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 | 857,327,074 | 24,888 | alexey-milovidov | `tests/queries/skip_list.json` | 2021-06-09T02:41:45 | 2021-06-09T02:41:45 | {} | MEMBER |
ClickHouse | ClickHouse | 857,328,642 | 24,728 | alexey-milovidov | @Mergifyio update | 2021-06-09T02:45:56 | 2021-06-09T02:45:56 | {
"+1": 1
} | MEMBER |
ClickHouse | ClickHouse | 856,358,399 | 24,763 | alexey-milovidov | @mergifyio update | 2021-06-08T00:54:04 | 2021-06-08T00:54:04 | {
"+1": 1
} | MEMBER |
ClickHouse | ClickHouse | 856,712,164 | 23,991 | tangxueming | > you need to upgrade to 20.12.8.5
> or set read_backoff_min_latency_ms=0
>
> duplicate #18137
Hi， what is the influence when I set read_backoff_min_latency_ms= 0 ？
I don't found this param in document.
Thanks! | 2021-06-08T12:13:33 | 2021-06-08T12:13:33 | {} | NONE |
ClickHouse | ClickHouse | 856,724,918 | 25,070 | den-crane | >the partitioning key for all my tables is of type (UUID, UInt32) (e.g PARTITION BY (Id, toYYYYMM(timestamp)) )
Can you share an example of such "create table" ?
>In the system tables, some tables have data from before the upgrade, while others seem to have been truncated
No, check `system.trace_log_0, system.... | 2021-06-08T12:31:55 | 2021-06-08T12:31:55 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 857,217,221 | 25,077 | alexey-milovidov | @filimonov `Map` data type is an experimental feature not ready for production usage. | 2021-06-08T22:20:50 | 2021-06-08T22:20:50 | {} | MEMBER |
ClickHouse | ClickHouse | 857,323,495 | 25,072 | alexey-milovidov | > test_distributed_queries_stress
WTF | 2021-06-09T02:32:01 | 2021-06-09T02:32:01 | {} | MEMBER |
ClickHouse | ClickHouse | 857,326,957 | 24,888 | alexey-milovidov | The test needs to be added into the skip list for ANTLR and parallel checks. | 2021-06-09T02:41:23 | 2021-06-09T02:41:23 | {} | MEMBER |
ClickHouse | ClickHouse | 857,327,482 | 24,358 | alexey-milovidov | Let's reopen it later. | 2021-06-09T02:42:43 | 2021-06-09T02:42:43 | {} | MEMBER |
ClickHouse | ClickHouse | 857,332,900 | 25,068 | qingping209 | > 21.4 is out of support.
> It's not the official build? Did you build Clickhouse yourself?
Yes. I built Clickhouse myself. | 2021-06-09T02:57:37 | 2021-06-09T02:57:37 | {} | NONE |
ClickHouse | ClickHouse | 856,138,989 | 24,158 | genzgd | The first crash was on the buffer table (and underlying data table) at the top of the thread.
