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 | 653,693,005 | 11,990 | phillipcheng | I have tried all your suggestions.
but it still happens and happens a lot.
<img width="1398" alt="截屏2020-07-04 上午7 05 45" src="https://user-images.githubusercontent.com/7891324/86500606-e4516280-bdc4-11ea-9229-4327d96c0a85.png">
i just insert once (3 records) into the distributed table, but the result is 5 records... | 2020-07-03T23:07:42 | 2020-07-03T23:07:42 | {} | NONE |
ClickHouse | ClickHouse | 653,667,090 | 12,122 | iameugenejo | I just found a workaround, adding a huge http_headers_progress_interval_ms value to suppress the progress header for a long time seems to work for my case. `http_headers_progress_interval_ms=3600000`.
I would still like to know if there's a better or a correct way to do this. | 2020-07-03T20:12:55 | 2020-07-04T03:33:02 | {} | NONE |
ClickHouse | ClickHouse | 653,785,218 | 12,128 | KochetovNicolai | If you just need to easily support `LowCardinality(T)`, you may set setting `low_cardinality_allow_in_native_format=0` for client session and work with data the same way as if it has type `T`. Server will automatically convert `LowCardinality(T) -> T` for selects and `T -> LowCardinality(T)` for inserts if needed.
I... | 2020-07-04T16:21:03 | 2020-07-04T16:21:03 | {
"+1": 2
} | MEMBER |
ClickHouse | ClickHouse | 653,954,228 | 12,142 | azat | Allocation distribution (dumb analysis):
```
$ docker run -v $PWD:/wrk:ro -w /wrk -it --rm --name ch yandex/clickhouse-server clickhouse-local -q "select floor(log2(size)) lg, min(size) min, max(size) max, count() cnt from file ('left-trace-log.tsv', TSVWithNamesAndTypes, '$(cat left-trace-log.tsv.columns)') where ... | 2020-07-05T23:42:05 | 2020-07-05T23:42:05 | {} | MEMBER |
ClickHouse | ClickHouse | 653,976,449 | 8,159 | wbigdata |
> @wbigdata解决了！
@gubinjie 兄弟，方便给我说一下咋解决的吗，困扰了两天了。。。 | 2020-07-06T01:56:18 | 2020-07-06T01:56:18 | {} | NONE |
ClickHouse | ClickHouse | 653,982,850 | 8,159 | gubinjie | @wbigdata
you need add "export LIBHDFS3_CONF=/etc/clickhouse-server/hdfs-client.xml " to /etc/init.d/clickhouse-server;
1、你把搭建hadoop的时候的hdfs-client.xml 的文件复制一份到装有ClickHouse的服务器的/etc/clickhouse-server文件夹里面。
2、然后在装有ClickHouse的服务器的/etc/init.d/clickhouse-server的文件里面加一行export LIBHDFS3_CONF=/etc/clickhouse-server/hdfs-cl... | 2020-07-06T02:28:58 | 2020-07-06T02:28:58 | {} | NONE |
ClickHouse | ClickHouse | 653,818,326 | 12,101 | alexey-milovidov | 
| 2020-07-04T22:18:03 | 2020-07-04T22:18:03 | {} | MEMBER |
ClickHouse | ClickHouse | 653,820,265 | 11,564 | bobrik | My understanding is that the strategy with PK range search is to do a tree search and cut off branches that cannot contain the requested PK range. For a query that matches the whole part, we need to recurse into every leaf, making it cost `O(n)`.
I think it would be more efficient to:
1. Do a binary search for th... | 2020-07-04T22:48:21 | 2020-07-04T22:50:02 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 653,907,066 | 12,136 | alexey-milovidov | > Yandex synchronization check (only for Yandex employees)
Our collegues has found a bug in this check. Actually everything is Ok, will merge. | 2020-07-05T16:10:01 | 2020-07-05T16:10:01 | {} | MEMBER |
ClickHouse | ClickHouse | 653,930,991 | 12,144 | mzealey | Is there a way to debug why the merge is not being run in the background as often as I might expect? Are there any guarantees around when clickhouse will run this operation?
