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 | 756,250,486 | 17,656 | alexey-milovidov | @tavplubix is on vacation until Jan 11. | 2021-01-07T17:09:20 | 2021-01-07T17:09:20 | {
"+1": 1
} | MEMBER |
ClickHouse | ClickHouse | 756,539,465 | 18,785 | l1t1 | i guess it is https://clickhouse-builds.s3.yandex.net/0/19ad9e7a5161821e3dcfe0b4ea1f7d85209000fc/clickhouse_special_build_check/report.html
and saw link failed | 2021-01-08T04:30:19 | 2021-01-08T04:30:19 | {} | NONE |
ClickHouse | ClickHouse | 756,123,336 | 18,825 | den-crane | Any engine is supported.
```sql
CREATE MATERIALIZED VIEW test_mv
ENGINE = MergeTree() ORDER BY tuple() AS
SELECT table_test.value, dimension.name
FROM table_test
INNER JOIN dimension
ON table_test.value = dimension.value;
create table table_test (value Int64) Engine = MergeTree() order by value;... | 2021-01-07T13:41:52 | 2021-01-07T13:41:52 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 756,185,763 | 18,832 | azat | Looks like CI stuck already, will rebase against upstream/master | 2021-01-07T15:27:52 | 2021-01-07T15:28:15 | {} | MEMBER |
ClickHouse | ClickHouse | 756,381,138 | 17,838 | alexey-milovidov | It failed 4 out of 16 times. | 2021-01-07T21:00:09 | 2021-01-07T21:00:09 | {} | MEMBER |
ClickHouse | ClickHouse | 756,491,206 | 18,822 | JunjianS | > > clickhouse 20.10.tgz
>
> it seems you did not install clickhouse binary.
> Why did you use tgz instead of recommended way with yum/rpm ?
@den-crane thanks ,i have to use tgz because the server cannot access the internet, i reinstall it and no problem now, I wonder if it is necessary to do anything with OS ... | 2021-01-08T01:34:10 | 2021-01-08T01:41:31 | {} | NONE |
ClickHouse | ClickHouse | 756,538,988 | 18,742 | alexey-milovidov | @zhang2014 @BohuTANG
> test_materialize_mysql_database/test.py::test_select_without_columns_8_0[clickhouse_node1]
Failed without a reason. | 2021-01-08T04:28:06 | 2021-01-08T04:28:06 | {} | MEMBER |
ClickHouse | ClickHouse | 755,812,055 | 18,786 | vegastar002 | 用右联(right join) | 2021-01-07T00:49:38 | 2021-01-07T00:49:38 | {} | NONE |
ClickHouse | ClickHouse | 755,839,738 | 18,784 | shadowDy | > Re-work / reload data later if needed.
In this mode, when adding columns to the table that need high-frequency query, you need to adjust the order of the sort key, and you need to replay all the data. It's too troublesome | 2021-01-07T02:19:26 | 2021-01-07T02:19:26 | {} | NONE |
ClickHouse | ClickHouse | 756,276,131 | 18,845 | azat | >Yandex synchronization check (only for Yandex employees) - Failed
>Yandex third-party checks (only for Yandex employees) - Failed
Hope that it is missing file in the `ya.make`
And seems that style check is broken - https://clickhouse-test-reports.s3.yandex.net/18845/798fb4e842ed88bfe788c5de9ccac9d9986586d3/styl... | 2021-01-07T17:53:23 | 2021-01-07T17:53:55 | {} | MEMBER |
ClickHouse | ClickHouse | 756,477,827 | 18,454 | zlx19950903 | Thanks. I will handle this question, assign to me. | 2021-01-08T00:54:26 | 2021-01-08T00:54:26 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 756,545,595 | 18,818 | alexey-milovidov | We also need to add a test that will check that this function is used correctly and initial issue is fixed.
See `tests/queries/0_stateless`. | 2021-01-08T04:54:58 | 2021-01-08T04:54:58 | {} | MEMBER |
ClickHouse | ClickHouse | 755,868,667 | 18,807 | hexiaoting | @vegastar002
you want to skip column b (because sum(b) == 0) to get the following result?
