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 | 551,139,909 | 6,112 | alesapin | No more releases for this branch. | 2019-11-07T15:53:05 | 2019-11-07T15:53:05 | {} | MEMBER |
ClickHouse | ClickHouse | 551,198,607 | 7,650 | yamrzou | Thank you for looking into it.
Indeed we have about 200 parquet files in the directory. Also the compression used is snappy. | 2019-11-07T18:11:56 | 2019-11-07T18:11:56 | {
"+1": 1
} | NONE |
ClickHouse | ClickHouse | 551,404,497 | 7,675 | guanlisheng | thanks @BayoNet | 2019-11-08T06:35:15 | 2019-11-08T06:35:15 | {} | NONE |
ClickHouse | ClickHouse | 552,127,949 | 7,694 | alexey-milovidov | 1. Make sure that `<query_thread_log>` is configured in `config.xml`. This is by default.
2. Make sure that the settings `log_queries` and `log_query_threads` are enabled for your queries. By default, `log_queries` is 0 and `log_query_threads` is 1. It means that if you enable `log_queries`, both tables `system.query_... | 2019-11-09T19:05:01 | 2019-11-09T19:05:01 | {} | MEMBER |
ClickHouse | ClickHouse | 552,248,812 | 7,648 | alexey-milovidov | DDLWorker.cpp, `isSupportedAlterType`:
FREEZE is marked as unsupported. Is there any reason for it? | 2019-11-10T23:55:34 | 2019-11-10T23:55:34 | {} | MEMBER |
ClickHouse | ClickHouse | 552,249,686 | 3,556 | alexey-milovidov | One possibility is to allow more than one LIMIT BY sections:
`LIMIT 10 BY a, b LIMIT 5 BY b`
This is not so hard to implement.
But you can already write it with subqueries and we don't have to introduce unusual syntax. | 2019-11-11T00:04:57 | 2019-11-11T00:04:57 | {} | MEMBER |
ClickHouse | ClickHouse | 552,273,988 | 2,253 | alexey-milovidov | Fixed in 19.14. | 2019-11-11T03:00:05 | 2019-11-11T03:00:05 | {
"+1": 1
} | MEMBER |
ClickHouse | ClickHouse | 551,125,311 | 3,902 | pradeshc | the answers here
https://stackoverflow.com/questions/34479040/how-to-install-wkhtmltopdf-with-patched-qt | 2019-11-07T15:22:24 | 2019-11-07T15:22:24 | {} | NONE |
ClickHouse | ClickHouse | 551,515,076 | 7,681 | alesapin | Fix https://github.com/ClickHouse/ClickHouse/pull/7685 | 2019-11-08T10:14:00 | 2019-11-08T10:14:00 | {} | MEMBER |
ClickHouse | ClickHouse | 552,226,894 | 6,355 | alexey-milovidov | Possible reason is that shared libraries are used and there is wrong linking order (either in compile time or runtime):
```
... .6.0.1 /usr/lib/llvm/6/lib64/libLLVMDlltoolDriver.so.6.0.1 /usr/lib/llvm/6/lib64/libLLVMLibDriver.so.6.0.1 /usr/lib/llvm/6/lib64/libLLVMXRay.so.6.0.1 /usr/lib/llvm/6/lib64/libLLVMWindowsMa... | 2019-11-10T19:44:41 | 2019-11-10T19:48:49 | {} | MEMBER |
ClickHouse | ClickHouse | 552,267,513 | 2,545 | alexey-milovidov | Fixed. | 2019-11-11T02:24:48 | 2019-11-11T02:24:48 | {} | MEMBER |
ClickHouse | ClickHouse | 551,035,581 | 7,667 | chu1070y | Thanks for your help. | 2019-11-07T11:13:57 | 2019-11-07T11:22:21 | {} | NONE |
ClickHouse | ClickHouse | 551,058,955 | 7,663 | filimonov | > > Maybe something like system.zookeeper?
>
> You just reminded me that `system.zookeeper` is also unsafe :) I'm sure that our information security team is not happy with it.
