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
420,014,454
3,091
alesapin
Test passing `00290_shard_aggregation_memory_efficient` locally.
2018-09-10T18:30:54
2018-09-10T18:31:11
{}
MEMBER
ClickHouse
ClickHouse
420,178,103
3,096
blinkov
You should figure out which of those queries will be more frequent. For larger periods people usually use partitioning by date/week/month, so ClickHouse reads only partitions that are needed, and sort data inside partitions by something else that is in filters often. But partitioning by small periods of time might lead...
2018-09-11T07:38:35
2018-09-11T07:38:35
{}
CONTRIBUTOR
ClickHouse
ClickHouse
420,592,400
3,111
zhang2014
thank you, view predicate optimization has a logical error, and I'm fixing that : )
2018-09-12T10:15:12
2018-09-12T10:15:12
{ "+1": 1 }
CONTRIBUTOR
ClickHouse
ClickHouse
420,801,807
3,091
alexey-milovidov
It's `std::exception`, it doesn't have stack trace. Note, that we can enable stack traces even for these exceptions by hacking C++ ABI functions, like it is done in Facebook's "folly" library: https://github.com/facebook/folly/blob/3af92dbe6849c4892a1fe1f9366306a2f5cbe6a0/folly/experimental/exception_tracer/Exceptio...
2018-09-12T21:20:10
2018-09-12T21:20:10
{}
MEMBER
ClickHouse
ClickHouse
420,802,225
3,111
alexey-milovidov
We will try to prepare the next release at the end of this week or on next week.
2018-09-12T21:21:37
2018-09-12T21:21:37
{ "hooray": 1 }
MEMBER
ClickHouse
ClickHouse
421,077,400
3,124
alesapin
Fix race condition, thanks to failed test.
2018-09-13T16:56:47
2018-09-13T16:56:47
{}
MEMBER
ClickHouse
ClickHouse
419,570,101
3,078
alexey-milovidov
Ok. As far as I have figured out, `capnpc` is Cap'n'Proto compiler and `capnp` is common Cap'n'Proto library.
2018-09-07T21:26:09
2018-09-07T21:26:09
{}
MEMBER
ClickHouse
ClickHouse
419,671,546
2,922
alexey-milovidov
The release job has been repaired and now testing releases are available at https://repo.yandex.ru/clickhouse/deb/testing/main/
2018-09-08T20:43:16
2018-09-08T20:43:16
{}
MEMBER
ClickHouse
ClickHouse
419,722,079
2,763
thesurlydev
I ran into the same issue until I added the key (without Ansible). Should instructions from @Tigro be added to the quick start page? https://clickhouse.yandex/#quick-start
2018-09-09T15:04:50
2018-09-09T15:04:50
{}
NONE
ClickHouse
ClickHouse
419,882,651
3,086
ztlpn
This is a duplicate of #2900 The bug is fixed in master and the fix will be released shortly.
2018-09-10T11:35:46
2018-09-10T11:35:46
{ "+1": 1 }
CONTRIBUTOR
ClickHouse
ClickHouse
420,622,852
3,108
alesapin
Stupid question: Does file `/var/lib/…/instantclient_12_2/lib/libsqora.so.12.1` exist on file system? Can you provide your version of unixodbc, `/etc/odbc.ini` and `/etc/odbcinst.ini`?
2018-09-12T12:02:43
2018-09-12T12:19:23
{}
MEMBER
ClickHouse
ClickHouse
420,658,776
873
sartor
Is there any progress for standard join syntax?
2018-09-12T13:59:44
2018-09-12T13:59:44
{}
NONE
ClickHouse
ClickHouse
420,782,291
3,106
dennissindibad
Uz už IX rrr tě d už Kč už už rrgrr už už Je IX rd R už už RC Už už i g IX úzce I RC r už rrr RC rty v IX rrrr dá rR už f DVĚ R EX CF R IT YE IT Z I TAX R,OR IX RRGG IF Rr dá gttrr dvě rrrrr RC r Odesláno ze smartphonu Sony Xperia™ ---- alexey-milovidov napsal/a ---- Merged #3106<https://github.com/yandex/Click...
