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 | 631,992,866 | 11,084 | qoega | @akuzm you can check memory diff will be found in performance test for this PR | 2020-05-21T09:46:38 | 2020-05-21T09:46:38 | {} | MEMBER |
ClickHouse | ClickHouse | 632,057,019 | 11,101 | alexey-milovidov | Yes, case insensitive regexps are simply not supported due to the use of OptimizedRegularExpression. | 2020-05-21T12:24:38 | 2020-05-21T12:24:38 | {} | MEMBER |
ClickHouse | ClickHouse | 632,064,231 | 11,102 | geekyouth | https://repo.yandex.ru/clickhouse/rpm/stable/x86_64/
change url | 修正地址 | 2020-05-21T12:41:40 | 2020-05-21T12:41:40 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 632,250,732 | 11,092 | den-crane | > yes,my table support sample data,the table has sample key.
then it's another issue. Not #5733 | 2020-05-21T17:50:05 | 2020-05-21T17:50:05 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 631,627,645 | 9,139 | alexey-milovidov | Continued in https://github.com/ClickHouse/ClickHouse/pull/10351 | 2020-05-20T17:49:40 | 2020-05-20T17:49:40 | {} | MEMBER |
ClickHouse | ClickHouse | 631,676,097 | 11,022 | azat | @alexey-milovidov
> >Functional stateless tests (ubsan) — fail: 1, passed: 1848, skipped: 6
>Can be fixed with some hack, but it will go away anyway after #11029
I can rebase against master to make tests green if you not on this right now?
And one more questions from the big [comment](https://github.com/Clic... | 2020-05-20T19:23:56 | 2020-05-20T19:23:56 | {} | MEMBER |
ClickHouse | ClickHouse | 632,005,566 | 11,003 | haarchri | so it looks like the same error ... i set the both ENV variables - but in real we are in an openshift environment | 2020-05-21T10:17:20 | 2020-05-21T10:17:20 | {} | NONE |
ClickHouse | ClickHouse | 632,009,665 | 11,074 | azat | >Thie leak issue can be detected with test_replicated_merge_tree_s3/test.py::test_insert_select_replicated (can be found on current master).
@alesapin AFAIU it will trigger the issue only when s3 (minio) will not be available, am I missing something?
For example [this](https://clickhouse-test-reports.s3.yandex.ne... | 2020-05-21T10:26:52 | 2020-05-21T10:26:52 | {} | MEMBER |
ClickHouse | ClickHouse | 632,347,578 | 11,094 | den-crane | SimpleAggregateFunction could not support argMax, because argMax require State and SimpleAggregateFunction made for functions without State (sum/min/max)
But AggregateFunction(argMax can: https://gist.github.com/den-crane/d03524eadbbce0bafa528101afa8f794
CH has designed for fast queries against immutable data tha... | 2020-05-21T21:09:56 | 2020-05-21T21:10:24 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 632,415,801 | 11,119 | akuzm | To sum up our discussion on TG, the main problems with the test are:
* Requires exclusive access to the entire server -- can't run it in parallel with other tests, as is often done during development.
* Uses summation over memory profiler data -- I'm not sure it can work correctly at all. The memory profiler performs... | 2020-05-22T00:38:13 | 2020-05-22T00:43:17 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 632,490,590 | 11,123 | filimonov | > Determine File columns from the file header for TSVWithNamesAndTypes. Goodbye to endlessly writing out the table structure, a real imporvement to the quality of life.
