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 | 789,245,096 | 21,384 | alexey-milovidov | In the first query numbers are parsed as Float64 before being casted to Int128.
The right way is to put them in quotes as in the second query.
We will check what's wrong with the second query.
PS. Big integers is experimental feature (in development). | 2021-03-02T21:50:27 | 2021-03-02T21:50:37 | {} | MEMBER |
ClickHouse | ClickHouse | 788,150,695 | 19,885 | vdimir | > Should we merge?
I found one issue, query doesn't work if syntax `JOIN ON` used with some constant in `WHERE`, e.g. `SELECT * FROM t1 FULL JOIN t2 ON t1.a = t2.a WHERE 1`. It shows error `DB::Exception: Not found column t2.a in block. There are only columns: CAST(t2.a, Int32)`. If type conversion for `t1.a` and `t... | 2021-03-01T18:01:02 | 2021-03-01T18:01:02 | {} | MEMBER |
ClickHouse | ClickHouse | 788,216,675 | 19,836 | sevirov | @akuzm Please, review this PR. | 2021-03-01T19:39:11 | 2021-03-01T19:39:11 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 788,816,418 | 21,339 | filimonov | admin / superadmin - does it came from some real life example?
Normally if you root / admin, it's only God have more privileges :D So when I give the other user right to be root, then he is 'as good as me' and of course he CAN revoke/grant everything, remove some of the permissions of himself, modify the permission... | 2021-03-02T10:50:54 | 2021-03-02T10:54:04 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 789,052,724 | 21,349 | stigsb | This was built on x86_64 Ubuntu 20.04 with a fresh master checkout. I'll try with a fresh clone and nuked ccache.
| 2021-03-02T16:52:52 | 2021-03-02T16:52:52 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 789,249,267 | 21,384 | smagellan | Thank you! | 2021-03-02T21:58:13 | 2021-03-02T21:58:13 | {} | NONE |
ClickHouse | ClickHouse | 788,190,439 | 21,356 | alexey-milovidov | I'm trying to fix Website workers right now. | 2021-03-01T19:00:16 | 2021-03-01T19:00:16 | {} | MEMBER |
ClickHouse | ClickHouse | 788,584,901 | 21,358 | amosbird | Maybe we can provide a new equal operator `==` that invalidates index analysis. | 2021-03-02T04:56:09 | 2021-03-02T04:56:09 | {} | COLLABORATOR |
ClickHouse | ClickHouse | 788,690,574 | 21,358 | amosbird | Or maybe we can provide a `noIndexHint()` function that will do the opposite of `indexHint()` | 2021-03-02T07:38:25 | 2021-03-02T07:38:44 | {} | COLLABORATOR |
ClickHouse | ClickHouse | 788,705,063 | 21,239 | Teeeto | Maintaining an external dictionary is extra code one needs to have at the ETL process populating the DB. Additionally as I understand a select query would have to utilize the dictionary explicitly or a view would be required (not sure about efficiency).
Sharing a LowCardinality dictionary is much more elegant and sea... | 2021-03-02T08:05:09 | 2021-03-02T08:05:09 | {} | NONE |
ClickHouse | ClickHouse | 788,800,283 | 14,740 | javisantana | This bug is fixed in 21.3.1
```
SELECT *
FROM
(
WITH
(
SELECT groupArray(a)
FROM
(
SELECT 1 AS a
)
) AS keys,
(
SELECT groupArray(a)
FROM
(
... | 2021-03-02T10:25:22 | 2021-03-02T10:25:22 | {
"hooray": 1
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 788,931,446 | 14,995 | vitlibar | > Is **Allow squash merging** enabled for the repo?
> If yes, you can choose to squash during the merge, actually.
> If not, I will squash manually.
I can allow squashing when I merge it. But it's always difficult not to forget about it when it's time to merge. I prefer to squash my own PR because in my experience... | 2021-03-02T14:03:45 | 2021-03-02T14:03:45 | {} | MEMBER |
ClickHouse | ClickHouse | 789,080,525 | 21,361 | filimonov | > We have ALTER TABLE database.<temp_table_name> MOVE PARTITION <partitionN> TO TABLE database.table. There are 100 partitions and each was flushed once a minute on average, hence +/-100 move partition requests a minute.