```
┌─t1─┬─t2─┐
│ 0 │ 3 │
│ 1 │ 0 │
│ 0 │ 0 │
└────┴────┘
```
| 2021-01-07T04:01:59 | 2021-01-07T04:01:59 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 755,938,148 | 18,805 | fastio | Hi, Please assign it to me. | 2021-01-07T07:25:48 | 2021-01-07T07:25:48 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 756,116,320 | 10,473 | godliness | @huang--yijian did you find any solutions about this? | 2021-01-07T13:28:10 | 2021-01-07T13:28:10 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 756,130,358 | 18,827 | azat | >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?
```sh
nm -D /path/to/libc.so.... | 2021-01-07T13:55:09 | 2021-01-07T13:55:09 | {} | MEMBER |
ClickHouse | ClickHouse | 756,138,588 | 18,822 | den-crane | >clickhouse 20.10.tgz
it seems you did not install clickhouse binary.
Why did you use tgz instead of recommended way with yum/rpm ? | 2021-01-07T14:09:46 | 2021-01-07T14:09:46 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 756,159,163 | 18,819 | den-crane | Yes you can.
But only in this order:
1. add columns to MergeTree tables at ALL nodes.
2. add columns to Distributed tables at ALL nodes.
3. add columns to Buffer_engine tables at ALL nodes.
| 2021-01-07T14:44:25 | 2021-01-07T14:44:59 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 756,253,605 | 18,845 | alexey-milovidov | > P.S. I've also have pending patches for fsync support for Distributed engine, by they require this patch set.
What are the typical reasons of using async inserts to Distributed tables over sync inserts? | 2021-01-07T17:14:43 | 2021-01-07T17:14:43 | {} | MEMBER |
ClickHouse | ClickHouse | 756,333,041 | 16,947 | alexey-milovidov | @azat I did not take this PR for review. | 2021-01-07T19:31:47 | 2021-01-07T19:31:47 | {} | MEMBER |
ClickHouse | ClickHouse | 756,386,659 | 18,822 | den-crane | @alexey-milovidov I know. But @JunjianS have installed clickhouse-server.tgz without clickhouse-common-static.tgz
and now unable to start CH without /usr/bin/clickhouse | 2021-01-07T21:11:10 | 2021-01-07T21:11:10 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 756,472,480 | 18,369 | azat | >because file header has length prefix
Not the header, but each compressed block, but yes. | 2021-01-08T00:35:57 | 2021-01-08T00:36:09 | {} | MEMBER |
ClickHouse | ClickHouse | 756,554,952 | 18,716 | sagarsumit | @alexey-milovidov Any other thing, you need to know? | 2021-01-08T05:26:54 | 2021-01-08T05:26:54 | {} | NONE |
ClickHouse | ClickHouse | 755,860,343 | 18,807 | alexey-milovidov | I don't understand your question quite well. | 2021-01-07T03:32:03 | 2021-01-07T03:32:03 | {} | MEMBER |
ClickHouse | ClickHouse | 755,909,897 | 18,778 | xiaobao3315 | @den-crane CH server crash | 2021-01-07T06:14:11 | 2021-01-07T06:14:11 | {} | NONE |
ClickHouse | ClickHouse | 756,384,338 | 18,849 | den-crane | >up_count ALIAS countMerge(updates_count_state)
select without group by will fail in case. | 2021-01-07T21:06:31 | 2021-01-07T21:06:31 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 756,479,632 | 18,785 | l1t1 | > ClickHouse special build check
what is the url of `ClickHouse special build check`? | 2021-01-08T00:58:43 | 2021-01-08T00:58:43 | {} | NONE |
ClickHouse | ClickHouse | 756,533,146 | 18,855 | UnamedRus | If we would replace arrayMap(a...) with arrayMap(x...) the result would be consistent.