2018-09-12T20:15:38
2018-09-12T20:15:38
{}
NONE
ClickHouse
ClickHouse
420,842,959
3,100
AlexanderJLiu
Thanks all of you, I've got it. 💯
2018-09-13T00:32:41
2018-09-13T00:32:41
{}
NONE
ClickHouse
ClickHouse
420,900,594
873
blinkov
@zhang2014 syntax and execution strategies are separate stories. You can write multi-way join even right now, but it requires explicit additional subqueries with two-way joins of inner subquery and Nth table. Execution improvements are also planned, but in previous comment I meant only syntax.
2018-09-13T06:41:05
2018-09-13T06:41:41
{}
CONTRIBUTOR
ClickHouse
ClickHouse
421,005,246
3,108
EvgenyVinogradov
Thank you! I find misconfiguration. I have added library path to ld.so.conf and CH started successfully
2018-09-13T13:23:54
2018-09-13T13:23:54
{}
NONE
ClickHouse
ClickHouse
419,496,561
3,061
sjoerdmulder
Yes I understand, its more in the scenario when using the UUIDStringToNum(str) and UUIDNumToString(str) with FixedString(16) to mimic the new UUID type. Op vr 7 sep. 2018 18:33 schreef Artem Zuikov <notifications@github.com>: > Nope. If you use FixedString representation than it would be stored as is. > It should use...
2018-09-07T16:37:53
2018-09-07T16:37:53
{}
CONTRIBUTOR
ClickHouse
ClickHouse
419,499,143
3,061
4ertus2
#3074 P.S. It based on changes from #3071
2018-09-07T16:46:29
2018-09-07T16:46:29
{}
CONTRIBUTOR
ClickHouse
ClickHouse
420,477,177
3,105
alexey-milovidov
``` SELECT * FROM (SELECT 1) WHERE `1` = 1; ``` It's Ok to have no support for this case. It's also Ok to throw exception when optimization is enabled. Reason: automatic names of columns with complex expressions is an implementation specific behaviour. For example, today when you type `SELECT x + y`, the colum...
2018-09-12T01:08:07
2018-09-12T01:08:07
{}
MEMBER
ClickHouse
ClickHouse
420,783,825
3,091
alesapin
Don't know how to reproduce error, locally this test is always ok. Stack trace from Autotests: ``` 2018.09.12 15:44:25.743511 [ 24 ] {0429c40d-ac08-47f8-bc80-1cee82703c26} <Error> executeQuery: Code: 1001, e.displayText() = DB::Exception: Received from 127.0.0.3:9000. DB::Exception: _Map_base::at. Stack trace: 0. ...
2018-09-12T20:20:46
2018-09-12T20:20:46
{}
MEMBER
ClickHouse
ClickHouse
420,704,375
3,111
TimonKK
Thank you too. Could you tell me, when I can get this fix in stable?
2018-09-12T16:04:11
2018-09-12T16:04:11
{}
NONE
ClickHouse
ClickHouse
420,954,660
1,067
blinkov
@ramazanpolat do you have any further questions?
2018-09-13T10:07:16
2018-09-13T10:07:16
{}
CONTRIBUTOR
ClickHouse
ClickHouse
420,958,014
1,821
blinkov
Looks like you have successfully went through this.