So you rewrite 100 partitions fully every minute (sounds bad, maybe at least remove the partitio... | 2021-03-02T17:32:35 | 2021-03-02T17:33:46 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 788,191,170 | 21,356 | alexey-milovidov | People complained that the link on https://www.youtube.com/watch?v=ltg8vstuHUU does not work. | 2021-03-01T19:01:18 | 2021-03-01T19:01:18 | {} | MEMBER |
ClickHouse | ClickHouse | 788,985,941 | 21,307 | akuzm | > @kitaisreal There is already a test https://github.com/ClickHouse/ClickHouse/blob/master/tests/performance/modulo.xml for this. Maybe the difference is not evident.
There was an obvious difference but somehow the test was not marked as failed. I'll investigate. | 2021-03-02T15:20:16 | 2021-03-02T15:20:16 | {
"+1": 1
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 788,251,359 | 7,195 | Alesh | > Workaround:
> `.inner.a2`
This workaround can no longer be used. The internal table for the materialized view generates a random name like this '.inner_id.2c1d2be9-d408-4845-9449-0b5ae5383ba4'. Are there any other options?
| 2021-03-01T20:29:17 | 2021-03-01T20:29:17 | {} | NONE |
ClickHouse | ClickHouse | 788,753,453 | 21,366 | nikitamikhaylov | There is no bug here. | 2021-03-02T09:15:24 | 2021-03-02T09:15:24 | {} | MEMBER |
ClickHouse | ClickHouse | 788,761,233 | 21,368 | nikitamikhaylov | @ypg521 FYI Copier uses Distributed tables to copy from source to destination cluster, that's why there are some limitations | 2021-03-02T09:26:24 | 2021-03-02T09:26:24 | {} | MEMBER |
ClickHouse | ClickHouse | 788,825,085 | 21,339 | filimonov | And that sounds super nonintuitive: I have permission to grant/revoke. I successfully revoked some permission, but effective permission was not changed 'because other SUPERadmin grant is more important' than my revoke.
So I would need to contact that SUPERadmin (he's on vacation for the next 3 months :) ) asking hi... | 2021-03-02T11:04:44 | 2021-03-02T11:04:44 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 789,164,653 | 21,369 | den-crane | 21.3.1.6054
```
SELECT field2
FROM agg_view
WHERE field1 = 'test'
Query id: c699659e-445a-450b-8d42-c5c6f6422f00
┌─field2─┐
│ 33 │
└────────┘
┌─field2─┐
│ 33 │
└────────┘
``` | 2021-03-02T19:42:23 | 2021-03-02T19:42:23 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 788,113,885 | 21,349 | alexey-milovidov | Maybe you are using modified source code? Or do you run tests on non x86_64 Linux? | 2021-03-01T17:08:03 | 2021-03-01T17:08:03 | {} | MEMBER |
ClickHouse | ClickHouse | 788,756,147 | 21,368 | pg-yang | I found a question that Similar to this question
https://github.com/ClickHouse/ClickHouse/issues/19384 | 2021-03-02T09:19:12 | 2021-03-02T09:19:12 | {} | NONE |
ClickHouse | ClickHouse | 788,758,548 | 21,368 | pg-yang | I delete index In table ;
It work OK
| 2021-03-02T09:22:43 | 2021-03-02T09:24:59 | {} | NONE |
ClickHouse | ClickHouse | 788,869,086 | 14,995 | zvonand | > @zvonand could you please take a look why all builds are failing? Can it be due to docker image related changes?
I guess no, the first fail occured after merge from master, not my commit
I modified nothing outside testflows | 2021-03-02T12:20:09 | 2021-03-02T12:20:09 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 788,870,286 | 19,885 | akuzm | > It shows error `DB::Exception: Not found column t2.a in block. There are only columns: CAST(t2.a, Int32)`.
