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 | 694,287,188 | 14,876 | alexey-milovidov | > Yandex synchronization check (only for Yandex employees)
Already fixed in master. | 2020-09-17T14:49:20 | 2020-09-17T14:49:20 | {} | MEMBER |
ClickHouse | ClickHouse | 694,287,976 | 14,863 | den-crane | 1. you can develop own application which process data from RocksDB into RowBinary or Native.
You can get a description of RowBinary format from CH documentation and Native almost the same as RowBinary the same value encodings but in columns not rows. You can copy Native implementation from JDBC driver or GO-driver.... | 2020-09-17T14:50:33 | 2020-09-17T14:54:30 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 694,344,134 | 14,931 | den-crane | https://clickhouse.tech/docs/en/sql-reference/statements/kill/#kill-mutation
kill mutation where not is_done | 2020-09-17T16:21:59 | 2020-09-17T16:21:59 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 694,344,729 | 13,574 | alesapin | As @filimonov sad we had a very similar discussion. I think fingerprints for parts is a good idea, but they should be hidden from the user. Query `ADD FINGERPRINT` seems too low-level. Also, I don't understand how fingerprints will interact with merges. What will be a result of the merge of two parts with different fin... | 2020-09-17T16:23:06 | 2020-09-17T16:23:06 | {} | MEMBER |
ClickHouse | ClickHouse | 694,421,027 | 9,516 | azat | >Functional stateless tests (debug) — fail: 1, passed: 2247, skipped: 15
#14957 | 2020-09-17T18:33:29 | 2020-09-17T18:33:29 | {} | MEMBER |
ClickHouse | ClickHouse | 694,108,064 | 13,156 | akuzm | > @akuzm looks like performance tests hanged, can you take a look please?
They are waiting in queue (even now, 16 hour later...). I increased the priority. | 2020-09-17T09:17:34 | 2020-09-17T09:17:34 | {
"+1": 1
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 694,116,425 | 14,813 | KochetovNicolai | Inlining slowed build time of `GatherUtils` up to 2 times. I've reverted it.
It have changed binary size only in 2MB anyway. | 2020-09-17T09:33:46 | 2020-09-17T09:35:22 | {} | MEMBER |
ClickHouse | ClickHouse | 694,203,622 | 14,918 | KochetovNicolai | Not waiting for tests cause code have not changed. | 2020-09-17T12:41:01 | 2020-09-17T12:41:01 | {} | MEMBER |
ClickHouse | ClickHouse | 694,229,598 | 14,930 | somewheve | [ 10, 120, 300, 20, 136, 70, 60, 20]
will be split into [ 10, 120, 300] [20, 136], [70, 60, 20] if value is set to 150.
I mean the sum of every array will more than 150, you can see the end of this issue [issue](https://github.com/ClickHouse/ClickHouse/issues/13493) | 2020-09-17T13:22:59 | 2020-09-17T13:22:59 | {} | NONE |
ClickHouse | ClickHouse | 694,317,311 | 14,909 | bgranvea | Thanks for your response. We are using these special values as a kind of hack to store in a single Float64 either a "normal" value or an error flag.
I don't know what should be the correct behavior for Clickhouse. Anyway we might consider other options to achieve what we want. | 2020-09-17T15:35:42 | 2020-09-17T15:35:42 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 694,392,262 | 14,221 | alexey-milovidov | +1 leaf is Ok. | 2020-09-17T17:41:34 | 2020-09-17T17:41:34 | {} | MEMBER |
ClickHouse | ClickHouse | 694,034,552 | 14,436 | peaksnail | @den-crane It happened when I alter table bill or goods_invocing, I show the last ch log when I alter table goods_invocing.
it's ReplacingMergeTree Engine | 2020-09-17T07:27:04 | 2020-09-17T07:27:04 | {} | NONE |
ClickHouse | ClickHouse | 694,058,232 | 14,904 | alesapin | Cherry-picked manually. | 2020-09-17T07:50:02 | 2020-09-17T07:50:02 | {} | MEMBER |
ClickHouse | ClickHouse | 694,114,841 | 14,889 | akuzm | > It also looks very bad...
