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 | 444,196,233 | 3,445 | proller | netlink dmesg problem solved - https://github.com/yandex/ClickHouse/pull/3749 | 2018-12-04T17:59:22 | 2018-12-04T17:59:22 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 444,530,975 | 3,735 | 4ertus2 | It looks like we need to realize reinterpretAsDecimal*. It's possible to make some dirty workaround. Something like this:
```
SELECT out
FROM
(
SELECT
'02b5e3af16b1880000' AS x,
length(x) AS xlen,
substring(x, 1, xlen - 16) AS first,
substring(x, (xlen - 16) + 1, 16) ... | 2018-12-05T15:42:33 | 2018-12-05T15:46:46 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 444,749,888 | 3,764 | shangshujie365 | the wrong result are always occurred with 'Aggregator: Converting aggregation data to two-level.' | 2018-12-06T05:07:20 | 2018-12-06T05:07:20 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 444,813,811 | 3,735 | TrollStation | Thanks for the reply! Glad to hear that big numbers will be nativly supported in near future because Clickhouse is really cool DB. | 2018-12-06T09:52:50 | 2018-12-06T09:52:50 | {} | NONE |
ClickHouse | ClickHouse | 445,158,794 | 3,713 | blinkov | Fixed in #3774 | 2018-12-07T08:27:19 | 2018-12-07T08:27:19 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 445,202,354 | 3,780 | filimonov | May be related: #3501 #3610 | 2018-12-07T11:17:20 | 2018-12-07T11:17:20 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 445,329,951 | 3,786 | alexey-milovidov | Yes, it is acceptable. | 2018-12-07T18:54:21 | 2018-12-07T18:54:21 | {} | MEMBER |
ClickHouse | ClickHouse | 445,109,695 | 1,582 | den-crane | you can combine columns into tuple and separate on another level
select p, V.1 t, V.2 x, V.3 y from (
select p, argMin((t,x,y),t) V from test
group by p)
https://gist.github.com/den-crane/fcc95468f37e40f9ff2a22cf3e250c0f | 2018-12-07T03:18:48 | 2018-12-07T03:18:48 | {
"+1": 3
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 445,388,738 | 3,768 | 4ertus2 | Hi,
Decimals are integers inside. You can write fist 8 bytes to 0xFF and negative Int64 after them for 64-bit values. For full 128-bit math you need 128-bit integers support (native or self-made). Then you could write Int128 values to the column as is. | 2018-12-07T22:44:42 | 2018-12-07T22:44:42 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 445,800,474 | 3,695 | filimonov | And also, while we are here - it's better to change those lines:
https://github.com/yandex/ClickHouse/blob/0466f6beafc7d1c0f20b4309e420ab3ab8b98a0f/docker/server/entrypoint.sh#L9-L15
to use `clickhouse-extract-from-config` | 2018-12-10T12:35:12 | 2018-12-10T12:35:12 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 444,841,624 | 3,764 | shangshujie365 | actually I have write a small fix to this problem | 2018-12-06T11:31:47 | 2018-12-06T11:31:47 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 444,859,225 | 3,759 | proller | NOTES:
1. fail in unbundled mode woth old double-cosnversion:
FAILED: contrib/arrow-cmake/CMakeFiles/arrow_static.dir/__/arrow/cpp/src/arrow/csv/converter.cc.o
/usr/bin/g++-7 -DBOOST_SYSTEM_NO_DEPRECATED -Icontrib/arrow-cmake -I../contrib/arrow/cpp/src -I../contrib/arrow-cmake/cpp/src -isystem ../contrib/boost -fd... | 2018-12-06T12:45:11 | 2018-12-06T12:45:11 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 445,228,934 | 3,782 | ztlpn | Currently it is only possible manually.
For non-replicated MergeTree:
1. `DETACH TABLE <table>`
2. Remove the `mutation_xxx.txt` file from the `/var/lib/data/<database>/<table>/` directory. Exact file name can be determined from the `mutation_id` column of the `system.mutations` table.
3. `ATTACH TABLE <table>`
... | 2018-12-07T13:13:30 | 2018-12-07T13:13:30 | {
"+1": 3
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 445,347,803 | 3,784 | alexey-milovidov | > Question: Is it possible to change user_files_path in a clickhouse-client session with an sql command ?