Looks kinda bad, probably better to investigate. Do I remember correctly that you keep the name of the column, but replace it with the casted one? So it should have been named just `t2.a` and not `cast...` if... | 2021-03-02T12:22:27 | 2021-03-02T12:22:27 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 788,969,520 | 14,995 | traceon | @zvonand **Style Check** still complains. | 2021-03-02T14:59:30 | 2021-03-02T14:59:30 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 789,043,588 | 14,995 | traceon | @zvonand still failing. | 2021-03-02T16:40:04 | 2021-03-02T16:40:04 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 788,065,081 | 20,213 | MaxJPowers | @nikitamikhaylov Even after changing <query_wait_timeout_milliseconds>120000</query_wait_timeout_milliseconds> it still gives the 60000ms error. Is there really no way of increasing this??? | 2021-03-01T16:02:52 | 2021-03-01T16:02:52 | {} | NONE |
ClickHouse | ClickHouse | 788,157,482 | 21,356 | alexey-milovidov | https://support.cloudflare.com/hc/en-us/articles/200172286-Configuring-URL-forwarding-or-redirects-with-Cloudflare-Page-Rules
Let's try:
| 2021-03-01T18:10:26 | 2021-03-01T18:10:26 | {} | MEMBER |
ClickHouse | ClickHouse | 788,190,254 | 21,356 | alexey-milovidov | @blinkov What you say is simply untrue. | 2021-03-01T18:59:54 | 2021-03-01T18:59:54 | {} | MEMBER |
ClickHouse | ClickHouse | 788,292,263 | 6,093 | devicenull | Any progress here? Even making this return instantly would be better then the current behavior.
I'm trying to use optimize from a script (it massively reduces disk space usage on our setup), and I'm having to catch and ignore timeout errors here just to actually execute optimize. | 2021-03-01T21:13:37 | 2021-03-01T21:13:37 | {} | NONE |
ClickHouse | ClickHouse | 788,901,216 | 14,995 | traceon | @zvonand **Style Check** indicates some stylistic problems in testflows code. Could you please take a look? | 2021-03-02T13:16:40 | 2021-03-02T13:16:40 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 788,904,271 | 21,339 | vitlibar | > This functionality does seem strange. Maybe to address original issue it is better to add something like `WITH REVOKE [OPTION] ALL ...` similar how we to have grant `WITH GRANT OPTION`?
It just seemed the `GRANTED BY` way is much more standard. | 2021-03-02T13:21:35 | 2021-03-02T13:23:29 | {} | MEMBER |
ClickHouse | ClickHouse | 789,183,791 | 21,301 | alesapin | Fake race IMergeTreeDataPart::setState. Need to cherry-pick manually. | 2021-03-02T20:14:42 | 2021-03-02T20:14:42 | {} | MEMBER |
ClickHouse | ClickHouse | 789,237,331 | 18,583 | alexey-milovidov | Thanks.
Btw, is it true that loop unrolling (I mean main loops with SSE/AVX copy) does not make sense? | 2021-03-02T21:37:18 | 2021-03-02T21:37:18 | {} | MEMBER |
ClickHouse | ClickHouse | 789,238,757 | 21,035 | Avogar | Failed test is not related. | 2021-03-02T21:39:35 | 2021-03-02T21:39:35 | {
"+1": 1
} | MEMBER |
ClickHouse | ClickHouse | 789,260,902 | 21,327 | alexey-milovidov | > test_replace_partition/test.py::test_replace_after_replace_failover
There are 5 failures in the last 100 days - flaky test. | 2021-03-02T22:18:47 | 2021-03-02T22:18:47 | {} | MEMBER |
ClickHouse | ClickHouse | 788,177,829 | 21,356 | blinkov | As far as I remember they have always been on a separate subdomain like https://presentations.clickhouse.tech/yatalks_2019/ and it's still working | 2021-03-01T18:40:26 | 2021-03-01T18:40:26 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 788,644,661 | 14,995 | vitlibar | Why have you updated `brotli`? What about `contrib/brotli-cmake`, shouldn't it be updated too? | 2021-03-02T06:14:08 | 2021-03-02T06:14:26 | {} | MEMBER |
ClickHouse | ClickHouse | 788,659,371 | 21,361 | cyberhuman | We have `ALTER TABLE database.<temp_table_name> MOVE PARTITION <partitionN> TO TABLE database.table`. There are 100 partitions and each was flushed once a minute on average, hence +/-100 move partition requests a minute. | 2021-03-02T06:44:09 | 2021-03-02T06:44:09 | {} | NONE |
ClickHouse | ClickHouse | 788,742,693 | 21,079 | tavplubix | Integration tests (thread) - `test_storage_kafka/test.py::test_kafka_virtual_columns2` - flaky | 2021-03-02T08:59:24 | 2021-03-02T08:59:24 | {} | MEMBER |
ClickHouse | ClickHouse | 788,792,388 | 21,362 | nikitamikhaylov | No :( Try to run another test, for example 'test_logs_level' (i don't exactly remember its name), because it is very lightweight, it starts only on ClickHouse instance. | 2021-03-02T10:13:06 | 2021-03-02T10:13:06 | {} | MEMBER |
ClickHouse | ClickHouse | 789,120,087 | 18,583 | alexey-milovidov | Thank you!
