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 | 748,651,032 | 18,238 | alexey-milovidov | A link to upstream PR: https://github.com/jemalloc/jemalloc/pull/2005 | 2020-12-20T19:34:44 | 2020-12-20T19:34:44 | {} | MEMBER |
ClickHouse | ClickHouse | 748,775,631 | 17,838 | weeds085490 | > > > @weeds085490 The code looks Ok but the tests deadlocked. I don't see the reason.
> > > Could you please merge/rebase with master?
> >
> >
> > Thanks for reviewing. I have merged with master. PTAL
>
> @alexey-milovidov `DROP TABLE test_table` in my function test still hung. It passed when test it locally.... | 2020-12-21T05:54:36 | 2020-12-21T05:54:36 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 749,007,787 | 18,223 | alesapin | Stress test failures are not related to changes. | 2020-12-21T14:42:04 | 2020-12-21T14:42:04 | {} | MEMBER |
ClickHouse | ClickHouse | 748,636,869 | 18,153 | sevirov | @akuzm Please, review this PR. | 2020-12-20T17:34:03 | 2020-12-20T17:34:03 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 748,667,312 | 17,847 | fibersel | 1. There is not level option for LZ4 into clickhouse-compressor tool.The only option for LZ4 is using HC codec.
2. Is default build type debug?If it is so, I am going to rebuild it...
3. Yes, units are mb/s | 2020-12-20T20:34:02 | 2020-12-20T20:34:02 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 748,669,156 | 17,847 | fibersel | Is it possible that files like URL.dat (600 mb) are too lightweight to measure compression speed? | 2020-12-20T20:50:38 | 2020-12-20T20:50:38 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 748,709,621 | 17,838 | weeds085490 | > @weeds085490 The code looks Ok but the tests deadlocked. I don't see the reason.
> Could you please merge/rebase with master?
Thanks for reviewing. I have merged with master. PTAL | 2020-12-21T01:30:21 | 2020-12-21T01:30:21 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 748,846,971 | 17,955 | BhavyaRajSharma | hi @SaltTan ,
I didn't use any DB engine while creating the database so it was taking the default database. I get to the below link while getting info about database creation in click house and here there is nothing about ordinary engine.
> https://clickhouse.tech/docs/en/engines/database-engines/
I used the below ... | 2020-12-21T08:38:13 | 2020-12-21T08:48:52 | {} | NONE |
ClickHouse | ClickHouse | 749,053,453 | 15,167 | alexey-milovidov | No longer needed (everything works fine). | 2020-12-21T16:11:04 | 2020-12-21T16:11:04 | {} | MEMBER |
ClickHouse | ClickHouse | 748,472,902 | 17,397 | compasses | @KochetovNicolai Oh, thank you very much. "but probably slower than String" I suppose should be always slower than the original column string? Since the LowCardinality type has a hidden index array, when the distinct values too many that would not do any help for compute and IO read. Am I right?
BTW, What's the plan ... | 2020-12-19T13:09:28 | 2020-12-19T13:09:28 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 748,562,287 | 18,214 | amosbird | > Ok.
>
> Now UInt8 is also a supertype for UInt8 and Enum8?
>
Enum8 is treated as Int8 so the super type would be Int16 | 2020-12-20T04:44:21 | 2020-12-20T04:44:21 | {} | COLLABORATOR |
ClickHouse | ClickHouse | 748,570,201 | 15,191 | alexey-milovidov | The issue was fixed in #16895. | 2020-12-20T06:34:08 | 2020-12-20T06:34:08 | {} | MEMBER |
ClickHouse | ClickHouse | 748,741,713 | 9,996 | zhang2014 | @alexey-milovidov I agree with @geldot too.
> PS. This feature is useful and it is appreciated.
As of now, we already have play web UI but I will be happy to also see clickhouse-client interface exposed via web terminal - it's good for demo purposes :)
(Also it's ok to just close this task if you don't want to both... | 2020-12-21T03:49:41 | 2020-12-21T03:49:41 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 748,742,575 | 17,838 | weeds085490 | > > @weeds085490 The code looks Ok but the tests deadlocked. I don't see the reason.