No, because the setting is security limit: to allow to read only restricted subset of files from server's filesystem. But you can specify `user_files_path` as `/` (and use path relative to filesystem root) if yo... | 2018-12-07T19:57:02 | 2018-12-07T19:59:26 | {} | MEMBER |
ClickHouse | ClickHouse | 444,196,361 | 3,198 | proller | netlink dmesg problem solved - https://github.com/yandex/ClickHouse/pull/3749 | 2018-12-04T17:59:47 | 2018-12-04T17:59:47 | {
"+1": 1
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 444,253,225 | 3,583 | alexey-milovidov | Released in 18.14.18. | 2018-12-04T20:51:14 | 2018-12-04T20:51:14 | {} | MEMBER |
ClickHouse | ClickHouse | 444,316,627 | 3,753 | sundy-li | ok | 2018-12-05T00:54:32 | 2018-12-05T00:54:32 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 444,496,334 | 3,758 | alex-zaitsev | @healiseu
SHOW CREATE TABLE.
Only overrides are shown. Global settings can be viewed in system.merge_tree_settings table (as of 18.10.3 release and later) | 2018-12-05T14:05:47 | 2018-12-05T14:05:47 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 444,533,223 | 3,716 | filimonov | See #3288
Still appears in 18.14.17. Fix only in master? | 2018-12-05T15:48:05 | 2018-12-05T15:48:05 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 444,853,105 | 1,784 | viongpanzi | @savelov
#3742 will work for ReplicatedMergeTree, but I'm not sure also work for ReplicatedReplacingMergeTree | 2018-12-06T12:20:10 | 2018-12-06T12:20:10 | {} | NONE |
ClickHouse | ClickHouse | 444,864,592 | 3,617 | alesapin | @chenxing-xc Where did you get libhdfs3? There are a lot of forks on github and we would like to have commits history in our Clickhouse-Extras repo. | 2018-12-06T13:07:00 | 2018-12-06T13:07:00 | {} | MEMBER |
ClickHouse | ClickHouse | 445,226,445 | 3,781 | excitoon | See ClickHouse-Extras/poco#2 | 2018-12-07T13:03:13 | 2018-12-07T13:03:29 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 444,235,960 | 2,601 | alexey-milovidov | > According to docs: "Keep in mind that it does not make much sense for PREWHERE to only specify those columns that have an index, because when using an index, only the data blocks that match the index are read."
This is obsolete statement. I have removed it from the docs.
Sometimes it does make sense to use PREW... | 2018-12-04T19:56:18 | 2018-12-04T19:56:18 | {} | MEMBER |
ClickHouse | ClickHouse | 444,515,817 | 3,758 | healiseu | OK thanks Alex, issue is closed | 2018-12-05T15:02:21 | 2018-12-05T15:02:21 | {} | NONE |
ClickHouse | ClickHouse | 444,786,715 | 1,784 | savelov | > Further - Trying again on a ReplicatedReplacingMergeTree across two servers. Optimize returns OK on one machine, but doesn't do anything, and on the 2nd machine it complains that the user password is not correct (use the same users.xml, and all DDL 'on cluster' statements work fine, and the data has replicated, so no... | 2018-12-06T08:19:13 | 2018-12-06T08:19:13 | {} | NONE |
ClickHouse | ClickHouse | 445,160,435 | 3,696 | blinkov | @yuliujuan do you have any further questions? | 2018-12-07T08:34:17 | 2018-12-07T08:34:17 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 445,233,524 | 3,783 | simPod | Version 18.14.18 | 2018-12-07T13:32:17 | 2018-12-07T13:32:17 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 445,435,600 | 3,784 | healiseu | Hi @alexey-milovidov, thank you for your answers, you covered me sufficiently for the moment. | 2018-12-08T06:25:32 | 2018-12-08T06:25:32 | {} | NONE |
ClickHouse | ClickHouse | 444,742,563 | 3,764 | shangshujie365 | this problem is very wired, keep max_threads=52, set max_block_size=65529, the result sometimes is right, sometimes is wrong, if set max_block_size larger, the result is always wrong, is set max_block_size smaller, the result is always right. btw, our defult max_block_size = 5242880.