Do I understand correctly that the header file with macro is needed to use the fact that less number of registers are clobbered in memcpy than it's usually expected?
Sadly this macro is incompatible with libc++ code:
```
/home/milovidov/work/ClickHouse/contrib/libcxx/include/utility:973:12: error: ex... | 2021-03-02T18:31:42 | 2021-03-02T18:33:12 | {} | MEMBER |
ClickHouse | ClickHouse | 788,082,109 | 21,097 | PhantomPhreak | > Hi! Unfortunately, version 20.11.3 doesn't have the fix for this problem #18844. You should update to the latest stable version, or at least the latest 20.11.
Thanks! I'm considering the stable version now, could you please a suggestion about the good option for now?
> Your <column>.bin file became longer than ... | 2021-03-01T16:25:32 | 2021-03-01T16:25:32 | {} | NONE |
ClickHouse | ClickHouse | 788,113,029 | 21,349 | alexey-milovidov | We ensure that it does not fail in our CI:
| 2021-03-01T17:06:55 | 2021-03-01T17:06:55 | {} | MEMBER |
ClickHouse | ClickHouse | 788,636,202 | 21,340 | filimonov | > Complex operations on each record with merging, requiring x2-x3 partition sizes on the disk?
It is wrong. Clickhouse always (w/o ttl too) reserve double disk space of src part set size during the merge.
Running moves/mutations also do disk reservations.
So for ex. if merge of 5 parts which overall size is 10... | 2021-03-02T05:57:39 | 2021-03-02T05:57:39 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 788,884,290 | 21,355 | xm-repo | Перенес в odbc, кажется, там проблема https://github.com/ClickHouse/clickhouse-odbc/issues/348 | 2021-03-02T12:48:33 | 2021-03-02T12:48:33 | {} | NONE |
ClickHouse | ClickHouse | 789,090,092 | 18,978 | tavplubix | Possibly we should change Changelog category to "Backward incompatible" and add a note that `ATTACH PART[ITION]` queries may not work during cluster upgrade. | 2021-03-02T17:47:38 | 2021-03-02T17:47:38 | {} | MEMBER |
ClickHouse | ClickHouse | 788,160,421 | 21,356 | alexey-milovidov | We already have this rule but it wasn't working. | 2021-03-01T18:14:43 | 2021-03-01T18:14:43 | {} | MEMBER |
ClickHouse | ClickHouse | 788,671,407 | 21,361 | amosbird | PS. you can do `SELECT * EXCEPT (query_id, database, table) ...` | 2021-03-02T07:05:40 | 2021-03-02T07:05:40 | {
"+1": 2
} | COLLABORATOR |
ClickHouse | ClickHouse | 788,672,731 | 21,097 | alesapin | I recommend one of the following version policies:
1. Choose the latest stable version to use and always update to the latest patch (third number). Update to the next stable if new features required or support of your stable version ended.