Also something I cannot find in the manual; is there any way to alter the partitioning scheme or do i have to create a new table? | 2020-07-05T19:48:00 | 2020-07-05T19:48:00 | {} | NONE |
ClickHouse | ClickHouse | 653,938,762 | 12,142 | alexey-milovidov | Looks like it doesn't affect performance. So we can merge :) | 2020-07-05T21:01:38 | 2020-07-05T21:01:38 | {} | MEMBER |
ClickHouse | ClickHouse | 653,678,858 | 12,104 | alexey-milovidov | @akuzm I will write comments in this PR and then will address all of them by myself. | 2020-07-03T21:15:51 | 2020-07-03T21:16:04 | {} | MEMBER |
ClickHouse | ClickHouse | 653,687,523 | 11,317 | bobrik | I'm getting the following error in `v20.6.1.3925-testing`:
```
[6397/7735] Building CXX object src/CMakeFiles/dbms.dir/DataStreams/CollapsingFinalBlockInputStream.cpp.o
FAILED: src/CMakeFiles/dbms.dir/DataStreams/CollapsingFinalBlockInputStream.cpp.o
/usr/bin/g++ -DAWS_SDK_VERSION_MAJOR=1 -DAWS_SDK_VERSION_MINOR... | 2020-07-03T22:16:46 | 2020-07-03T22:16:46 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 653,866,750 | 11,590 | azat | After patching tcmalloc to use **front-end** (allocations can be done w/o locking) **up to 4MB** (in upstream it is only possible to use it for allocations up to 256K), **perf tests became a little bit faster**, but still not enough:
- default (lock-free up to 8<<10): 1 errors, 8 faster, 232 slower, 215 unstable
- ... | 2020-07-05T09:50:50 | 2020-07-05T09:50:50 | {
"+1": 1
} | MEMBER |
ClickHouse | ClickHouse | 653,920,901 | 10,299 | alexey-milovidov | Ok.
Another (less efficient) option is to do
```
SELECT
warehouse, arrayJoin([product, '']) AS product, SUM(quantity)
FROM
inventory
GROUP BY warehouse, product;
``` | 2020-07-05T18:13:53 | 2020-07-05T18:13:53 | {} | MEMBER |
ClickHouse | ClickHouse | 653,950,616 | 12,146 | alexey-milovidov | https://github.com/ClickHouse/ClickHouse/pull/9516 | 2020-07-05T23:07:42 | 2020-07-05T23:07:42 | {} | MEMBER |
ClickHouse | ClickHouse | 653,950,767 | 12,146 | alexey-milovidov | `SELECT * FROM remote('127..2', 'a.');` | 2020-07-05T23:09:12 | 2020-07-05T23:09:12 | {} | MEMBER |
ClickHouse | ClickHouse | 653,747,532 | 12,125 | myrrc | > Need to check how it will work with UTF-8 in case of regexp.
> `'ёЁё' ILIKE 'Ё%Ё'`
I'll push the tests today. | 2020-07-04T10:11:46 | 2020-07-04T10:11:46 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 653,804,972 | 8,871 | alexey-milovidov | It means that ZooKeeper server has closed connection on its own.
ClickHouse will reestablish connection with different ZooKeeper server and continue to work.
This is normal behaviour unless it happens too frequently - in this case you have to investigate the issue in ZooKeeper configuration.
Whenever you got an ex... | 2020-07-04T19:55:40 | 2020-07-04T19:57:03 | {} | MEMBER |
ClickHouse | ClickHouse | 653,825,939 | 12,119 | alexey-milovidov | `select intDiv(materialize(toInt32(1)), -0x100000000)` | 2020-07-05T00:17:45 | 2020-07-05T00:17:45 | {} | MEMBER |
ClickHouse | ClickHouse | 653,830,029 | 12,134 | alexey-milovidov | Network issue when connecting to Docker registry. | 2020-07-05T01:23:14 | 2020-07-05T01:23:14 | {} | MEMBER |
ClickHouse | ClickHouse | 653,886,031 | 7,316 | zhicwu | Tried DateTime64 in 20.3.12.112 but wasn't able to save date like `1970-01-01 00:00:00.0` or older, and value looks weird when converting from a negative number(please refer to the last query below).