```
SELECT
uuid,
arrayFirstIndex((a, b) -> ((arraySum(arrayMap(x -> if(x >= 5, 0, 1), arraySlice(data_f, b + 1, 4))) = 0) AND (a >= 5)), arraySlice(data_f, 1, length(data_f) - 3), range(toUInt64(length(data_f) - 3))) ... | 2021-01-08T04:05:33 | 2021-01-08T04:05:33 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 755,870,187 | 18,810 | tenderisthenightGM | Thanks, but could you please tell me why it did not introduce the cost-base optimizer likes GPDB or Spark. | 2021-01-07T04:07:51 | 2021-01-07T04:07:51 | {} | NONE |
ClickHouse | ClickHouse | 755,906,657 | 18,810 | tenderisthenightGM | @den-crane Thanks very much. | 2021-01-07T06:05:32 | 2021-01-07T06:05:32 | {} | NONE |
ClickHouse | ClickHouse | 756,143,581 | 16,947 | azat | Any chance that someone will pick this in a the nearest future? | 2021-01-07T14:17:27 | 2021-01-07T14:17:27 | {} | MEMBER |
ClickHouse | ClickHouse | 756,295,277 | 18,847 | alexey-milovidov | I can overcome this limitation by calling _exit after kill. | 2021-01-07T18:26:19 | 2021-01-07T18:26:19 | {} | MEMBER |
ClickHouse | ClickHouse | 755,859,517 | 18,810 | alexey-milovidov | Mostly rule-based. Also simple cost-based for one case ("move-to-prewhere" optimization).
Also optimization with bayesian bandits for lz4 decompression and for some functions dispatch. | 2021-01-07T03:29:19 | 2021-01-07T03:29:19 | {} | MEMBER |
ClickHouse | ClickHouse | 755,972,009 | 18,785 | l1t1 | ```
:~/tpc-ds-maste/queries$ pgrep clickhouse-serv
:~/tpc-ds-maste/queries$ pgrep clickhouse
4565
:~/tpc-ds-maste/queries$ wc -l /proc/$(pgrep clickhouse)/maps
152 /proc/4565/maps
:~/tpc-ds-maste/queries$ head /proc/4565/maps
00200000-06a2a000 r--p 00000000 08:02 548284 /home/a/ch/c... | 2021-01-07T08:41:18 | 2021-01-07T08:41:18 | {} | NONE |
ClickHouse | ClickHouse | 756,234,792 | 18,844 | filimonov | As an immediate workaround: try to set `min_merge_bytes_to_use_direct_io` to zero.
1) What is the filesystem?
2) Is is a fresh install? Or older ClickHouse version was working properly before? | 2021-01-07T16:46:23 | 2021-01-07T16:46:23 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 755,833,982 | 17,838 | weeds085490 | @alexey-milovidov I have changed the test method. All functional tests have passed. PTAL | 2021-01-07T01:59:46 | 2021-01-07T01:59:46 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 755,871,869 | 18,720 | hexiaoting | assign to me | 2021-01-07T04:13:59 | 2021-01-07T04:14:12 | {
"rocket": 2
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 755,872,950 | 18,810 | den-crane | @tenderisthenightGM
1. because mostly CBO is needed for choosing of the most efficient access path: indexes / joins (choosing what type of join to use. Right now CH does not support inverted indexes and supports mostly only HJ. So the access path is straightforward and does not require even RBO.
2. there are tons of ... | 2021-01-07T04:17:53 | 2021-01-07T04:20:43 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 755,932,243 | 18,813 | tkionshao | > Need to upgrade to the latest ClickHouse release first.
But I already use 20.8.7.15, is it not enough? | 2021-01-07T07:11:26 | 2021-01-07T07:11:26 | {} | NONE |
ClickHouse | ClickHouse | 755,953,257 | 18,785 | azat | I guess that you reached some limits:
- either total memory limit
- or, and I think that is more likely, maximum number of vmas (`vm.max_map_count`, default is 65k), you can verify this with smth like `wc -l /proc/$(pgrep clickhouse-serv)/maps`, you can attach `/proc/pid/maps` or/and `/proc/pid/smaps` for further ana... | 2021-01-07T08:01:09 | 2021-01-07T08:01:09 | {} | MEMBER |
ClickHouse | ClickHouse | 756,127,682 | 18,827 | ucasfl | > @ucasFL Are you using bundled build?