2. Use the LTS version and always update to the latest patch version. Update t... | 2021-03-02T07:07:55 | 2021-03-02T07:08:16 | {
"+1": 1
} | MEMBER |
ClickHouse | ClickHouse | 788,698,295 | 14,995 | vitlibar | Could you fix the build? It looks broken. | 2021-03-02T07:52:56 | 2021-03-02T07:52:56 | {} | MEMBER |
ClickHouse | ClickHouse | 788,764,952 | 21,362 | nikitamikhaylov | You can run integration test in docker with `runner` script in `tests/integration` directory. | 2021-03-02T09:31:41 | 2021-03-02T09:31:41 | {} | MEMBER |
ClickHouse | ClickHouse | 788,799,184 | 21,362 | hssrs | > No :( Try to run another test, for example 'test_logs_level' (i don't exactly remember its name), because it is very lightweight, it starts only on ClickHouse instance.
Yes, it contains only one item, but same error occurs againe. I have no idea how to check the environment is deployed successful or not. | 2021-03-02T10:23:30 | 2021-03-02T10:23:30 | {} | NONE |
ClickHouse | ClickHouse | 788,872,192 | 14,995 | traceon | > Why have you updated brotli?
Probably a merge artifact, I will fix it - it should not be any different from what is in the `master`. | 2021-03-02T12:26:09 | 2021-03-02T12:26:09 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 788,902,988 | 21,339 | vitlibar | > Very confusing.
Yeah, maybe. | 2021-03-02T13:19:32 | 2021-03-02T13:19:32 | {} | MEMBER |
ClickHouse | ClickHouse | 789,158,442 | 18,583 | alexey-milovidov | BTW, is there significant benefit of using AVX for memcpy?
We still use SSE4.2 in default build and we can use AVX and further only via dynamic dispatch.
(so, the initialization function can be still needed) | 2021-03-02T19:32:04 | 2021-03-02T19:32:04 | {} | MEMBER |
ClickHouse | ClickHouse | 789,160,200 | 21,385 | alexey-milovidov | But this test is unfair because the data is not used after copying (and non-temporary stores are always better). | 2021-03-02T19:35:09 | 2021-03-02T19:35:34 | {} | MEMBER |
ClickHouse | ClickHouse | 789,159,788 | 21,385 | alexey-milovidov | At least in some cases it's not faster than the current implementation.
On AMD Ryzen 3950:
Old:
```
milovidov-desktop :) SELECT count() FROM system.numbers WHERE NOT ignore(repeat('Hello, world' || toString(number), 1000))
SELECT count()
FROM system.numbers
WHERE NOT ignore(repeat(concat('Hello, world', to... | 2021-03-02T19:34:25 | 2021-03-02T19:47:16 | {} | MEMBER |
ClickHouse | ClickHouse | 788,590,596 | 21,340 | Oinari | I guess I misunderstand the logic behind the TTL tables.
Expectation:
Just move the partition to another disk after a specified time in TTL, without cloning to moving dir.
Reality:
Complex operations on each record with merging, requiring x2-x3 partition sizes on the disk?
Disabling TTL TO DISK and setting 'mo... | 2021-03-02T05:08:18 | 2021-03-02T05:08:57 | {} | NONE |
ClickHouse | ClickHouse | 788,699,041 | 21,097 | alesapin | If you have any other questions just reopen this issue. | 2021-03-02T07:54:24 | 2021-03-02T07:54:24 | {} | MEMBER |
ClickHouse | ClickHouse | 788,734,221 | 17,623 | melin | > Or:
>
> ```
> {
> "test": "Turn this to JSON",
> "test@attrib": "moretest"
> }
> ```
https://github.com/Cheedoong/xml2json/blob/master/README.md
<table>
<tbody><tr>
<td bgcolor="#CCCCCC"><strong>Pattern</strong>
</td>
<td bgcolor="#f0f0f0"><strong>XML</strong>
</td>
<td bgcolor="#CCCCCC"><... | 2021-03-02T08:46:04 | 2021-03-02T08:46:24 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 788,786,937 | 21,362 | hssrs | > You can run integration test in docker with `runner` script in `tests/integration` directory.