```bash
clickhouse :) select toDate('1970-01-01'), toDateTime('1970-01-01 00:00:00'), toDateTime64('1970-01-01 00:00... | 2020-07-05T12:58:54 | 2020-07-05T12:58:54 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 653,907,469 | 9,577 | alexey-milovidov | @excitoon The issue needs to be fixed, why so long? | 2020-07-05T16:13:45 | 2020-07-05T16:13:45 | {} | MEMBER |
ClickHouse | ClickHouse | 653,587,750 | 6,682 | alesapin | Fixed in CI, but we have to check in the next release. | 2020-07-03T15:04:15 | 2020-07-03T15:04:15 | {} | MEMBER |
ClickHouse | ClickHouse | 653,684,717 | 12,123 | alexey-milovidov | The commands
`cat /proc/$(pidof clickhouse-server)/maps`
`sudo lsof -p $(pidof clickhouse-server)`
can show what kind of things are used from the system. | 2020-07-03T21:55:07 | 2020-07-03T21:55:07 | {} | MEMBER |
ClickHouse | ClickHouse | 653,810,149 | 12,092 | alexey-milovidov | PS. I see three invocations of `addArena` in `Aggregator.cpp`. | 2020-07-04T20:44:58 | 2020-07-04T20:44:58 | {} | MEMBER |
ClickHouse | ClickHouse | 653,818,191 | 12,101 | alexey-milovidov | @blinkov If it is not possible to disable scanning of subdirectories (contrib), you should create an issue in the repository of "CodeQL security scanner" with a link to this PR. | 2020-07-04T22:15:58 | 2020-07-04T22:15:58 | {} | MEMBER |
ClickHouse | ClickHouse | 653,921,538 | 8,706 | alexey-milovidov | This behaviour is by design - to gain some performance of parsing by allowing to have some implementation specific behaviour on incorrect input.
But we already support multiple parsing modes in code and we allow to switch between them with `date_time_input_format` setting. Modes `basic` and `best_effort` are availab... | 2020-07-05T18:19:38 | 2020-07-05T18:21:22 | {} | MEMBER |
ClickHouse | ClickHouse | 653,662,808 | 12,073 | alexey-milovidov | > Maybe the execution of 'ALTER TABLE test_table MODIFY COLUMN nested_column.x FIRST' should be rejected ?
Yes, it's reasonable to forbid moving of nested columns.
BTW, do we support MODIFY COLUMN ... AFTER ...?
Upd. I see. | 2020-07-03T19:48:40 | 2020-07-03T19:54:00 | {} | MEMBER |
ClickHouse | ClickHouse | 653,683,714 | 12,123 | alexey-milovidov | ```
$ mkdir proc
$ sudo mount --bind /proc proc
$ sudo chroot . /clickhouse local --query "SELECT 1"
1
```
clickhouse-client fails with a cryptic error message:
```
$ sudo chroot . /clickhouse client
ClickHouse client version 20.6.1.1.
Connecting to localhost:9000 as user default.
Poco::Exception. Code: ... | 2020-07-03T21:47:10 | 2020-07-03T21:47:10 | {} | MEMBER |
ClickHouse | ClickHouse | 653,690,163 | 12,072 | den-crane | 19.15.7.3 it's the same as 19.15.7.30
https://hub.docker.com/layers/yandex/clickhouse-server/19.15.7.3/images/sha256-718d4f52449061e2fe6e2ff90610896476fe729be922f79c11d2f5d52af22a4f?context=explore | 2020-07-03T22:39:02 | 2020-07-03T22:39:02 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 653,808,688 | 12,021 | alexey-milovidov | > SELECT count() FROM test.hits WHERE NOT ignore(least(URL, Referer))
This test result is unrelated. | 2020-07-04T20:27:25 | 2020-07-04T20:27:25 | {} | MEMBER |
ClickHouse | ClickHouse | 653,817,697 | 12,102 | alexey-milovidov | Docker checks appeared to be 99% useless.
They report that we have wrong version of Python inside container (but we don't need Python at all), we have wrong version of Perl inside container (but we don't need Perl at all) and so on.
Maybe we can overcome these issues by avoiding the usage of Ubuntu: https://github.... | 2020-07-04T22:08:32 | 2020-07-04T22:08:32 | {} | MEMBER |
ClickHouse | ClickHouse | 653,823,497 | 12,101 | alexey-milovidov | CodeQL is useful, one bug has been found: #12138 | 2020-07-04T23:40:47 | 2020-07-04T23:40:47 | {} | MEMBER |
ClickHouse | ClickHouse | 653,905,378 | 12,144 | mzealey | Thanks, trying that now.