1) jemalloc shouldn't have an assertion in this place -- a creation of new arena *can* return null in case we are out of memory
2) there are a lot of messages like these:
```
It looks like that the process is near the limit on number of virtual memory mappings.
Current number of mapping... | 2020-09-17T09:30:38 | 2020-09-17T09:30:38 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 694,115,572 | 14,853 | KochetovNicolai | I suppose it shouldn't affect perftests at all. This code changes only index analysis.
We could possibly create a table with a lot of small parts, but I suppose it will mostly test other parts of our code. | 2020-09-17T09:32:06 | 2020-09-17T09:32:06 | {} | MEMBER |
ClickHouse | ClickHouse | 694,152,394 | 14,265 | qoega | Yes. It is correct. We forced to use default server configuration and it was not so when tests replaced whole config directory. It is related only to `add_instance`. Configurations of other services nodes can be done in any convenient way. | 2020-09-17T10:48:23 | 2020-09-17T10:48:23 | {
"+1": 1
} | MEMBER |
ClickHouse | ClickHouse | 694,265,440 | 14,893 | alexey-milovidov | The link to C++ library: https://github.com/apache/incubator-datasketches-cpp | 2020-09-17T14:14:31 | 2020-09-17T14:14:31 | {
"+1": 1
} | MEMBER |
ClickHouse | ClickHouse | 694,292,635 | 14,786 | liangli1213 | @zhang2014 @BohuTANG , thanks for your effort to fix this bug, here I record how to 100% reproduce this issue, it is a little complex, I believe there are other better methods, but I don't have much time to dig.
1. Setup 2 MySQL server: mysql_a (master) and mysql_b (slave), then make the replication relationship
... | 2020-09-17T14:56:51 | 2020-09-17T14:56:51 | {} | NONE |
ClickHouse | ClickHouse | 694,300,484 | 14,930 | den-crane | I think it's impossible now in CH. | 2020-09-17T15:07:59 | 2020-09-17T15:07:59 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 694,415,399 | 14,776 | alexey-milovidov | There is something related to DatabaseAtomic... | 2020-09-17T18:22:55 | 2020-09-17T18:22:55 | {} | MEMBER |
ClickHouse | ClickHouse | 694,266,781 | 14,893 | alexey-milovidov | The library looks Ok, it supports all the required parts of AggregateFunction interface.
(the serialize/deserialize functions may be inefficient) | 2020-09-17T14:16:27 | 2020-09-17T14:16:27 | {
"rocket": 1
} | MEMBER |
ClickHouse | ClickHouse | 694,334,184 | 14,941 | moweonlee | @den-crane
Thank you for your response.
Now I understand that my data in aggregatingMergeTree is not well merged because I have so many columns in ORDER BY.
By the way in my application, every column should be provided to get the result except for the time unit change.
Only thing that I want to be merg... | 2020-09-17T16:04:13 | 2020-09-17T16:06:39 | {} | NONE |
ClickHouse | ClickHouse | 694,371,031 | 14,367 | doom369 | @den-crane thanks. It helped. Now memory dropped to 14% and to 28% during query execution. Look like it now defaults due to https://github.com/ClickHouse/ClickHouse/pull/11139. Thanks for pointing out. | 2020-09-17T17:06:59 | 2020-09-17T17:06:59 | {} | NONE |
ClickHouse | ClickHouse | 694,258,582 | 14,931 | simplemagik | maybe i could delete mutation file from
clickhouse/data/database/table directory?
will it help? | 2020-09-17T14:04:20 | 2020-09-17T14:04:42 | {} | NONE |
ClickHouse | ClickHouse | 694,268,648 | 14,928 | akuzm | Builds passed, will merge now. | 2020-09-17T14:19:18 | 2020-09-17T14:19:18 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 694,319,203 | 14,931 | simplemagik | seems like nothing changed
also noticed that
SELECT * FROM system.mutations
also timeouts
maybe it's a hint that mutations are still in progress or stuck? | 2020-09-17T15:38:54 | 2020-09-17T15:39:06 | {} | NONE |
ClickHouse | ClickHouse | 694,380,514 | 14,221 | azat | >leaf is a better name. Query for shard can be served by multiple replicas in parallel
But this is still shards anyway (and remote_servers contains shards too)
> and the shard naming doesn't make sense as the limit is applied locally.