@nikitamikhaylov Hi, I tried using the Docker earlier based on the instructions in [README.md](https://github.com/ClickHouse/ClickHouse/tree/master/tests/integration#running-with-runner-script) , but the tests themselves... | 2021-03-02T10:05:11 | 2021-03-02T10:05:11 | {} | NONE |
ClickHouse | ClickHouse | 788,804,116 | 11,003 | sag1 | Any update on this ? When will it allow arbitary uid ? | 2021-03-02T10:30:53 | 2021-03-02T10:30:53 | {} | NONE |
ClickHouse | ClickHouse | 788,893,126 | 14,995 | vitlibar | 180 commits are a lot. If you can squash them it would be nice. | 2021-03-02T13:03:38 | 2021-03-02T13:03:38 | {} | MEMBER |
ClickHouse | ClickHouse | 788,083,389 | 19,885 | akuzm | Integration test failure doesn't look relevant.
MSan fuzzer looks like this one https://github.com/ClickHouse/ClickHouse/issues/20046
Should we merge? | 2021-03-01T16:27:16 | 2021-03-01T16:27:16 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 788,197,123 | 21,356 | alexey-milovidov | I've fixed it. | 2021-03-01T19:10:11 | 2021-03-01T19:10:11 | {
"+1": 1
} | MEMBER |
ClickHouse | ClickHouse | 788,319,877 | 21,358 | alexey-milovidov | A separate setting like `use_data_skipping_indices`, default true, will be better solution. | 2021-03-01T21:45:52 | 2021-03-01T21:45:52 | {} | MEMBER |
ClickHouse | ClickHouse | 788,405,548 | 21,350 | den-crane | By Excerpts I mean fragments not exceptions. Exceptions are useless here. Need to see memorytracker messages and a source (previous lines) of these messages. | 2021-03-01T23:45:05 | 2021-03-01T23:45:05 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 788,641,568 | 21,361 | filimonov | Can you explain where those REPLACE_RANGE come from? Do you use smth like replace partition commands? How often? | 2021-03-02T06:08:01 | 2021-03-02T06:08:01 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 788,653,263 | 21,355 | filimonov | You closed it bacause ...? | 2021-03-02T06:32:05 | 2021-03-02T06:32:05 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 788,948,804 | 21,340 | Oinari | >reserve double disk space of src part set size
Thanks. Changing partitioning key and reducing partition size is my way.
Another question about the directory moving.
It looks like duplicates of clones have accumulated there.
How can I clean it safely?
I try before manually moving to temporary storage
```
c... | 2021-03-02T14:29:37 | 2021-03-02T14:31:18 | {} | NONE |
ClickHouse | ClickHouse | 788,063,789 | 20,182 | alexey-milovidov | https://github.com/ClickHouse/ClickHouse/issues/21170 | 2021-03-01T16:01:11 | 2021-03-01T16:01:11 | {} | MEMBER |
ClickHouse | ClickHouse | 788,070,822 | 21,340 | Oinari | This is it
DiskSpaceReservedForMerge │ 930866379664 │ Disk space reserved for currently running background merges. It is slightly more than the total size of currently merging parts.
Why so much? How much should be reserved? And how can they be reduced at the moment?
SYSTEM STOP TTL MERGES ? | 2021-03-01T16:10:43 | 2021-03-01T16:10:43 | {} | NONE |
ClickHouse | ClickHouse | 788,135,661 | 21,350 | joein | @den-crane
There are several kinds of errors which I found in error log:
1.
```
2021.02.05 16:29:21.494300 [ 1855 ] {9ff8193c-a81e-4aef-814c-c7693a10b6d7} <Error> executeQuery: Code: 241, e.displayText() = DB::Exception: Memory limit (for query) exceeded: would use 9.32 GiB (attempt to allocate chunk of 4718592 ... | 2021-03-01T17:38:26 | 2021-03-01T17:38:26 | {} | NONE |
ClickHouse | ClickHouse | 788,472,534 | 18,583 | alexey-milovidov | ```
pxor %xmm4,%xmm4
pxor %xmm3,%xmm3
```
```
vxorps %ymm4,%ymm4,%ymm4
vxorps %ymm3,%ymm3,%ymm3
```
Why do we zeroing these registers before return? Is it required by ABI? | 2021-03-02T01:05:00 | 2021-03-02T01:05:00 | {} | MEMBER |
ClickHouse | ClickHouse | 788,688,576 | 21,358 | mzealey | I think a setting is more appropriate as there are many different operators which could use indexes, in our case it was in but with trigram bloom filters it could also be like etc. | 2021-03-02T07:35:00 | 2021-03-02T07:35:00 | {} | NONE |
ClickHouse | ClickHouse | 788,697,320 | 21,364 | amosbird | > It looks a bit of overkill to me. Wouldn't it be better to add something like 'CREATE CLUSTER' instead and allow to define clusters programmatically?
