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 | 721,948,459 | 16,330 | UnamedRus | Does it make sense in case of lowCardinality(Nullable(String)) reserve one dictionary value for Null instead of creating another Nullable dictionary?
As low Cardinality usually has specific optimizations based on dictionary, Nullable can be treated specially. | 2020-11-04T20:08:41 | 2020-11-04T20:09:11 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 722,214,988 | 16,395 | macEar | @Jokser
We have several other services which use s3 + proxy, and there aren't problems with them. Could you please clarify about this issue? What is tunnel mode? Our Squid proxy works as a transparent proxy. I cannot understand why there are problems with Clickhouse while there are no problems with other systems. Inc... | 2020-11-05T08:09:25 | 2020-11-05T08:09:25 | {} | NONE |
ClickHouse | ClickHouse | 722,217,621 | 9,769 | Slach | @catalina-ch no progress
I propose to you use `remote()` or `remoteSecure()` table functions instead of `mysql()`
https://clickhouse.tech/docs/en/sql-reference/table-functions/remote/ | 2020-11-05T08:14:48 | 2020-11-05T08:14:48 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 721,714,312 | 7,895 | alexey-milovidov | @George3d6 It may happen due to parallel query execution, if every thread get the same `account_id`s in their aggregation state. You can lower the number of threads. | 2020-11-04T12:53:02 | 2020-11-04T12:53:02 | {} | MEMBER |
ClickHouse | ClickHouse | 721,795,027 | 16,287 | alexey-milovidov | Yes. | 2020-11-04T15:20:53 | 2020-11-04T15:20:53 | {} | MEMBER |
ClickHouse | ClickHouse | 721,802,440 | 16,214 | alexey-milovidov | @minaevmike WITH TOTALS works exactly as it supposed to work. | 2020-11-04T15:33:07 | 2020-11-04T15:33:07 | {} | MEMBER |
ClickHouse | ClickHouse | 721,963,072 | 16,671 | alexey-milovidov | Race condition in `DB::HashJoin::getTotalRowCount()`, @4ertus2
| 2020-11-04T20:39:41 | 2020-11-04T20:39:41 | {} | MEMBER |
ClickHouse | ClickHouse | 722,248,095 | 6,749 | baibaichen | @den-crane see https://github.com/ClickHouse/ClickHouse/pull/2217, looks replace is atomicity | 2020-11-05T09:13:45 | 2020-11-05T09:13:45 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 722,267,059 | 16,620 | akuzm | > Ok.
>
> I'd also upload raw query run times from all runs (so we can compare with various methods).
> Also it will be nice to upload all trace_log and query_log (but we need to upgrade our server in Yandex.Cloud to something reasonably powerful). And metric_log and asynchronous_metric_log if you feel greedy.
Q... | 2020-11-05T09:48:51 | 2020-11-05T09:48:51 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 721,802,779 | 16,214 | alexey-milovidov | It is not going to calculate sum unless you have an aggregate function sum in your query. | 2020-11-04T15:33:36 | 2020-11-04T15:33:36 | {} | MEMBER |
ClickHouse | ClickHouse | 721,831,464 | 14,190 | alexey-milovidov | @qoega You have copied the wrong query.
I checked it manually and found that this query cannot lead to this error. | 2020-11-04T16:22:34 | 2020-11-04T16:22:34 | {} | MEMBER |
ClickHouse | ClickHouse | 721,845,002 | 16,647 | gyuton | @alexey-milovidov It's not done yet, and also I haven't added changes to Russian version, so merging was a little bit early. | 2020-11-04T16:45:42 | 2020-11-04T16:45:42 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 721,875,028 | 7,088 | alexey-milovidov | Join is a key-value data structure for lookups. This is exactly what is needed in `transform`.
Currently `transform` builds very specialized hash table. Join data structure looks like a generalization. | 2020-11-04T17:39:21 | 2020-11-04T17:39:21 | {} | MEMBER |
ClickHouse | ClickHouse | 721,948,600 | 15,670 | SaltTan | It seems that in some cases CH removes parts from disk, but not from ZooKeeper.
In this example a few parts were merged into a bigger part, and then the table was detached less than `old_parts_lifetime` seconds later.