By "locally" you mean on shards and on the initiator?
P.S. leaf is ok to ... | 2020-09-17T17:23:24 | 2020-09-17T17:23:24 | {} | MEMBER |
ClickHouse | ClickHouse | 694,372,375 | 13,617 | akuzm | ERROR: type should be string, got "https://clickhouse-test-reports.s3.yandex.net/13617/9a3627103693d215a6a02c2d74f13911d9384f50/performance_comparison/report.html#test-performance-changes\r\n\r\nI disabled the profiler, and now clang is 1% slower... not sure what's going on." | 2020-09-17T17:09:23 | 2020-09-17T17:24:08 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 694,401,281 | 6,286 | alexey-milovidov | The task is still to do, but there is a lack of motivation from both of involved contributors.
But we really need it and there are multiple use-cases:
- indexing by geographical coordinates (now you can index by geohash but you will have to calculate possible geohashes and specify all of them in query);
- indexing... | 2020-09-17T17:57:29 | 2020-09-17T17:57:29 | {
"+1": 5,
"rocket": 3
} | MEMBER |
ClickHouse | ClickHouse | 694,095,550 | 13,574 | alesapin | As @filimonov sad we had a very similar discussion. I think fingerprints for parts is a good idea, but they should be hidden from the user. Query `ADD FINGERPRINT` seems too low-level. Also, I don't understand how fingerprints will interact with merges. What will be a result of the merge of two parts with different fin... | 2020-09-17T08:55:00 | 2020-09-17T08:55:00 | {} | MEMBER |
ClickHouse | ClickHouse | 694,163,016 | 14,925 | alesapin | Maybe we can do it similar to `ccache` compiler launcher, but using `timeout 600 ...` https://github.com/clickhouse/ClickHouse/blob/master/cmake/find/ccache.cmake#L33-L34? | 2020-09-17T11:12:24 | 2020-09-17T11:12:43 | {} | MEMBER |
ClickHouse | ClickHouse | 694,164,469 | 8,429 | alexey-milovidov | Need to rebase on master and open another PR when it will be more ready. | 2020-09-17T11:15:45 | 2020-09-17T11:15:45 | {} | MEMBER |
ClickHouse | ClickHouse | 694,167,137 | 14,221 | nvartolomei | @KochetovNicolai I think `leaf` is a future proof name. Even we add support for hierarchical query execution, similar to Scuba. 👍
<img width="1648" alt="Screen Shot 2020-09-17 at 12 21 12 PM" src="https://user-images.githubusercontent.com/543193/93463944-5e7f8680-f8e0-11ea-8bf8-cbecab952448.png">
| 2020-09-17T11:22:05 | 2020-09-17T11:22:05 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 694,176,870 | 13,583 | CurtizJ | Do I understand right, that merges with TTL stopped after table rename? What time did pass from moment beetween moment, when data should be deleted and moment, when you noticed, that is's actually not deleted? | 2020-09-17T11:44:51 | 2020-09-17T11:44:51 | {} | MEMBER |
ClickHouse | ClickHouse | 694,175,585 | 13,574 | nvartolomei | @alesapin take a look at #14860 which attempts to implement it for MergeTree engine.
> Query ADD FINGERPRINT seems too low-level.
In my initial POC these queries weren't present and ClickHouse would have been responsible for everything (a worker), but getting it right is hard. From my last comment:
> Orchestra... | 2020-09-17T11:42:00 | 2020-09-17T12:57:49 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 694,244,696 | 14,914 | alexey-milovidov | I confirm.