Extra options:
- take all columns as strings
- determinie minimal type of the column
- auto name columns as col1,..., colN
That will also all... | 2020-05-22T05:29:36 | 2020-05-22T05:29:36 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 632,537,669 | 11,128 | blinkov | @f7753 actually, those look like the function names and they're supposed to be all in English as they are used in SQL queries. | 2020-05-22T07:22:38 | 2020-05-22T07:22:38 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 631,892,884 | 3,271 | jakesylvestre | Any update? We’re running into this as well | 2020-05-21T05:42:05 | 2020-05-21T05:42:05 | {} | NONE |
ClickHouse | ClickHouse | 632,014,743 | 11,082 | Felixoid | Are failures something which my changes have caused? | 2020-05-21T10:39:21 | 2020-05-21T10:39:21 | {} | MEMBER |
ClickHouse | ClickHouse | 632,268,522 | 10,771 | jakesylvestre | Awesome, thanks @BayoNet | 2020-05-21T18:27:31 | 2020-05-21T18:27:31 | {} | NONE |
ClickHouse | ClickHouse | 632,286,182 | 11,112 | den-crane | You do understand that a proper insert (**without functions in values**) uses a stream parser and does not require max_query_size/max_parser_depth/stack/ast/etc and can be **any length**???
see parameter : **--input_format_values_interpret_expressions arg For Values format: if the field could not... | 2020-05-21T19:02:58 | 2020-05-21T19:10:46 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 632,290,861 | 11,112 | jakesylvestre | @den-crane - still running into this after setting that option. Will get you reproduction code now | 2020-05-21T19:11:54 | 2020-05-21T19:11:54 | {} | NONE |
ClickHouse | ClickHouse | 632,451,246 | 11,092 | den-crane | Then I guess it is related to select count() optimization which uses parts metadata | 2020-05-22T02:59:07 | 2020-05-22T02:59:07 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 632,549,505 | 11,004 | KochetovNicolai | Test fails are the same as on master. | 2020-05-22T07:47:08 | 2020-05-22T07:47:08 | {} | MEMBER |
ClickHouse | ClickHouse | 631,652,503 | 11,075 | abraithwaite | I have not! We will try that :+1: | 2020-05-20T18:39:18 | 2020-05-20T18:39:18 | {} | NONE |
ClickHouse | ClickHouse | 631,940,113 | 11,061 | asurovenko | **den-crane**, thanks. Could you say me how I can drop all partitions older that month? As I search in Google, there is no way to drop partitions by condition? | 2020-05-21T07:48:13 | 2020-05-21T07:48:13 | {} | NONE |
ClickHouse | ClickHouse | 632,206,895 | 11,084 | akuzm | `MemoryStatisticsOS` could be useful, but it's in `system.asynchronous_metrics`, and there is no log for them yet https://github.com/ClickHouse/ClickHouse/issues/9429 | 2020-05-21T16:39:45 | 2020-05-21T16:39:45 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 632,306,548 | 479 | socketpair | #1765 | 2020-05-21T19:45:00 | 2020-05-21T19:45:00 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 632,538,856 | 11,003 | haarchri | hello, the problem in openshift is that the clickhouse image is with root and openshift sets a other id like 1016890000 and if we map a volume (pvc) to the container then it looks like:
`2 drwxrwsr-x. 3 root 1016890000 4.0K May 18 09:24 clickhouse`
here is a example of an working docker image for clickhouse:
... | 2020-05-22T07:25:17 | 2020-05-22T07:25:17 | {} | NONE |
ClickHouse | ClickHouse | 631,835,133 | 11,067 | den-crane | >Which UI based tools do you use to keep
zabbix.
>DB::Exception: Table system.metric_log doesn't exist..
it's enabled by default. But this table exists only in the recent versions 20.3 or something. | 2020-05-21T02:08:17 | 2020-05-21T02:37:38 | {
"+1": 1
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 631,929,318 | 11,022 | alexey-milovidov | First for me is to fix unrelated integration test failure... then proceed with PRs. | 2020-05-21T07:22:17 | 2020-05-21T07:22:17 | {} | MEMBER |
ClickHouse | ClickHouse | 632,053,176 | 11,023 | alexey-milovidov | It's Ok. | 2020-05-21T12:14:17 | 2020-05-21T12:14:17 | {} | MEMBER |
ClickHouse | ClickHouse | 632,107,272 | 10,973 | nikitamikhaylov | This feature is awesome! | 2020-05-21T14:09:19 | 2020-05-21T14:09:56 | {} | MEMBER |
ClickHouse | ClickHouse | 632,150,188 | 11,084 | akuzm | What we could see in performance tests is probably the reduction in global memory usage -- metric_log is exported, but I'm not sure which metric we need. | 2020-05-21T15:28:07 | 2020-05-21T15:28:07 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 632,237,216 | 11,084 | alexey-milovidov | Both PRs are Ok according to the performance tests.