if set max_threads=1, the result i... | 2018-12-06T04:13:54 | 2018-12-06T04:13:54 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 444,766,166 | 3,198 | cyfdecyf | @proller Thanks. The netlink error message disappears after applying your fix. | 2018-12-06T06:45:09 | 2018-12-06T06:45:09 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 445,244,994 | 3,781 | proller | maybe name flag POCO_SKIP_OPENSSL_FIND ? | 2018-12-07T14:15:11 | 2018-12-07T14:15:11 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 445,419,485 | 3,617 | alexey-milovidov | @chenxing-xc Congratulations! | 2018-12-08T01:44:19 | 2018-12-08T01:44:19 | {
"hooray": 5
} | MEMBER |
ClickHouse | ClickHouse | 445,841,772 | 3,780 | filimonov | 1) pid file is created only in daemon mode (so just add --daemon command line parameter if you really need that):
https://github.com/yandex/ClickHouse/blob/f1791e94e209b238b28a89c25e3a5f81882785a6/libs/libdaemon/src/BaseDaemon.cpp#L957-L958
2) folder for pid-file shoud exists i.e. run first
```bash
mkdir -p /var/r... | 2018-12-10T14:50:30 | 2018-12-10T14:51:41 | {
"+1": 1,
"heart": 1,
"hooray": 1,
"rocket": 1
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 445,922,535 | 3,801 | alexey-milovidov | Note: you have not updated your branch. | 2018-12-10T18:34:13 | 2018-12-10T18:34:13 | {} | MEMBER |
ClickHouse | ClickHouse | 445,931,408 | 3,800 | alexey-milovidov | Missing functional test. Look at `dbms/tests/queries` directory. | 2018-12-10T19:00:48 | 2018-12-10T19:00:48 | {} | MEMBER |
ClickHouse | ClickHouse | 444,573,956 | 3,596 | 4ertus2 | #3531 reopened | 2018-12-05T17:36:22 | 2018-12-05T17:36:22 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 445,229,079 | 3,782 | ztlpn | Reopening as this is a common request and we should implement a more convenient method to do it. | 2018-12-07T13:14:08 | 2018-12-07T13:14:08 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 445,361,885 | 3,759 | alexey-milovidov | I kindly ask to squash all your commits but to keep all commits of @luc1ph3r as separate items (unsquashed). | 2018-12-07T20:52:14 | 2018-12-07T20:52:14 | {} | MEMBER |
ClickHouse | ClickHouse | 444,180,057 | 3,753 | blinkov | @zhang2014 @sundy-li this touches few zh articles too | 2018-12-04T17:13:26 | 2018-12-04T17:13:26 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 444,903,855 | 3,770 | filimonov | Confirm. The problem appears for **any** netmask in CIDR format.
Test case:
```xml
<?xml version="1.0"?>
<yandex>
<users>
<readonly>
<password></password>
<networks>
<ip>1.0.0.1/24</ip>
</networks>
<profile>readonly</profile>
... | 2018-12-06T15:10:57 | 2018-12-06T16:13:23 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 445,227,620 | 3,782 | simPod | Removed mutation txt file, just took a while | 2018-12-07T13:08:12 | 2018-12-07T13:08:12 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 445,330,461 | 3,786 | alexey-milovidov | Both min/max date and min/max time columns look like collection of special cases. Not good in general.
But as we already have min/max date, adding min/max time looks perfectly reasonable.
I appreciate if you will add this enhancement! | 2018-12-07T18:56:04 | 2018-12-07T18:56:23 | {} | MEMBER |
ClickHouse | ClickHouse | 445,563,582 | 3,695 | xzkostyan | I can make another PR with port readiness probe within 10 retries. @alexey-milovidov should I? | 2018-12-09T19:19:48 | 2018-12-09T19:19:48 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 445,799,519 | 3,764 | shangshujie365 | Hi, @filimonov, you can see my fix at https://github.com/shangshujie365/ClickHouse/commit/4f143af9a00d36c115c239d2466131d8509016d5.
