing the owner/group of /var/lib/clickhouse:
```
$> chown -R clickhouse:clickhouse /var/lib/clickhouse
```
For us /var/lib/clickhouse is a named volume. Somehow the user id for clickhouse went from 102 to 103 and messagebus took 102. This had the effect of changing the owner of all th... | 2020-01-27T17:00:41 | 2020-01-27T17:00:41 | {
"+1": 1
} | NONE |
ClickHouse | ClickHouse | 578,444,709 | 8,695 | alexey-milovidov | From the description, it is very similar to cached dictionaries with `http` or `https` sources, and with dictionaries we also have connection pool and result cache. Now we can create dictionaries with `CREATE DICTIONARY` query.
Though I didn't look at the code yet...
The code is Ok, but we should limit hosts with... | 2020-01-25T21:30:15 | 2020-01-25T21:40:37 | {} | MEMBER |
ClickHouse | ClickHouse | 578,780,661 | 8,738 | filimonov | > CSV, which can have any amount of newlines in data
AFAIR you just need to replace any newline with `\ndata: ` (need to reread the specs to ensure)
| 2020-01-27T14:43:00 | 2020-01-27T14:43:17 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 578,894,437 | 5,770 | Avogar | I created a pull-request:
https://github.com/ClickHouse/ClickHouse/pull/8860 | 2020-01-27T18:47:58 | 2020-01-27T18:47:58 | {} | MEMBER |
ClickHouse | ClickHouse | 578,446,578 | 8,833 | alexey-milovidov | `override` with presense of `final` isn't required for gcc 9.2
and gcc 9.1 is already obsolete. | 2020-01-25T21:58:18 | 2020-01-25T21:58:43 | {} | MEMBER |
ClickHouse | ClickHouse | 578,453,804 | 7,207 | alexey-milovidov | > Looking at the mark cache I can see it going up over time
Until recently (version 20.1) we had unusual behaviour of mark cache. It kept records not only up to `mark_cache_size` but also all the records for `mark_cache_min_lifetime` (one hour by default). | 2020-01-25T23:41:24 | 2020-01-25T23:41:34 | {} | MEMBER |
ClickHouse | ClickHouse | 578,483,630 | 8,826 | filimonov | Do you have a single clickhouse process running and it's zombie?
Can you please list all system processes (with threads) and find a zombie parent?
Check those:
https://stackoverflow.com/a/44752341/1555175
https://askubuntu.com/a/290012/971776 | 2020-01-26T09:07:38 | 2020-01-26T09:09:58 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 578,490,247 | 8,826 | alexey-milovidov | @MeteHanC
`pstree` also displays all threads of a processs.
No, it is not related to client sessions, because it is clickhouse-server, not clickhouse-client. | 2020-01-26T10:53:48 | 2020-01-26T10:53:48 | {} | MEMBER |
ClickHouse | ClickHouse | 578,626,015 | 8,695 | alex-krash | > @alex-krash Don't you mind if we close?
I will try to explain my use case, and if HTTP dictionaries would solve it - let's close :)
My need is to decompose timeseries into seasonality, trend, etc. I was going to group values by timeseries_id, send it as array to remote http endpoint, and get back Array(Array(Floa... | 2020-01-27T07:40:18 | 2020-01-27T07:40:18 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 578,802,365 | 8,573 | dstruck | After a while HTTPS client connections stopped working and we saw the following error message:
```
2020.01.27 15:19:08.159844 [ 104 ] {} <Error> ServerErrorHandler: Poco::Exception. Code: 1000, e.code() = 0, e.displayText() = SSL Exception: error:2400006E:random number generator::error retrieving entropy0. 0xbc3a63... | 2020-01-27T15:28:31 | 2020-01-27T15:28:31 | {} | NONE |
ClickHouse | ClickHouse | 578,994,075 | 8,823 | alexey-milovidov | @KochetovNicolai Any idea why `logs_level_bugfix` test has failed? | 2020-01-27T22:52:27 | 2020-01-27T22:52:27 | {} | MEMBER |
ClickHouse | ClickHouse | 579,009,380 | 8,815 | den-crane | Connected to ClickHouse server version 20.1.2 revision 54431.