As a tactical approach we can just introduce `<allow_server_management>` user setting which will allow access to system.zookeeper, system.... | 2019-11-07T12:27:14 | 2019-11-07T12:27:36 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 551,071,289 | 7,611 | den-crane | >CREATE TABLE test_gogo_dist ON CLUSTER
**DDL on cluster won't work.** -- expected behavior. | 2019-11-07T13:07:09 | 2019-11-07T13:07:09 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 551,265,358 | 7,553 | vitlibar | > @vitlibar
> min - Minimum value of setting.
> max - Maximum value of setting.
> readonly - Flag showing that the setting can't be changed by user.
>
> Am I right?
Yes.
> If I'm right, I want to know why some settings in this table can't be changed?
Maybe it's because the setting `readonly` is set:
`... | 2019-11-07T21:09:57 | 2019-11-07T21:11:22 | {} | MEMBER |
ClickHouse | ClickHouse | 551,649,347 | 7,686 | akuzm | Nice, I was fixing the exact same thing right now :)
You probably should remove the space after "WITH FILL " instead -- the general idea is that there is no space at the end, and you add one when you append something. | 2019-11-08T11:46:27 | 2019-11-08T11:46:27 | {
"+1": 1
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 551,695,886 | 7,687 | filimonov | Try that:
```
SELECT ... WHERE geohash IN (SELECT arrayJoin(geohashesInBox(...)) AS box_geohash)
```
Related: https://github.com/ClickHouse/ClickHouse/pull/5342
| 2019-11-08T12:19:44 | 2019-11-08T12:19:44 | {
"+1": 3
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 552,249,849 | 3,682 | alexey-milovidov | We can do it under `--progress` option. This will not break backward compatibility. | 2019-11-11T00:06:29 | 2019-11-11T00:06:29 | {} | MEMBER |
ClickHouse | ClickHouse | 552,251,436 | 4,963 | alexey-milovidov | That will be useful. Although quotas mechanism is not under development right now. We hope that we can do major refactoring on quotas/events/limits, but please don't wait for it. Minor changes are welcome. | 2019-11-11T00:21:46 | 2019-11-11T00:21:46 | {} | MEMBER |
ClickHouse | ClickHouse | 550,991,388 | 7,667 | chu1070y | Thank you so much. Your comments are very helpful.
If you don`t mind, I want to ask you just one more question.
I used different databases per replica.
for example,
```xml
<clickhouse_cluster>
<shard>
<weight>10</weight>
<internal_replication>false</internal_replication>
... | 2019-11-07T09:13:55 | 2019-11-07T09:17:38 | {} | NONE |
ClickHouse | ClickHouse | 551,137,403 | 7,603 | alesapin | @kwai-code Maybe it's possible to make a similar paste as text, not screenshot? Seems like you don't use our `clickhouse-common-static-dbg` package, so line numbers of source code files are hidden in the trace. It's possible to use addresses to deduce lines with `addr2line` tool, but copy them from the screenshot is no... | 2019-11-07T15:47:33 | 2019-11-07T15:47:52 | {} | MEMBER |
ClickHouse | ClickHouse | 551,842,187 | 7,689 | akuzm | uptime is already marked as non-deterministic, while version is deterministic. Maybe we should mark it as deterministic in scope of query, and only ship deterministic functions to replicas. | 2019-11-08T14:11:46 | 2019-11-08T14:11:46 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 551,909,075 | 7,691 | alesapin | I have asked our repository SRE to check this. | 2019-11-08T17:02:39 | 2019-11-08T17:02:39 | {} | MEMBER |
ClickHouse | ClickHouse | 552,163,691 | 7,546 | Mrfogg | I got another error
`1 warning generated.
[215/2319] Building CXX object contrib..._/simdjson/src/stage1_find_marks.cpp.o
warning: unknown warning option '-Wno-ctad-maybe-unsupported' [-Wunknown-warning-option]
1 warning generated.