> > Could you please merge/rebase with master?
>
> Thanks for reviewing. I have merged with master. PTAL
@alexey-milovidov `DROP TABLE test_table` in my function test still hung. It passed when test it locally. | 2020-12-21T03:53:17 | 2020-12-21T03:53:17 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 748,853,548 | 18,275 | akuzm | This looks like a particular use case that makes sense only for `uniq` functions over a particular `GROUP BY` key for a particular sharding scheme. It's not something that should be in the general interface of aggregate functions.
What's the alternative, `distributed_group_by_no_merge` + second `GROUP BY` on initiat... | 2020-12-21T08:52:20 | 2020-12-21T08:52:20 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 748,855,428 | 18,242 | k0st1an | 1. `attach`/`detach` helped, thanks.
2. I would not say that String fields are large. For example `bid_id` is no more than 16 characters. Maybe the `user_agent` field contains a lot of characters, but it is not all filled. There are many fields of type String themselves in this form: `tracker LowCardinality (String)` ... | 2020-12-21T08:56:11 | 2020-12-21T08:56:11 | {} | NONE |
ClickHouse | ClickHouse | 748,934,250 | 18,095 | spongedu | @alexey-milovidov PTAL~ :) | 2020-12-21T11:50:38 | 2020-12-21T11:50:38 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 748,984,343 | 17,857 | fastio | > @fastio
>
> ```
> 2020.12.15 16:44:02.408249 [ 10671 ] {} <Fatal> BaseDaemon: Address: NULL pointer. Access: read. Address not mapped to object.
> 2020.12.15 16:44:02.408407 [ 10671 ] {} <Fatal> BaseDaemon: Stack trace: 0x1a88e374 0x1a886245 0x1a8898ba 0x1aaeade5 0x1aaeb3f7 0x1aee1e5f 0x1aef09db 0x1aef095b 0x1ae... | 2020-12-21T13:52:38 | 2020-12-21T13:52:38 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 748,506,144 | 18,241 | alexey-milovidov | That's a good idea! | 2020-12-19T18:07:04 | 2020-12-19T18:07:04 | {} | MEMBER |
ClickHouse | ClickHouse | 748,673,207 | 17,847 | alexey-milovidov | Also you can measure only userspace CPU time (`time` will output `user` field). | 2020-12-20T21:28:25 | 2020-12-20T21:28:34 | {} | MEMBER |
ClickHouse | ClickHouse | 748,972,311 | 17,719 | mzealey | i've tried this sort of thing with a few different schema (like `(pk1, pk2, ..., toHashInt(time))`) and really don't see any difference using `sample by` compared to a full range scan, except that `sample by` takes about 3* as long - basically the same number of rows read. I assume this is because `pk1, pk2, ...` is to... | 2020-12-21T13:26:30 | 2020-12-21T13:26:30 | {} | NONE |
ClickHouse | ClickHouse | 749,034,513 | 17,359 | ildus | @Avogar hi, could you look at https://github.com/ClickHouse-Extras/dragonbox/pull/2? I suspect that build errors are related with `dragonbox`. | 2020-12-21T15:32:46 | 2020-12-21T15:32:46 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 749,058,429 | 7,815 | alexey-milovidov | Works fine on master.