The question is - what is the actual difference in memory fragmentation that you are experiencing. | 2020-05-21T17:23:41 | 2020-05-21T18:29:04 | {} | MEMBER |
ClickHouse | ClickHouse | 632,293,098 | 11,112 | den-crane | max_parser_depth -- was introduced to address an issue with really complicated queries with huge recursion level like 10000 ((((((((( )))))) | 2020-05-21T19:16:01 | 2020-05-21T19:16:01 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 632,306,687 | 10,376 | socketpair | #1765 | 2020-05-21T19:45:17 | 2020-05-21T19:45:17 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 631,687,307 | 11,022 | azat | @alexey-milovidov also what do you think about enabling `background_thread:true` by default (or limit number of arenas? or even use percpu arena, but there are some some known performance issues for it) ? | 2020-05-20T19:45:55 | 2020-05-20T19:45:55 | {} | MEMBER |
ClickHouse | ClickHouse | 631,991,913 | 10,591 | alexey-milovidov | But there is no test for this feature :( | 2020-05-21T09:44:20 | 2020-05-21T09:44:20 | {} | MEMBER |
ClickHouse | ClickHouse | 632,110,141 | 11,106 | alesapin | It's already tested by `test_version_update_after_mutation` integration test. But we run them with retries and sometimes it helps. | 2020-05-21T14:14:34 | 2020-05-21T14:14:34 | {} | MEMBER |
ClickHouse | ClickHouse | 632,392,879 | 11,114 | azat | Will undraft it once the test failures will be investigated | 2020-05-21T23:16:49 | 2020-05-21T23:16:49 | {} | MEMBER |
ClickHouse | ClickHouse | 632,440,685 | 11,059 | templarzq | it need people work days.. | 2020-05-22T02:18:05 | 2020-05-22T02:18:49 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 632,466,763 | 2,352 | hongyuzhou | > column
Thanks for your answer... Make a new `timestamp` column sound a good idea, but could you please explain `a1 * c1 + a2 * c2 + a3 * c3..` specifically more? What do these `a1, c1, a2, c2...` parameters represent?
For example, here is a demo:
| pid | event | event_time |
|---|---|---|
| 123 | A |... | 2020-05-22T04:05:07 | 2020-05-22T04:05:07 | {} | NONE |
ClickHouse | ClickHouse | 631,988,206 | 10,858 | KochetovNicolai | I've only checked that tests are ok. | 2020-05-21T09:35:54 | 2020-05-21T09:35:54 | {} | MEMBER |
ClickHouse | ClickHouse | 632,112,985 | 11,061 | den-crane | @asurovenko there is no such feature in CH. I use bash script - 3 lines.
```
sql="select distinct partition_id from system.parts
where database = '$db' and table='$table' and active
and min_date < today()-$days format TabSeparatedRaw
"
for partition in `$ch_client "$sql"`
do
sql="alter table \`$table... | 2020-05-21T14:19:55 | 2020-05-21T14:19:55 | {
"heart": 1
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 632,173,516 | 11,092 | weiyongyuan | > Do you have sampling on the table ? `Sample by` section?
>
> max_parallel_replicas -- supposed to work against sampled tables. Each replica processes part of samples.