The implementation of correlated subqueries (and only for the case when they can be rewritten to JOINs) is not in current roadmap and most likely will be planned for 2021. | 2020-09-17T13:43:59 | 2020-09-17T13:43:59 | {} | MEMBER |
ClickHouse | ClickHouse | 694,239,751 | 14,796 | 4ertus2 | `joinSortedBlock<false, true>` from stack trace says thet it's join on disk. | 2020-09-17T13:36:19 | 2020-09-17T13:50:27 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 694,373,891 | 14,941 | moweonlee | Thank you very much for your help 👍 | 2020-09-17T17:11:54 | 2020-09-17T17:11:54 | {} | NONE |
ClickHouse | ClickHouse | 694,409,991 | 14,367 | alexey-milovidov | @doom369 There is ongoing PR for improvement: https://github.com/ClickHouse/ClickHouse/pull/14929 | 2020-09-17T18:13:20 | 2020-09-17T18:13:20 | {
"+1": 1
} | MEMBER |
ClickHouse | ClickHouse | 694,414,455 | 14,937 | alexey-milovidov | > Yandex third-party checks (only for Yandex employees)
Does not matter - ignore. | 2020-09-17T18:21:25 | 2020-09-17T18:21:25 | {} | MEMBER |
ClickHouse | ClickHouse | 694,139,304 | 14,221 | akuzm | > Looks fine to me.
>
> Probably, we could use a better name for settings?
> Logically, it is just a limit on single server.
> Could it be like `max_rows_to_read_for_single_server` or `max_rows_to_read_one_instance` or something like that?
> @azat what do you think?
`max_rows_to_read_per_shard` ? | 2020-09-17T10:18:55 | 2020-09-17T10:18:55 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 694,142,544 | 14,221 | nvartolomei | leaf is a better name. Query for shard can be served by multiple replicas
in parallel and the shard naming doesn't make sense as the limit is applied
locally.
Instance can be both root (initial) and leaf (remote).
On Thu, 17 Sep 2020, 11:21 Azat Khuzhin, <notifications@github.com> wrote:
> max_rows_to_read_per_shard... | 2020-09-17T10:25:55 | 2020-09-17T10:25:55 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 694,139,122 | 14,859 | 4ertus2 | PVS error looks like false positive: if (bit_shift) => (overflow_shift < 64).
Fuzzer asserts in dictionaries:
`Assertion `key_columns.size() == dict_struct.key->size()' failed Received signal 6 Received signal Aborted (6)` | 2020-09-17T10:18:30 | 2020-09-17T10:53:15 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 694,275,758 | 14,931 | den-crane | if table is not Replicated then
```
system stop merges TABLE;
system start merges TABLE;
```
if it replicated then need to remove this task from ZK before stop/start | 2020-09-17T14:30:56 | 2020-09-17T14:30:56 | {
"+1": 1,
"confused": 1
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 694,343,281 | 14,931 | den-crane | OPTIMIZE TABLE FINAL has no relation to mutations | 2020-09-17T16:20:34 | 2020-09-17T16:20:34 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 694,372,159 | 14,832 | mikolajsnioch | ```I would say that we need to forbid a tz argument for toDate but I am not sure that it always does nothing.```
That's an option too. | 2020-09-17T17:08:59 | 2020-09-17T17:08:59 | {} | NONE |
ClickHouse | ClickHouse | 694,050,785 | 14,899 | keliss | But they don't get merged. We have such parts for over half a year. And if it is intended behaviour and the order matters, will there be any problems if we just use DROP PARTITION instead? | 2020-09-17T07:44:03 | 2020-09-17T07:44:03 | {} | NONE |
ClickHouse | ClickHouse | 694,148,264 | 14,265 | ilejn | > There were no problems in current integration tests, so it feels possible to implement everything in current way.
>
> You just need to take into account that all main_configs files will be moved to confid.d and user_configs to users.d respectively if you set paths in your configs.
@qoega This is about ClickHou... | 2020-09-17T10:38:54 | 2020-09-17T10:38:54 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 694,204,418 | 14,844 | den-crane | >But I won't be able to write populate query
Not true. Give me example.