Test needed. | 2020-12-21T16:21:39 | 2020-12-21T16:21:39 | {} | MEMBER |
ClickHouse | ClickHouse | 748,588,931 | 18,265 | alexey-milovidov | ```
SELECT
arrayJoin(NER) AS k,
count() AS c
FROM recipes
GROUP BY k
ORDER BY c DESC
LIMIT 50
┌─k────────────────────┬──────c─┐
│ salt │ 890741 │
│ sugar │ 620027 │
│ butter │ 493823 │
│ flour │ 466110 │
│ eggs │ 401276 │... | 2020-12-20T10:24:21 | 2020-12-20T10:24:21 | {} | MEMBER |
ClickHouse | ClickHouse | 748,590,145 | 18,265 | alexey-milovidov | ```
SELECT
title,
length(NER),
length(directions)
FROM recipes
WHERE has(NER, 'strawberry')
ORDER BY length(directions) DESC
LIMIT 10
┌─title────────────────────────────────────────────────────────────┬─length(NER)─┬─length(directions)─┐
│ Chocolate-Strawberry-Orange Wedding Cake ... | 2020-12-20T10:37:08 | 2020-12-20T10:37:08 | {} | MEMBER |
ClickHouse | ClickHouse | 748,651,374 | 18,238 | alexey-milovidov | @azat Let's add a reference to the description of upstream PR, like this:
`This change is needed for [ClickHouse](https://github.com/ClickHouse/ClickHouse/) to enable tests of [AArch64 builds under QEMU](https://github.com/ClickHouse/ClickHouse/issues/15174).` | 2020-12-20T19:37:31 | 2020-12-20T19:37:31 | {} | MEMBER |
ClickHouse | ClickHouse | 748,741,491 | 18,280 | awesomeleo | There is already has one.
https://github.com/ClickHouse/ClickHouse/issues/15182 | 2020-12-21T03:48:27 | 2020-12-21T03:48:27 | {
"+1": 1
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 748,564,145 | 6,286 | alexey-milovidov | @Akazz I still dream to get this feature as a New Year present. | 2020-12-20T05:13:38 | 2020-12-20T05:13:38 | {
"heart": 3,
"hooray": 1
} | MEMBER |
ClickHouse | ClickHouse | 748,588,264 | 18,264 | alexey-milovidov | @azat You're right, it's my wrong thought. | 2020-12-20T10:16:58 | 2020-12-20T10:16:58 | {} | MEMBER |
ClickHouse | ClickHouse | 748,631,158 | 17,538 | olgarev | @tavplubix Thanks for the comments, I updated the file, please, review it. | 2020-12-20T16:43:36 | 2020-12-20T16:43:36 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 748,926,477 | 18,192 | stigsb | @alexey-milovidov works now - I removed the `fail-build: true`, but can put it back if you want the build to fail if issues are found (seems like it fails regardless of severity, could create a lot of noise). | 2020-12-21T11:31:15 | 2020-12-21T11:32:00 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 748,564,437 | 3,775 | alexey-milovidov | I will open another PR when the time will come. | 2020-12-20T05:17:21 | 2020-12-20T05:17:21 | {} | MEMBER |
ClickHouse | ClickHouse | 748,604,292 | 17,350 | olgarev | @alexey-milovidov Sorry, this PR was published as ready by mistake. It was still at work. | 2020-12-20T12:56:02 | 2020-12-20T12:56:02 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 748,660,530 | 17,847 | alexey-milovidov | @fibersel Yandex Metrica should not be used as a single dataset. It contains multiple columns and different columns have very different data distribution. Testing over all columns together will give no insights. | 2020-12-20T19:57:49 | 2020-12-20T19:57:49 | {} | MEMBER |
ClickHouse | ClickHouse | 748,784,311 | 18,275 | alexey-milovidov | I think it's still worth adding: uniq, count, min, max, sum will work. And you will be able to fine tune some queries. | 2020-12-21T06:17:32 | 2020-12-21T06:17:32 | {} | MEMBER |
ClickHouse | ClickHouse | 748,864,539 | 18,275 | amosbird | > To me it looks good enough
Me same. | 2020-12-21T09:14:37 | 2020-12-21T09:14:37 | {} | COLLABORATOR |
ClickHouse | ClickHouse | 748,985,336 | 18,293 | alesapin | Flaky check passed. | 2020-12-21T13:54:38 | 2020-12-21T13:54:38 | {} | MEMBER |
ClickHouse | ClickHouse | 748,441,490 | 17,977 | azat | >Stress test (thread) — Sanitizer assert (in stderr.log)
https://clickhouse-test-reports.s3.yandex.net/17977/2b3fcd6d02533665e510e4201a995c31a7bc9607/stress_test_(thread).html#fail1
gcc's TSAN reports data race in `ColumnVector::insertRangeFrom`, can't reproduce for now
<details>
```
WARNING: ThreadSanitiz... | 2020-12-19T08:30:11 | 2020-12-19T08:30:32 | {} | MEMBER |
ClickHouse | ClickHouse | 748,529,179 | 18,244 | den-crane | They are just informational columns.