What would the lag usually be before data is merged? We have quite high insert rate on this table and from what I can see it hasn't been merged in the past 4-5 days even though most rollup merges are specified to happen within 1d of the data being created. Anything we can do to speed this up... | 2020-07-05T15:54:58 | 2020-07-05T15:54:58 | {} | NONE |
ClickHouse | ClickHouse | 653,683,022 | 12,123 | alexey-milovidov | And without procfs neither:
```
$ mkdir tmp
$ sudo chroot . /clickhouse local --query "SELECT 1"
Code: 468, e.displayText() = DB::ErrnoException: Cannot pthread_getattr_np, errno: 2, strerror: No such file or directory (version 20.6.1.1)
``` | 2020-07-03T21:42:34 | 2020-07-03T21:42:34 | {} | MEMBER |
ClickHouse | ClickHouse | 653,766,885 | 4,457 | ucasfl | This issue seems to have been fixed:
```
Poco::Exception. Code: 1000, e.code() = 0, e.displayText() = System exception: cannot get current directory, Stack trace (when copying this messag
e, always include the lines below):
```
@alexey-milovidov | 2020-07-04T13:32:05 | 2020-07-04T13:32:05 | {
"+1": 1
} | COLLABORATOR |
ClickHouse | ClickHouse | 653,816,891 | 11,995 | azat | @nikitamikhaylov friendly ping | 2020-07-04T21:58:07 | 2020-07-04T21:58:07 | {
"+1": 1
} | MEMBER |
ClickHouse | ClickHouse | 653,905,590 | 12,125 | alexey-milovidov | (Will merge with squash) | 2020-07-05T15:56:58 | 2020-07-05T15:56:58 | {} | MEMBER |
ClickHouse | ClickHouse | 653,918,202 | 12,144 | den-crane | Check this project https://github.com/innogames/graphite-ch-optimizer | 2020-07-05T17:49:00 | 2020-07-05T17:49:00 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 653,557,483 | 11,913 | alexey-milovidov | Ok, I understand the reason. | 2020-07-03T13:48:52 | 2020-07-03T13:48:52 | {} | MEMBER |
ClickHouse | ClickHouse | 653,688,593 | 11,317 | bobrik | Nevermind, I forgot to do `--delete` while doing `rsync` to my build machine. | 2020-07-03T22:24:32 | 2020-07-03T22:24:32 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 653,734,010 | 12,121 | alesapin | Obvious bugfix, I'll merge it | 2020-07-04T07:35:13 | 2020-07-04T07:35:13 | {} | MEMBER |
ClickHouse | ClickHouse | 653,780,967 | 12,133 | alexey-milovidov | Arcadia is broken, fix: https://a.yandex-team.ru/review/1332342/details | 2020-07-04T15:43:32 | 2020-07-04T15:43:32 | {} | MEMBER |
ClickHouse | ClickHouse | 653,797,866 | 12,128 | vahid-sohrabloo | Thanks @KochetovNicolai. you saved my day | 2020-07-04T18:31:51 | 2020-07-04T18:31:51 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 653,824,272 | 12,119 | alexey-milovidov | Because `libdivide` is using `exit` instead of `abort`:
```
#0 __GI_raise (sig=sig@entry=6) at ../sysdeps/unix/sysv/linux/raise.c:50
#1 0x00007ffff7da8899 in __GI_abort () at abort.c:79
#2 0x000000000d9329b1 in terminate_handler () at ../base/daemon/BaseDaemon.cpp:387
#3 0x0000000010dced33 in std::__terminat... | 2020-07-04T23:54:58 | 2020-07-04T23:54:58 | {} | MEMBER |
ClickHouse | ClickHouse | 653,827,776 | 12,138 | alexey-milovidov | > Yandex synchronization check (only for Yandex employees)
"Arcadia" build is using wrong build system, will fix. | 2020-07-05T00:46:09 | 2020-07-05T00:46:09 | {} | MEMBER |
ClickHouse | ClickHouse | 653,908,477 | 12,076 | alexey-milovidov | Ok, I cannot wait to see some test results... | 2020-07-05T16:22:46 | 2020-07-05T16:22:46 | {} | MEMBER |
ClickHouse | ClickHouse | 653,933,066 | 12,144 | den-crane | Merges have another goal -- minimization a number of part with minimum CPU / disk I/O.
They (MergeScheduler) is not aware about a rollup. Rollup is just a byproduct.
>Are there any guarantees around when clickhouse will run this operation?
No. On the contrary. CH never does a final merge automatically and after... | 2020-07-05T20:07:10 | 2020-07-05T23:08:10 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 653,554,167 | 12,113 | den-crane | Nested means that all properties must be filled
cmdty_id, cmdty_name, cmdty_group_id, cmdty_group_name
[5], ['myname'], [], ['fafa']
you have to fill cmdty_group_id , you can not pass empty []
and empty array does not mean Null.