2018-09-13T10:20:15
2018-09-13T10:20:15
{}
CONTRIBUTOR
ClickHouse
ClickHouse
419,819,601
3,061
4ertus2
fixed in master
2018-09-10T07:47:36
2018-09-10T07:47:36
{}
CONTRIBUTOR
ClickHouse
ClickHouse
419,849,788
3,016
lfundaro
Thank you @alexey-milovidov . Looking forward to see the addition of _total time spent in queries execution_
2018-09-10T09:31:37
2018-09-10T09:31:37
{}
NONE
ClickHouse
ClickHouse
420,406,329
3,105
alexey-milovidov
@zhang2014 FYI
2018-09-11T20:04:11
2018-09-11T20:04:11
{}
MEMBER
ClickHouse
ClickHouse
420,548,766
1,464
jsnow42
I have exactly the same issue. Please reopen it.
2018-09-12T07:52:05
2018-09-12T07:52:05
{}
NONE
ClickHouse
ClickHouse
420,844,954
873
zhang2014
By the way, does this task introduce a cost model ? or how do you determine the access path for the base table ?
2018-09-13T00:47:00
2018-09-13T00:47:00
{}
CONTRIBUTOR
ClickHouse
ClickHouse
419,573,568
2,881
alexey-milovidov
Looks all right.
2018-09-07T21:42:41
2018-09-07T21:42:41
{}
MEMBER
ClickHouse
ClickHouse
419,837,322
3,081
4ertus2
I've added this test in regular checks.
2018-09-10T08:50:32
2018-09-10T08:50:32
{}
CONTRIBUTOR
ClickHouse
ClickHouse
419,869,850
3,082
heyciao
Great, very useful to know, thanks a lot!
2018-09-10T10:43:59
2018-09-10T10:43:59
{}
NONE
ClickHouse
ClickHouse
419,917,654
2,682
alexey-milovidov
Now you can also use `--send_logs_level=trace` option for clickhouse-client. It is available in master.
2018-09-10T13:46:39
2018-09-10T13:46:39
{}
MEMBER
ClickHouse
ClickHouse
419,958,608
2,958
alexey-milovidov
The fix https://github.com/yandex/ClickHouse/commit/5a096ee0992ce139b4982a6edbade5a5c9beca3b is from 18.12.5. I recommend to install latest testing, currently it is 18.12.11.
2018-09-10T15:38:58
2018-09-10T15:38:58
{ "+1": 1 }
MEMBER
ClickHouse
ClickHouse
420,339,047
3,096
champtar
you can also use something like `toStartOfMinute(datetime)` in primary key ``` CREATE TABLE IF NOT EXISTS thetable ( date Date, dateTime DateTime, accountId String, appId String, hostname String, cpuUsed UInt32, memoryUsed UInt32 ) ENGINE = MergeTree PARTITION BY date ORDER BY (accountId, appId, date, ...
2018-09-11T16:40:28
2018-09-11T16:40:28
{}
CONTRIBUTOR
ClickHouse
ClickHouse
420,509,914
3,097
alexey-milovidov
:+1:
2018-09-12T04:41:14
2018-09-12T04:41:14
{}
MEMBER
ClickHouse
ClickHouse
420,961,466
856
blinkov
@bobrik do you have any further questions?
2018-09-13T10:34:04
2018-09-13T10:34:04
{}
CONTRIBUTOR
ClickHouse
ClickHouse
419,495,252
3,061
4ertus2
Nope. If you use FixedString representation than it would be stored as is. It should use 128 bit only after conversion to UUID.
2018-09-07T16:32:58
2018-09-07T16:32:58
{}
CONTRIBUTOR
ClickHouse
ClickHouse
419,887,225
3,088
sarg
@Chenjp, in fact there is `today()` function which is the same as `toDate(now())` according to official docs. But the point is that the way clickhouse handles this situation violates principle of least astonishment. Pg and mysql correctly truncates the datetime to fit into date column.
2018-09-10T11:57:07
2018-09-10T11:57:26
{}
NONE
ClickHouse
ClickHouse
419,892,179
188
VladRassokhin
Also `Ctrl^U` helps
2018-09-10T12:16:48
2018-09-10T12:16:48
{}
CONTRIBUTOR
ClickHouse
ClickHouse
420,261,465
3,100
filimonov
Compression ratio depends on the nature of the data. If you have almost random column (for example we have one column with md5) compression ratio is very bad for it (like 99.9%). If you have 'mostly constant' column compression ratio for it would be really good - close 0%. For 'regular' data it's usually it's abou...