It does not related to partition pruning/elimination (pruning works with DateTime64 perfectly).
| 2020-12-19T21:40:33 | 2020-12-19T21:41:02 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 748,564,395 | 10,484 | alexey-milovidov | Please open a new PR. | 2020-12-20T05:16:41 | 2020-12-20T05:16:41 | {} | MEMBER |
ClickHouse | ClickHouse | 748,571,446 | 17,857 | alexey-milovidov | @fastio
```
2020.12.15 16:44:02.408249 [ 10671 ] {} <Fatal> BaseDaemon: Address: NULL pointer. Access: read. Address not mapped to object.
2020.12.15 16:44:02.408407 [ 10671 ] {} <Fatal> BaseDaemon: Stack trace: 0x1a88e374 0x1a886245 0x1a8898ba 0x1aaeade5 0x1aaeb3f7 0x1aee1e5f 0x1aef09db 0x1aef095b 0x1aef08fb 0x1... | 2020-12-20T06:54:49 | 2020-12-20T06:54:49 | {} | MEMBER |
ClickHouse | ClickHouse | 748,637,284 | 18,074 | sevirov | @filimonov Please, review this PR. | 2020-12-20T17:37:20 | 2020-12-20T17:37:20 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 748,644,421 | 17,847 | fibersel | Yandex metrics
| 2020-12-20T18:36:59 | 2020-12-20T18:36:59 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 748,742,385 | 16,007 | IVitamin-C | 20.8.6.6-lts,this version is still not available.
DB::Exception: Method loadKeys is unsupported for RedisDictionarySource.
Thanks
20.10,this version is available.
| 2020-12-21T03:52:23 | 2020-12-21T03:52:23 | {} | NONE |
ClickHouse | ClickHouse | 748,781,535 | 18,275 | amosbird | > We can modify the IAggregationFunction interface a little bit to make it possible to support different ways of merging finalized result on initiator node for different aggregation functions.
It's too specific. I cannot think of any other non-simple aggregate functions except `count*` that can be extended. | 2020-12-21T06:10:48 | 2020-12-21T06:10:48 | {} | COLLABORATOR |
ClickHouse | ClickHouse | 748,792,679 | 18,275 | amosbird | `min, max, sum` are simple aggregates and it makes no sense to do `distributed_group_by_no_merge`.
> I think it's still worth adding
Ok... | 2020-12-21T06:37:59 | 2020-12-21T06:37:59 | {} | COLLABORATOR |
ClickHouse | ClickHouse | 748,824,017 | 18,238 | azat | >@azat Let's add a reference to the description of upstream PR, like this:
Sure, done (although this does not show the whole problem - any program with jemalloc under qemu is broken now) | 2020-12-21T07:50:02 | 2020-12-21T07:50:02 | {} | MEMBER |
ClickHouse | ClickHouse | 749,056,707 | 17,359 | Avogar | > @Avogar hi, could you look at [ClickHouse-Extras/dragonbox#2](https://github.com/ClickHouse-Extras/dragonbox/pull/2)? I suspect that build errors are related with `dragonbox`.