```
2020.11.03 22:17:21.722468 [ 13312 ] {} <Trace> db.table1: Executing log entry to merge parts 2... | 2020-11-04T20:08:59 | 2020-11-04T20:08:59 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 722,068,125 | 16,676 | den-crane | >config.xml: <max_threads>12</max_threads>
A setting 'max_threads' appeared at top level in config /etc/clickhouse-server/config.xml.
But it is user-level setting that should be located in users.xml inside section for specific profile.
You can add it to if you want to change default value of this setting.
You ... | 2020-11-05T01:30:45 | 2020-11-05T01:30:45 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 722,079,366 | 7,088 | amosbird | Ah, I thought the `join` layout means the JOIN engine that needs to be somehow passed into `transform`. | 2020-11-05T02:08:16 | 2020-11-05T02:08:16 | {} | COLLABORATOR |
ClickHouse | ClickHouse | 722,163,887 | 16,690 | wmaa0002 | thank u so much for ur answer~ | 2020-11-05T06:07:19 | 2020-11-05T06:07:19 | {} | NONE |
ClickHouse | ClickHouse | 722,250,723 | 16,658 | alexey-milovidov | > 00840_long_concurrent_select_and_drop_deadlock
Flaky test.
> Stress test (undefined) — Server failed to start
Known issue, @tavplubix | 2020-11-05T09:18:17 | 2020-11-05T09:18:17 | {} | MEMBER |
ClickHouse | ClickHouse | 722,284,924 | 16,701 | amosbird | It looks like a pre-calculated predicate. Perhaps we can have it generatlized.
For example,
```
alter table add index i_1 has(b) type predicate;
alter table add index i_2 equals(a) type predicate;
```
| 2020-11-05T10:21:53 | 2020-11-05T10:21:53 | {
"+1": 1
} | COLLABORATOR |
ClickHouse | ClickHouse | 722,308,444 | 16,330 | alexey-milovidov | @UnamedRus It is already implemented this way. | 2020-11-05T11:06:59 | 2020-11-05T11:06:59 | {
"+1": 1
} | MEMBER |
ClickHouse | ClickHouse | 721,850,364 | 7,088 | amosbird | > @amosbird Function `transform` allows to specify simple inline transformation without creating a dictionary.
> But currently it does not support all the ClickHouse types.
Ok.. but how can `Join` layout apply to the `transform` function? | 2020-11-04T16:55:12 | 2020-11-04T16:55:12 | {} | COLLABORATOR |
ClickHouse | ClickHouse | 722,082,167 | 16,283 | hexiaoting | The behaviour is correct in version 20.11 testing. so close this pr | 2020-11-05T02:17:12 | 2020-11-05T02:17:12 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 722,084,537 | 6,953 | lizhichao | > Sounds interesting.
>
> Actually log analysis a VERY popular usecase for ClickHouse.
> BUT: usually you have logs to analyze on one hosts (set of hosts), and clickhouse on another.
> Because of that there are tools which do something like 'tail' on one host and pass the logs to remote server.
>
> * logstash +... | 2020-11-05T02:24:37 | 2020-11-05T02:24:37 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 722,127,914 | 16,690 | den-crane | No table comments, only columns.
```sql
create table x( a Int64 comment 'some comment') Engine=Memory;
desc table x
┌─name─┬─type──┬─default_type─┬─default_expression─┬─comment──────┬─codec_expression─┬─ttl_expression─┐
│ a │ Int64 │ │ │ some comment │ │ ... | 2020-11-05T04:04:39 | 2020-11-05T04:04:56 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 722,253,698 | 15,087 | fastio | Hi team, Maybe we have another solution for clickhouse to support multi zookeeper clusters.
We can pass the zookeeper cluster information by extending more parameters when creating a replicated merge tree table.
For instance,
```
CREATE TABLE test (a String) ENGINE=ReplicatedMergeTree('/clickhouse/table/test', '{r... | 2020-11-05T09:23:42 | 2020-11-05T09:23:42 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 722,290,189 | 16,020 | alexey-milovidov | @hexiaoting Query Fuzzer is a tool that actively searches for random weird corner cases, most of them don't even have any sense :) | 2020-11-05T10:31:46 | 2020-11-05T10:31:46 | {} | MEMBER |
ClickHouse | ClickHouse | 722,310,703 | 16,622 | alexey-milovidov | BTW, it's very difficult to say confidently that it is faster.