The second stack trace is from the following buffer table:
```CREATE TABLE comcast_xcr.origin_avail_1m
(
`datetime` DateTime,
`svc_type` LowCardinality(String),
`routing_type` LowCardinality(String),
... | 2021-06-07T17:49:00 | 2021-06-07T17:49:00 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 856,356,270 | 25,032 | juyonychen | thanks！ | 2021-06-08T00:47:36 | 2021-06-08T00:47:36 | {} | NONE |
ClickHouse | ClickHouse | 856,400,802 | 24,932 | den-crane | >DB::Exception: There is no subcolumn format in type Nested(topic Int32)
Can you clarify how did fix the structure? | 2021-06-08T03:02:21 | 2021-06-08T03:02:21 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 856,405,273 | 25,063 | anneji-dev | `When creating a database, first create the data directory metadata_path, and then create metadata_file_path，but when create metadata_file_path failed, then the data directory always there,cause the next creation to report an error, so, we should not judge the metadata_path exists.` | 2021-06-08T03:14:46 | 2021-06-08T03:14:46 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 856,518,226 | 19,970 | r-zenine | It's seems that the integration with `rust` was attempted by the community already here
https://github.com/NeowayLabs/ClickHouse/commits/fulltext-21.3
Would you welcome such approach in Clickhouse ? | 2021-06-08T07:13:21 | 2021-06-08T07:13:21 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 856,686,054 | 24,885 | orloffv | let's merge? | 2021-06-08T11:33:08 | 2021-06-08T11:33:08 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 856,779,372 | 25,022 | alexey-milovidov | @mergifyio update | 2021-06-08T13:42:38 | 2021-06-08T13:42:38 | {
"+1": 1
} | MEMBER |
ClickHouse | ClickHouse | 856,844,386 | 24,920 | KochetovNicolai | All fails in flacky check are timeouts. | 2021-06-08T15:00:09 | 2021-06-08T15:00:09 | {} | MEMBER |
ClickHouse | ClickHouse | 856,958,151 | 24,932 | khuenth | >
>
> > DB::Exception: There is no subcolumn format in type Nested(topic Int32)
>
> Can you clarify how did fix the structure?
My colleague just change from nested to every single column. Then It fixed. | 2021-06-08T17:31:09 | 2021-06-08T17:31:09 | {} | NONE |
ClickHouse | ClickHouse | 857,211,688 | 25,068 | alexey-milovidov | @qingping209 This is your own alternative build of ClickHouse.
The official build does not have this issue. | 2021-06-08T22:13:17 | 2021-06-08T22:13:17 | {} | MEMBER |
ClickHouse | ClickHouse | 857,264,905 | 25,067 | alexey-milovidov | The potential harm is that when people will see a file named VersionNumber they may start to think that we use the same class for ClickHouse version numbers. | 2021-06-08T23:45:24 | 2021-06-08T23:45:24 | {} | MEMBER |
ClickHouse | ClickHouse | 857,265,123 | 25,067 | alexey-milovidov | So, to solve this issue, we can bury this class into some file where no one will look. | 2021-06-08T23:46:00 | 2021-06-08T23:46:00 | {} | MEMBER |
ClickHouse | ClickHouse | 857,333,524 | 25,068 | qingping209 | > @qingping209 This is your own alternative build of ClickHouse.
> The official build does not have this issue.
I followed the build instruction. So which version do you recommend to use. And what's the difference of the official build from the build instruction ? anything need special attention ? | 2021-06-09T02:59:32 | 2021-06-09T02:59:32 | {} | NONE |
ClickHouse | ClickHouse | 856,618,887 | 22,535 | CurtizJ | 15 failed tests are related to implementation specific details: checksums, number of files, etc. | 2021-06-08T09:35:13 | 2021-06-08T09:35:13 | {} | MEMBER |
ClickHouse | ClickHouse | 856,723,195 | 25,071 | den-crane | CH needs to have a list of parts which should be mutated before a mutation started to execute because only active parts which are exists now should mutated.
Partition pruning is not implemented for Mutations yet.
You can do "pruning" manually:
UPDATE IN PARTITION
DELETE IN PARTITION
https://clickhouse.tech/do... | 2021-06-08T12:29:33 | 2021-06-08T12:29:33 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 856,736,001 | 23,518 | nikitamikhaylov | I think we can merge with failed flaky test check. I don't know exactly what is the reason of the fault. But if it is a really flaky test, then we will see it failed often because of parallelized tests execution. | 2021-06-08T12:47:41 | 2021-06-08T12:47:41 | {} | MEMBER |
ClickHouse | ClickHouse | 856,859,013 | 18,740 | filimonov | > Old configuration continue to work.
Not all of them. Some people were using substitutions file and were relying on existing of those `incl` in default config.