No.
> There is a fix for unbundled build - #18764, but if this is a similar error you should have errors during compiling too not only during linking
I don't see other errors during build. | 2021-01-07T13:49:56 | 2021-01-07T13:49:56 | {} | COLLABORATOR |
ClickHouse | ClickHouse | 756,164,534 | 18,784 | den-crane | > > Re-work / reload data later if needed.
>
> In this mode, when adding columns to the table that need high-frequency query, you need to adjust the order of the sort key, and you need to replay all the data. It's too troublesome
The problem that you don't not have options with CH.
There is no magic. Terra-byte... | 2021-01-07T14:53:37 | 2021-01-07T14:53:37 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 756,302,447 | 18,848 | alexey-milovidov | @MyroTk It fixes #18847 but no tests will be provided for minor issue. Need to check manually. | 2021-01-07T18:38:45 | 2021-01-07T18:38:45 | {} | MEMBER |
ClickHouse | ClickHouse | 755,732,379 | 16,669 | den-crane | The same issue with 20.8.11 and 20.13.1.5579
ping to mysql time=155 ms
and no issues with Engine=MySQL it seems it fails on a metadata discovery? | 2021-01-06T21:43:51 | 2021-01-06T21:48:21 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 755,308,753 | 18,785 | l1t1 | my version https://builds.clickhouse.tech/master/aarch64/clickhouse
and after running `strip clickhouse`
```sql
SELECT version()
Query id: 44414303-c4d8-45a4-84c8-522285d36e8e
┌─version()────┐
│ 20.13.1.5583 │
└──────────────┘
``` | 2021-01-06T13:49:12 | 2021-01-07T01:50:32 | {} | NONE |
ClickHouse | ClickHouse | 755,897,058 | 18,807 | vegastar002 | > @vegastar002
> you want to skip column b (because sum(b) == 0) to get the following result?
>
> ```
> ┌─t1─┬─t2─┐
> │ 0 │ 3 │
> │ 1 │ 0 │
> │ 0 │ 0 │
> └────┴────┘
> ```
Yes !!! | 2021-01-07T05:39:14 | 2021-01-07T05:39:14 | {} | NONE |
ClickHouse | ClickHouse | 756,086,607 | 18,808 | alexey-milovidov | But I feel that the presence of build-id is more important. | 2021-01-07T12:26:45 | 2021-01-07T12:26:45 | {} | MEMBER |
ClickHouse | ClickHouse | 756,241,091 | 18,844 | jorisgio | Changed settings :
```
┌─name──────────────────────────┬─value────────┬─changed─┐
│ max_insert_block_size │ 83886080 │ 1 │
│ use_uncompressed_cache │ 0 │ 1 │
│ background_pool_size │ 96 │ 1 │
│ background_fetches_pool_size │ 16 │ ... | 2021-01-07T16:57:00 | 2021-01-07T16:57:00 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 756,306,609 | 18,845 | alexey-milovidov | Yes, this is the part that @qoega missed in #18463 (git repository is not present in new check). | 2021-01-07T18:46:11 | 2021-01-07T18:46:11 | {} | MEMBER |
ClickHouse | ClickHouse | 756,308,183 | 18,845 | qoega | git repository is present. There is some issue with ignored paths. Contribs should not be checked | 2021-01-07T18:49:00 | 2021-01-07T18:49:00 | {} | MEMBER |
ClickHouse | ClickHouse | 756,491,789 | 18,854 | den-crane | All table functions require RW privileges by design.
You can use `select count(*) from system.numbers where number <= 10000000` instead. | 2021-01-08T01:36:10 | 2021-01-08T01:36:39 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 756,494,594 | 17,838 | weeds085490 | > It failed 4 out of 16 times.