2018-09-11T12:46:45
2018-09-11T12:46:45
{}
CONTRIBUTOR
ClickHouse
ClickHouse
420,281,287
3,096
fizerkhan
@blinkov - you had mentioned `For larger periods people usually use partitioning by date/week/month` How could we do that? I assumed that Clickhouse only allows partitioning by `month`
2018-09-11T13:49:40
2018-09-11T13:49:40
{}
NONE
ClickHouse
ClickHouse
420,612,490
3,111
zhang2014
fixed in https://github.com/yandex/ClickHouse/pull/3107/commits/e6e376e86afd9121f48cd72ab84637347fe91e6b
2018-09-12T11:20:14
2018-09-12T11:20:14
{ "+1": 1 }
CONTRIBUTOR
ClickHouse
ClickHouse
420,631,196
3,108
EvgenyVinogradov
Yes, sure library file exists and it works on 18.5.1 without any ODBC reconfiguration. root@clickhouse-1:logs# apt-cache madison unixodbc unixodbc | 2.3.4-1 | http://deb.debian.org/debian stretch/main amd64 Packages cat /etc/odbc.ini [customers] Driver=/usr/local/lib/libmaodbc.so Server=mysql-slave-rep...
2018-09-12T12:33:30
2018-09-12T12:33:30
{}
NONE
ClickHouse
ClickHouse
420,713,193
873
simPod
Do you know it is possible to multi join already, right? Queries are just a bit ugly but it works.
2018-09-12T16:30:30
2018-09-12T16:30:30
{}
CONTRIBUTOR
ClickHouse
ClickHouse
420,871,431
3,109
alexey-milovidov
@ztlpn Clarification is needed.
2018-09-13T03:27:15
2018-09-13T03:27:15
{}
MEMBER
ClickHouse
ClickHouse
419,627,727
2,014
Slind14
At this point one could also consider adding a function for a T-Test (mean, std. dev., count, confidence) as well as T-Test(x, confidence)
2018-09-08T09:45:09
2018-09-08T09:45:35
{}
NONE
ClickHouse
ClickHouse
419,815,810
3,054
igor-sh8
Hi, Alex. This is great news. Thanks.
2018-09-10T07:31:25
2018-09-10T07:31:25
{}
NONE
ClickHouse
ClickHouse
419,827,183
3,081
4ertus2
Can't reproduce on master. > clickhouse-client -q "select toString(toUUID('3f1ed72e-f7fe-4459-9cbe-95fe9298f845'))" > 3f1ed72e-f7fe-4459-9cbe-95fe9298f845
2018-09-10T08:16:41
2018-09-10T08:16:41
{}
CONTRIBUTOR
ClickHouse
ClickHouse
419,971,959
3,088
Chenjp
@sarg According to official doc, `DateTime` stored in 4-bytes indicates unix-timestamp while `Date` in 2-bytes indicates days from 1970-01-01. Maybe implicit convertion between `Date` and `DateTime` is a new feature / engine enhancement request. Currently maxinum supported `Date` is 2105, extended date range support i...
2018-09-10T16:17:49
2018-09-10T16:17:49
{}
NONE
ClickHouse
ClickHouse
420,323,816
3,096
filimonov
> for e.g. when we have it grouped by month - and if we query without hostname then it would end up parsing the whole set of rows.. (if hostname is behind timestamp). Is this assumption correct? if you have not too many hostnames, and each group defined by tuple (`accountId, appId, hostname`) will still have >100K (...