Such changes should be mentioned in changelog. | 2021-06-08T15:13:53 | 2021-06-08T15:15:02 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 857,076,088 | 25,085 | NAlexandrov | Ok... Maybe then at least the ClickHouse will report an error and not crash | 2021-06-08T20:00:21 | 2021-06-08T20:00:21 | {} | NONE |
ClickHouse | ClickHouse | 857,216,655 | 25,058 | alexey-milovidov | if max_memory_usage is 10 GB, I recommend to set max_bytes_before_external_group_by to 5 GB. | 2021-06-08T22:20:00 | 2021-06-08T22:20:00 | {} | MEMBER |
ClickHouse | ClickHouse | 856,212,518 | 25,057 | vzakaznikov |
Even without a Nullable column we can get random values from `leadInFrame`.
```
user-host :) select empno, salary, leadInFrame(salary,1,8) OVER (ORDER BY salary) AS lead FROM empsalary ORDER BY empno
SELECT
empno,
salary,
leadInFrame(salary, 1, 8) OVER (ORDER BY salary ASC) AS lead
FROM empsala... | 2021-06-07T19:54:45 | 2021-06-07T19:54:45 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 856,357,733 | 23,976 | alexey-milovidov | @mergifyio update | 2021-06-08T00:52:04 | 2021-06-08T00:52:04 | {
"+1": 1
} | MEMBER |
ClickHouse | ClickHouse | 856,357,764 | 23,976 | mergify[bot] | **Command `update`: failure**
> **Pull request can't be updated with latest base branch changes**
> Mergify needs the author permission to update the base branch of the pull request.
> excitoon-favorites needs to [authorize modification on its head branch](https://help.github.com/articles/allowing-changes-to-a-pull-req... | 2021-06-08T00:52:12 | 2021-06-08T00:52:12 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 856,563,698 | 24,207 | danlark1 | Still looks like a proof of concept | 2021-06-08T08:19:46 | 2021-06-08T08:19:46 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 856,739,731 | 25,070 | larspars | > > the partitioning key for all my tables is of type (UUID, UInt32) (e.g PARTITION BY (Id, toYYYYMM(timestamp)) )
>
> Can you share an example of such "create table" ?
>
```sql
CREATE TABLE mydb.counts
(
`id` UUID,
`thing` LowCardinality(String),
`timestamp` DateTime CODEC(DoubleDelta, LZ4),
... | 2021-06-08T12:52:45 | 2021-06-08T12:52:45 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 856,994,428 | 25,085 | den-crane | Dictionaries do not support Arrays, it's not implemented https://github.com/ClickHouse/ClickHouse/issues/2135 | 2021-06-08T18:23:41 | 2021-06-08T18:23:41 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 857,114,670 | 25,059 | alesapin | Ok, no related failures. | 2021-06-08T20:36:42 | 2021-06-08T20:36:42 | {} | MEMBER |
ClickHouse | ClickHouse | 857,300,743 | 25,071 | zhangjmruc | @den-crane Thanks Denny for the reply. By the way, is there any process on the light delete and update implimentation? | 2021-06-09T01:26:02 | 2021-06-09T01:26:02 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 856,356,597 | 24,538 | alexey-milovidov | @tavplubix It is broken here: https://github.com/ClickHouse/ClickHouse/pull/24523 | 2021-06-08T00:48:37 | 2021-06-08T00:48:37 | {} | MEMBER |
ClickHouse | ClickHouse | 857,322,828 | 25,103 | alexey-milovidov | The changelog entry must be readable for ClickHouse users. | 2021-06-09T02:30:18 | 2021-06-09T02:30:18 | {} | MEMBER |
ClickHouse | ClickHouse | 856,744,399 | 24,069 | vzakaznikov | > We need to either remove TestFlows or to fix them @vzakaznikov
> Keeping flaky tests in master is not an option.
TestFlows check failed due to extended precision data type test. We will fix it.
```
Assertion values
assert that(snapshot("\n" + r.output.strip() + "\n", "tests", name=name, encoder=str... | 2021-06-08T12:58:41 | 2021-06-08T12:58:41 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 856,779,336 | 25,070 | larspars | All the timestamps were essentially set to `now()`, so the problem shouldn't be that in this case. The lost partitions are from the previous month, and the month before (I've not been using CH very long).