[216/2319] Building CXX object contrib...n.dir/__/simdjson/src/parsedjson.cpp.o
wa... | 2019-11-10T05:05:39 | 2019-11-10T05:05:39 | {} | NONE |
ClickHouse | ClickHouse | 552,196,186 | 7,695 | CurtizJ | The error is caused by optimization of order by. Before we will fix it, you can temporarly disable the optimization to avoid error by setting `optimize_read_in_order=0`. | 2019-11-10T13:44:35 | 2019-11-10T13:45:22 | {} | MEMBER |
ClickHouse | ClickHouse | 552,248,206 | 7,668 | alexey-milovidov | @filimonov The query looks simple enough to not exceed max parse depth, but we use recursive descend parser that can go tens levels deep to parse every nesting level for some kind of expressions, that is obvious downside of our parser. | 2019-11-10T23:48:07 | 2019-11-10T23:48:07 | {} | MEMBER |
ClickHouse | ClickHouse | 552,248,353 | 7,663 | alexey-milovidov | About access restriction for `system.zookeeper` table. Just hold before @vitlibar will implement RBAC and the issue solved. | 2019-11-10T23:49:51 | 2019-11-10T23:49:51 | {} | MEMBER |
ClickHouse | ClickHouse | 552,249,330 | 6,774 | alexey-milovidov | @nikvas0 is currently developing different but related task named "key-value dictionaries on local SSD". | 2019-11-11T00:01:32 | 2019-11-11T00:01:32 | {} | MEMBER |
ClickHouse | ClickHouse | 552,262,794 | 7,700 | den-crane | ```
create table calendar (Day Date, is_holiday Int8) Engine=MergeTree order by Day;
insert into calendar select today()+number, number%7 from numbers(1000);
select count() from calendar where Day between '2019-12-30' and '2020-12-30' and not is_holiday;
``` | 2019-11-11T01:54:56 | 2019-11-11T01:54:56 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 552,262,945 | 7,383 | alexey-milovidov | This task is assigned to @laplab | 2019-11-11T01:55:53 | 2019-11-11T01:55:53 | {} | MEMBER |
ClickHouse | ClickHouse | 551,139,666 | 5,966 | alesapin | No more patch-releases for this branch. | 2019-11-07T15:52:30 | 2019-11-07T15:52:30 | {} | MEMBER |
ClickHouse | ClickHouse | 551,141,331 | 7,607 | alesapin | If the upgrade doesn't help just reopen this. | 2019-11-07T15:56:21 | 2019-11-07T15:56:21 | {} | MEMBER |
ClickHouse | ClickHouse | 551,160,591 | 7,650 | alesapin | Just tried to query some parquet files from local FS and also small file (about 150MB, 29M rows) from local HDFS installation.
```
SELECT countDistinct(BannerID)
FROM file('huge_event_log*.parq', 'Parquet', 'EventTime DateTime, BannerID UInt64, Cost UInt64, CounterType UInt8')
┌─uniqExact(BannerID)─┐
│ ... | 2019-11-07T16:38:38 | 2019-11-07T16:44:52 | {} | MEMBER |
ClickHouse | ClickHouse | 551,681,609 | 7,248 | filimonov | > It is possible, but it seems confusing as the authentication will work/not work depending only on whether the password is empty/non-empty.
I would rather exprect the for all plain (unencrypted) passwords, like `<password>qwerty123</password>` (also including empty password) use sha1 in mysql. | 2019-11-08T12:09:27 | 2019-11-08T12:09:27 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 551,925,560 | 7,691 | Tasselmi | Problem has been solved. Thanks
| |
梁凡
邮箱：tasselmi@yeah.net
|
签名由 网易邮箱大师 定制
On 11/09/2019 01:03, alesapin wrote:
I have asked our repository SRE to check this.
—
You are receiving this because you authored the thread.