Hi, I don't have rights to merge PR in ClickHouse-Extras, you can ask @alexey-milovidov. I had a look at build errors, it seems that splitt... | 2020-12-21T16:17:44 | 2020-12-21T16:17:44 | {} | MEMBER |
ClickHouse | ClickHouse | 748,460,386 | 17,639 | ucasfl | Looks ok now. | 2020-12-19T11:21:29 | 2020-12-19T11:21:29 | {} | COLLABORATOR |
ClickHouse | ClickHouse | 748,560,428 | 18,238 | alexey-milovidov | > Functional stateless tests flaky check (address) — fail: 0, passed: 0, skipped: 100
It's marked as failed due to "passed: 0", actually it's Ok. | 2020-12-20T04:12:05 | 2020-12-20T04:12:05 | {} | MEMBER |
ClickHouse | ClickHouse | 748,564,009 | 10,376 | alexey-milovidov | Migrated here: #17495 | 2020-12-20T05:11:57 | 2020-12-20T05:11:57 | {
"+1": 1,
"heart": 1
} | MEMBER |
ClickHouse | ClickHouse | 748,903,640 | 18,305 | akuzm | Duplicate https://github.com/ClickHouse/ClickHouse/issues/18300 | 2020-12-21T10:39:02 | 2020-12-21T10:39:02 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 749,040,368 | 17,463 | nikitamikhaylov | @alexey-milovidov This PR broke Arcadia :( How can I create a unique_ptr on `AsynchronousMetrics`? Can I pass empty vector as `servers_to_start_before_tables_` and `servers_`? How do I get the old behavior in the new interface now? | 2020-12-21T15:44:44 | 2020-12-21T15:44:44 | {} | MEMBER |
ClickHouse | ClickHouse | 748,565,633 | 16,243 | alexey-milovidov | @FawnD2 Should we continue working on this PR or better to close it? | 2020-12-20T05:27:14 | 2020-12-20T05:27:14 | {} | MEMBER |
ClickHouse | ClickHouse | 748,588,504 | 18,265 | alexey-milovidov | ```
CREATE TABLE recipes
(
title String,
ingredients Array(String),
directions Array(String),
link String,
source LowCardinality(String),
NER Array(String)
) ENGINE = MergeTree ORDER BY title;
```
`clickhouse-client --query "INSERT INTO recipes SELECT title, JSONExtract(ingredients,... | 2020-12-20T10:19:52 | 2020-12-20T10:22:46 | {} | MEMBER |
ClickHouse | ClickHouse | 748,642,011 | 17,639 | Akazz | All of the [failed performance tests](https://clickhouse-test-reports.s3.yandex.net/17639/ccbd03122273dbd9e2e2501c33290aac72e5e59e/performance_comparison/report.html#fail1) are due to "Unexpected Query Duration" and that's irrelevant to this PR | 2020-12-20T18:17:02 | 2020-12-20T18:17:29 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 748,667,932 | 17,847 | alexey-milovidov | The default build type for ClickHouse is release with debug info (that's what we need).
But if you are using CLion, it has a quirk that sets the build mode to debug by default (that is a common source of confusion). | 2020-12-20T20:39:35 | 2020-12-20T20:39:35 | {} | MEMBER |
ClickHouse | ClickHouse | 748,673,126 | 17,847 | alexey-milovidov | 600 MB should be enough, it will take hundreds of milliseconds. You can repeat multiple times to get more statistics... | 2020-12-20T21:27:30 | 2020-12-20T21:27:30 | {} | MEMBER |
ClickHouse | ClickHouse | 748,673,177 | 17,847 | alexey-milovidov | Actually running multiple times is required for data to fit in page cache (otherwise it can be IO bound). | 2020-12-20T21:28:05 | 2020-12-20T21:28:05 | {} | MEMBER |
ClickHouse | ClickHouse | 748,707,261 | 17,838 | alexey-milovidov | @weeds085490 The code looks Ok but the tests deadlocked. I don't see the reason.
Could you please merge/rebase with master? | 2020-12-21T01:19:05 | 2020-12-21T01:19:05 | {} | MEMBER |
ClickHouse | ClickHouse | 748,473,856 | 18,221 | vzakaznikov | > Could you please resolve conflicts?