Here are the other definitions I've used. One table is defined like this:
```sql
CREATE TABLE mydb.measures... | 2021-06-08T13:42:35 | 2021-06-08T13:42:35 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 857,328,919 | 23,311 | alexey-milovidov | @Avogar it still needs your changes... | 2021-06-09T02:46:38 | 2021-06-09T02:46:38 | {} | MEMBER |
ClickHouse | ClickHouse | 857,329,218 | 23,976 | alexey-milovidov | @excitoon You did not provide access to your fork. | 2021-06-09T02:47:28 | 2021-06-09T02:47:28 | {} | MEMBER |
ClickHouse | ClickHouse | 857,329,220 | 24,728 | mergify[bot] | **Command `update`: success**
> **Branch has been successfully updated**
>
| 2021-06-09T02:47:28 | 2021-06-09T02:47:28 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 857,338,693 | 23,991 | tangxueming | @den-crane Thanks!
In this case , Ch return less data, due to some threads read slowly ? | 2021-06-09T03:13:39 | 2021-06-09T03:13:39 | {} | NONE |
ClickHouse | ClickHouse | 856,358,490 | 24,763 | mergify[bot] | **Command `update`: success**
> **Branch has been successfully updated**
>
| 2021-06-08T00:54:18 | 2021-06-08T00:54:18 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 856,364,773 | 17,623 | melin | The use of clickhouse distributed tables is complex and does not support dynamic expansion, which brings great maintenance costs to the use of clickhouse, especially in traditional enterprises, such as banks, who do not support dynamic expansion will be rejected. To solve these two problems, clickhouse use scenarios wi... | 2021-06-08T01:12:55 | 2021-06-08T01:12:55 | {
"+1": 1
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 856,382,998 | 24,958 | den-crane | ```
select uniq(number) from numbers(1000000000);
┌─uniq(number)─┐
│ 1001803777 │
└──────────────┘
1 rows in set. Elapsed: 2.466 sec. Processed 1.00 billion rows, 8.00 GB (405.58 million rows/s., 3.24 GB/s.)
SELECT uniqExact(number) FROM numbers(1000000000);
┌─uniqExact(number)─┐
│ 1000000000 │... | 2021-06-08T02:09:24 | 2021-06-08T02:09:24 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 856,397,704 | 24,932 | khuenth | I found out what problem i got. I tried to query the table system.replication_queue. It showed many row with last_exception col which show my issue. My issue is : **Code: 44, e.displayText() = DB::Exception: There is no subcolumn format in type Nested(topic Int32) (version 21.4.7.3 (official build))**
I tried to chang... | 2021-06-08T02:53:24 | 2021-06-08T02:53:24 | {} | NONE |
ClickHouse | ClickHouse | 856,398,588 | 18,721 | davidshtian | > > > Hi @davidshtian I have the similar certificate issue and tested on multiple versions including the latest one. But all failed.
> > > I wonder if there is any issue with my s3 bucket configuration. Could help share some information on your s3 bucket config? Does it have to be a public bucket?