Reply to this email directly, view it on GitHub, or unsubscribe. | 2019-11-08T17:48:45 | 2019-11-08T17:48:45 | {} | NONE |
ClickHouse | ClickHouse | 477,378,584 | 4,785 | alexey-milovidov | - [x] LowCardinality: done
- [x] Per column custom compression codecs: done
- [x] Non-trivial default expressions for missing values in JSONEachRow format: done
- [x] Parquet format: done
- [x] HDFS: done
- [ ] S3: moved to Q2
- [x] ALTER ORDER BY: done
- [x] Compact metadata in ZooKeeper: done
- [x] DDL for di... | 2019-03-27T22:57:36 | 2019-11-09T23:56:01 | {} | MEMBER |
ClickHouse | ClickHouse | 552,248,419 | 7,663 | alexey-milovidov | > <password hide="1">secret</password>
I don't like this solution, it introduces something to keep in mind. | 2019-11-10T23:50:46 | 2019-11-10T23:50:46 | {} | MEMBER |
ClickHouse | ClickHouse | 552,274,114 | 2,278 | alexey-milovidov | Now data parts are loaded in parallel. | 2019-11-11T03:00:44 | 2019-11-11T03:00:44 | {} | MEMBER |
ClickHouse | ClickHouse | 551,025,387 | 7,668 | filimonov | The query which is sent actuallty to the data_shard looks like that:
```
SELECT arraySort(arrayMap((x, y) -> (x, y), groupArray(event_index), groupArray(url))) AS events, arrayFilter(event -> ((event.1) = ((((arrayFilter(event -> ((event.1) = ((((arrayFilter(event -> ((event.1) = ((((arrayFilter(event -> ((event.1) =... | 2019-11-07T10:44:58 | 2019-11-07T10:47:00 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 551,037,968 | 7,665 | 4ertus2 | Perf tests are not good | 2019-11-07T11:21:08 | 2019-11-07T11:21:08 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 551,344,408 | 6,985 | JTCunning | I somehow missed this in the changelog. Thanks! | 2019-11-08T01:35:03 | 2019-11-08T01:35:03 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 552,248,731 | 7,648 | alexey-milovidov | CC @sundy-li | 2019-11-10T23:54:31 | 2019-11-10T23:54:31 | {} | MEMBER |
ClickHouse | ClickHouse | 552,267,130 | 4,172 | alexey-milovidov | `1970-01-01 00:00:00` is before unix epoch in timezones that have positive UTC offset at that point of time. The reason of this behaviour is that it overflows to 2106.
This behaviour has to be fixed. | 2019-11-11T02:22:18 | 2019-11-11T02:22:30 | {
"+1": 2
} | MEMBER |
ClickHouse | ClickHouse | 551,167,574 | 7,644 | BayoNet | https://clickhouse.yandex/docs/en/operations/system_tables/#system_tables-asynchronous_metrics | 2019-11-07T16:54:29 | 2019-11-07T16:54:29 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 551,948,206 | 7,407 | alesapin | Failed test is ok, because we don't support dictionaries DDL in this version, but use it in test. | 2019-11-08T18:55:12 | 2019-11-08T18:55:12 | {} | MEMBER |
ClickHouse | ClickHouse | 552,250,633 | 4,881 | alexey-milovidov | This is minor issue.
I have quickly checked the code and it looks Ok.
If you really want to debug, you can add debug print every time, `CurrentMetrics::ZooKeeperWatch` is incremented or decremented in ZooKeeperImpl.cpp. | 2019-11-11T00:14:34 | 2019-11-11T00:14:34 | {} | MEMBER |
ClickHouse | ClickHouse | 552,263,551 | 5,860 | alexey-milovidov | Table may not attach if it doesn't present on filesystem. It can be a configuration issue. Could you please also provide logs from server startup? | 2019-11-11T01:59:52 | 2019-11-11T01:59:52 | {} | MEMBER |
ClickHouse | ClickHouse | 552,266,549 | 4,374 | alexey-milovidov | `clickhouse-copier` can be run in multiple instances in parallel if there are multiple tasks. Each task is single partition for single shard of destination cluster.
There is no support to run single task in parallel with one `clickhouse-copier` instance.
There is a feature request to allow to split single partiti... | 2019-11-11T02:18:43 | 2019-11-11T02:18:43 | {} | MEMBER |
ClickHouse | ClickHouse | 552,270,092 | 86 | alexey-milovidov | Fixed. | 2019-11-11T02:38:48 | 2019-11-11T02:38:48 | {} | MEMBER |
ClickHouse | ClickHouse | 551,080,796 | 7,673 | filimonov | @den-crane it's not the case here.
It looks like it just tries to take String from Nullable(String), ignoring the Nullable.
```
SELECT toDateTime(CAST(NULL, 'Nullable(String)'))
Received exception from server (version 19.16.1):
Code: 41. DB::Exception: Received from localhost:9000. DB::Exception: Cannot pars... | 2019-11-07T13:34:13 | 2019-11-07T13:34:13 | {
"+1": 2
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 551,179,746 | 7,651 | alexey-milovidov | @achimbab Is this function applicable for strings with characters not from Basic Multilingual Plane? If they are represented not as UTF-16 (surrogate pairs) in Java, then another function like `javaHashUTF32LE` will be needed. | 2019-11-07T17:24:03 | 2019-11-07T17:24:03 | {} | MEMBER |
ClickHouse | ClickHouse | 551,785,670 | 7,680 | filimonov | Work on access control is currently in progress.