We test on various datasets: integers, strings, small and large, highly and low compressable.
And on different CPU models: there are multiple differencies on what will be actually faster.
https://habr.com/en/company/yandex/blog/457612/ | 2020-11-05T11:11:21 | 2020-11-05T11:11:21 | {} | MEMBER |
ClickHouse | ClickHouse | 722,334,982 | 16,706 | alexey-milovidov | @yiguolei S3 as a disk is experimental feature, it is currently in beta.
ReplicatedMergeTree should not be used with DiskS3 in current version. If you use it, you will get excessive data copies.
@alex-zaitsev it is not merged yet. | 2020-11-05T12:02:12 | 2020-11-05T12:03:00 | {} | MEMBER |
ClickHouse | ClickHouse | 722,227,351 | 16,020 | KochetovNicolai | @hexiaoting this query was found by fuzz test. There is no meaning in query.
Multiplication of `AggregateFunction(groupArray(262145), UInt8)` and `UInt32` looks strange indeed. However, no query should crash server. | 2020-11-05T08:34:16 | 2020-11-05T08:34:16 | {} | MEMBER |
ClickHouse | ClickHouse | 722,247,892 | 16,679 | alexey-milovidov | I have already merged it yesterday with my edits. | 2020-11-05T09:13:21 | 2020-11-05T09:13:21 | {} | MEMBER |
ClickHouse | ClickHouse | 722,248,448 | 16,673 | alexey-milovidov | Fuzzer: #13281 | 2020-11-05T09:14:24 | 2020-11-05T09:14:24 | {} | MEMBER |
ClickHouse | ClickHouse | 722,163,138 | 16,689 | weiqxu | use cast is a workroud.
However it should be fixed since the behaviour should be the same when the table is local table or remote/distributed table. | 2020-11-05T06:05:22 | 2020-11-05T12:10:03 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 721,793,084 | 16,400 | alexey-milovidov | > Oracle allows adjusting that behavior
Oracle does not support argMin and argMax functions. | 2020-11-04T15:17:36 | 2020-11-04T15:17:36 | {} | MEMBER |
ClickHouse | ClickHouse | 721,970,527 | 16,487 | stumitchell | Is it changing the timezone of the result (i.e. always outputting `UTC`) would the solution be to preserve the timezone on the outputs? | 2020-11-04T20:55:55 | 2020-11-04T20:55:55 | {} | NONE |
ClickHouse | ClickHouse | 722,058,134 | 16,626 | den-crane | > if I set join_algorithm=auto,set default_max_bytes_in_join=max_memory_usage, hash can be used when memory is enough, and merge join is used when memory is not enough, is it right?
Is not.
CH uses HashJoin until it predicts that HashJoin will overuse memory in the next memory allocation. Sometimes this prediction... | 2020-11-05T00:58:10 | 2020-11-05T01:01:42 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 722,130,767 | 9,195 | den-crane | @baibaichen yes | 2020-11-05T04:16:38 | 2020-11-05T04:16:38 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 722,289,636 | 16,020 | alexey-milovidov | @KochetovNicolai Not strange at all. We support adding and multiplying aggregation states in a natural way. | 2020-11-05T10:30:39 | 2020-11-05T10:30:39 | {} | MEMBER |
ClickHouse | ClickHouse | 722,294,121 | 16,176 | alexey-milovidov | Yes, this is the way of implementation.
Care should be taken to adjust existing limit and offset if any.
Also a special case when WITH TIES or WITH FILL are present. | 2020-11-05T10:39:12 | 2020-11-05T10:39:12 | {} | MEMBER |
ClickHouse | ClickHouse | 722,314,869 | 16,400 | filimonov | > > Oracle allows adjusting that behavior
>
> Oracle does not support argMin and argMax functions.
It have LAST_VALUE(...) OVER. | 2020-11-05T11:19:30 | 2020-11-05T11:19:30 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 722,394,511 | 16,669 | StpMax | Hello
I have same issue with CH 20.1, config as default.