Done, but still need to check tests and remove CFB1 and CFB8 modes that are no longer supported. | 2020-12-19T13:18:16 | 2020-12-19T13:18:16 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 748,565,337 | 11,777 | alexey-milovidov | Implemented in another PR. | 2020-12-20T05:24:57 | 2020-12-20T05:24:57 | {} | MEMBER |
ClickHouse | ClickHouse | 748,636,874 | 17,847 | fibersel | Here are first results of testing:
silesia dataset.
| 2020-12-20T17:34:05 | 2020-12-20T17:34:05 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 748,810,273 | 18,284 | alexey-milovidov | > Docs check — Docs check passed
Will see if it helps. | 2020-12-21T07:19:30 | 2020-12-21T07:19:30 | {} | MEMBER |
ClickHouse | ClickHouse | 748,969,668 | 18,312 | alesapin | Need build for coverage with tidy. After that tests should be stopped. | 2020-12-21T13:20:09 | 2020-12-21T13:20:09 | {} | MEMBER |
ClickHouse | ClickHouse | 748,455,139 | 18,100 | allen12921 | You save my days,It works like a charm
Will we fix it on the latest version in further? | 2020-12-19T10:33:55 | 2020-12-19T10:33:55 | {} | NONE |
ClickHouse | ClickHouse | 748,502,084 | 18,241 | mzealey | Yeah I figured we could duplicate. I don't know the code, but one way to work around easily may be allow the `<key>` attributes to also be specified as `<attribute>` so effectively the same as the duplicating workaround but without having to do it in the base data? | 2020-12-19T17:31:55 | 2020-12-19T17:31:55 | {} | NONE |
ClickHouse | ClickHouse | 748,530,500 | 18,221 | vzakaznikov | @alexey-milovidov, for GCM mode the IV is expected to be 12 bytes based on the error message but there is no error when it less than 12. Is this the expected behavior?
```
user-host :) SELECT encrypt('aes-192-gcm', 'hello', '123456789012345612345678')
SELECT encrypt('aes-192-gcm', 'hello', '1234567890123456123456... | 2020-12-19T21:55:36 | 2020-12-19T21:55:36 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 748,532,887 | 18,221 | vzakaznikov | @alexey-milovidov, I've updated requirements and tests. When TestFlows check is green this PR is ready for merge. | 2020-12-19T22:20:14 | 2020-12-19T22:20:14 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 748,562,432 | 16,405 | alexey-milovidov | Merged to master in #17829. | 2020-12-20T04:47:27 | 2020-12-20T04:47:27 | {} | MEMBER |
ClickHouse | ClickHouse | 748,566,577 | 18,249 | alexey-milovidov | @vzakaznikov Only constant string is supported for `mode` argument. | 2020-12-20T05:37:07 | 2020-12-20T05:37:07 | {} | MEMBER |
ClickHouse | ClickHouse | 748,809,661 | 18,278 | alexey-milovidov | > Functional stateless tests flaky check (address) — fail: 1, passed: 91
@Enmk #17846 | 2020-12-21T07:18:00 | 2020-12-21T07:18:00 | {} | MEMBER |
ClickHouse | ClickHouse | 748,931,647 | 18,102 | AnaUvarova | @alexey-milovidov
Hi, Alexey. Okay, what should the user do to disable it? And in what cases users need to do this?
We can rewrite the text so:
- the setting is enabled by default;
- the user can enable it manually (in a case of something) by described manipulations;
- how to disable and when it is needed to ... | 2020-12-21T11:43:48 | 2020-12-21T11:44:37 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 748,563,430 | 9,996 | alexey-milovidov | @zhang2014 I agree with @geldot.
The current implementation duplicates half of clickhouse-client logic and half of HTTPHandler logic.
It is quite well engineered, but we can do it a more simple way...
What if we send and receive terminal commands between the server and client directly?