You can follow those PR
https://github.com/ClickHouse/ClickHouse/pull/6663
https://github.com/ClickHouse/ClickHouse/pull/7257
https://github.com/ClickHouse/ClickHouse/pull/7310
You can also check that:
https://github.com/ClickHouse/ClickHouse/pull/4792/files | 2019-11-08T13:23:19 | 2019-11-08T13:23:19 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 551,979,361 | 5,241 | alexey-milovidov | @den-crane
For two reasons:
- for consistency with other SQL dialects (example: https://dev.mysql.com/doc/refman/8.0/en/create-table.html) - they have INDEX and CONSTRAINT near columns;
- because there can be multiple indices. | 2019-11-08T20:31:04 | 2019-11-08T20:31:04 | {} | MEMBER |
ClickHouse | ClickHouse | 552,252,088 | 3,236 | alexey-milovidov | We already have (and always had) retries for dictionaries loading.
Plase clarify your scenario. | 2019-11-11T00:28:37 | 2019-11-11T00:28:37 | {} | MEMBER |
ClickHouse | ClickHouse | 551,777,880 | 7,248 | yurriy | > > It is possible, but it seems confusing as the authentication will work/not work depending only on whether the password is empty/non-empty.
>
> I would rather exprect the for all plain (unencrypted) passwords, like `<password>qwerty123</password>` (also including empty password) use sha1 in mysql.
@alexey-milo... | 2019-11-08T13:17:43 | 2019-11-08T13:17:43 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 552,169,331 | 7,671 | yuzhichang | I see that for ClickHouse servers at +08:00 timezone,
- for Date(stored as UInt16), user input '1970-01-01' is equivalent to ck output '0000-00-00'.
- for DateTime(stored as Int32), user input '1970-01-01 08:00:00' is equivalent to ck output '0000-00-00 00:00:00'.
@den-crane Thanks for your clarification.
| 2019-11-10T07:07:24 | 2019-11-10T07:10:54 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 552,268,904 | 7,494 | alexey-milovidov | What is the stack trace? The issue looks critical. | 2019-11-11T02:32:15 | 2019-11-11T02:32:15 | {} | MEMBER |
ClickHouse | ClickHouse | 551,020,652 | 7,663 | nvartolomei | I like @filimonov's proposal as long as we limit access to this table to a administrator who would have access to the config anyway. | 2019-11-07T10:32:14 | 2019-11-07T10:32:14 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 551,074,650 | 7,671 | den-crane | CH epoch is in UTC!!!!! "1970-01-01 00:00:00"
```
SELECT
toDateTime(0),
toDate(0),
toDateTime('1970-01-01 00:00:00'),
toDate('1970-01-01')
┌───────toDateTime(0)─┬──toDate(0)─┬─toDateTime('1970-01-01 00:00:00')─┬─toDate('1970-01-01')─┐
│ 0000-00-00 00:00:00 │ 0000-00-00 │ 0000... | 2019-11-07T13:16:58 | 2019-11-07T13:23:58 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 551,177,446 | 7,412 | BayoNet | @4ertus2 What user cases does this setting cover? | 2019-11-07T17:18:27 | 2019-11-07T17:18:27 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 551,391,174 | 7,546 | cdoco | I have a same mistake.
```
[100%] Linking CXX executable clickhouse
Undefined symbols for architecture x86_64:
"_thread_local_rng", referenced from:
Allocator<false, false>::getMmapHint() in libclickhouse-benchmark-libd.a(Benchmark.cpp.o)
Allocator<false, false>::getMmapHint() in libclickhouse-ser... | 2019-11-08T05:32:18 | 2019-11-08T05:33:01 | {} | NONE |
ClickHouse | ClickHouse | 551,456,136 | 7,653 | filimonov | I think that changing way of serialization and have a possibility to ignore unknown settings is good .