```
2020.10.29 08:17:35.951881 [ 90 ] {dad48d5e-77e3-4667-a237-236cd4cdc1c6} <Error> DynamicQueryHandler: Poco::Exception. Code: 1000, e.code() = 2026, e.displayText() = mysqlxx::ConnectionFailed: Unknown SSL error ((nullptr):47335), Stack trace (when copying ... | 2020-11-05T13:58:03 | 2020-11-05T13:58:03 | {} | NONE |
ClickHouse | ClickHouse | 721,735,817 | 16,654 | den-crane | No. Use the first example and join with these calendar | 2020-11-04T13:36:40 | 2020-11-04T13:36:40 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 721,795,893 | 16,287 | alexey-milovidov | You can use Managed ClickHouse on
- Yandex Cloud;
- Altinity Cloud;
- Mail.ru Cloud;
- Alibaba Cloud;
- Tencent Cloud.
You can use ClickHouse Kubernetes operator.
https://github.com/Altinity/clickhouse-operator
You can use ClickHouse Docker image. | 2020-11-04T15:22:17 | 2020-11-04T15:22:17 | {} | MEMBER |
ClickHouse | ClickHouse | 721,798,224 | 16,231 | alexey-milovidov | The behaviour is correct in version 20.11 testing. | 2020-11-04T15:26:05 | 2020-11-04T15:26:05 | {} | MEMBER |
ClickHouse | ClickHouse | 722,127,959 | 16,570 | JacobHayes | Are the failing tests flakey (guessing so from other PRs)? | 2020-11-05T04:04:51 | 2020-11-05T04:05:00 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 722,214,477 | 9,769 | catalina-ch | is there any progress? | 2020-11-05T08:08:24 | 2020-11-05T08:08:24 | {} | NONE |
ClickHouse | ClickHouse | 721,782,708 | 16,667 | den-crane | Such optimization has been implemented only for scalar subq (and documentation states about scalars).
```sql
SELECT (SELECT count(*) AS x FROM numbers(1000000) WHERE has(range(100), number)) AS main,
(SELECT count(*) AS x FROM numbers(1000000) WHERE has(range(100), number)) AS snd
┌─main─┬─snd─┐
│ 100 │ ... | 2020-11-04T15:00:04 | 2020-11-04T15:01:37 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 721,785,018 | 4,910 | marrocksd | This is really neccessary | 2020-11-04T15:03:53 | 2020-11-04T15:03:53 | {} | NONE |
ClickHouse | ClickHouse | 721,848,605 | 15,641 | alexey-milovidov | Duplicate #14974 | 2020-11-04T16:52:00 | 2020-11-04T16:52:00 | {} | MEMBER |
ClickHouse | ClickHouse | 722,025,548 | 16,635 | vzakaznikov | > @vzakaznikov Please help to debug TestFlows check.
Another timeout during `ALTER` command.
```
✘ 7m 57s [ Error ] /clickhouse/rbac/privileges/alter index/table_type='MergeTree'/user with privileges from role with grant option
ExpectTimeoutError
Traceback (most recent call last):
File "/usr... | 2020-11-04T23:15:44 | 2020-11-04T23:15:44 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 722,055,800 | 16,626 | chang0391 | if I set join_algorithm=auto,set default_max_bytes_in_join=max_memory_usage, hash can be used when memory is enough, and merge join is used when memory is not enough, is it right? | 2020-11-05T00:49:58 | 2020-11-05T00:49:58 | {} | NONE |
ClickHouse | ClickHouse | 722,066,107 | 16,676 | trollhe | > @trollhe Please provide an output of
>
> `cd /etc/clickhouse-server/; grep -Rl max_threads *`
bellow:
```
$:/etc/clickhouse-server# grep max_threads *
grep: config.d: Is a directory
config.xml: <max_threads>12</max_threads>
grep: preprocessed: No such file or directory
grep: users.d: Is a directory
... | 2020-11-05T01:23:47 | 2020-11-05T01:23:47 | {} | NONE |
ClickHouse | ClickHouse | 722,172,773 | 16,691 | amosbird | I've observed a similar issue but only with HTTPConnection in version 20.8.1.1. ClickHouse suddenly can only have one active HTTPConnection and queues up any other requests. However this issue is very rare and I haven't seen it for more than one month now. | 2020-11-05T06:31:02 | 2020-11-05T06:31:02 | {} | COLLABORATOR |
ClickHouse | ClickHouse | 722,126,749 | 16,689 | den-crane | use cast
```
select sum(cast(null,'Nullable(Float64)')) from remote('127.0.0.1,127.0.0.2', 'system', 'one') ;
┌─sum(cast(NULL, 'Nullable(Float64)'))─┐
│ ᴺᵁᴸᴸ │
└──────────────────────────────────────┘
``` | 2020-11-05T04:00:05 | 2020-11-05T04:00:05 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 722,244,762 | 6,168 | baibaichen | @YiuRULE
Is move partition atomicity as https://github.com/ClickHouse/ClickHouse/issues/4729 required? | 2020-11-05T09:07:27 | 2020-11-05T09:07:39 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 722,250,123 | 16,664 | alexey-milovidov | Why don't also update Arrow to the latest version? | 2020-11-05T09:17:18 | 2020-11-05T09:17:18 | {} | MEMBER |
ClickHouse | ClickHouse | 722,303,345 | 16,701 | fastio | @amosbird It's actually a good idea.