`INSERT INTO cmdts_test VALUES ([5], ['myname'], [Null], ['fafa'])`
| 2020-07-03T13:40:36 | 2020-07-03T13:51:55 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 653,686,557 | 12,123 | alexey-milovidov | ```
$ mkdir root
$ sudo chroot . /clickhouse client
ClickHouse client version 20.6.1.1.
Connecting to localhost:9000 as user default.
Connected to ClickHouse server version 20.6.1 revision 54436.
milovidov-desktop :) SELECT 1
SELECT 1
┌─1─┐
│ 1 │
└───┘
1 rows in set. Elapsed: 0.002 sec.
milovid... | 2020-07-03T22:09:06 | 2020-07-03T22:09:06 | {} | MEMBER |
ClickHouse | ClickHouse | 653,697,777 | 11,541 | bobrik | @CurtizJ, I'm still having trouble querying the table. My goal is to read the data in index order, so my query looks like this:
```sql
select traceID
from jaeger_index_v2
where service = 'edgeworker'
and -toUnixTimestamp(timestamp) <= -toUnixTimestamp(now() - 60)
order by service, -toUnixTimestamp(timest... | 2020-07-03T23:58:11 | 2020-07-03T23:58:11 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 653,700,727 | 12,124 | alexey-milovidov | @blinkov Missing description for changelog. | 2020-07-04T00:29:42 | 2020-07-04T00:29:42 | {} | MEMBER |
ClickHouse | ClickHouse | 653,746,295 | 12,074 | marius3000 | **How much memory server has?**
8GB RAM + 1GB SWAP (tried disabling SWAP as well)
**How long rows in your table?**
GBsize: 49
nrows: 8252394
avg_row_size_bytes: 40267
parts_count: 226
partition_count: 145
**Solution**
I had to lower merge_max_block_size all the way to 512 to stop seeing errors, with ... | 2020-07-04T09:56:49 | 2020-07-04T09:56:49 | {} | NONE |
ClickHouse | ClickHouse | 653,768,226 | 4,457 | alexey-milovidov | @ucasFL That's great, thank you for validating! | 2020-07-04T13:44:45 | 2020-07-04T13:44:45 | {} | MEMBER |
ClickHouse | ClickHouse | 653,807,674 | 12,021 | alexey-milovidov | It is strange that I cannot reproduce performance regression on my today clang release build. | 2020-07-04T20:20:17 | 2020-07-04T20:20:17 | {} | MEMBER |
ClickHouse | ClickHouse | 653,827,461 | 4,211 | alexey-milovidov | It works perfectly in the latest versions:
```
milovidov-desktop :) SET connections_with_failover_max_tries = 5
SET connections_with_failover_max_tries = 5
Ok.
0 rows in set. Elapsed: 0.001 sec.
milovidov-desktop :) SELECT hostName() FROM remote('128.1.2.3', default.tmp)
SELECT hostName()
FROM remo... | 2020-07-05T00:40:53 | 2020-07-05T00:40:53 | {} | MEMBER |
ClickHouse | ClickHouse | 653,830,081 | 11,815 | alexey-milovidov | Split build smoke test:
https://github.com/ClickHouse/ClickHouse/pull/12134#issuecomment-653829989
Need to add retries for all docker run / docker pull, etc. | 2020-07-05T01:24:11 | 2020-07-05T01:24:11 | {} | MEMBER |
ClickHouse | ClickHouse | 653,949,320 | 7,749 | alexey-milovidov | You can set `mutations_sync` setting to 1 for the query. When this setting enabled, the query will return success to the client only after mutation will be executed on every replica.
But if query will not return success (e.g. one replica is unavailable), mutation will be executed nevertheless. In that case, you have... | 2020-07-05T22:55:37 | 2020-07-05T22:55:37 | {
"heart": 1
} | MEMBER |
ClickHouse | ClickHouse | 653,977,619 | 12,150 | alexey-milovidov | The initial password configuration is implemented via "debconf".