Example: as we already impl... | 2020-12-20T05:05:00 | 2020-12-20T05:11:13 | {} | MEMBER |
ClickHouse | ClickHouse | 748,566,318 | 15,409 | alexey-milovidov | @liangli1213 It should allow to replicate one or a few selected tables from MySQL,
in contrast to MaterializeMySQL feature that replicates whole databases. | 2020-12-20T05:33:50 | 2020-12-20T05:33:50 | {} | MEMBER |
ClickHouse | ClickHouse | 748,570,664 | 16,559 | alexey-milovidov | Outdated. | 2020-12-20T06:42:05 | 2020-12-20T06:42:05 | {} | MEMBER |
ClickHouse | ClickHouse | 748,645,002 | 18,238 | azat | >@azat I think it is ready to merge.
@alexey-milovidov this should be a correct fix, however I want to wait for the feedback from maintainers, so if i.e. there will some wish to implement it in a different way or some cosmetic changes would rebase the backported patch (to avoid further conflicts and similar issues)
... | 2020-12-20T18:42:07 | 2020-12-20T18:42:07 | {} | MEMBER |
ClickHouse | ClickHouse | 748,903,387 | 18,300 | akuzm | Fixed here https://github.com/ClickHouse/ClickHouse/pull/18167 | 2020-12-21T10:38:31 | 2020-12-21T10:38:31 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 748,951,906 | 18,275 | fuwhu | @alexey-milovidov @amosbird @akuzm
Hi All,
Thanks a lot for your review on this idea.
It is not only for countDistinct. In our case, at least '**groupBitmapOr**' can also benefit from this change.
And we are still trying to find out other agg functions that can be improved by this change in our use cases.
| 2020-12-21T12:35:09 | 2020-12-21T13:14:45 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 748,564,711 | 10,872 | alexey-milovidov | Closing due to lack of demand.
16bit floating point is good for analysis of coefficients of machine learning models.
We can continue with some minimal implementation, e.g. bfloat16 can be enough. | 2020-12-20T05:19:47 | 2020-12-20T05:20:02 | {} | MEMBER |
ClickHouse | ClickHouse | 748,571,199 | 17,350 | alexey-milovidov | @dfenelonov This task was not done. | 2020-12-20T06:50:47 | 2020-12-20T06:50:47 | {} | MEMBER |
ClickHouse | ClickHouse | 748,597,451 | 18,238 | alexey-milovidov | @azat I think it is ready to merge. | 2020-12-20T11:50:28 | 2020-12-20T11:50:28 | {} | MEMBER |
ClickHouse | ClickHouse | 748,635,275 | 17,538 | tavplubix | Docs check - passed | 2020-12-20T17:20:31 | 2020-12-20T17:20:31 | {} | MEMBER |
ClickHouse | ClickHouse | 748,638,720 | 18,131 | paul-nameless | Sorry for your time spent, there was my mistake. | 2020-12-20T17:50:12 | 2020-12-20T17:50:52 | {} | NONE |
ClickHouse | ClickHouse | 748,662,821 | 17,847 | alexey-milovidov | The results in table are unclear and most likely contain errors:
1. The "level" of lz4 is not specified.
2. If it is the default level (1), decompression speed looks unrealistic and most likely indicates that you have build it in debug mode (e.g. if you are using CLion, it enables debug mode by default). Typical deco... | 2020-12-20T19:59:37 | 2020-12-20T20:02:51 | {} | MEMBER |
ClickHouse | ClickHouse | 748,668,792 | 17,847 | fibersel | Hm, no, I build it from command line. | 2020-12-20T20:47:26 | 2020-12-20T20:47:26 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 748,704,900 | 17,043 | alexey-milovidov | Translation units became too heavy, they cannot build under the limits:
```
prlimit --rss=10000000 --cpu=600
```
Let's split them to multiple translation units. | 2020-12-21T01:05:26 | 2020-12-21T01:05:26 | {} | MEMBER |
ClickHouse | ClickHouse | 748,828,435 | 18,242 | alesapin | Your merge did not stop because it's already assigned in replication queue. You can just detach/attach partition `202005` on single replica to avoid this or try to delete this part/partition with `ALTER TABLE ... DROP PARTITION|PART name`.