I have implemented this feature in my branch but it's not generalized. | 2020-11-05T10:56:54 | 2020-11-05T11:06:57 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 722,324,883 | 16,591 | alexey-milovidov | @akuzm I'm 100% agree. | 2020-11-05T11:39:56 | 2020-11-05T11:39:56 | {} | MEMBER |
ClickHouse | ClickHouse | 721,713,707 | 7,895 | alexey-milovidov | @den-crane You are reproducing a completely different use case that whan is mentioned in the first message.
In your use case, we have exactly one value in uniqExact for each value of GROUP BY key.
The memory usage is consistent with expectations.
| 2020-11-04T12:51:36 | 2020-11-04T12:53:15 | {} | MEMBER |
ClickHouse | ClickHouse | 721,715,417 | 7,895 | alexey-milovidov | `uniqExact` with high cardinality is special, because the `merge` operation has similar complexity to doing `update` for all elements again and parallel aggregation may not improve performance much. | 2020-11-04T12:55:23 | 2020-11-04T12:56:02 | {} | MEMBER |
ClickHouse | ClickHouse | 721,875,560 | 16,667 | adrian17 | In that case, can this part in documentation be fixed? No mention of scalars here.
https://clickhouse.tech/docs/en/sql-reference/functions/#common-subexpression-elimination | 2020-11-04T17:40:25 | 2020-11-04T17:40:25 | {} | NONE |
ClickHouse | ClickHouse | 722,075,119 | 16,645 | amosbird | Btw, I feel that we can provide some additional features to `clickhouse-benchmark`. Sometimes we may like to provide a query template with some parameters and rules, and the benchmark tool should generate random queries based on that. | 2020-11-05T01:53:52 | 2020-11-05T01:53:52 | {} | COLLABORATOR |
ClickHouse | ClickHouse | 722,186,490 | 16,020 | hexiaoting | @KochetovNicolai
what does y * 1048576 mean in this query?
multipy a AggregateFunction(groupArray(262145), UInt8) and UInt32 ???
```
SELECT groupArrayMerge(y * 1048576) FROM (SELECT groupArrayState(262145)(x) AS y FROM (SELECT 1 AS x)) 【failed】
SELECT groupArrayMerge(y * 1048576) FROM (SELECT groupArrayState(2... | 2020-11-05T07:04:39 | 2020-11-05T07:04:39 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 722,249,524 | 16,675 | alexey-milovidov | > 00840_long_concurrent_select_and_drop_deadlock
Flaky test. | 2020-11-05T09:16:14 | 2020-11-05T09:16:14 | {} | MEMBER |
ClickHouse | ClickHouse | 722,251,938 | 9,769 | catalina-ch | actually what i want to do is; connect to clickhouse from prestodb.