>
> Duplicate #5733
yes,my table support sample data,the table has sample key. | 2020-05-21T15:55:19 | 2020-05-21T15:56:04 | {} | NONE |
ClickHouse | ClickHouse | 632,487,020 | 3,271 | iameugenejo | We're not seeing this issue anymore and we're at 19.14.10.16 now. | 2020-05-22T05:18:12 | 2020-05-22T05:18:12 | {} | NONE |
ClickHouse | ClickHouse | 631,683,858 | 10,917 | azat | >Jemalloc does not free allocated memory immediately
looks too huge even for jemalloc + thread pool (with taking into account that only allocations < 64MB done via jemalloc, other with direct mmap)
actually it is very easy to verify, @huang--yijian can you try use one of the following and get back with results?
... | 2020-05-20T19:39:05 | 2020-05-20T19:39:05 | {} | MEMBER |
ClickHouse | ClickHouse | 631,721,571 | 11,022 | azat | >Let's test all three variants.
Agree
P.S. But I don't think that `nareans:16` (or anyt other) should be tested, other two variants is enough.
> You can send three draft pull requests, we will wait for performance tests and look.
Ok (although I don't like PR storm :) )
>If memory usage will be lower
... | 2020-05-20T20:55:07 | 2020-05-20T20:55:07 | {} | MEMBER |
ClickHouse | ClickHouse | 632,159,800 | 9,609 | sergey-buturlakin | I'm not aware of logging subsystem usage and not confident that the proposed timeout options pass method is optimal. Thought that the subsystem author could apply my suggestions most optimally :)
I hope that I'll create PRs in the future after improving the codebase understanding and preparing an appropriate developme... | 2020-05-21T15:35:48 | 2020-05-21T15:35:48 | {} | NONE |
ClickHouse | ClickHouse | 632,431,163 | 2,352 | sundy-li | > My suggestion is do a little change
In this case, your advice may make the versatility lost. I don't think we should modify the `ComparePairFirst` function, yet we can make a new column, such as `a1 * c1 + a2 * c2 + a3 * c3..` to be another kind of `timestamp` in source data.
| 2020-05-22T01:39:45 | 2020-05-22T01:40:24 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 632,552,832 | 11,106 | alesapin | Deadlock happens here https://github.com/clickhouse/ClickHouse/blob/master/src/Storages/MergeTree/MergeTreeDataMergerMutator.cpp#L186 | 2020-05-22T07:53:46 | 2020-05-22T07:53:46 | {} | MEMBER |
ClickHouse | ClickHouse | 631,652,256 | 11,075 | blinkov | @abraithwaite have you tried altering the underlying inner table instead of materialized view itself? | 2020-05-20T18:38:49 | 2020-05-20T18:38:49 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 631,892,484 | 11,090 | alexey-milovidov | Test query:
```
SELECT max(1) FROM test.hits
```
Additional issue:
- this query requires the user to have access to some column that is unknown in advance:
```
Access granted: SELECT(`ParsedParams.Key5`) ON test.hits
``` | 2020-05-21T05:40:37 | 2020-05-21T05:40:37 | {} | MEMBER |
ClickHouse | ClickHouse | 631,685,505 | 11,078 | den-crane | 20.3.9
```
INSERT INTO test VALUES (2, 'ruuuu');
Exception on client:
Code: 131. DB::Exception: String too long for type FixedString(2)
SELECT * FROM test
┌─id─┬─val─┐
│ 1 │ ru │
└────┴─────┘
```
| 2020-05-20T19:42:26 | 2020-05-21T14:53:06 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 632,231,669 | 11,084 | azat | @akuzm `MemoryTracking` from `system.metrics` periodically resets to the RSS, but I doubt that this is enough or even useful
https://github.com/ClickHouse/ClickHouse/blob/f98ffdbc4ca7ff32350c4b2fd5f672d70168cb2d/src/Interpreters/AsynchronousMetrics.cpp#L155
Maybe there is some internal monitoring for this machine... | 2020-05-21T17:13:02 | 2020-05-21T17:13:02 | {} | MEMBER |
ClickHouse | ClickHouse | 632,262,528 | 10,745 | Brutalis64 | Actually not yet, may be in 20.4, i didn't try it. | 2020-05-21T18:14:47 | 2020-05-21T18:14:47 | {} | NONE |
ClickHouse | ClickHouse | 632,420,833 | 11,112 | jakesylvestre | Hi,
Your correct. I'm inadvertently attempting to create a recursive MV here. All set! | 2020-05-22T00:59:26 | 2020-05-22T00:59:26 | {} | NONE |
ClickHouse | ClickHouse | 632,528,977 | 8,889 | jackpgao | The error appears with 20.3.8.53.