>sum(count()) is not supported.
You are mistaken. You don't need `sum(count())`. Try with only `count()`. MV will calculate count. Let's say 301. And will write 301 into a `eval SimpleAggregateFunction`. SimpleAggregateFunction does not re... | 2020-09-17T12:42:26 | 2020-09-17T12:43:51 | {
"+1": 3
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 694,338,595 | 14,367 | den-crane | @doom369 try to set in config.xml `<mlock_executable>false</mlock_executable>` | 2020-09-17T16:12:13 | 2020-09-17T16:12:13 | {
"+1": 1
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 694,361,597 | 14,941 | den-crane | I think you don't need MV & aggregatingMergeTree. You can't get any gain from it.
Just query your original table. | 2020-09-17T16:49:17 | 2020-09-17T16:49:17 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 694,136,616 | 14,920 | alexey-milovidov | Startup can take significant amount of time if you have too many tables or too many data parts inside your tables. It may happen if you partition data too much. The reason why startup takes a time will be clearly visible in server logs.
In recent versions we have made database initialization more parallel and it wil... | 2020-09-17T10:12:53 | 2020-09-17T10:12:53 | {} | MEMBER |
ClickHouse | ClickHouse | 694,189,792 | 14,920 | den-crane | @flyfoxCI Check this https://github.com/ClickHouse/ClickHouse/issues/10087 | 2020-09-17T12:12:58 | 2020-09-17T12:12:58 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 694,201,917 | 14,897 | den-crane | CH is an opensource product. There is no exact planing/date.
Eventually. Maybe never.
Check --join_algorithm=partial_merge | 2020-09-17T12:37:41 | 2020-09-17T12:37:51 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 694,296,779 | 14,221 | KochetovNicolai | Ok, let's keep leaf name | 2020-09-17T15:01:49 | 2020-09-17T15:01:49 | {} | MEMBER |
ClickHouse | ClickHouse | 694,136,968 | 14,846 | alesapin | `01193_metadata_loading` :( | 2020-09-17T10:13:39 | 2020-09-17T10:13:39 | {} | MEMBER |
ClickHouse | ClickHouse | 694,141,477 | 14,265 | qoega | >It seems that dealing with "secrets" directory does not fit well with recent config_d_dir changes. Advices?
There were no problems in current integration tests, so it feels possible to implement everything in current way.
You just need to take into account that all main_configs files will be moved to confid.d and us... | 2020-09-17T10:23:39 | 2020-09-17T10:23:39 | {} | MEMBER |
ClickHouse | ClickHouse | 694,164,328 | 14,367 | Slach | @doom369 you have total different error
for @peaksnail use case we have just changed system.mertics value behavior
in your case probably you need look to
```
SELECT * FROM system.settings WHERE name LIKE '%max_mem%' OR name LIKE '%max_thead%' OR name LIKE '%background%';
```
and decrease `max_threads` and `bac... | 2020-09-17T11:15:22 | 2020-09-17T11:15:22 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 694,170,663 | 14,367 | doom369 | @Slach thanks for the answer.
> and decrease max_threads and background_pool_size values
That's what I did and that helped. However, top command still show 100% memory usage for the CH server. That's wasn't like that for the previous versions. It was at the level of 40-50%.
> 1 GB RAM is too low for ClickHouse... | 2020-09-17T11:30:18 | 2020-09-17T11:32:04 | {} | NONE |
ClickHouse | ClickHouse | 472,731,070 | 4,491 | etah000 | clickhouse-client --multiquery < script.sql | 2019-03-14T07:14:08 | 2020-09-17T13:01:02 | {
"+1": 3
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 694,218,319 | 14,930 | den-crane | I don't understand your question.