Ok. Let me take a look | 2021-01-08T01:46:14 | 2021-01-08T01:46:14 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 755,921,614 | 18,780 | ClownfishYang | While I was working with UDFs, it was great to realize that Functions supported nesting.
```
SELECT
uuid,
arrayFirstIndex((a, b) -> ((a >= 5) AND (arraySum(arrayMap(a -> if(a >= 5, 1, 0), arraySlice(data_f, b, 4))) = 4)), data_f, arrayEnumerate(data_f)) AS freq_idx
FROM
(
SELECT
'id1' AS... | 2021-01-07T06:44:39 | 2021-01-07T06:44:39 | {} | NONE |
ClickHouse | ClickHouse | 756,122,805 | 17,838 | alexey-milovidov | Sorry, the test is still flaky, I will revert this PR and we need to resubmit it. | 2021-01-07T13:40:49 | 2021-01-07T13:40:49 | {} | MEMBER |
ClickHouse | ClickHouse | 756,126,110 | 18,785 | azat | @alexey-milovidov
>my version https://builds.clickhouse.tech/master/aarch64/clickhouse
| 2021-01-07T13:46:52 | 2021-01-07T13:46:52 | {} | MEMBER |
ClickHouse | ClickHouse | 756,272,696 | 18,775 | azat | >@alexey-milovidov Arcadia does not support distributed tables
BTW is there any clue (I've looked for `ARCADIA_BUILD`/`grep -v` for `ya.make` - nothing) in the code that will show this? | 2021-01-07T17:47:27 | 2021-01-07T17:47:27 | {} | MEMBER |
ClickHouse | ClickHouse | 755,856,435 | 18,684 | alexey-milovidov | For some reason all `test_materialize_mysql_database` have failed. | 2021-01-07T03:18:33 | 2021-01-07T03:18:33 | {} | MEMBER |
ClickHouse | ClickHouse | 756,101,210 | 18,785 | azat | @l1t1 thanks for details!
Indeed the problem was not in the mmap limits of memory limits but in the size of alternative stack, should be fixed in #18832.
You can try to get new binaries from the `ClickHouse special build check` check report page (once it will be ready) | 2021-01-07T12:58:44 | 2021-01-07T12:58:44 | {} | MEMBER |
ClickHouse | ClickHouse | 756,235,249 | 18,844 | UnamedRus | Which filesystem it is?
AIO actually has some issues:
https://github.com/ClickHouse/ClickHouse/issues/14093
And was disabled in 20.13
https://github.com/ClickHouse/ClickHouse/pull/18481
> Disable write with AIO during merges because it can lead to extremely rare data corruption of primary key columns during me... | 2021-01-07T16:47:11 | 2021-01-07T16:47:11 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 756,265,124 | 18,845 | azat | >What are the typical reasons of using async inserts to Distributed tables over sync inserts?
One of the main advantage is batching, asynchronous inserts into Distributed allows batching, so you get something like Buffer table but on disk.
Also it allows to not fail INSERT if remote node is unavailable, files will ... | 2021-01-07T17:34:10 | 2021-01-07T17:34:10 | {} | MEMBER |
ClickHouse | ClickHouse | 756,295,154 | 18,847 | alexey-milovidov | It correctly sends the signal but it has no effect due to Linux limitation:
```
$ docker run --rm ubuntu:latest /bin/bash -c 'echo $$ && kill -9 $$ && echo test'
1
test
``` | 2021-01-07T18:26:05 | 2021-01-07T18:26:05 | {} | MEMBER |
ClickHouse | ClickHouse | 756,335,435 | 18,822 | alexey-milovidov | @den-crane `tgz` is valid installation option, the same as `deb` or `rpm`. | 2021-01-07T19:36:04 | 2021-01-07T19:36:04 | {} | MEMBER |
ClickHouse | ClickHouse | 756,386,150 | 16,947 | azat | >Fast test — fail: 1, passed: 2320, skipped: 278. Other tests won't be run before fast test will be fixed.