It seems releated to MySQL engine.
When it occours, server load will blow to more than 1000.
```
2020.05.22 15:37:32.357591 [ 37327 ] {} <Fatal> BaseDaemon: ########################################
2020.05.22 15:37:32.357638 [ 37327 ] {} <Fatal> BaseDaemon: (version 20.3.8... | 2020-05-22T07:05:17 | 2020-05-22T07:58:42 | {} | NONE |
ClickHouse | ClickHouse | 631,806,383 | 11,067 | sundarv85 | Thanks @den-crane
`InsertQuery` returns a number - which I assume it is the number of inserts at this point in time. Which UI based tools do you use to keep a tab on how the inserts are done over a period of time.
For the other query, we get the error
```
DB::Exception: Table system.metric_log doesn't exist.... | 2020-05-21T00:23:24 | 2020-05-21T00:23:24 | {} | NONE |
ClickHouse | ClickHouse | 632,314,476 | 5,387 | akuzm | I added some very basic support here: https://github.com/ClickHouse/ClickHouse/pull/10915 Had to rewrite from scatch because of conflicts.
@manish53 please check if that is suitable for you. I'll close this one. | 2020-05-21T20:03:27 | 2020-05-21T20:03:27 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 631,889,835 | 11,089 | alexey-milovidov | Yes, we are planning to support these queries.
The proposed implementation is the following:
1. Support `@@identifier` expressions in parser and translate them to functions `globalVariable('identifier')`. (doing it directly in parser is Ok)
2. Implement this function and add support for MySQL global variables that... | 2020-05-21T05:30:46 | 2020-05-21T05:31:15 | {} | MEMBER |
ClickHouse | ClickHouse | 631,917,123 | 10,973 | qoega | You can also add optional way to generate data using randomGenerate table function. To add specific logic about certain cases you can transform this random values to more specific ones straight in INSERT SELECT QUERY | 2020-05-21T06:50:11 | 2020-05-21T06:50:11 | {} | MEMBER |
ClickHouse | ClickHouse | 632,054,206 | 10,678 | alexey-milovidov | We need to add a setting `allow_experimental_geo_types`. | 2020-05-21T12:16:59 | 2020-05-21T12:16:59 | {} | MEMBER |
ClickHouse | ClickHouse | 632,069,415 | 11,102 | alexey-milovidov | @blinkov FYI | 2020-05-21T12:54:17 | 2020-05-21T12:54:17 | {} | MEMBER |
ClickHouse | ClickHouse | 632,123,862 | 11,092 | den-crane | Do you have sampling on the table ? `Sample by` section?
max_parallel_replicas -- supposed to work against sampled tables. Each replica processes part of samples.
Duplicate https://github.com/ClickHouse/ClickHouse/issues/5733 | 2020-05-21T14:40:31 | 2020-05-21T14:40:31 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 632,235,177 | 11,096 | alexey-milovidov | Tests broken in master, if something is wrong, @qoega will fix. | 2020-05-21T17:19:37 | 2020-05-21T17:19:37 | {} | MEMBER |
ClickHouse | ClickHouse | 632,301,117 | 11,003 | azat | >so it looks like the same error ... i set the both ENV variables
Are you sure you are using 20.3? Can you post the command output and `ls -l /path/to/data/volume/*`?