Check https://clickhouse.tech/docs/en/sql-reference/aggregate-functions/combinators/#agg-functions-combinator-resample | 2020-09-17T13:05:18 | 2020-09-17T13:05:18 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 694,296,906 | 14,796 | 4ertus2 | There's also one issue: 01142_join_lc_and_nullable_in_key fails with `join_algorithm='partial_merge'`. It affects RIGHT and FULL JOIN with `join_use_nulls = 1`. | 2020-09-17T15:02:02 | 2020-09-17T17:12:44 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 694,394,962 | 14,832 | alexey-milovidov | > I would say that we need to forbid a tz argument for toDate
When you type `toDate('2019-01-13 01:02:03', 'UTC')` (a string literal), the timezone argument does nothing and we can forbid it.
When you type `toDate(datetime, 'UTC')` or `toDate(timestamp, 'UTC')` (an argument of type DateTime or integer), the timezon... | 2020-09-17T17:46:22 | 2020-09-17T17:46:44 | {} | MEMBER |
ClickHouse | ClickHouse | 694,137,084 | 14,846 | alesapin | Performance build with gcc-10, ok. | 2020-09-17T10:13:56 | 2020-09-17T10:13:56 | {} | MEMBER |
ClickHouse | ClickHouse | 694,137,730 | 14,221 | KochetovNicolai | Looks fine to me.
Probably, we could use a better name for settings?
Logically, it is just a limit on single server.
Could it be like `max_rows_to_read_for_single_server` or `max_rows_to_read_one_instance` or something like that?
@azat what do you think?
| 2020-09-17T10:15:26 | 2020-09-17T10:15:26 | {} | MEMBER |
ClickHouse | ClickHouse | 694,140,355 | 14,221 | azat | >max_rows_to_read_per_shard
Sounds great | 2020-09-17T10:21:11 | 2020-09-17T10:21:11 | {} | MEMBER |
ClickHouse | ClickHouse | 694,155,995 | 14,221 | KochetovNicolai | Yep, shard is not quite correct. We can read form several replicas of one shard in parallel, and limits will be applied separately.
> Instance can be both root (initial) and leaf (remote).
That's probably ok as long as settings are applied for both. Only rows read from remote replica on initial instance are not c... | 2020-09-17T10:56:39 | 2020-09-17T10:56:39 | {} | MEMBER |
ClickHouse | ClickHouse | 694,259,768 | 14,678 | alexey-milovidov | Yes, it is very valuable change. Thank you! | 2020-09-17T14:06:07 | 2020-09-17T14:06:07 | {} | MEMBER |
ClickHouse | ClickHouse | 694,042,971 | 14,846 | alesapin | Wait only for a performance build to be sure that it works with gcc-10. | 2020-09-17T07:36:10 | 2020-09-17T07:36:10 | {} | MEMBER |
ClickHouse | ClickHouse | 694,105,382 | 14,899 | keliss | Ok, thank you. | 2020-09-17T09:12:45 | 2020-09-17T09:12:45 | {} | NONE |
ClickHouse | ClickHouse | 694,107,150 | 14,853 | akuzm | Do we have any perf test for this? Maybe it doesn't show on our `hits` dataset and needs some weird setup... | 2020-09-17T09:15:56 | 2020-09-17T09:15:56 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 694,127,022 | 13,059 | filimonov | Possibility for naive fix (w/o dependencies resolution/topology sort etc)
1) use SQL file creation time to determine load order
2) if during loading some of the table we get an exception related to unsatisfied dependencies (lack of dictionary or similar) we can put the problematic table at the end of the queue to do... | 2020-09-17T09:54:47 | 2020-09-17T09:55:00 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 694,182,723 | 14,925 | alexey-milovidov | `timeout` will calculate wall clock time - it's unreliable. | 2020-09-17T11:57:39 | 2020-09-17T11:57:39 | {} | MEMBER |
ClickHouse | ClickHouse | 694,194,420 | 14,914 | den-crane | CH does not support this.
https://clickhouse.tech/docs/en/sql-reference/ansi/
E061-13 | Correlated subqueries | No
| 2020-09-17T12:22:36 | 2020-09-17T12:22:36 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 694,229,961 | 14,909 | alexey-milovidov | The IEEE double precision floating point numbers less than 2^-1022 are called "subnormal" (or "denormal") numbers.