> > > Kindly lookin... | 2021-06-08T02:55:58 | 2021-06-08T02:55:58 | {} | NONE |
ClickHouse | ClickHouse | 856,704,671 | 25,077 | Avogar | I know what the problem is, will fix | 2021-06-08T12:03:24 | 2021-06-08T12:03:24 | {} | MEMBER |
ClickHouse | ClickHouse | 857,164,447 | 25,028 | alexey-milovidov | Ok, I've installed 21.7 (master):
```
milovidov@github-explorer:~$ wget https://clickhouse-builds.s3.yandex.net/0/9291834e3691115f5fa238c9ae34dac7a1745141/clickhouse_build_check/clang-11_relwithdebuginfo_none_bundled_unsplitted_disable_False_binary/clickhouse
--2021-06-08 21:18:28-- https://clickhouse-builds.s3.y... | 2021-06-08T21:22:44 | 2021-06-08T21:22:44 | {} | MEMBER |
ClickHouse | ClickHouse | 857,263,893 | 15,065 | pennyfea | @alexey-milovidov No, in regards to help from the university I attended. I'm solely attempting to do this to gain real world experience and improve my C++ skills. | 2021-06-08T23:42:51 | 2021-06-08T23:42:51 | {
"+1": 1
} | NONE |
ClickHouse | ClickHouse | 857,312,910 | 25,028 | l1t1 | thanks | 2021-06-09T02:02:17 | 2021-06-09T02:02:17 | {} | NONE |
ClickHouse | ClickHouse | 856,356,040 | 25,030 | juyonychen | tks。 | 2021-06-08T00:46:57 | 2021-06-08T00:46:57 | {} | NONE |
ClickHouse | ClickHouse | 856,728,637 | 23,991 | den-crane | @tangxueming it allows to equalize I/O among threads if a query needs to read data from several disks (multi-disk volume / JBOD). Multiple threads read from different disks the same time.
```
clickhouse-client --help|grep backoff
--read_backoff_min_latency_ms arg Setting to reduce ... | 2021-06-08T12:37:28 | 2021-06-08T12:37:28 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 857,064,111 | 18,740 | alexey-milovidov | Не припоминаю изменений в этой области. | 2021-06-08T19:48:30 | 2021-06-08T19:48:30 | {} | MEMBER |
ClickHouse | ClickHouse | 856,358,738 | 24,897 | alexey-milovidov | > Stress test (thread) — Fatal message in clickhouse-server.log
What is this message and why it is appeared? | 2021-06-08T00:54:59 | 2021-06-08T00:54:59 | {} | MEMBER |
ClickHouse | ClickHouse | 856,519,333 | 18,721 | lingcuv | > > > > Hi @davidshtian I have the similar certificate issue and tested on multiple versions including the latest one. But all failed.
> > > > I wonder if there is any issue with my s3 bucket configuration. Could help share some information on your s3 bucket config? Does it have to be a public bucket?
> > > > Kindly ... | 2021-06-08T07:15:02 | 2021-06-08T07:15:02 | {} | NONE |
ClickHouse | ClickHouse | 857,250,241 | 24,341 | buyology | Thanks so much for working on this.
I tried this PR as I thought it could help us to load some of our Parquet files into ClickHouse.
### The table
```sql
:) CREATE TABLE t ( v Nested(a String, b String) ) ENGINE=Memory()
```
### An array of structs
```bash
» parquet-tools schema f1
message schema {
... | 2021-06-08T23:11:32 | 2021-06-08T23:11:32 | {} | NONE |
ClickHouse | ClickHouse | 857,264,458 | 25,067 | alexey-milovidov | Then let's make it support arbitrary number of components (simple std::vector) | 2021-06-08T23:44:14 | 2021-06-08T23:44:14 | {} | MEMBER |
ClickHouse | ClickHouse | 857,326,073 | 25,082 | alexey-milovidov | Please do not merge before looking at the checks. | 2021-06-09T02:39:18 | 2021-06-09T02:39:18 | {} | MEMBER |
ClickHouse | ClickHouse | 857,343,795 | 25,058 | sherrychenzhang | > @sherrychenzhang There is nothing surprising if 1 GB of compressed data requires 90 GB of in memory data structures for aggregation.