> - but in real we are in an openshift environment
I did not use openshift and don't know what restrictions it has? You cannot pass environment va... | 2020-05-21T19:32:28 | 2020-05-21T19:32:28 | {} | MEMBER |
ClickHouse | ClickHouse | 632,428,555 | 11,092 | weiyongyuan | In my tests,only count query result is not correct. | 2020-05-22T01:29:11 | 2020-05-22T01:29:11 | {} | NONE |
ClickHouse | ClickHouse | 631,674,804 | 11,075 | abraithwaite | That worked for the inner tables (alter table replicated across replicas), but we don't see those changes reflected on the normal tables still. Is that expected? | 2020-05-20T19:21:27 | 2020-05-20T19:22:03 | {} | NONE |
ClickHouse | ClickHouse | 631,784,138 | 11,022 | azat | Rebased, previous HEAD was 47965fbf5cf2edbdb13c3fd31caceef0cb246ac5 | 2020-05-20T23:08:03 | 2020-05-20T23:08:03 | {} | MEMBER |
ClickHouse | ClickHouse | 631,860,953 | 2,352 | hongyuzhou | > In 2017, there was a wonderful [share](https://events.yandex.com/lib/talks/247/) by `Mariya Mansurova` who taught us how to calculate complex funnel metrics in ClickHouse by `sequenceMatch`.
>
> But it was less efficient (pattern matching) and it lacks the sliding timestamp window support, which is very important ... | 2020-05-21T03:42:54 | 2020-05-21T03:42:54 | {} | NONE |
ClickHouse | ClickHouse | 632,025,337 | 11,025 | alexey-milovidov | It's totally Ok as they are intended for Unicode strings.
We support Unicode strings in our String type with the http://utf8everywhere.org/ principle. | 2020-05-21T11:06:03 | 2020-05-21T11:06:03 | {} | MEMBER |
ClickHouse | ClickHouse | 632,258,088 | 8,438 | jakesylvestre | How do we increase the stack size? | 2020-05-21T18:05:09 | 2020-05-21T18:05:09 | {} | NONE |
ClickHouse | ClickHouse | 632,333,296 | 11,084 | azat | Using latest upstream/master (19067621190677a0589ae3af84514f9a8cd94356)
And started server with `max_server_memory_usage=8G`:
```
clickhouse-server -- --max_server_memory_usage=$((8<<30))
```
Then I get the test from #10818
```
clickhouse-benchmark --max_threads 16 <<< "select WatchID from hits_100m_obfu... | 2020-05-21T20:39:42 | 2020-05-21T20:39:42 | {
"+1": 1
} | MEMBER |
ClickHouse | ClickHouse | 632,430,067 | 11,098 | weiyongyuan | BTW,the replicated table has a dependency on zookeeper,can we provide some tools or commands revover the replica status from readonly to active .And in my tests,there is a chance two replicas is nethier leader nor active,this seems like what brain splits .It willi be very useful if we can provide sool tools or commands... | 2020-05-22T01:35:26 | 2020-05-22T01:35:26 | {} | NONE |
ClickHouse | ClickHouse | 631,990,309 | 11,074 | alesapin | Thie leak issue can be detected with `test_replicated_merge_tree_s3/test.py::test_insert_select_replicated` (can be found on current master). | 2020-05-21T09:40:28 | 2020-05-21T09:41:03 | {} | MEMBER |
ClickHouse | ClickHouse | 632,388,363 | 3,271 | filimonov | > Any update? We’re running into this as well
We need a reproducable test case. Try to create minimał sample. | 2020-05-21T23:01:11 | 2020-05-21T23:01:11 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 632,441,169 | 11,059 | templarzq | > Also we need a warp drive and laser guns
so why not build warp ship and laser guns? | 2020-05-22T02:19:49 | 2020-05-22T02:19:49 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 632,455,105 | 3,271 | den-crane | >And we also notice the running aggregation for current day(during ingestion) are matching usually.