https://unix.stackexchange.com/questions/106552/apt-get-install-without-debconf-prompt
PS. If you don't have debconf in your system (e.g. you install `rpm` or `tgz` package on non-debian system) it will not ask. | 2020-07-06T02:02:37 | 2020-07-06T02:03:49 | {
"+1": 1
} | MEMBER |
ClickHouse | ClickHouse | 653,808,734 | 12,021 | alexey-milovidov | > SELECT arraySlice(arrayFill(x -> ((x % 2) >= 2), range(100000000)), 1, 10) FORMAT Null
This neither. | 2020-07-04T20:27:52 | 2020-07-04T20:27:52 | {} | MEMBER |
ClickHouse | ClickHouse | 653,908,638 | 12,057 | alexey-milovidov | Duplicate of #12060. | 2020-07-05T16:24:13 | 2020-07-05T16:24:13 | {} | MEMBER |
ClickHouse | ClickHouse | 653,926,801 | 12,144 | den-crane | > does it make sense to change partition by to toyyyymmdd()
Probably daily partitions would be easier to merge (CPU/disk IO) but still you need to call `optimize` | 2020-07-05T19:08:10 | 2020-07-05T19:08:10 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 653,685,653 | 12,123 | alexey-milovidov | ```
$ cp /lib/x86_64-linux-gnu/libnss_dns.so.2 lib
$ cp /lib/x86_64-linux-gnu/libresolv.so.2 lib
$ mkdir etc
$ cp /etc/resolv.conf ./etc
$ sudo chroot . /clickhouse client
Poco::Exception. Code: 1000, e.code() = 2, e.displayText() = File not found: /root/.clickhouse-client-history (version 20.6.1.1)
``` | 2020-07-03T22:02:45 | 2020-07-03T22:02:45 | {} | MEMBER |
ClickHouse | ClickHouse | 653,816,951 | 11,995 | nikitamikhaylov | > @nikitamikhaylov friendly ping
😉 | 2020-07-04T21:58:57 | 2020-07-04T21:58:57 | {} | MEMBER |
ClickHouse | ClickHouse | 653,825,154 | 12,119 | alexey-milovidov | `select materialize(toInt32(1)) % -9223372036854775808` | 2020-07-05T00:08:44 | 2020-07-05T00:08:44 | {} | MEMBER |
ClickHouse | ClickHouse | 653,828,529 | 8,564 | alexey-milovidov | `from_zk` is the answer on how to store cluster configuration in ZooKeeper.
Automatic discovery most likely will not be implemented soon due to considerations that @den-crane has mentioned. | 2020-07-05T00:59:48 | 2020-07-05T00:59:48 | {} | MEMBER |
ClickHouse | ClickHouse | 653,921,678 | 8,706 | alexey-milovidov | @ucasFL Maybe you are interested? | 2020-07-05T18:20:48 | 2020-07-05T18:20:48 | {} | MEMBER |
ClickHouse | ClickHouse | 653,951,462 | 12,144 | den-crane | One more thing.
>PARTITION BY toYYYYMM(Date) ORDER BY (Path, Time)
I don't know how your queries where section looks.
If it does not have `Date` filter. For example where Path = .... and Time between ....
Then consider Time as partitioning column.
`PARTITION BY toYYYYMM(toDateTime(Time)) ORDER BY (Path, Time... | 2020-07-05T23:16:12 | 2020-07-05T23:16:12 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 653,977,783 | 12,150 | alexey-milovidov | Reference: https://github.com/ClickHouse/ClickHouse/blob/master/debian/clickhouse-server.postinst#L131 | 2020-07-06T02:03:35 | 2020-07-06T02:03:35 | {} | MEMBER |
ClickHouse | ClickHouse | 653,686,799 | 12,123 | alexey-milovidov | After these steps, clickhouse-server and clickhouse-client work successfully in chroot. | 2020-07-03T22:10:32 | 2020-07-03T22:12:51 | {} | MEMBER |
ClickHouse | ClickHouse | 653,813,652 | 11,564 | bobrik | After adding some logging, I was able to catch what looks like a pathological case:
```
$ echo "SELECT DISTINCT traceID FROM jaeger_index_v2 PREWHERE service = 'nginx-fl' AND antiTimestamp between -toUnixTimestamp(now()) and -toUnixTimestamp(now() - 60) ORDER BY service, antiTimestamp LIMIT 20" | docker exec -i cli... | 2020-07-04T21:15:16 | 2020-07-04T21:15:16 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 653,924,744 | 12,144 | mzealey | I had tried that before but it also had said nothing to do.
I had to `optimize table final` to get it to run but it has now done the aggregations.