The bug has been fixed, but introduce another bug,
if gender or platform has 'null' value, the partial cube result (gender=null, platform-cubed-to-null), (gender-cubed-to-null, platform... | 2018-12-10T12:31:30 | 2018-12-10T12:31:30 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 445,860,468 | 3,804 | alexey-milovidov | `00719_format_datetime_rand` - unrelated (tech debt). | 2018-12-10T15:40:59 | 2018-12-10T15:40:59 | {} | MEMBER |
ClickHouse | ClickHouse | 444,505,229 | 3,597 | ztlpn | Yes, this is by design - mutations are executed strictly in order, so if one of them cannot be finished, all mutations that come after it unfortunately cannot finish too. | 2018-12-05T14:32:59 | 2018-12-05T14:32:59 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 444,769,257 | 3,764 | shangshujie365 | you can reproduce my problem by 'set group_by_two_level_threshold=1;'
using a table like following:
select * from test.r2;
┌───────date─┬─id─┐
│ 2018-10-01 │ 1 │
│ 2018-10-02 │ 2 │
│ 2018-11-01 │ 1 │
│ 2018-11-03 │ 2 │
└────────────┴────┘
set group_by_two_level_threshold=1;
select date,id,count() from t... | 2018-12-06T07:01:27 | 2018-12-06T07:01:27 | {
"+1": 1
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 444,784,021 | 3,765 | alex-zaitsev | +1
Naming conventions wise, I suggest a function arrayMapElement or arrayElementMap(<keys_array>,<values_array>,'key') to show the connection to sumMap and arrayElement. | 2018-12-06T08:09:10 | 2018-12-06T09:23:12 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 445,205,079 | 3,764 | filimonov | > here is my small fix, maybe you can check whether they are right:
Great that you find the exact reason and even create fix for that!
Could you create a PR with your changes (also with a test-case, please)? That will allow running the tests suite with your changes and will allow making a code review (and do any ... | 2018-12-07T11:29:36 | 2018-12-07T11:29:36 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 445,520,131 | 3,750 | fancyqlx | why not limit the usage of memory. I found there is a parameter for limiting disk usage. Can we add some parameters for limiting memory usage when selecting parts to merge. | 2018-12-09T08:37:17 | 2018-12-09T08:37:17 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 445,911,840 | 3,806 | alexey-milovidov | > waiting for server start (via http pings)
The only concern is that the user may disable HTTP completely (they can leave HTTPS only).
(For example, this is the case for Yandex.Cloud). | 2018-12-10T18:02:09 | 2018-12-10T18:02:09 | {} | MEMBER |
ClickHouse | ClickHouse | 445,927,915 | 3,799 | alexey-milovidov | BTW both Set and Join can benefit from simple and efficient implementation for LowCardinality (project right hand side dictionary onto left hand side dictionary). | 2018-12-10T18:50:26 | 2018-12-10T18:50:26 | {} | MEMBER |
ClickHouse | ClickHouse | 444,531,209 | 2,640 | bretthoerner | I think this was fixed by #3673? | 2018-12-05T15:43:09 | 2018-12-05T15:43:09 | {
"+1": 1
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 444,858,528 | 3,764 | shangshujie365 | here is my small fix, maybe you can check whether they are right:
add these fields to CubeBlockInputStream.h:
BlocksList ori_blocks;
bool hasMasked = false;
add new read method of CubeBlockInputStream.cpp:
````
Block CubeBlockInputStream::readImpl()
{
if (hasMasked){
return {};
}el... | 2018-12-06T12:42:22 | 2018-12-07T03:53:13 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 445,211,988 | 3,764 | shangshujie365 | sure, let me try | 2018-12-07T12:00:32 | 2018-12-07T12:00:32 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 445,484,638 | 3,791 | blinkov | > I don't check the final web page in local development environment : (
> Because I didn't build it successfully.
It's ok: even if there are any issues, they are usually easy to fix right before release. | 2018-12-08T19:42:27 | 2018-12-08T19:42:39 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 445,927,197 | 3,799 | alexey-milovidov | It looks like first minimal solution: no special optimization for LowCardinality, it just works better than default "hashed" method. Correct me if I'm wrong. | 2018-12-10T18:48:16 | 2018-12-10T18:48:16 | {} | MEMBER |
ClickHouse | ClickHouse | 444,389,668 | 3,753 | blinkov | @sundy-li thanks | 2018-12-05T07:44:46 | 2018-12-05T07:44:46 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 444,421,717 | 3,597 | healiseu | Hi Alex (@ztlpn), thank you for your reply, I confirm that **all my update queries work if I use the full table name**. These include sub-queries, views, materialized views and sets as well as drop/add column. I have also asked you about the execution of mutations in a queue. I noticed that if any of these update quer... | 2018-12-05T09:40:23 | 2018-12-05T09:42:41 | {} | NONE |
ClickHouse | ClickHouse | 444,592,940 | 3,271 | rajadayalan | can you please provide update on the issue? | 2018-12-05T18:33:15 | 2018-12-05T18:33:15 | {} | NONE |
ClickHouse | ClickHouse | 444,823,687 | 2,640 | ztlpn | Yes, looks like it is the same issue. So it should be fixed in 18.14.17 and later versions. | 2018-12-06T10:25:06 | 2018-12-06T10:25:06 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 445,250,921 | 3,781 | excitoon | Done. | 2018-12-07T14:35:18 | 2018-12-07T14:35:18 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 445,331,360 | 3,780 | filimonov | BTW why do you run `clickhouse-server` directly?