Now JOIN part is broken with #15162 (and is blocked by #8802), so I've replaced the test cases to expect 403 error, but patches are still relevant, since it fixes GROUP BY (and not related to JOIN actually, jus... | 2021-01-07T21:10:10 | 2021-01-07T21:10:10 | {} | MEMBER |
ClickHouse | ClickHouse | 755,883,513 | 18,742 | alexey-milovidov | Waiting on condvars with timeout is broken. | 2021-01-07T04:55:19 | 2021-01-07T04:55:19 | {} | MEMBER |
ClickHouse | ClickHouse | 756,102,966 | 18,827 | azat | @ucasFL Are you using bundled build?
There is a fix for unbundled build - #18764, but if this is a similar error you should have errors during compiling too not only during linking | 2021-01-07T13:02:12 | 2021-01-07T13:02:12 | {} | MEMBER |
ClickHouse | ClickHouse | 756,333,647 | 16,947 | alexey-milovidov | Let's merge with master again. | 2021-01-07T19:32:51 | 2021-01-07T19:32:51 | {} | MEMBER |
ClickHouse | ClickHouse | 756,367,876 | 18,845 | filimonov | > > P.S. I've also have pending patches for fsync support for Distributed engine, by they require this patch set.
>
> What are the typical reasons of using async inserts to Distributed tables over sync inserts?
Proper retries even if first attempt failed. | 2021-01-07T20:34:33 | 2021-01-07T20:34:33 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 756,491,597 | 18,814 | peaksnail | for example ,I have the population of every province and city
And I need store them with array
so I have the data like
```
select
['provinceA', 'provinceB'] as `data.province`,
[['provinceA_cityA','provinceA_cityB'], ['provinceB_cityC', 'provinceB_cityD']] as `data.city`,
[[10000,2000], [500,300000]] as... | 2021-01-08T01:35:29 | 2021-01-08T01:36:38 | {} | NONE |
ClickHouse | ClickHouse | 756,537,387 | 18,854 | l1t1 | thanks | 2021-01-08T04:21:43 | 2021-01-08T04:21:43 | {} | NONE |
ClickHouse | ClickHouse | 756,550,180 | 18,839 | fastio | > @fastio Thank you! Some edits required for build...
The compiling error is fixed now. | 2021-01-08T05:12:06 | 2021-01-08T05:12:06 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 755,883,914 | 18,813 | alexey-milovidov | Need to upgrade to the latest ClickHouse release first. | 2021-01-07T04:56:45 | 2021-01-07T04:56:45 | {} | MEMBER |
ClickHouse | ClickHouse | 755,910,430 | 18,807 | hexiaoting | @vegastar002
As I know, it is difficult to write the sql when we don't know the select schema.
maybe column transformer can help, like `select * except xxx from ....`
But in this case, I'm afraid whether it can be written in sql or not.
Waiting for others to give you the sql you want. | 2021-01-07T06:15:47 | 2021-01-07T06:15:47 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 755,942,863 | 15,806 | ahutsunshine | @Enmk @CurtizJ @akuzm May I ask when and what's the plan to build a new version to release the feature? | 2021-01-07T07:36:49 | 2021-01-07T07:36:49 | {} | NONE |
ClickHouse | ClickHouse | 756,190,772 | 18,832 | alexey-milovidov | BTW, what is the value of MINSIGSTKSZ on AArch64? | 2021-01-07T15:35:39 | 2021-01-07T15:35:39 | {} | MEMBER |
ClickHouse | ClickHouse | 756,239,635 | 17,656 | bharatnc | @tavplubix @azat just following up - let me know if you have any other inputs, I have implemented the changes based on your latest reviews. | 2021-01-07T16:54:34 | 2021-01-07T16:54:34 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 756,123,158 | 18,832 | alexey-milovidov | @azat #18834 | 2021-01-07T13:41:31 | 2021-01-07T13:41:31 | {} | MEMBER |
ClickHouse | ClickHouse | 756,124,971 | 18,785 | alexey-milovidov | @azat BTW, How did you know that this is AArch64 build? | 2021-01-07T13:44:39 | 2021-01-07T13:44:39 | {} | MEMBER |
ClickHouse | ClickHouse | 756,162,257 | 18,778 | den-crane | @xiaobao3315 can you provide logs with crash messages? | 2021-01-07T14:49:48 | 2021-01-07T14:49:48 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 756,244,123 | 18,832 | alexey-milovidov | Reasoning:
https://patchwork.ozlabs.org/project/glibc/patch/mvmwq1v2aqw.fsf@hawking.suse.de/ | 2021-01-07T17:02:00 | 2021-01-07T17:02:00 | {} | MEMBER |
ClickHouse | ClickHouse | 756,246,304 | 18,844 | alexey-milovidov | We already addressed this issue in recent patch releases (20.12.5 was released on 28 Dec 2020). | 2021-01-07T17:05:37 | 2021-01-07T17:05:37 | {} | MEMBER |
ClickHouse | ClickHouse | 756,496,261 | 18,824 | tangxueming | > because `ORDER BY (timestamp, duration, ` timestamp & duration are high cardinality columns.