`CREATE CLUSTER` is good for endpoint definition. ExplicitDistributed is for table-level cluster (sharding). It's a bit different. We can have multiple table shard o... | 2021-03-02T07:51:05 | 2021-03-02T07:51:05 | {} | COLLABORATOR |
ClickHouse | ClickHouse | 788,891,244 | 21,339 | vzakaznikov | This functionality does seem strange. Maybe to address original issue it is better to add something like `WITH REVOKE [OPTION] ALL ...` similar how we to have grant `WITH GRANT OPTION`? | 2021-03-02T13:00:40 | 2021-03-02T13:00:40 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 788,163,128 | 21,356 | alexey-milovidov | It because of "workers" (Run JavaScript on the edge) being configured.
Need to untangle it... | 2021-03-01T18:18:42 | 2021-03-01T18:18:42 | {} | MEMBER |
ClickHouse | ClickHouse | 788,524,071 | 7,195 | den-crane | @Alesh
* TO table? https://youtu.be/1LVJ_WcLgF8?list=PLO3lfQbpDVI-hyw4MyqxEk3rDHw95SzxJ&t=7597 https://den-crane.github.io/Everything_you_should_know_about_materialized_views_commented.pdf
* disabling atomic? https://github.com/ClickHouse/ClickHouse/issues/12135#issuecomment-653932557 | 2021-03-02T02:18:24 | 2021-03-02T02:18:24 | {
"+1": 2
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 788,655,955 | 21,339 | filimonov | > If two users A and B grants the same permission, and after that the user A revokes it, the permission is considered as still granted (by the user B).
Sounds weird. | 2021-03-02T06:37:29 | 2021-03-02T06:37:29 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 788,661,267 | 21,103 | pg-yang | @filimonov
How to remove TTL in table ?
in 20.8 | 2021-03-02T06:47:41 | 2021-03-02T06:51:05 | {} | NONE |
ClickHouse | ClickHouse | 788,694,568 | 21,364 | alex-zaitsev | It looks a bit of overkill to me. Wouldn't it be better to add something like 'CREATE CLUSTER' instead and allow to define clusters programmatically? | 2021-03-02T07:46:06 | 2021-03-02T07:46:06 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 788,712,082 | 21,339 | vitlibar | > Sounds weird.
Otherwise it won't solve the following problem. Imagine there are two users `admin` and `superadmin`, and `superadmin` grants `admin` to do something with grant option, so `admin` will able to give grants and revoke them. The problem is that in this case `admin` happens to be able to revoke any gran... | 2021-03-02T08:16:30 | 2021-03-02T08:20:50 | {} | MEMBER |
ClickHouse | ClickHouse | 788,867,163 | 14,995 | traceon | @zvonand could you please take a look why all builds are failing? Can it be due to docker image related changes? | 2021-03-02T12:16:25 | 2021-03-02T12:16:25 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 788,899,721 | 14,995 | traceon | > 180 commits are a lot. If you can squash them it would be nice.
Is **Allow squash merging** enabled for the repo?
If yes, you can choose to squash during the merge, actually.