If you install clickhouse-server from the packages it should be registered as a systemd service, which should be started automatically. | 2018-12-07T18:59:15 | 2018-12-07T18:59:15 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 445,548,300 | 2,135 | amosbird | You can have a dictionary-like structure with any kind of value after this pr https://github.com/yandex/ClickHouse/pull/3728
```
201.nobida.cn :) insert into js values (2, ['c++', 'ruby']);
INSERT INTO js VALUES
Ok.
1 rows in set. Elapsed: 0.001 sec.
201.nobida.cn :) select joinGet('js', 'ss', toInt32(1... | 2018-12-09T15:49:22 | 2018-12-09T15:49:22 | {} | COLLABORATOR |
ClickHouse | ClickHouse | 445,745,023 | 3,798 | blinkov | Not critical, but please change /en/ -> /zh/ in those links at bottom of each page in the future.
cc @zhang2014 | 2018-12-10T09:19:22 | 2018-12-10T09:19:22 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 444,456,458 | 3,755 | ztlpn | @blinkov Thanks for the comments. https://github.com/yandex/ClickHouse/commit/64bdb2a4040a6a10ce311d16e8553dcfc210cdf3 | 2018-12-05T11:41:12 | 2018-12-05T11:41:12 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 444,490,724 | 3,758 | healiseu | @alex-zaitsev I have used a SETTINGS directive in CREATE TABLE e.g.
``` sql
CREATE TABLE Test (
dim3 UInt16,
dim2 UInt16,
dim1 UInt16,
val Float32,
sel UInt8)
ENGINE = MergeTree()
PARTITION BY (dim3, dim2)
ORDER BY (dim3, dim2, dim1)
SETTINGS old_parts_lifetime = 120
''');
... | 2018-12-05T13:48:07 | 2018-12-05T13:48:07 | {
"+1": 3
} | NONE |
ClickHouse | ClickHouse | 444,920,839 | 2,601 | filimonov | > This is obsolete statement. I have removed it from the docs.
> Sometimes it does make sense to use PREWHERE even for indexed columns.
Index allows to select ranges of data (granules that contain index_granularity number of rows) that can match the condition.
> possibility to read ranges less than one granule was i... | 2018-12-06T15:54:42 | 2018-12-06T16:45:35 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 445,184,175 | 3,617 | chenxing-xc | > @chenxing-xc Where did you get libhdfs3? There are a lot of forks on github and we would like to have commits history in our Clickhouse-Extras repo.
I got it from [https://github.com/Pivotal-DataFabric/attic-libhdfs3.git] | 2018-12-07T10:06:57 | 2018-12-07T10:06:57 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 445,327,402 | 3,787 | alexey-milovidov | Yes.
clickhouse-client doesn't need and doesn't use any clickhouse-server other than the server it is connected to. | 2018-12-07T18:45:50 | 2018-12-07T18:45:50 | {} | MEMBER |
ClickHouse | ClickHouse | 445,360,846 | 3,695 | alexey-milovidov | > I don't know which approach is better sleep or this cycle.
With current approach we have race condition.
```
while true; do
timeout 1 bash -c 'cat < /dev/null > /dev/tcp/localhost/9000' 2>1 /dev/null;
rv=$?
if [[ $rv = 0 ]]; then
break
fi
done
```
With this approach we have po... | 2018-12-07T20:47:51 | 2018-12-07T20:48:17 | {} | MEMBER |
ClickHouse | ClickHouse | 445,361,476 | 3,592 | alexey-milovidov | @abyss7 is doing refactoring of `StorageDistributedShardsOptimizer` right now.