There is no max_threads_for_parallel_parsing
SET input_format_parallel_parsing = 1
Ok.
SET max_threads_for_parallel_parsing = 1
Received exception from server (version 20.1.2):
Code: 115. DB::Exception: Received from localhost:9000. DB::Except... | 2020-01-27T23:41:28 | 2020-01-27T23:41:28 | {
"+1": 1
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 579,094,127 | 8,736 | dingxiangfei2009 | @filimonov Sorry, I put a wrong digit on the issue number I intend to fix.
I think I figured out why tests were failing. There is a reason why signalling must happen after destruction of `thread_status`. `thread_status` may borrow some `Context` for the duration of the work, and another thread such as `TCPHandler` m... | 2020-01-28T06:01:49 | 2020-01-28T06:01:49 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 579,119,144 | 7,528 | velimovich | Fixed in 20.1.2.4. | 2020-01-28T07:40:48 | 2020-01-28T07:40:48 | {} | NONE |
ClickHouse | ClickHouse | 578,525,905 | 8,826 | MeteHanC | > does the process become zombie if you successfully run clickhouse-server from terminal and exit with Ctrl+C after while?
No, on a clean server, I could not reproduce this. I will just restart the server. I thought it could be related to CH not being able to kill the parent process but this issue seems invalid. Tha... | 2020-01-26T17:50:11 | 2020-01-26T17:50:11 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 578,744,720 | 8,661 | filimonov | Maybe metadata should be also included in backups? | 2020-01-27T13:25:07 | 2020-01-27T13:25:07 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 578,875,703 | 8,677 | alexey-milovidov | Ok. https://github.com/ClickHouse/ClickHouse/commit/774a98b4061c71fff99244879899eeecff25b8d5 | 2020-01-27T18:04:14 | 2020-01-27T18:04:14 | {} | MEMBER |
ClickHouse | ClickHouse | 578,444,871 | 8,695 | alexey-milovidov | We should also consider the following three upcoming features for dictionaries:
- `direct` layout that will query remote source without cache;
- asynchronous mode of update for cache dictionaries;
- it will be possible to define custom names for dictionary functions. | 2020-01-25T21:32:42 | 2020-01-25T21:35:11 | {} | MEMBER |
ClickHouse | ClickHouse | 578,452,639 | 7,477 | alexey-milovidov | @Vinatorul is everything all right? | 2020-01-25T23:22:22 | 2020-01-25T23:22:22 | {} | MEMBER |
ClickHouse | ClickHouse | 578,490,514 | 8,826 | alexey-milovidov | > And yes it happens when I run CH server from terminal too.
No. My question is:
- does the process become zombie if you successfully run clickhouse-server from terminal and exit with Ctrl+C after while?
You can check it on clean server where you don't already have zombie process. | 2020-01-26T10:57:30 | 2020-01-26T10:57:50 | {} | MEMBER |
ClickHouse | ClickHouse | 578,516,748 | 8,695 | filimonov | >From the description, it is very similar to cached dictionaries with http or https sources, and with dictionaries we also have connection pool and result cache.
It works only for cached layout, and is very slow (i think slowness is caused by syncronized updates of cache, mostly it works single-thread-alike).
Im... | 2020-01-26T16:10:51 | 2020-01-26T16:10:51 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 578,615,069 | 8,846 | azat | Not ready yet, please ignore for now | 2020-01-27T06:56:04 | 2020-01-27T06:56:04 | {} | MEMBER |
ClickHouse | ClickHouse | 578,711,883 | 8,813 | CurtizJ | Benchmark results below are irrelevant. My mistake was that I have been comparing release gcc build against my locally builded release clang build. With builds on the same compilers performace is almost identical. I will add a performance test later. | 2020-01-27T11:47:02 | 2020-01-27T11:47:35 | {} | MEMBER |
ClickHouse | ClickHouse | 578,903,340 | 8,834 | alexey-milovidov | Request for the next PR:
- add a setting `allow_nullable_key`, false by default;
- if it is set or if we attach table, allow Nullable key;
- add a test where comparison with null is covered. | 2020-01-27T19:08:15 | 2020-01-27T19:08:15 | {} | MEMBER |
ClickHouse | ClickHouse | 578,452,877 | 8,814 | alexey-milovidov | Duplicate: https://github.com/ClickHouse/ClickHouse/issues/6833 | 2020-01-25T23:26:21 | 2020-01-25T23:26:21 | {} | MEMBER |
ClickHouse | ClickHouse | 578,490,699 | 8,826 | alexey-milovidov | If this issue happened just a single time, then:
- restart the server;
- update linux kernel to newer version; and if it is already fresh enough, just try another version.