They are supported by CPU but they can lead to significant slowdown:
https://randomascii.wordpress.com/2012/05/20/thats-not-normalthe-performance-of-odd-floats/
There are two different parsers in your... | 2020-09-17T13:23:36 | 2020-09-17T13:24:19 | {} | MEMBER |
ClickHouse | ClickHouse | 694,340,289 | 14,931 | simplemagik | deleting file with mutation on disk helped | 2020-09-17T16:15:12 | 2020-09-17T16:15:12 | {} | NONE |
ClickHouse | ClickHouse | 694,431,918 | 14,776 | tavplubix | > @tavplubix about DETACH/ATTACH sequence in DatabaseAtomic - is there any caveats, what should we do?
Test failure can be fixed by using short attach syntax (`ATTACH TABLE set;` instead of `ATTACH TABLE set (x String) ENGINE = Set() SETTINGS disable_persistency=1;`). Another way to fix it is to specify table UUID e... | 2020-09-17T18:53:24 | 2020-09-17T18:53:24 | {} | MEMBER |
ClickHouse | ClickHouse | 694,058,050 | 14,906 | alesapin | Cherry-picked manually. | 2020-09-17T07:49:53 | 2020-09-17T07:49:53 | {} | MEMBER |
ClickHouse | ClickHouse | 694,054,032 | 14,899 | alesapin | > But they don't get merged. We have such parts for over half a year. And if it is intended behaviour and the order matters, will there be any problems if we just use DROP PARTITION instead?
It's strange that they don't get merged, but it's possible if already existing non-empty parts are large enough. There will be... | 2020-09-17T07:46:23 | 2020-09-17T07:50:55 | {} | MEMBER |
ClickHouse | ClickHouse | 694,110,371 | 14,889 | akuzm | > AST Fuzzer:
>
> ```
> 56 Aborted (core dumped) ./clickhouse-server --config-file db/config.xml -- --path db 2>&1
> ```
>
> But no errors in the server log and in the fuzzer log.
No errors would be bad, better investigate. But I found the error in server log:
```
2020.09.17 03:44:13.65482... | 2020-09-17T09:22:01 | 2020-09-17T09:22:01 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 694,118,094 | 14,915 | filimonov | BTW: that looks like duplicate of #4491 | 2020-09-17T09:37:04 | 2020-09-17T09:37:04 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 694,134,702 | 14,920 | alexey-milovidov | What version do you use? | 2020-09-17T10:08:52 | 2020-09-17T10:08:52 | {} | MEMBER |
ClickHouse | ClickHouse | 694,304,751 | 14,941 | den-crane | It seems your aggregatingMergeTree has only 2 times fewer rows than source table.
Your aggregatingMergeTree aggregates almost nothing because ORDER BY is long.
uniqMerge works slower than uniq.
Make aggregatingMergeTree with shorter ORDER BY and 100 times fewer rows then you get a boost. | 2020-09-17T15:14:59 | 2020-09-17T15:15:23 | {
"+1": 1
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 694,207,126 | 14,832 | den-crane | >I would argue that toDate(date, tz) producing a tz-naive date is confusing.
Date by nature does not have a TZ attribute in the real world and in CH as well. By design.
I would say that we need to forbid a tz argument for `toDate` but I am not sure that it always does nothing.