At the same time the decision "to ignore / not ignore" particular (or all) settings should be controlled by user IMO (from both ends - client or server).
So client can send smth like 'ignore_unknown_settings=0'... | 2019-11-08T09:33:38 | 2019-11-08T09:41:12 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 551,840,132 | 7,689 | KochetovNicolai | @akuzm I think it is strange to use functions like `uptime()` and `version()` in mutations indeed. Result also won't be deterministic. Probably we need another `may-be-not-deterministic-on-different-replicas` flag? | 2019-11-08T14:05:50 | 2019-11-08T14:05:50 | {} | MEMBER |
ClickHouse | ClickHouse | 551,844,438 | 7,689 | KochetovNicolai | Yes, it seems reasonable. | 2019-11-08T14:17:51 | 2019-11-08T14:17:51 | {} | MEMBER |
ClickHouse | ClickHouse | 551,930,989 | 7,248 | alexey-milovidov | @yurriy Yes, we can activate SHA1 authentication for plaintext passwords.
Motivation: if the user uses plaintext passwords for ClickHouse, it's already unsafe within native ClickHouse protocol (unless SSL is used), so it can be as safe in MySQL protocol. | 2019-11-08T18:04:23 | 2019-11-08T18:04:45 | {} | MEMBER |
ClickHouse | ClickHouse | 551,983,298 | 7,653 | filimonov | > It seems the client will be able to calculate that `can_ignore` flag without extra settings.
Does it mean that the decision allowing / forbidding ignorance will be hardcoded?
What if i prefer not to ignore that? | 2019-11-08T20:44:39 | 2019-11-08T20:44:39 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 552,148,406 | 4,785 | alexey-milovidov | November 2019. | 2019-11-09T23:56:18 | 2019-11-09T23:56:18 | {} | MEMBER |
ClickHouse | ClickHouse | 552,266,177 | 3,992 | alexey-milovidov | @vicdashkov ClickHouse will collapse data even if it is located in one data part if there is background merge process involved this data part among others. And background merge process is initiated if there is more than a single data part. | 2019-11-11T02:16:20 | 2019-11-11T02:16:20 | {} | MEMBER |
ClickHouse | ClickHouse | 551,180,304 | 7,644 | ofek | @BayoNet The meaning of each metric is still undocumented. Can you please re-open this? | 2019-11-07T17:25:15 | 2019-11-07T17:25:15 | {} | NONE |
ClickHouse | ClickHouse | 551,813,945 | 7,687 | yamrzou | Hi @Enmk,
For now the workaround proposed by @filimonov works perfectly. Thanks a lot ! | 2019-11-08T13:42:56 | 2019-11-08T13:42:56 | {} | NONE |
ClickHouse | ClickHouse | 552,181,562 | 7,696 | alexey-milovidov | Suspect a bug in libressl (something like infinite loop of retries) or retries of syscalls in a loop after a signal without proper adjustment of timeouts (e.g. each second `recv` is retried with 300 seconds timeout).
We still have not reproduced this issue in controllable environment. | 2019-11-10T10:21:03 | 2019-11-10T10:21:03 | {} | MEMBER |
ClickHouse | ClickHouse | 552,181,812 | 7,696 | alexey-milovidov | Please check if the issue can be solved by sending a signal to each thread:
```
for i in $(ls -1 /proc/$(pidof clickhouse-server)/task/); do kill -PROF $i; done
``` | 2019-11-10T10:24:36 | 2019-11-10T10:24:36 | {} | MEMBER |
ClickHouse | ClickHouse | 552,252,615 | 5,931 | alexey-milovidov | That's not easy to fix. These arguments are expected to be constants to call dictionary functions in a batch fashion.