There is no legal Clickhouse connector for prestodb but Mysql connector is.
when connecting to clickhouse from prestodb using mysql protocol ; is perfect. But gets error while querying tables
Database INFORMATION_SCHEMA doesn't exist ...
ch ver... | 2020-11-05T09:20:33 | 2020-11-05T09:20:33 | {} | NONE |
ClickHouse | ClickHouse | 722,321,307 | 16,591 | akuzm | I'm wondering, does it make any sense for queries with File table engine to use the format settings from the current query context? I think not, the format settings are a property of the table, and you just break the table if you change that. OTOH, `file` table function should always use settings from current context. | 2020-11-05T11:32:20 | 2020-11-05T11:32:20 | {
"+1": 1
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 722,332,066 | 16,706 | alex-zaitsev | Yes, see https://github.com/ClickHouse/ClickHouse/pull/16240 | 2020-11-05T11:56:06 | 2020-11-05T11:56:06 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 721,813,121 | 9,248 | alexey-milovidov | (1) and (3) have been fixed. | 2020-11-04T15:51:25 | 2020-11-04T15:51:25 | {} | MEMBER |
ClickHouse | ClickHouse | 721,829,042 | 16,676 | den-crane | A setting 'max_threads' appeared at top level in config /etc/clickhouse-server/config.xml.
But it is user-level setting that should be located in users.xml inside <profiles> section for specific profile.
You can add it to <profiles><default> if you want to change default value of this setting.
You can also disabl... | 2020-11-04T16:18:34 | 2020-11-04T16:18:34 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 712,392,617 | 15,670 | den-crane | >Removing locally missing part from ZooKeeper and queuing a fetch: 1594965600_1_6_1
I've seen complains about this issue before. It hard to say is it a bug it was an issue in some version of CH.
Probably one of CH versions could leave info about partitions after drop partition.
I was never able to reproduce it. | 2020-10-19T19:27:41 | 2020-11-04T20:12:10 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 722,272,996 | 15,542 | hrissan | This pull request does not fix uncontrolled growth of TDigest and is only workaround. Here is short example of QuantileTDigest uncontrolled growth (though artifical):
DB::QuantileTDigest<double> qt0;
qt0.add(5, 4294967300);
for (size_t i = 0; i != 5000; ++i) {
qt0.add(5 + i, 1);
qt0.serialize(temp);
}
// qt0... | 2020-11-05T09:59:57 | 2020-11-05T09:59:57 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 722,303,295 | 16,624 | filimonov | Well, it may limit the IO a bit. And since Join / Set tables have that (they are also in-memory) persistence it would be more consistent to be able to enable the same approach for engine=Memory too.
But it's not an important task - actually I don't care when/if it will be implemented (I just wrote it down because i... | 2020-11-05T10:56:47 | 2020-11-05T10:56:47 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 721,791,943 | 16,487 | alexey-milovidov | The function timeSlots is using simple arithmetic on unix timestamp, it was not meant to take timezone into account. | 2020-11-04T15:15:37 | 2020-11-04T15:15:37 | {} | MEMBER |
ClickHouse | ClickHouse | 721,903,664 | 14,319 | vkingnew | ```
SELECT
roundDown(scores, [0, 60, 80, 90]) AS p,
count(user_id)
FROM scores
GROUP BY p
ORDER BY p ASC
┌─────p─┬─count(user_id)─┐
│ 0.00 │ 2 │
│ 60.00 │ 3 │
│ 80.00 │ 2 │
│ 90.00 │ 5 │
└───────┴────────────────┘
4 rows in set. Elapsed: 0.... | 2020-11-04T18:37:37 | 2020-11-04T18:37:37 | {
"+1": 1
} | NONE |
ClickHouse | ClickHouse | 721,918,953 | 16,678 | den-crane | This feature will be released in 20.11.
Documentation reflects the master code branch.
```
ClickHouse client version 20.11.1.5065.
Connecting to localhost:9000 as user default.
Connected to ClickHouse server version 20.11.1 revision 54442.
WITH t1 AS (SELECT 1) SELECT 2
WITH t1 AS
(
SELECT... | 2020-11-04T19:08:41 | 2020-11-04T19:08:41 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 722,248,584 | 16,673 | alexey-milovidov | > Stress test (undefined) — Server failed to start
Known issue, @tavplubix | 2020-11-05T09:14:39 | 2020-11-05T09:14:39 | {} | MEMBER |
ClickHouse | ClickHouse | 721,803,411 | 16,171 | alexey-milovidov | @KochetovNicolai
> DB::Exception: Expected ColumnLowCardinality, gotUInt8: While executing ReplacingSorted.