We are going to merge at next week. | 2018-12-07T20:50:28 | 2018-12-07T20:50:28 | {} | MEMBER |
ClickHouse | ClickHouse | 445,482,284 | 3,791 | zhang2014 | I don't check the final web page in local development environment : (
Because I didn't build it successfully. | 2018-12-08T19:05:43 | 2018-12-08T19:07:11 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 445,576,361 | 2,684 | gajus | @filimonov Is this issue specific to the `IN` operator? | 2018-12-09T22:07:36 | 2018-12-09T22:07:36 | {} | NONE |
ClickHouse | ClickHouse | 445,740,799 | 3,569 | amosbird | Isn't that deliberately allowed? https://github.com/yandex/ClickHouse/blob/master/dbms/src/IO/ReadHelpers.h#L304 | 2018-12-10T09:04:46 | 2018-12-10T09:04:46 | {} | COLLABORATOR |
ClickHouse | ClickHouse | 445,851,734 | 3,790 | healiseu | Hi @filimonov thanks for the reply, I was a bit confused with the session/user/global settings. I have edited users.xml for my user and I confirm that I was able to get query log for the last N queries. It also seems the type setting I used in the query I wrote above is the most appropriate in my case. I am copying th... | 2018-12-10T15:17:22 | 2018-12-10T15:55:40 | {} | NONE |
ClickHouse | ClickHouse | 444,743,313 | 3,764 | shangshujie365 | somehow, if these 2 lines are in one block, like following, the result is right,
┌─gender─┬─platform─┬─count()─┐
│ M │ IPHONE │ Num1 │
│ F │ IPHONE │ Num2 │
└────────┴──────────┴─────────┘
if these 2 lines are in different blocks, the result is wrong, like following:
┌─gender─┬─platform─┬─count()─... | 2018-12-06T04:19:42 | 2018-12-06T04:19:42 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 444,854,829 | 3,715 | blinkov | fixed in #3763 | 2018-12-06T12:27:26 | 2018-12-06T12:27:26 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 445,150,389 | 3,711 | blinkov | Alias added in #3774, but `position` don't support startPos optional arg at the moment. | 2018-12-07T07:46:59 | 2018-12-07T07:46:59 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 445,350,744 | 2,427 | alexey-milovidov | Superseded by #3759. | 2018-12-07T20:08:31 | 2018-12-07T20:08:31 | {} | MEMBER |
ClickHouse | ClickHouse | 445,736,224 | 3,730 | 4ertus2 | There's only quantileExact supported for now. We probably need quantile and quantileTDigest as well. | 2018-12-10T08:48:38 | 2018-12-10T08:48:38 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 445,744,767 | 3,780 | ghost | I want to run clickhouse in a docker container for testing, so that's not an option. | 2018-12-10T09:18:22 | 2018-12-10T09:18:22 | {} | NONE |
ClickHouse | ClickHouse | 445,752,665 | 3,790 | filimonov | 1. Use `show processlist`, or `select * from system.processes`
2. `log_queries` is a user/session level setting. You can enable it in a session by sending `set log_queries = 1`; or for the particular user by editing `users.xml` section; or for default user profile if you want to enable that globally. | 2018-12-10T09:44:41 | 2018-12-10T09:44:41 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 445,786,398 | 3,611 | vglocus | Having the same issue.
On replica leader
`OPTIMIZE TABLE fake_table_1d_local_10 FINAL;
Received exception from server (version 18.14.9):
Code: 49. DB::Exception: Received from localhost:9000, 127.0.0.1. DB::Exception: Unknown log entry type: 0.`
As well as on non leader
`OPTIMIZE TABLE fake_table_1d_local_10 ... | 2018-12-10T11:38:35 | 2018-12-10T11:45:26 | {} | NONE |
ClickHouse | ClickHouse | 445,834,441 | 3,540 | Perlovka | Additionally, if you have single IN parameter, query is passed to mysql, but with wrong syntax (no parentheses for IN clause):
ClickHouse server version 18.14.15 revision 54409
CH:
```
SELECT * FROM metrics.advertisers_list WHERE advertiser_name IN ('XXX') OR campaign_name IN ('XXX')
```
MySQL:
```
SELEC... | 2018-12-10T14:29:41 | 2018-12-10T14:29:41 | {} | NONE |
ClickHouse | ClickHouse | 444,444,917 | 3,758 | alex-zaitsev | It is controlled by merge tree setting old_parts_lifetime. Default is 480s, or 8 mins. You can change it at server level or per table. | 2018-12-05T10:57:13 | 2018-12-05T10:57:13 | {
"+1": 3
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 444,868,039 | 3,767 | ketbra | I just tried putting NULL within the list of values and it also produces the wrong results. For the example in my post, the following query produces two rows where it should produce none.