>
> Check this example https://gist.github.com/den-crane/42bf77285b53762f11d88d73a46179de
> https://stackoverflow.com/questions/64369622/using-sample-in-clickhouse-seems-to-read-all-rows-and-more-bytes-is-this-expect
>
... | 2021-01-08T01:52:12 | 2021-01-08T01:52:12 | {} | NONE |
ClickHouse | ClickHouse | 756,538,811 | 18,785 | l1t1 | > @l1t1 thanks for details!
>
> Indeed the problem was not in the mmap limits of memory limits but in the size of alternative stack, should be fixed in #18832.
> You can try to get new binaries from the `ClickHouse special build check` check report page (once it will be ready)
there is no Arch64 build at https:/... | 2021-01-08T04:27:21 | 2021-01-08T04:27:21 | {} | NONE |
ClickHouse | ClickHouse | 756,543,921 | 18,839 | alexey-milovidov | @fastio Thank you! Some edits required for build... | 2021-01-08T04:48:27 | 2021-01-08T04:48:27 | {} | MEMBER |
ClickHouse | ClickHouse | 755,718,826 | 7,649 | alexey-milovidov | @deripas We are not going to add custom function to calculate shingles, but there is another way to go:
First preprocess text by extracting meaningful words with normalization and write them separated by whitespaces. This has to be done before loading data into ClickHouse. Then save the preprocessed text to a separa... | 2021-01-06T21:22:59 | 2021-01-06T21:23:28 | {
"+1": 1
} | MEMBER |
ClickHouse | ClickHouse | 755,776,301 | 18,794 | alexey-milovidov | > Linux is not using fast "TSC" timer;
`cat /sys/devices/system/clocksource/clocksource0/current_clocksource`
> memory overcommit is set to 2;
`cat /proc/sys/vm/overcommit_memory`
> CPU scaling governor is not "performance";
`cat /sys/devices/system/cpu/cpu*/cpufreq/scaling_governor`
> transparent hug... | 2021-01-06T23:25:40 | 2021-01-06T23:25:40 | {} | MEMBER |
ClickHouse | ClickHouse | 755,790,165 | 18,210 | azat | >repro
@den-crane Thanks you for the reproducer and bisect!
>related: #16574 ?
Indeed, it has bee introduced in #16610
Fix is here - #18806 | 2021-01-07T00:12:46 | 2021-01-07T00:12:46 | {} | MEMBER |
ClickHouse | ClickHouse | 755,884,089 | 18,720 | alexey-milovidov | @hexiaoting Ok. Thank you in advance! | 2021-01-07T04:57:21 | 2021-01-07T04:57:21 | {} | MEMBER |
ClickHouse | ClickHouse | 756,190,862 | 18,813 | den-crane | @tkionshao upgrade to 20.8.11.17 | 2021-01-07T15:35:48 | 2021-01-07T15:35:48 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 756,236,561 | 18,844 | jorisgio | Will disable yes, thanks. (strangely, and luckily, it seems to happen only on one of our database that is thankfully not in prod, but i'm not sure what is special about this one yet).