Given that most of my partitions are based around days rather than months & I'm taking about 10b row/day, does it make sense to change `partition by` to `toyyyymmdd()`... | 2020-07-05T18:49:53 | 2020-07-05T18:49:53 | {} | NONE |
ClickHouse | ClickHouse | 653,949,451 | 7,749 | alexey-milovidov | @simPod
Unfortunately, @blinkov made himself a robot. | 2020-07-05T22:56:42 | 2020-07-05T22:56:42 | {} | MEMBER |
ClickHouse | ClickHouse | 653,957,364 | 8,962 | alexey-milovidov | We have implemented SQL compatible data types that are also case-insensitive. Example:
```
CREATE TEMPORARY TABLE test (x char varying not null)
``` | 2020-07-06T00:06:49 | 2020-07-06T00:06:49 | {} | MEMBER |
ClickHouse | ClickHouse | 653,802,547 | 7,660 | alexey-milovidov | @filimonov Isn't it fixed? | 2020-07-04T19:28:00 | 2020-07-04T19:28:00 | {} | MEMBER |
ClickHouse | ClickHouse | 653,815,123 | 11,590 | azat | FWIW CI adds status for the incorrect commit, the latest commit is 19c4aa66e2c00c1bde2aa6b1d2e3ab9b0037c039 but marked by CI - 24dca801959b241268f8c3608affd26eaa831d7c, looks like it get the last from the list w/o sort (but I'm pretty sure that it uses sources from the correct commit)
Oh, and maybe that's why it does ... | 2020-07-04T21:34:35 | 2020-07-04T21:34:35 | {} | MEMBER |
ClickHouse | ClickHouse | 653,906,203 | 12,021 | alexey-milovidov | @akuzm Performance test showed misleading result and delayed the fix for 5 days. | 2020-07-05T16:02:17 | 2020-07-05T16:02:17 | {} | MEMBER |
ClickHouse | ClickHouse | 653,936,177 | 12,144 | mzealey | Many thanks for your help! | 2020-07-05T20:37:37 | 2020-07-05T20:37:37 | {} | NONE |
ClickHouse | ClickHouse | 653,946,143 | 12,142 | azat | >Looks like it doesn't affect performance.
Indeed, this is interesting, I mean that jemalloc+memory sampling (although it uses pipe w/ O_NONBLOCK, so not sure that all the allocations was written) works faster then tcmalloc (need to look at data)
> So we can merge :)
But maybe worth adjust the `max_untracked_m... | 2020-07-05T22:20:35 | 2020-07-05T22:20:35 | {} | MEMBER |
ClickHouse | ClickHouse | 653,808,990 | 12,021 | alexey-milovidov | > SELECT count() FROM test.hits WHERE NOT ignore(rand() % 2 ? URL : Referer)
Function names in profiles are absolutely the same. | 2020-07-04T20:31:27 | 2020-07-04T20:31:27 | {} | MEMBER |
ClickHouse | ClickHouse | 653,824,438 | 11,990 | phillipcheng | i changed to insert to local tables, i am doing the host and db selection on the client side, but still, there is duplication and missing problems. i am doing parallel insertion to the same table.
clickhouse does not support this? | 2020-07-04T23:58:25 | 2020-07-04T23:58:25 | {} | NONE |
ClickHouse | ClickHouse | 653,682,845 | 12,123 | alexey-milovidov | CC @akuzm | 2020-07-03T21:41:32 | 2020-07-03T21:41:32 | {} | MEMBER |
ClickHouse | ClickHouse | 653,817,108 | 12,136 | alexey-milovidov | I've added NOLINT to skip clang-tidy check about function names. | 2020-07-04T22:00:57 | 2020-07-04T22:00:57 | {} | MEMBER |
ClickHouse | ClickHouse | 653,829,989 | 12,134 | alexey-milovidov | Split build:
```
Executing cmd: 'docker run --network=host --volume=/place/sandbox-data/tasks/4/5/716107354/clang-10_relwithdebuginfo_none_bundled_splitted_disable_False_binary:/package_folder --volume=/place/sandbox-data/tasks/5/0/716112705/server_log:/var/log/clickhouse-server yandex/clickhouse-split-build-smoke-te... | 2020-07-05T01:22:36 | 2020-07-05T01:22:36 | {} | MEMBER |
ClickHouse | ClickHouse | 653,895,570 | 11,990 | den-crane | CH supports parallel insertion. Probably something wrong with your client. How do you handle insertion errors and how re-tries implemented?