SELECT value FROM null_test WHERE value NOT IN (NULL);
<pre>
┌─value────┐
│ A String │
│ ᴺᵁᴸᴸ │
└──────────┘
</pre>
| 2018-12-06T13:19:54 | 2018-12-06T13:19:54 | {} | NONE |
ClickHouse | ClickHouse | 445,629,691 | 3,795 | sundy-li | LGTM | 2018-12-10T02:16:59 | 2018-12-10T02:16:59 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 445,769,414 | 2,684 | filimonov | > @filimonov Is this issue specific to the IN operator?
In spring 2018 there were few issues connected with using (or rather "not using") primary key for IN operators. They are fixed right now (dig the changelog somewhere in March-May 2018 for details). | 2018-12-10T10:39:46 | 2018-12-10T10:39:46 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 445,780,615 | 3,797 | gajus | Thank you. I will report back what solution we ended up using. | 2018-12-10T11:15:45 | 2018-12-10T11:15:45 | {} | NONE |
ClickHouse | ClickHouse | 445,753,843 | 3,790 | filimonov | > I am trying to understand what is this type field
Please check: https://github.com/yandex/ClickHouse/issues/847#issuecomment-306013716
> ... how the following settings affect the logged queries ?
```sql
SELECT *
FROM system.settings
WHERE name LIKE '%log%'
┌─name──────────────────────┬─value──┬─c... | 2018-12-10T09:48:37 | 2018-12-10T12:28:47 | {
"+1": 2
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 445,912,111 | 3,806 | alexey-milovidov | > greps changed to clickhouse extract-from-config
format_schema_path added
Great! Thank you! | 2018-12-10T18:03:00 | 2018-12-10T18:03:00 | {} | MEMBER |
ClickHouse | ClickHouse | 444,770,263 | 3,764 | shangshujie365 | In line 62 of CubeBlockInputStream.cpp,
source_block = children[0]->read();
CH get a block from its underlying stream, and do some mask on it, but we can't guarantee all related data (with the same group by key) is in one block, maybe there are too many rows or some related data crossed blocks. | 2018-12-06T07:06:45 | 2018-12-06T07:12:52 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 445,071,457 | 1,582 | simPod | Using the same as @muravjov . I suppose it finds max value for `ts` and returns the column `size` in the same row as the found `ts`. | 2018-12-06T23:44:52 | 2018-12-06T23:44:52 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 445,759,976 | 3,797 | filimonov | There is no 'officially recommended' way here, different companies use different products, some of those integrations are open-sourced, some not, not all of them are mature enough. ClickHouse developers do not test/certify those solutions.
Please check
https://clickhouse.yandex/docs/en/interfaces/third-party/integ... | 2018-12-10T10:07:55 | 2018-12-10T10:28:18 | {
"+1": 4
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 445,229,430 | 3,782 | simPod | Kk, cool. But my MergeTree is not replicated and only thing I had to do was `rm mutation_xxx.txt` and then wait few minutes, no need for detaching table. | 2018-12-07T13:15:38 | 2018-12-07T13:15:38 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 445,420,558 | 3,789 | viongpanzi | Try `truncate table default.MyTable`?
More details: [TRUNCATE](https://clickhouse.yandex/docs/en/query_language/misc/#truncate) | 2018-12-08T01:58:02 | 2018-12-08T01:58:02 | {
"+1": 2
} | NONE |
ClickHouse | ClickHouse | 445,485,704 | 3,791 | blinkov | Looks fine in test environment, on it's way to production website. | 2018-12-08T19:58:58 | 2018-12-08T19:58:58 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 445,770,562 | 2,684 | filimonov | And 'key-value' usage doesn't suit ClickHouse well independently of using IN operator. ClickHouse does the best in processing ranges of data (like do some daily aggregations), not in single row access. | 2018-12-10T10:43:05 | 2018-12-10T10:43:05 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 445,780,896 | 3,695 | filimonov | > As alternative port readiness check for the client can be performed in something like this:
>
> ```cat < /dev/null > /dev/tcp/localhost/9000```
That will leave "warning" in the log file, something like that:
```
2018.12.10 12:08:49.548198 [ 25 ] {} <Trace> TCPHandlerFactory: TCP Request. Address: [::1]:58626
... | 2018-12-10T11:16:51 | 2018-12-10T12:16:01 | {} | CONTRIBUTOR |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.