It could be that TZ is the reason. `substring(partition_code, 1, 10) as partition_code` | 2020-05-22T03:15:15 | 2020-05-22T03:15:15 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 631,726,508 | 11,082 | Felixoid | Here's a list of remainings before `LGTM`:
- Tests in `tests/integration/test_distributed_ddl/test_replicated_alter.py`
- Documentation for `distributed_ddl_replication_check` | 2020-05-20T21:05:22 | 2020-05-20T21:05:22 | {} | MEMBER |
ClickHouse | ClickHouse | 631,753,781 | 9,503 | 4ertus2 | Continuation is here #11083 | 2020-05-20T22:08:36 | 2020-05-20T22:54:01 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 632,216,387 | 10,973 | nikitamikhaylov | ```sql
SELECT uniqExact(multiIf(ticket_status_name IN ('closed', 'new', 'processing', 'registered'), ticket_id, NULL))
FROM table_name
Received exception from server (version 20.4.1):
Code: 53. DB::Exception: Received from localhost:9000. DB::Exception: Type mismatch in IN or VALUES section. Expected: Int64. Got:... | 2020-05-21T16:44:44 | 2020-05-21T16:44:44 | {} | MEMBER |
ClickHouse | ClickHouse | 632,231,922 | 10,825 | azat | Maybe a trigger-ci button then (to avoid random commits) ? | 2020-05-21T17:13:29 | 2020-05-21T17:13:29 | {
"+1": 1
} | MEMBER |
ClickHouse | ClickHouse | 632,294,101 | 11,112 | den-crane | Maybe `x.yz` is a view ? But CH does not support recursive views | 2020-05-21T19:18:16 | 2020-05-21T19:18:16 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 632,440,347 | 11,059 | templarzq | i've finished the optimization of the merge algorithm of replacingmerge tree and get 30% speedup on final query , and the remained work is not very difficult. | 2020-05-22T02:16:44 | 2020-05-22T02:16:44 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 632,504,671 | 10,621 | azat | Test report summary:
- 01281_group_by_limit_memory_tracking - #11119
- test_version_update_after_mutation - looks flacky
- test_replicated_merge_tree_s3
```
"Instance `{}' failed to start. Container status: {}, logs: {}".format(self.name, status,
> ... | 2020-05-22T06:08:00 | 2020-05-22T06:08:00 | {} | MEMBER |
ClickHouse | ClickHouse | 631,642,223 | 11,074 | alexey-milovidov | +1, let's add a test that can reproduce the leak before this change (no need to validate it, just let's add this test case). | 2020-05-20T18:18:47 | 2020-05-20T18:18:47 | {} | MEMBER |
ClickHouse | ClickHouse | 631,668,937 | 11,076 | akuzm | This is more of a quick fix, and in the long run we need to have structured error messages (like error/context/hint in postgres), and scoped error hints. | 2020-05-20T19:09:37 | 2020-05-20T19:09:37 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 632,093,105 | 10,973 | ZeDRoman | > You can also add optional way to generate data using randomGenerate table function. To add specific logic about certain cases you can transform this random values to more specific ones straight in INSERT SELECT QUERY
Oh, that's better way to create random values instead of mine funcs. I'll add it | 2020-05-21T13:42:08 | 2020-05-21T13:42:08 | {} | NONE |
ClickHouse | ClickHouse | 632,118,331 | 11,105 | den-crane | duplicate https://github.com/ClickHouse/ClickHouse/issues/8469
already fixed
```
20.3.9.70
┌─val─┐
│ 99 │
│ 199 │
│ 299 │
│ 399 │
│ 499 │
│ 599 │
│ 699 │
│ 799 │
│ 899 │
│ 991 │
│ 993 │
│ 995 │
│ 997 │
│ 999 │
└─────┘
``` | 2020-05-21T14:29:57 | 2020-05-21T14:29:57 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 632,460,437 | 11,092 | weiyongyuan | > Then I guess it is related to select count() optimization which uses parts metadata
Yes,i think so! | 2020-05-22T03:37:09 | 2020-05-22T03:37:09 | {} | NONE |
ClickHouse | ClickHouse | 631,696,660 | 11,078 | asamr | Cool! It's time to upgrade. Thank you | 2020-05-20T20:04:41 | 2020-05-20T20:04:41 | {} | NONE |
ClickHouse | ClickHouse | 631,721,816 | 11,022 | azat | And what about this PR?