but we are using like 1GB of uncompressed data. | 2021-06-09T03:29:34 | 2021-06-09T03:29:34 | {} | NONE |
ClickHouse | ClickHouse | 857,221,137 | 24,901 | alexey-milovidov | Ok, now we have ssh access... | 2021-06-08T22:26:47 | 2021-06-08T22:26:47 | {} | MEMBER |
ClickHouse | ClickHouse | 857,265,477 | 25,067 | alexey-milovidov | And if we do it, I don't care that it only supports three-component versions. | 2021-06-08T23:46:58 | 2021-06-08T23:46:58 | {} | MEMBER |
ClickHouse | ClickHouse | 857,324,271 | 25,061 | alexey-milovidov | ```
milovidov@milovidov-desktop:~/work/ClickHouse/tests/queries$ rg LOGS_LEVEL
shell_config.sh
7:export CLICKHOUSE_CLIENT_SERVER_LOGS_LEVEL=${CLICKHOUSE_CLIENT_SERVER_LOGS_LEVEL:="warning"}
19:[ -v CLICKHOUSE_CLIENT_SERVER_LOGS_LEVEL ] && CLICKHOUSE_CLIENT_OPT0+=" --send_logs_level=${CLICKHOUSE_CLIENT_SERVER_LOGS_L... | 2021-06-09T02:34:26 | 2021-06-09T02:34:26 | {} | MEMBER |
ClickHouse | ClickHouse | 856,381,111 | 24,958 | den-crane | uniq function is approximate. It uses probabilistic structures / algorithms to store smaller number of bytes but gives decent result.
>SETTINGS index_granularity = 16384,
You need to use the opposite to achieve better performance with tables which store uniqStates. Try index_granularity=256.
| 2021-06-08T02:04:23 | 2021-06-08T02:04:23 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 856,527,966 | 25,058 | MrSsunlight | set max_memory_usage | 2021-06-08T07:27:40 | 2021-06-08T07:27:40 | {} | NONE |
ClickHouse | ClickHouse | 856,730,570 | 25,068 | den-crane | 21.4 is out of support.
It's not the official build? Did you build Clickhouse yourself? | 2021-06-08T12:40:16 | 2021-06-08T12:40:16 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 856,781,148 | 25,022 | mergify[bot] | **Command `update`: success**
> **Branch has been successfully updated**
>
| 2021-06-08T13:44:49 | 2021-06-08T13:44:49 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 857,077,558 | 18,740 | den-crane | да, это так было изначально, всегда.
Проблема что этот PR стрельнул и поменял поведение.
Этот PR "убрал" incl из первого вхождения тега. | 2021-06-08T20:01:41 | 2021-06-08T20:03:01 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 857,265,930 | 25,067 | alexey-milovidov | People also may mistakenly think that there is something good in this code, but actually it is not suitable for typical version numbers (e.g. OpenSSL version that have alphabetical components). | 2021-06-08T23:48:08 | 2021-06-08T23:48:08 | {} | MEMBER |
ClickHouse | ClickHouse | 857,276,294 | 22,398 | alexey-milovidov | ClickHouse vs. TimescaleDB:
https://twitter.com/ThisIsFernandez/status/1402369109594148865 | 2021-06-09T00:12:49 | 2021-06-09T00:12:49 | {} | MEMBER |
ClickHouse | ClickHouse | 857,330,883 | 25,058 | alexey-milovidov | @sherrychenzhang There is nothing surprising if 1 GB of compressed data requires 90 GB of in memory data structures for aggregation. | 2021-06-09T02:52:30 | 2021-06-09T02:52:30 | {} | MEMBER |
ClickHouse | ClickHouse | 856,785,423 | 25,070 | larspars | > PS. Partition by UUID is not always a good idea. It only makes sense if you have very small number of UUIDs.
Understood. It will be fewer than 1000 in my case.
Thanks for being so responsive here in the github issues. I've been thoroughly impressed by Clickhouse so far, and have seen performance that makes me f... | 2021-06-08T13:49:53 | 2021-06-08T13:49:53 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 857,213,981 | 25,086 | alexey-milovidov | Let's build with clang-11 or clang-12. | 2021-06-08T22:16:12 | 2021-06-08T22:16:12 | {} | MEMBER |
ClickHouse | ClickHouse | 857,215,040 | 25,062 | alexey-milovidov | Yes, it might be an issue with network connection.