If this issue happens only on single server across large fleet then it is hardware issue. | 2020-01-26T11:00:15 | 2020-01-26T11:00:15 | {} | MEMBER |
ClickHouse | ClickHouse | 578,517,796 | 8,814 | filimonov | version 19.13.2.19?
May be it's just another manifestation of https://github.com/ClickHouse/ClickHouse/issues/7383 (it shouldn't happen on 19.13 unless you enable trace_log manually - it was enabled by default in 19.14).
Normally clickhouse rather not create so many threads... So i would focus on looking for the ... | 2020-01-26T16:21:38 | 2020-01-26T16:22:18 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 578,554,795 | 5,379 | Chocorean | I am no longer able to give you logs/info as I' not using it anymore. Shouild we close it?
| 2020-01-26T23:35:01 | 2020-01-26T23:35:01 | {} | NONE |
ClickHouse | ClickHouse | 578,734,771 | 8,841 | alexey-milovidov | > Why tar?
Allow to manipulate backup as a single unit.
Otherwise we will end up with huge number of files.
But the main motivation is how backups are already implemented:
https://clickhouse-datasets.s3.yandex.net/hits/partitions/hits_100m_obfuscated_v1.tar.xz
And the users also tend to the same solution:
htt... | 2020-01-27T12:57:24 | 2020-01-27T12:57:24 | {} | MEMBER |
ClickHouse | ClickHouse | 578,768,651 | 8,830 | zlobober | Fixed segfault, added supports_replication and supports_deduplication.
> Maybe just array?
I have no strong opinion on this, but n columns seems OK for me.
| 2020-01-27T14:19:29 | 2020-01-27T14:19:46 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 578,448,040 | 8,833 | alexey-milovidov | gcc 8.3 is so 2010s.
Just disable this specific warning for gcc versions less than 9.2. | 2020-01-25T22:18:32 | 2020-01-25T22:18:38 | {
"+1": 1
} | MEMBER |
ClickHouse | ClickHouse | 578,452,318 | 8,826 | alexey-milovidov | Current status: I doubt that this is ClickHouse issue. | 2020-01-25T23:17:17 | 2020-01-25T23:17:17 | {} | MEMBER |
ClickHouse | ClickHouse | 578,452,815 | 8,814 | alexey-milovidov | This happens if total number of query processing threads is too large.
What is the size of your cluster? How many queries do you process concurrently?
What is the value of max_threads setting? | 2020-01-25T23:25:07 | 2020-01-25T23:25:07 | {} | MEMBER |
ClickHouse | ClickHouse | 578,454,704 | 7,187 | alexey-milovidov | We have similar optimization in mind: https://github.com/ClickHouse/ClickHouse/issues/5329
We will implement this optimization but it will be disabled by default, because we don't use the same table snapshot to process subqueries. | 2020-01-25T23:57:45 | 2020-01-25T23:57:45 | {} | MEMBER |
ClickHouse | ClickHouse | 579,062,462 | 8,159 | nickevin | > @filimonov I'd like to enhance the document this week.
Is this work in progress? If so, please provide the document url. Thank you! | 2020-01-28T03:27:57 | 2020-01-28T03:27:57 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 579,120,431 | 8,060 | alexmgit | Это не дубль задачи #6556. И дело не типе колонки clickhouse. Даже просто выполнив запрос SELECT * FROM mysql_db._test_binary; получаю ошибку Unsupported type FixedString(1) (version 20.2.1.2229 (official build)). Т.е. clickhouse не понимает тип binary(1) в mysql. | 2020-01-28T07:44:52 | 2020-01-28T07:44:52 | {} | NONE |
ClickHouse | ClickHouse | 578,444,760 | 8,830 | zlobober | > Those that always return true/false.