But high memory consumptions is very strange. Does your table contain huge St... | 2020-12-21T07:59:34 | 2020-12-21T07:59:34 | {} | MEMBER |
ClickHouse | ClickHouse | 748,496,205 | 18,241 | alexey-milovidov | Workaround: duplicate network in another attribute.
I'm not sure if it's possible to implement this feature for free, because networks may be split in parts in internal data structure. | 2020-12-19T16:39:17 | 2020-12-19T16:39:17 | {} | MEMBER |
ClickHouse | ClickHouse | 748,560,014 | 18,238 | alexey-milovidov | Performance: 7340839d6d | 2020-12-20T04:04:55 | 2020-12-20T04:04:55 | {} | MEMBER |
ClickHouse | ClickHouse | 748,560,440 | 18,238 | alexey-milovidov | > Yandex ... check
Don't care. | 2020-12-20T04:12:27 | 2020-12-20T04:12:27 | {} | MEMBER |
ClickHouse | ClickHouse | 748,566,166 | 15,409 | alexey-milovidov | This PR was abandoned. | 2020-12-20T05:32:26 | 2020-12-20T05:32:26 | {} | MEMBER |
ClickHouse | ClickHouse | 748,581,023 | 18,264 | alexey-milovidov | @ucasFL Also need to test that it already works if cluster already has its own sharding key. | 2020-12-20T08:57:59 | 2020-12-20T08:57:59 | {} | MEMBER |
ClickHouse | ClickHouse | 748,637,448 | 18,245 | sevirov | @alexey-milovidov Please, review this PR. | 2020-12-20T17:38:50 | 2020-12-20T17:38:50 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 748,808,117 | 18,278 | filimonov | I guess it resolves https://github.com/ClickHouse/ClickHouse/issues/14489 | 2020-12-21T07:14:23 | 2020-12-21T07:14:23 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 748,946,365 | 18,299 | alesapin | ```
2020.12.21 15:15:46.031783 [ 141 ] {} <Debug> ThreadFuzzer: Sleep probability 0.1
2020.12.21 15:15:46.031812 [ 141 ] {} <Debug> ThreadFuzzer: Cpu time period 1000
2020.12.21 15:15:46.031833 [ 141 ] {} <Debug> ThreadFuzzer: Sleep time us 100000.0
```
Works | 2020-12-21T12:21:09 | 2020-12-21T12:21:09 | {} | MEMBER |
ClickHouse | ClickHouse | 749,062,134 | 6,913 | alexey-milovidov | Requires to rewrite `primary.idx` during mutation. | 2020-12-21T16:29:19 | 2020-12-21T16:29:19 | {} | MEMBER |
ClickHouse | ClickHouse | 755,819,814 | 18,808 | alexey-milovidov | It is strange that build-id is present in all binaries built in CI. | 2021-01-07T01:11:43 | 2021-01-07T01:11:43 | {} | MEMBER |
ClickHouse | ClickHouse | 756,119,667 | 18,814 | den-crane | I don't see any logic, why `[[1], [3]]` 1 and 3 are arrays ?
Is it possible to be [[1,2,3,4,4,5,], [3,33,3,3,3,3,3,,,3,3,3,,,3,3,3]] ? | 2021-01-07T13:34:41 | 2021-01-07T13:34:41 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 756,203,008 | 18,827 | ucasfl | > > I don't see other errors during build.
>
> I think that the problem you see only error during linking is because clickhouse is shipped with some version of glibc headers, that provides `sys_siglist`, but there is no such symbol in your libc, maybe it is too old or some patched version?
>
> ```shell
> nm -D /... | 2021-01-07T15:54:32 | 2021-01-07T15:54:32 | {} | COLLABORATOR |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.