You have a bug: missing whitespace in exception message. | 2020-11-04T15:34:37 | 2020-11-04T15:34:37 | {} | MEMBER |
ClickHouse | ClickHouse | 722,127,667 | 16,176 | hexiaoting | @alexey-milovidov
hi, I think maybe we can implement in this way:
on clickhouse client execute below queries:
1. create table xxxx;
2. select xxxx [not effected by limit and offset settings]
3. set limit=50;
4. set offset=10;
5. select xxxx [this will be affected by the new settings's limit and offset]
We can... | 2020-11-05T04:03:36 | 2020-11-05T04:04:00 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 722,223,931 | 16,308 | KochetovNicolai | `00840_long_concurrent_select_and_drop_deadlock` is flappy | 2020-11-05T08:27:24 | 2020-11-05T08:27:24 | {} | MEMBER |
ClickHouse | ClickHouse | 722,311,267 | 7,088 | alexey-milovidov | Yes, it's about reuse of the internal data structure: basically a hash table pointing to blocks. | 2020-11-05T11:12:32 | 2020-11-05T11:12:32 | {} | MEMBER |
ClickHouse | ClickHouse | 722,346,370 | 16,642 | CurtizJ | Stress test:
```
DB::Exception: Syntax error (in file /var/lib/clickhouse/store/4c4/4c44cac1-9334-4846-8471-fbb0a18861f5/test_213E705C47512C7467516A414669482F207B657E4B3524482C7B236C506D505A5165636627335B53.sql): failed at position 21 (''5050d23e-4637-47e1-8e13-dc'): '5050d23e-4637-47e1-8e13-dc. Single quoted string... | 2020-11-05T12:26:02 | 2020-11-05T12:26:02 | {} | MEMBER |
ClickHouse | ClickHouse | 721,796,939 | 16,258 | alexey-milovidov | Most likely not going to be changed. This behaviour should be documented. | 2020-11-04T15:24:05 | 2020-11-04T15:24:05 | {} | MEMBER |
ClickHouse | ClickHouse | 722,128,899 | 9,195 | baibaichen | @den-crane if I use `insert into xx select` where xx is only a `MergeTree` without replication. what does ClickHouse guarantee in case of a failure? Does the following still hold?
> It could be atomic if this insert creates only one part.
>
> * only if it inserts into only one partition
> * only if number of ro... | 2020-11-05T04:09:03 | 2020-11-05T04:09:03 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 722,249,376 | 16,675 | alexey-milovidov | Fuzzer: #15889 @KochetovNicolai | 2020-11-05T09:15:59 | 2020-11-05T09:15:59 | {} | MEMBER |
ClickHouse | ClickHouse | 722,307,560 | 16,176 | alexey-milovidov | We can adjust existing limits if there are no WITH TIES, WITH FILL and wrap into another LimitTransform if there are. | 2020-11-05T11:05:12 | 2020-11-05T11:05:12 | {} | MEMBER |
ClickHouse | ClickHouse | 721,715,874 | 16,664 | FawnD2 | > Also you probably want to switch the commit sha of the submodule.
Thank you for your comment
Submodule has been synced | 2020-11-04T12:56:28 | 2020-11-04T12:56:28 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 715,337,468 | 16,278 | den-crane | Check /var/log/clickhouse-server/stderr.log and /var/log/clickhouse-server/clickhouse-server.log
Or start CH without daemonizing and check an output
sudo -u clickhouse clickhouse-server --pid-file=/var/run/clickhouse-server/clickhouse-server.pid --config-file=/etc/clickhouse-server/config.xml | 2020-10-23T13:22:21 | 2020-11-04T15:52:58 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 721,831,931 | 14,190 | alexey-milovidov | I can easily reproduce the ubsan issue with the following query:
```
SET max_threads = nan
``` | 2020-11-04T16:23:22 | 2020-11-04T16:23:22 | {} | MEMBER |
ClickHouse | ClickHouse | 721,860,953 | 15,641 | nvartolomei | @alexey-milovidov the stack traces look different /shrug | 2020-11-04T17:13:14 | 2020-11-04T17:13:14 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 722,152,230 | 15,419 | myrrc | 00840_long_concurrent_select_and_drop_deadlock is flapping | 2020-11-05T05:32:50 | 2020-11-05T05:32:50 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 722,258,901 | 16,338 | akuzm | I merged to master and fixed some incompatibility with recent changes.