If not, I will squash manually. | 2021-03-02T13:14:06 | 2021-03-02T13:14:06 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 788,933,668 | 21,325 | excitoon | @azat I added test for this in 02d4fd293f55d33fd87d8000ab72f928b1190a04 -- wrong credentials shall report an error but it doesn't happen when any compression method is chosen. | 2021-03-02T14:06:59 | 2021-03-02T14:06:59 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 788,160,769 | 21,356 | alexey-milovidov | The new rule does not work either. | 2021-03-01T18:15:11 | 2021-03-01T18:15:11 | {} | MEMBER |
ClickHouse | ClickHouse | 788,425,755 | 21,337 | alexey-milovidov | Need to rebase to upstream master to avoid
> AST fuzzer (ASan) — Received signal -2 | 2021-03-02T00:20:52 | 2021-03-02T00:20:52 | {} | MEMBER |
ClickHouse | ClickHouse | 788,895,558 | 21,339 | vitlibar | > So I would need to contact that SUPERadmin (he's on vacation for the next 3 months :) ) asking him to revoke something, despite the fact he gave me all the needed grants.
No, you can execute `REVOKE` on behalf of that admin if you use the full form
```
REVOKE ... GRANTED BY name_of_that_admin
```
| 2021-03-02T13:07:05 | 2021-03-02T13:07:26 | {} | MEMBER |
ClickHouse | ClickHouse | 788,975,221 | 21,307 | kitaisreal | @amosbird could you please add perf test from linked issue ? | 2021-03-02T15:06:56 | 2021-03-02T15:06:56 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 788,077,216 | 21,097 | alesapin | Hi! Unfortunately, version 20.11.3 doesn't have the fix for this problem #18844. You should update to the latest stable version, or at least the latest 20.11.
Your `<column>.bin` file became longer than expected `81322926 instead of 81265582`, so you can try to fix it and truncate it to the expected size. No guaran... | 2021-03-01T16:19:18 | 2021-03-01T16:19:18 | {} | MEMBER |
ClickHouse | ClickHouse | 788,550,018 | 18,583 | alexey-milovidov | @jart
I have managed to create standalone version of memcpy:
https://gist.github.com/alexey-milovidov/2c3b4e532b11bfb1708c89bccc7d6576
I tried to understand your code and write comments but I'm not sure if I get everthing right.
Could you please take a look at the comments in code?
Also I don't understand ho... | 2021-03-02T03:23:44 | 2021-03-02T03:23:44 | {} | MEMBER |
ClickHouse | ClickHouse | 788,970,616 | 14,995 | zvonand | > @zvonand **Style Check** still complains.
Didn't know it treats warnings as fail, too | 2021-03-02T15:00:58 | 2021-03-02T15:00:58 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 788,985,294 | 21,307 | amosbird | @kitaisreal There is already a test https://github.com/ClickHouse/ClickHouse/blob/master/tests/performance/modulo.xml for this. Maybe the difference is not evident.
| 2021-03-02T15:19:25 | 2021-03-02T15:19:25 | {} | COLLABORATOR |
ClickHouse | ClickHouse | 789,122,860 | 18,583 | alexey-milovidov | If someone will try to write `std::memcpy` it will also fail. | 2021-03-02T18:36:09 | 2021-03-02T18:36:09 | {} | MEMBER |
ClickHouse | ClickHouse | 779,054,737 | 20,249 | filimonov | Also: actually several options for handling errors were discussed:
1) put errors in the stream (the approach you implementing) - the simplest, but has disadvantages.
2) put errors in a fixed system table (system.kafka_errors: event_data, event_time, database, table, topic, partition, offset, error, raw_message)
3... | 2021-02-15T08:36:54 | 2021-02-15T08:42:44 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 779,113,462 | 20,087 | Felixoid | > to use distributed table or backend table of distributed table so that after delete it will get replicated to peer
Sorry, I don't get the question. `Distributed` table itself does not contain any data. It only sends the INSERT and SELECT queries to the cluster nodes. For `ALTER TABLE` you can use `ON CLUSTER name`... | 2021-02-15T10:10:14 | 2021-02-15T11:40:48 | {} | MEMBER |
ClickHouse | ClickHouse | 779,211,963 | 15,713 | ghost | @amosbird
I am sorry for the fact that I have no time to follow up this issue right away because I got a higher-priority problem in my company. So I close this issue.
I will retry later If this issue would not be resolved at that point. Sorry for the confusion. | 2021-02-15T13:06:25 | 2021-02-15T13:06:25 | {} | NONE |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.