Does it reproduce consistenly or only occasionally? | 2021-06-08T22:17:33 | 2021-06-08T22:17:33 | {} | MEMBER |
ClickHouse | ClickHouse | 857,216,404 | 25,058 | alexey-milovidov | If you set `max_bytes_before_external_group_by` to a very low value, large number of temporary files will be generated, and (paradoxically) the amount of memory needed to merge these temporary files and finish the query will be even higher. | 2021-06-08T22:19:37 | 2021-06-08T22:19:37 | {} | MEMBER |
ClickHouse | ClickHouse | 857,265,582 | 25,067 | azat | >Then let's make it support arbitrary number of components (simple std::vector)
Ok.
(I though about this initially, but decided to keep it simpler, by using tuple, i.e. compare is much simpler with it) | 2021-06-08T23:47:16 | 2021-06-08T23:47:16 | {} | MEMBER |
ClickHouse | ClickHouse | 857,275,797 | 22,398 | alexey-milovidov | ClickHouse vs TimescaleDB vs InfluxDB vs. QuasarDB:
https://blog.quasardb.net/benchmarking-timeseries-ingress | 2021-06-09T00:11:33 | 2021-06-09T00:11:33 | {} | MEMBER |
ClickHouse | ClickHouse | 857,322,715 | 25,103 | alexey-milovidov | One test reference has to be updated. | 2021-06-09T02:29:57 | 2021-06-09T02:29:57 | {} | MEMBER |
ClickHouse | ClickHouse | 857,328,907 | 25,058 | sherrychenzhang | meet this problem，too. seems the group by need a lot of memory. 1 G data 90G memory, hard to think. | 2021-06-09T02:46:37 | 2021-06-09T02:46:37 | {} | NONE |
ClickHouse | ClickHouse | 857,351,701 | 25,035 | ghost | > Do you connect to ClickHouse directly or it is behind a proxy like Cloudflare?
Directly, server and client both in one host.
> And do you use the unmodified version of ClickHouse?
Actually not, i have some changed in it. Maybe it's a self-introduced problem, will update here later if any clue. | 2021-06-09T03:55:29 | 2021-06-09T03:55:29 | {} | NONE |
ClickHouse | ClickHouse | 856,183,609 | 24,947 | sevirov | @kitaisreal Please, review this PR. | 2021-06-07T19:01:36 | 2021-06-07T19:01:36 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 856,250,929 | 23,902 | vzakaznikov | @akuzm, the bug is still present in master
```
2ms [clickhouse1] bash# echo -e "select empno, salary, leadInFrame(salary) OVER (ORDER BY salary) AS lead FROM (SELECT * FROM empsalary ORDER BY empno) FORMAT TabSeparatedWithNames" | clickhouse client -n --allow_experimental_window_... | 2021-06-07T20:55:20 | 2021-06-07T20:55:20 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 856,359,834 | 24,069 | alexey-milovidov | We need to either remove TestFlows or to fix them @vzakaznikov
Keeping flaky tests in master is not an option. | 2021-06-08T00:58:31 | 2021-06-08T00:58:31 | {} | MEMBER |
ClickHouse | ClickHouse | 856,360,604 | 24,830 | digdeep126 | Hi, is this bug fixed? | 2021-06-08T01:00:40 | 2021-06-08T01:00:40 | {} | NONE |
ClickHouse | ClickHouse | 856,843,049 | 24,938 | akuzm | ```
2021.06.06 20:24:41.235761 [ 184 ] {joins_in_memory.query9.prewarm0} <Debug> executeQuery: (from [::1]:51590, using production parser) SELECT COUNT() FROM ints l LEFT JOIN ints r USING i64,i32,i16,i8 WHERE i32 = 20042 settings query_plan_filter_push_down = 0
...