https://clickhouse.tech/docs/en/engines/table-engines/mergetree-family/replication/
_Data blocks are deduplicated. For multiple writes of the same data block (data blocks of t... | 2020-07-05T14:27:16 | 2020-07-05T14:27:16 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 653,951,952 | 12,092 | CurtizJ | Other invocations are done at intermidiate aggregation step and the columns are always `ColumnAggregateFunction`. | 2020-07-05T23:21:15 | 2020-07-05T23:21:25 | {} | MEMBER |
ClickHouse | ClickHouse | 653,967,623 | 11,469 | alexey-milovidov | The bug was introduced in https://github.com/ClickHouse/ClickHouse/pull/10342
CC @azat | 2020-07-06T01:11:53 | 2020-07-06T01:11:53 | {} | MEMBER |
ClickHouse | ClickHouse | 653,970,940 | 12,091 | alexey-milovidov | Still reproduces. | 2020-07-06T01:27:55 | 2020-07-06T01:27:55 | {} | MEMBER |
ClickHouse | ClickHouse | 653,556,817 | 11,913 | CurtizJ | It's possible, but we have to write compressed block for every column in granule. Won't blocks be too small? Also, in that case it will be possible to count per column sizes. | 2020-07-03T13:47:18 | 2020-07-03T13:48:59 | {} | MEMBER |
ClickHouse | ClickHouse | 653,685,043 | 12,123 | alexey-milovidov | The command
`sudo strace -f chroot . /clickhouse client`
will show us what is missing... | 2020-07-03T21:57:45 | 2020-07-03T21:57:45 | {} | MEMBER |
ClickHouse | ClickHouse | 653,765,903 | 3,501 | alexey-milovidov | @filimonov
Do we need pid file when using systemd?:
https://github.com/ClickHouse/ClickHouse/blob/master/debian/clickhouse-server.service#L13
We can remove request for its creation (the `--pid-file` argument to the server).
Why do we need pid file at all? | 2020-07-04T13:22:27 | 2020-07-04T13:25:02 | {} | MEMBER |
ClickHouse | ClickHouse | 653,798,192 | 12,133 | alexey-milovidov | Ok, it works perfectly. | 2020-07-04T18:36:15 | 2020-07-04T18:36:15 | {} | MEMBER |
ClickHouse | ClickHouse | 653,800,757 | 4,659 | vivarum | https://github.com/ClickHouse/ClickHouse/pull/12136 | 2020-07-04T19:07:02 | 2020-07-04T19:07:02 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 653,903,077 | 12,144 | den-crane | A rolling-up happens during a merge process.
Probably this part is not merged yet in background automatically (by any reason) you can force a merge manually using `optimize table` https://clickhouse.tech/docs/en/sql-reference/statements/misc/#misc_operations-optimize | 2020-07-05T15:33:50 | 2020-07-05T15:34:39 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 653,905,675 | 12,125 | alexey-milovidov | > Yandex synchronization check
Bogus result, our friends from Arcadia will investigate. | 2020-07-05T15:57:45 | 2020-07-05T15:57:45 | {} | MEMBER |
ClickHouse | ClickHouse | 653,932,557 | 12,135 | tavplubix | It's because with `Atomic` database engine `DROP TABLE` query just marks table as dropped. Table are actually removed later in background, when all queries to the table are finished. Even if no queries to the table are running, there is a delay between `DROP TABLE` and background removal, it can be configured by `<data... | 2020-07-05T20:02:21 | 2020-07-05T20:02:21 | {
"+1": 3,
"-1": 1
} | MEMBER |
ClickHouse | ClickHouse | 653,968,911 | 11,469 | alexey-milovidov | @azat This is also broken:
```
milovidov-desktop :) SELECT dictGetString('default' || '.countryId', 'country', toUInt64(number)) AS country FROM numbers(2) GROUP BY country;
SELECT dictGetString(concat('default', '.countryId'), 'country', toUInt64(number)) AS country
FROM numbers(2)
GROUP BY country
Recei... | 2020-07-06T01:17:42 | 2020-07-06T01:17:42 | {} | MEMBER |
ClickHouse | ClickHouse | 653,991,455 | 11,696 | alexey-milovidov | @CurtizJ Maybe we can enable group by in order optimization by default? | 2020-07-06T03:08:12 | 2020-07-06T03:08:12 | {} | MEMBER |
ClickHouse | ClickHouse | 654,141,407 | 12,142 | akuzm | > > https://twitter.com/trav_downs/status/1262058940831092738
> > A nice trick they use in another sampling memory profiler to record deallocation for the same blocks.
>
> Interesting, but if understand it right, it will work only for "large enough" allocations
Didn't read the code, but as I understood from the ... | 2020-07-06T10:06:48 | 2020-07-06T10:07:25 | {} | CONTRIBUTOR |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.