Instead of using multiple dictionaries sharded by `application_id`, I recommend to use single dictionary with `application_id` as the first column of complex key. | 2019-11-11T00:33:40 | 2019-11-11T00:33:40 | {} | MEMBER |
ClickHouse | ClickHouse | 552,266,683 | 7,543 | alexey-milovidov | @alesapin maybe you can fix it? | 2019-11-11T02:19:38 | 2019-11-11T02:19:38 | {} | MEMBER |
ClickHouse | ClickHouse | 551,167,163 | 7,650 | alesapin | Yes, just created 200 files and get OOM:
```
$ ls event_log* | wc -l
200
$ clickhouse-client
:) select countDistinct(BannerID) from file('event_log*.parq', 'Parquet', 'EventTime DateTime, BannerID UInt64, Cost UInt64, CounterType UInt8');
```
```
[7 19:52:13 2019] oom_kill_process+0x254/0x280
[7 19:52:13 2... | 2019-11-07T16:53:37 | 2019-11-07T16:57:34 | {} | MEMBER |
ClickHouse | ClickHouse | 551,191,091 | 7,650 | alexey-milovidov | @stavrolia Look at `narrowBlockInputStreams`. | 2019-11-07T17:52:37 | 2019-11-07T17:52:37 | {
"+1": 1
} | MEMBER |
ClickHouse | ClickHouse | 551,341,412 | 2,416 | tlightsky | @alexey-milovidov
will this end up like
```SQL
WITH (SELECT a, b FROM t) as t1
SELECT * FROM t1 WHERE a=1
``` | 2019-11-08T01:21:59 | 2019-11-08T01:21:59 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 551,732,185 | 7,689 | KochetovNicolai | Do not know how to create a good simple test. | 2019-11-08T12:45:29 | 2019-11-08T12:45:29 | {} | MEMBER |
ClickHouse | ClickHouse | 552,249,070 | 1,647 | alexey-milovidov | Still relevant. | 2019-11-10T23:59:22 | 2019-11-10T23:59:22 | {} | MEMBER |
ClickHouse | ClickHouse | 551,175,565 | 7,553 | BayoNet | @vitlibar
min - Minimum value of setting.
max - Maximum value of setting.
readonly - Flag showing that the setting can't be changed by user.
Am I right?
If I'm right, I want to know why some settings in this table can't be changed? Aren't there are only session settings that can be changed with the SET query? | 2019-11-07T17:13:52 | 2019-11-07T17:14:10 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 551,516,357 | 7,638 | alesapin | Will be available in all next stable releases. | 2019-11-08T10:14:52 | 2019-11-08T10:14:52 | {} | MEMBER |
ClickHouse | ClickHouse | 552,227,552 | 6,355 | alexey-milovidov | Also look at `man ld-linux`. | 2019-11-10T19:52:12 | 2019-11-10T19:52:12 | {} | MEMBER |
ClickHouse | ClickHouse | 552,248,535 | 7,663 | alexey-milovidov | Maybe we can have this table with all config entries but only if access will be restricted by default. | 2019-11-10T23:52:12 | 2019-11-10T23:52:12 | {} | MEMBER |
ClickHouse | ClickHouse | 552,250,112 | 7,538 | alexey-milovidov | This exception happens when Distributed table tries to process data for asynchronous insertion. The data was written to local filesystem before you had removed the shard and it tries to finish sending it to the shard.
To fix the issue, go to the data directory of the table: `/var/lib/clickhouse/data/db/table/` and r... | 2019-11-11T00:09:25 | 2019-11-11T00:09:25 | {} | MEMBER |
ClickHouse | ClickHouse | 551,079,370 | 7,673 | den-crane | Clickhouse calculates both branches (if and else) for PERFORMANCE.