>In other words, I believe toDate(d... | 2020-09-17T12:47:10 | 2020-09-17T16:52:31 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 694,058,112 | 14,905 | alesapin | Cherry-picked manually. | 2020-09-17T07:49:56 | 2020-09-17T07:49:56 | {} | MEMBER |
ClickHouse | ClickHouse | 694,119,586 | 14,221 | hagen1778 | Hi @azat @KochetovNicolai! I've update PR according to changes in https://github.com/ClickHouse/ClickHouse/pull/14815
Would appreciate your review! Pls let me know if it is worth to decline this PR and open a new one to make review easier. | 2020-09-17T09:40:03 | 2020-09-17T09:40:03 | {} | NONE |
ClickHouse | ClickHouse | 694,158,259 | 14,221 | hagen1778 | The `leaf` sounds natural to me. Pls also note the initial naming discussion here https://github.com/ClickHouse/ClickHouse/issues/13143. | 2020-09-17T11:01:44 | 2020-09-17T11:01:44 | {
"+1": 1
} | NONE |
ClickHouse | ClickHouse | 694,170,295 | 11,891 | alexey-milovidov | Every ClickHouse version is affected. | 2020-09-17T11:29:25 | 2020-09-17T11:29:25 | {
"+1": 1
} | MEMBER |
ClickHouse | ClickHouse | 694,047,951 | 14,899 | alesapin | It's not a bug and intended behavior. MergeTree relies on the order of blocks stored in parts, so we cannot easily remove any part from disk. BTW, empty parts should be merged soon with normal ones. | 2020-09-17T07:41:58 | 2020-09-17T07:41:58 | {} | MEMBER |
ClickHouse | ClickHouse | 694,057,900 | 14,907 | alesapin | Cherry-picked manually. | 2020-09-17T07:49:45 | 2020-09-17T07:49:45 | {} | MEMBER |
ClickHouse | ClickHouse | 694,165,343 | 7,287 | alexey-milovidov | Most likely fixed by https://github.com/ClickHouse/ClickHouse/pull/14444
| 2020-09-17T11:17:57 | 2020-09-17T11:17:57 | {} | MEMBER |
ClickHouse | ClickHouse | 694,176,637 | 14,813 | KochetovNicolai | Not waiting for other tests. | 2020-09-17T11:44:21 | 2020-09-17T11:44:21 | {} | MEMBER |
ClickHouse | ClickHouse | 694,307,858 | 14,852 | liangli1213 | @tavplubix , I saw you have approved the change, and when it will merge to master branch? | 2020-09-17T15:20:12 | 2020-09-17T15:20:12 | {} | NONE |
ClickHouse | ClickHouse | 694,325,475 | 14,931 | simplemagik | also
SELECT * FROM merges
returns 0
so there is no merges running
| 2020-09-17T15:49:27 | 2020-09-17T15:49:36 | {} | NONE |
ClickHouse | ClickHouse | 694,151,905 | 14,362 | alexey-milovidov | @bobrik This is just a temporary change to allow CI to run before we optimize the build.
Now the build is optimized and we are changing it back.
PS. We don't control the amount of memory on CI servers. I think, we should assume that they have 128 GiB. | 2020-09-17T10:47:13 | 2020-09-17T10:47:31 | {} | MEMBER |
ClickHouse | ClickHouse | 694,197,281 | 14,899 | den-crane | related https://github.com/ClickHouse/ClickHouse/issues/5491 | 2020-09-17T12:28:35 | 2020-09-17T12:28:35 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 694,246,995 | 14,436 | alexey-milovidov | I confirm, updating to 20.6 will help.
ALTER queries are 100% lock-free* since 20.6.
\* lock-free means that they don't use table-level locks, only small locks on data structures that cannot block queries. | 2020-09-17T13:47:28 | 2020-09-17T13:47:28 | {} | MEMBER |
ClickHouse | ClickHouse | 694,420,154 | 14,776 | alexey-milovidov | @tavplubix about DETACH/ATTACH sequence in DatabaseAtomic - is there any caveats, what should we do? | 2020-09-17T18:31:45 | 2020-09-17T18:31:45 | {} | MEMBER |
ClickHouse | ClickHouse | 695,217,651 | 11,778 | alexey-milovidov | This PR is empty. Close. | 2020-09-19T14:06:08 | 2020-09-19T14:06:08 | {} | MEMBER |
ClickHouse | ClickHouse | 695,219,451 | 10,748 | alexey-milovidov | Will be reimplemented by @alesapin. | 2020-09-19T14:23:07 | 2020-09-19T14:23:07 | {} | MEMBER |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.