1. filesystem is plain ext4, setup is jbod with storage policy on several nvme disks with ext4
2. This is old install, and was working... | 2021-01-07T16:49:31 | 2021-01-07T16:49:31 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 756,238,229 | 18,832 | azat | >BTW, what is the value of MINSIGSTKSZ on AArch64?
5120
Refs:
- https://elixir.bootlin.com/linux/v5.11-rc2/source/arch/arm64/include/uapi/asm/signal.h#L23
- https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=c9692657c0 | 2021-01-07T16:52:17 | 2021-01-07T16:52:17 | {} | MEMBER |
ClickHouse | ClickHouse | 756,525,022 | 18,451 | ElapsedSoul | I got through this by set `set(CMAKE_HAVE_LIBC_PTHREAD 1)` at the top of cmakelist.txt. hope to help. | 2021-01-08T03:31:49 | 2021-01-08T03:31:49 | {} | NONE |
ClickHouse | ClickHouse | 756,545,096 | 18,817 | alexey-milovidov | Yes, this is very typical optimization for aggregate functions!
I did not dig into `windowFunnel` implementation in more detail - maybe there are more possible optimizations.
This PR is Ok to merge, but let's add performance test to prove and keep the improvement, see `tests/performance`. | 2021-01-08T04:53:09 | 2021-01-08T04:53:09 | {} | MEMBER |
ClickHouse | ClickHouse | 755,852,158 | 18,794 | alexey-milovidov | > checksum of the loaded code segment does not correspond to some record in ELF file;
#18811 | 2021-01-07T03:02:50 | 2021-01-07T03:02:50 | {} | MEMBER |
ClickHouse | ClickHouse | 756,086,417 | 18,808 | alexey-milovidov | It may slowdown linking a little. | 2021-01-07T12:26:23 | 2021-01-07T12:26:23 | {} | MEMBER |
ClickHouse | ClickHouse | 756,120,807 | 18,832 | azat | >Fast test — fail: 1, passed: 2321, skipped: 278. Other tests won't be run before fast test will be fixed.
01600_count_of_parts_metrics is flaky, can someone add force-test label please? | 2021-01-07T13:36:58 | 2021-01-07T13:37:19 | {} | MEMBER |
ClickHouse | ClickHouse | 756,125,183 | 18,824 | den-crane | because `ORDER BY (timestamp, duration, ` timestamp & duration are high cardinality columns.
Check this example https://gist.github.com/den-crane/42bf77285b53762f11d88d73a46179de
https://stackoverflow.com/questions/64369622/using-sample-in-clickhouse-seems-to-read-all-rows-and-more-bytes-is-this-expect
Try
`... | 2021-01-07T13:45:04 | 2021-01-07T13:49:50 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 756,383,321 | 18,849 | den-crane | BTW, You can use SimpleAggregateFunction here.
```sql
create table test_alias (
timestamp UInt64,
up_count SimpleAggregateFunction(sum, UInt64) CODEC(T64, LZ4),
name String
) Engine = AggregatingMergeTree
ORDER BY
timestamp;
```
| 2021-01-07T21:04:34 | 2021-01-07T21:04:34 | {
"+1": 1
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 753,685,398 | 16,020 | alexey-milovidov | Maybe fixed in #18526? | 2021-01-03T22:28:53 | 2021-01-03T22:28:53 | {} | MEMBER |
ClickHouse | ClickHouse | 754,003,045 | 18,655 | alexey-milovidov | @sundy-li Maybe you want to help fixing the remaining use-after-free error? https://github.com/facebook/rocksdb/issues/7821 | 2021-01-04T14:23:11 | 2021-01-04T14:23:11 | {} | MEMBER |
ClickHouse | ClickHouse | 754,033,786 | 18,612 | siradjev | Probably would also need UTC_TIMESTAMP() implemented as per #17177 | 2021-01-04T15:15:02 | 2021-01-04T15:15:02 | {} | NONE |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.