```
SELECT parseDateTimeBestEffortOrNull(nullIf('', ''))
┌─parseDateTimeBestEffortOrNull(nullIf('', ''))─┐
│ ᴺᵁᴸᴸ │
└───────────────────────────────────────────────┘
``` | 2019-11-07T13:30:23 | 2019-11-07T15:05:12 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 551,169,339 | 7,675 | BayoNet | I have a corresponding task. I'll try to do it ASAP. | 2019-11-07T16:58:46 | 2019-11-07T16:58:46 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 551,684,259 | 7,685 | filimonov | Fixes https://github.com/ClickHouse/ClickHouse/issues/5465 | 2019-11-08T12:11:23 | 2019-11-08T12:11:23 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 552,224,493 | 7,700 | alexey-milovidov | Here is an example:
```
WITH
toDate('2019-01-01') AS d1,
today() AS d2
SELECT length(arrayFilter(x -> ((toDayOfWeek(d1 + x) <= 5) AND ((d1 + x) NOT IN ('2019-01-01', '2019-01-02'))), range(toUInt64(d2 - d1)))) AS num_work_days
┌─num_work_days─┐
│ 222 │
└───────────────┘
```
Here I cr... | 2019-11-10T19:17:10 | 2019-11-10T19:17:10 | {} | MEMBER |
ClickHouse | ClickHouse | 552,227,120 | 6,355 | alexey-milovidov | To debug the dynamic linker, you can use LD_DEBUG environment variable. | 2019-11-10T19:47:43 | 2019-11-10T19:47:43 | {} | MEMBER |
ClickHouse | ClickHouse | 552,251,598 | 4,549 | alexey-milovidov | Fixed in recent releases. | 2019-11-11T00:23:29 | 2019-11-11T00:23:29 | {} | MEMBER |
ClickHouse | ClickHouse | 552,252,358 | 515 | alexey-milovidov | @alesapin
Please verify if it's possible to trigger this issue with the new dictionaries DDL queries. | 2019-11-11T00:31:11 | 2019-11-11T00:31:11 | {} | MEMBER |
ClickHouse | ClickHouse | 551,573,254 | 7,653 | vitlibar | > A better solution in my opinion would be for the server to expose list of supported settings (optionally, supported values), and let the client decide whether or not these settings are enough to execute the query.
I'll try to implement a slightly different solution. The client will send `can_ignore` flag with ever... | 2019-11-08T10:53:49 | 2019-11-08T10:53:49 | {} | MEMBER |
ClickHouse | ClickHouse | 551,697,737 | 7,687 | Enmk | Hi @yamrzou !
Well, there was a plan to fix `operator IN` just for that, but it has low priority due to a workaround @filimonov just posted.
And right now there is no way to convert `Array` to `Tuple`. IIRC, there are no plans to do that either. | 2019-11-08T12:21:02 | 2019-11-08T12:22:51 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 551,819,163 | 7,689 | akuzm | Is this related to https://github.com/ClickHouse/ClickHouse/pull/7247 , e.g. should we prevent pushing down non-deterministic functions to remote server? | 2019-11-08T13:46:33 | 2019-11-08T13:46:33 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 551,933,541 | 7,688 | alexey-milovidov | +1.
After "DDL Dictionaries" feature is implemented, the user will have a table for every dictionary. And it will be very likely that the user will JOIN with this table instead of using dictGet. | 2019-11-08T18:12:06 | 2019-11-08T18:12:20 | {} | MEMBER |
ClickHouse | ClickHouse | 551,978,676 | 5,241 | alexey-milovidov | https://github.com/ClickHouse/ClickHouse/pull/6968 | 2019-11-08T20:28:44 | 2019-11-08T20:28:44 | {} | MEMBER |
ClickHouse | ClickHouse | 552,177,382 | 4,959 | dbaops-com | ClickHouse server version 19.14.6:
when i use mysql cilent, get this error:
2019.11.10 17:15:35.159193 [ 183 ] {} <Error> MySQLHandler: DB::Exception: Cannot read all data. Bytes read: 0. Bytes expected: 3. | 2019-11-10T09:20:40 | 2019-11-10T09:20:40 | {} | NONE |
ClickHouse | ClickHouse | 552,251,942 | 7,676 | alexey-milovidov | This is not supported right now.
@stavrolia will add `_file` and `_path` virtual columns to HDFS, File, S3 and URL storage engines and you will be able to extract the value of partition key from filenames while reading multiple files with wildcards. | 2019-11-11T00:27:08 | 2019-11-11T00:27:08 | {} | MEMBER |
ClickHouse | ClickHouse | 552,252,915 | 7,432 | alexey-milovidov | We don't enforce backward compatibility for system tables. But it should be mentioned in changelog and documentation. | 2019-11-11T00:36:47 | 2019-11-11T00:36:47 | {} | MEMBER |
ClickHouse | ClickHouse | 553,481,310 | 6,005 | richardartoul | @alex-zaitsev Do you have a moment to shed some light on how Array(LowCardinality(String)) columns work? My understanding (please correct me if I'm wrong) is that the storage engines in clickhouse can basically do whatever they want but when a query is executed they're responsible for generating "native format" blocks ... | 2019-11-13T16:30:37 | 2019-11-13T16:32:39 | {} | NONE |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.