>I can rebase against master to make tests green if you not on this right now? | 2020-05-20T20:55:39 | 2020-05-20T20:55:39 | {} | MEMBER |
ClickHouse | ClickHouse | 631,889,996 | 7,997 | shizhenchina | Have you solved the problem? | 2020-05-21T05:31:20 | 2020-05-21T05:31:20 | {} | NONE |
ClickHouse | ClickHouse | 631,922,482 | 11,088 | hczhcz | @alexey-milovidov
Arrow ("file" mode), ArrowStream, and Parquet are highly overlapped in type system, internal implementation, etc. The difference is, ArrowStream can be parsed without seeking. Thus, we can process it easily via network, pipe, etc without buffering. On the other hand, we will choose Arrow when random... | 2020-05-21T07:04:33 | 2020-05-21T07:04:33 | {
"+1": 1
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 631,995,360 | 10,858 | KochetovNicolai | Preftest failure is not caused by this pr. | 2020-05-21T09:52:40 | 2020-05-21T09:52:40 | {} | MEMBER |
ClickHouse | ClickHouse | 632,146,829 | 11,084 | akuzm | > @akuzm you can check memory diff will be found in performance test for this PR
This PR just changes where the memory is allocated, but won't change per-query memory usage. | 2020-05-21T15:22:28 | 2020-05-21T15:22:28 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 632,172,065 | 7,019 | champtar | This is still broken in latest lts (20.3.9), it should be in 20.3.10 (https://github.com/ClickHouse/ClickHouse/pull/10684#issuecomment-624912487)
commit: https://github.com/ClickHouse/ClickHouse/commit/ab3adc46519259d1be7b277630ac046ad9dbdf9b
```
$ podman run -d --network host --name clickhouse-server yandex/clickho... | 2020-05-21T15:53:51 | 2020-05-21T17:07:45 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 632,259,552 | 10,283 | jakesylvestre | what sets the `max_stack_size`? | 2020-05-21T18:08:16 | 2020-05-21T18:08:16 | {} | NONE |
ClickHouse | ClickHouse | 632,496,886 | 11,127 | blinkov | @samz406 as a workaround, you can just list all columns that need to be returned. | 2020-05-22T05:48:18 | 2020-05-22T05:48:18 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 632,524,209 | 11,119 | azat | >This PR fixes none of the above, but tweaks the test to stop failing for now -- well, this is also useful, no objections.
This was an attempt to make test green ASAP, but once I saw that it does not fix "anything", I've marked it as draft.
>A fix for the above issues could be to use query log data instead of mem... | 2020-05-22T06:55:48 | 2020-05-22T06:55:48 | {} | MEMBER |
ClickHouse | ClickHouse | 632,605,522 | 8,889 | jackpgao | > Does it crash if `enable_scalar_subquery_optimization ` is disabled?
- Yes.
```
SET enable_scalar_subquery_optimization = 0
Ok.
0 rows in set. Elapsed: 0.001 sec.
SELECT
xxxxx
GROUP BY date
FORMAT Null
↙ Progress: 0.00 rows, 0.00 B (0.00 rows/s., 0.00 B/s.)
Exception on client:
Code: 3... | 2020-05-22T09:53:40 | 2020-05-22T09:53:40 | {} | NONE |
ClickHouse | ClickHouse | 620,315,901 | 10,545 | azat | And seems that performance improved (since 20.3) maybe related to #8728 (it was backported to 20.1 but not fully) | 2020-04-28T01:04:41 | 2020-04-28T01:05:04 | {} | MEMBER |
ClickHouse | ClickHouse | 620,609,737 | 9,450 | azat | >Probably we need to add help or man for clickhouse-client with binds and other features.
Or maybe it will be better to make it fully readline compatible (since right now it is not), and help will not be needed (but a note that inpurc is not interpreted maybe a good thing) | 2020-04-28T13:31:21 | 2020-04-28T13:31:21 | {} | MEMBER |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.