This plan excludes almost all features (except for `supportsReplcation()` and `supportsDeduplication()`) as there is MaterializedView storage whose features depend on the target table.
Maybe we should introduce special value of 2 (or null) for this flag indicating that cert... | 2020-01-25T21:31:04 | 2020-01-25T21:31:04 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 578,458,538 | 6,833 | alexey-milovidov | The global limit is 10 000 threads. It should be enough in all but pathological cases.
Probably you are using `Merge` table on top of huge number of `Distributed` tables?
In that case, every `Distributed` table will create multiple connections and every connection will correspond to a thread on remote server.
(Y... | 2020-01-26T01:03:56 | 2020-01-26T01:04:39 | {} | MEMBER |
ClickHouse | ClickHouse | 577,758,138 | 8,303 | nikitamikhaylov | Benchmarking 16-cell cache dictionary.
#### Random binary from master:
``` bash
😼 @jakalletti:~/ClickHouse/clion-build-release/dbms/programs
↳ $ ./clickhouse-benchmark -c 4 -i 1000 < queries.txt
Loaded 18 queries.
Queries ... | 2020-01-23T16:26:08 | 2020-01-27T11:58:18 | {
"rocket": 1
} | MEMBER |
ClickHouse | ClickHouse | 578,728,934 | 8,692 | LittleMaverick | @mrxotey query was executed, but incorrectly )
see image below:
| 2020-01-27T12:40:45 | 2020-01-27T12:40:45 | {} | NONE |
ClickHouse | ClickHouse | 578,866,692 | 8,851 | kmatt | Does this PR support path style access in S3 ?
For example, if files are partitioned by Spark in a bucket as below, can wildcards be used to scan all files under a partial path ? For example `bucket/ymd=20191220/*/*.parquet` :
```
bucket/ymd=20191219/cid=1/part-01615-00100182-deb3-4f7d-ab32-a8d1cfe9f80a.c000.sna... | 2020-01-27T17:43:22 | 2020-01-27T17:43:22 | {} | NONE |
ClickHouse | ClickHouse | 579,098,840 | 8,812 | qoega | Diff is not only about query profiler. Can you split PR please? | 2020-01-28T06:22:39 | 2020-01-28T06:22:39 | {} | MEMBER |
ClickHouse | ClickHouse | 578,452,221 | 8,826 | alexey-milovidov | AFAIK, it may happen due to the following reasons:
1. The parent process did not read the exit code and status of child process (this is unlikely because the parent process is `init`).
2. OS spend some time freeing resources.
Does the issue fixes automatically after some time? How long does it take?
Does it hap... | 2020-01-25T23:16:18 | 2020-01-25T23:16:18 | {} | MEMBER |
ClickHouse | ClickHouse | 578,710,983 | 8,692 | mrxotey | @LittleMaverick could you try this query:
```sql
SELECT
status,
count() AS all
FROM temp.test ARRAY JOIN nested as nestedJoined
PREWHERE (id IN ('1', '2'))
WHERE (status IN (
SELECT status
FROM temp.test ARRAY JOIN nested as nestedJoined
GROUP BY status
ORDER BY count() DESC
L... | 2020-01-27T11:44:03 | 2020-01-27T11:44:03 | {} | NONE |
ClickHouse | ClickHouse | 578,845,655 | 8,677 | tchetwin | @alexey-milovidov could you indicate on the repository somewhere that you accept DCO and indicate how a user should indicate in a contribution that they agree to it. Thanks! | 2020-01-27T16:58:51 | 2020-01-27T16:58:51 | {} | NONE |
ClickHouse | ClickHouse | 578,964,038 | 7,851 | SaltTan | I've installed 20.1.2 on one node of the cluster that runs 19.16, and I'm facing the same problem.
What are your suggestions? | 2020-01-27T21:34:15 | 2020-01-27T21:34:15 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 593,993,078 | 9,492 | alexey-milovidov | @powturbo We will be happy to change it back whenever the fix is ready! | 2020-03-03T14:57:13 | 2020-03-03T14:57:13 | {} | MEMBER |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.