2021.06.06 20:24:42.469517 [ 7403 ] {joins_in_memo... | 2021-06-08T14:58:36 | 2021-06-08T14:58:36 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 856,962,771 | 24,069 | alexey-milovidov | Ok. BTW, the least significant bit depends on the implementation that is selected dynamically depending on the availability of AVX2 on the machine. | 2021-06-08T17:38:08 | 2021-06-08T17:38:08 | {} | MEMBER |
ClickHouse | ClickHouse | 856,131,525 | 24,053 | org0000h | i want to know how to drop the send in queue,because it aways failed again and again.DB::Exception: Too many partitions for single INSERT block | 2021-06-07T17:36:37 | 2021-06-07T17:36:37 | {} | NONE |
ClickHouse | ClickHouse | 857,218,099 | 24,909 | alexey-milovidov | The reason why Decimal is slower is that Float and Int code paths are excessively optimized.
We spent large amount of efforts and successfully optimized aggregation of Float and Int,
that's why you can observe that the aggregation of Float and Int is faster than Decimal. | 2021-06-08T22:22:11 | 2021-06-08T22:24:02 | {
"+1": 4
} | MEMBER |
ClickHouse | ClickHouse | 857,266,002 | 25,067 | azat | >The potential harm is that when people will see a file named VersionNumber they may start to think that we use the same class for ClickHouse version numbers.
That was one of my concerns, and this is the reason why it called VersionNumber not Version. | 2021-06-08T23:48:18 | 2021-06-08T23:48:18 | {} | MEMBER |
ClickHouse | ClickHouse | 856,167,232 | 4,238 | adevyatova | Internal documentation ticket: DOCSUP-10111 | 2021-06-07T18:33:18 | 2021-06-07T18:33:18 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 856,587,577 | 25,058 | zcw5116 | > set max_memory_usage
Setting this parameter is useless.
Note:
1. The default max_memory_usage parameter value is 10g. When I don't set max_bytes_before_external_group_by, everything is normal.
2. When I set the value of the max_bytes_before_external_group_by parameter to 2g, everything is normal.
3. When I set... | 2021-06-08T08:52:57 | 2021-06-08T08:52:57 | {} | NONE |
ClickHouse | ClickHouse | 856,986,447 | 18,740 | den-crane | incl работает только для первого упоминания тега?
раньше: config.xml `<remote_servers incl="clickhouse_remote_servers" />` и incl= работал
теперь: config.xml `<remote_servers />`
если добавить в собственном конфиге `<remote_servers incl="clickhouse_remote_servers" />` то работать не будет.
надо добавлять ... | 2021-06-08T18:13:28 | 2021-06-08T18:15:56 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 857,064,913 | 25,072 | alexey-milovidov | Why do we need some limited size of this pool? Why do we need this pool at all? | 2021-06-08T19:49:26 | 2021-06-08T19:49:26 | {
"+1": 1
} | MEMBER |
ClickHouse | ClickHouse | 847,940,702 | 24,491 | den-crane | I guess `<path>/var/lib/clickhouse/</path>` in config.xml | 2021-05-25T14:56:47 | 2021-05-25T14:56:47 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 848,017,860 | 24,350 | alexey-milovidov | @MyroTk ClickHouse is using `libc` from the system. Different versions of `libc` may have different behaviour.
We will work on static linking with `libc` to avoid this issue. | 2021-05-25T16:17:19 | 2021-05-25T16:17:19 | {} | MEMBER |
ClickHouse | ClickHouse | 848,381,557 | 24,505 | vfilatov | No, it doesn't help at all :(
Could the problem be in `PARTITION BY` size? Or TTL?
```
CREATE TABLE default.syslog
(
`severity` UInt16,
`facility` UInt16,
`timestamp` DateTime,
`hostname` LowCardinality(String),
`tag` String,
`message` String
)
ENGINE = MergeTree
PARTITION BY toYY... | 2021-05-26T01:06:17 | 2021-05-26T01:09:33 | {} | NONE |
ClickHouse | ClickHouse | 848,432,562 | 24,461 | filimonov | Of course crash is not acceptable, but building state on the top of some other state is kinda odd. Do you have any practical use case when it may be needed?
Maybe we can just forbid it, instead of fixing it? | 2021-05-26T03:49:29 | 2021-05-26T03:49:29 | {} | CONTRIBUTOR |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.