I noticed that now the union list is not flattened, so there is some weird duality where we have different plans for logically equivalent queries, depending on the brackets. E.g., `UNION` is a left-associative binary operator, so the queries `sel... | 2020-11-05T09:33:29 | 2020-11-05T09:33:29 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 722,350,339 | 16,700 | alexey-milovidov | Arcadia check Ok. | 2020-11-05T12:34:30 | 2020-11-05T12:34:30 | {} | MEMBER |
ClickHouse | ClickHouse | 721,735,089 | 16,666 | den-crane | ```
SELECT d, t from ( WITH
[1, 2, 3] AS d1,
['a', 'e', 'f'] AS t1 select d1, t1)
array Join d1 as d, t1 as t
┌─d─┬─t─┐
│ 1 │ a │
│ 2 │ e │
│ 3 │ f │
└───┴───┘
``` | 2020-11-04T13:35:20 | 2020-11-04T13:35:20 | {
"+1": 1
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 721,829,820 | 16,676 | den-crane | @trollhe Please provide an output of
`cd /etc/clickhouse-server/; grep -Rl max_threads *` | 2020-11-04T16:19:53 | 2020-11-04T16:19:53 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 721,926,915 | 16,678 | den-crane | I am wrong.
It is supported in the stable already.
```sql
ClickHouse client version 20.10.3.30 (official build).
Connecting to localhost:9000 as user default.
Connected to ClickHouse server version 20.10.3 revision 54441.
8a861f168fda :) WITH t1 AS (SELECT 1) SELECT 2
WITH t1 AS
(
SELECT 1... | 2020-11-04T19:23:54 | 2020-11-04T19:23:54 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 721,962,524 | 16,647 | den-crane | @gyuton
To create a pull request that is ready for review, click Create Pull Request.
**To create a draft pull request**, use the drop-down and select Create Draft Pull Request, then click Draft Pull Request. For more information about draft pull requests,
https://docs.github.com/en/free-pro-team@latest/github... | 2020-11-04T20:38:26 | 2020-11-04T20:38:26 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 722,219,765 | 16,605 | Slach | @zhang2014 could you look to https://github.com/ClickHouse/ClickHouse/issues/16605#issuecomment-721468826
and help us resolve this issue? | 2020-11-05T08:18:59 | 2020-11-05T08:18:59 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 722,275,139 | 9,769 | Slach | @catalina-ch clickhouse doesn't support MySQL SQL dialect and internal tables, it supports only MySQL wire protocol
could you try to connect via JDBC from Prestro?
| 2020-11-05T10:03:49 | 2020-11-05T10:03:49 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 716,138,805 | 16,265 | l1t1 | the numbers here are all very close to -2^31 , the source data minus the converted data are all the same value.
```python
>>> 2**31
2147483648
>>> 0.3333 + 2147483653.2202
2147483653.5535
>>> 0.5000 + 2147483653.0535
2147483653.5535
>>> 0.1667 + 2147483653.3868
2147483653.5534997
>>> 0.0833 + 2147483653.4702... | 2020-10-25T12:18:41 | 2020-10-25T12:23:48 | {} | NONE |
ClickHouse | ClickHouse | 716,377,622 | 16,260 | vpanfilov | Function `minus` is not working either (tested on version 20.9.2):
```sql
SELECT now64() - now64()
Received exception from server (version 20.9.2):
Code: 43. DB::Exception: Received from localhost:9000. DB::Exception: Illegal types DateTime64(3) and DateTime64(3) of arguments of function minus.
```
though it wo... | 2020-10-26T07:55:38 | 2020-10-26T07:55:38 | {} | NONE |
ClickHouse | ClickHouse | 716,424,934 | 16,095 | KochetovNicolai | Closing after #16242 | 2020-10-26T09:28:47 | 2020-10-26T09:28:47 | {} | MEMBER |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.