2018-09-11T15:54:03
2018-09-11T15:54:28
{}
CONTRIBUTOR
ClickHouse
ClickHouse
420,837,834
3,004
reflection
Any movement on this issue? Also, probably related, simple `WHERE <low cardinality string column> IN (<list of strings>)` doesn't work either. ``` SET allow_experimental_low_cardinality_type = 1; CREATE TABLE lc_test2 ( ts DateTime, hero LowCardinality(String) ) ENGINE = MergeTree PARTITION BY toYYYYMMD...
2018-09-13T00:01:38
2018-09-13T00:01:38
{}
CONTRIBUTOR
ClickHouse
ClickHouse
420,957,549
1,826
blinkov
В последние недели ODBC драйвер был сильно доработан, стоит попробовать свежие версии.
2018-09-13T10:18:31
2018-09-13T10:18:31
{}
CONTRIBUTOR
ClickHouse
ClickHouse
420,972,165
3,109
ztlpn
We need to get rid of unnecessary quoting.
2018-09-13T11:17:45
2018-09-13T11:17:45
{}
CONTRIBUTOR
ClickHouse
ClickHouse
420,973,822
3,112
ztlpn
Is the connection to host 192.168.1.24 port 8123 possible from the host where you run your application? What does `curl http://192.168.1.24:8123/` say?
2018-09-13T11:25:30
2018-09-13T11:25:30
{}
CONTRIBUTOR
ClickHouse
ClickHouse
420,985,042
3,107
zhang2014
@alexey-milovidov Done. BTW, sorry for my negligence.
2018-09-13T12:13:52
2018-09-13T12:16:23
{}
CONTRIBUTOR
ClickHouse
ClickHouse
421,019,228
3,091
alesapin
something strange with PR commits
2018-09-13T14:05:03
2018-09-13T14:05:03
{}
MEMBER
ClickHouse
ClickHouse
421,036,953
3,110
ztlpn
Which ClickHouse version do you use?
2018-09-13T14:55:27
2018-09-13T14:55:27
{}
CONTRIBUTOR
ClickHouse
ClickHouse
419,697,516
3,082
filimonov
>I'm doing some experiments with indexing of data; I want to check how the amount of data and/or the order or the columns impact the singlecolumn&total compression ratio of the data. 1) are you aware about existence of `system.parts` , `system.columns`, `system.parts_columns`? They have different metadata, including...
2018-09-09T07:49:01
2018-09-09T07:49:01
{ "+1": 1 }
CONTRIBUTOR
ClickHouse
ClickHouse
419,713,404
3,082
heyciao
Thank you, no I wasn't aware of "system.parts_columns", great! I will use the sum of the bytes directly from that table when I test changes to the order of columns or other stuff :) . On the other hand, if I compare "sum(column_data_compressed_bytes)" of a specific column and partition against the compressed size of...
2018-09-09T12:44:36
2018-09-09T19:17:19
{}
NONE
ClickHouse
ClickHouse
419,774,928
3,082
alexey-milovidov
ClickHouse use zstd compression level 1 by default. You can change it by adding `<level>3</level>` in `<case>` near `<method>`. And yes, ClickHouse use smaller compressed blocks by default (from 64 KB to 1 MB, depending on your data and index granularity). This is controlled by `min_compress_block_size` and ...
2018-09-10T03:08:59
2018-09-10T03:08:59
{}
MEMBER
ClickHouse
ClickHouse
420,173,905
3,096
blinkov
By time range you mean something narrow like few minutes? Or more like "last day/week/month"?
2018-09-11T07:22:18
2018-09-11T07:22:18
{}
CONTRIBUTOR
ClickHouse
ClickHouse
420,509,638
3,099
alexey-milovidov
Consider adding `partition_id` also to `part_log`.
2018-09-12T04:39:08
2018-09-12T04:39:08
{}
MEMBER
ClickHouse
ClickHouse
420,700,389
873
sartor
Nice to here it. I think this is last important feature, that prevents migration to ClickHouse from traditional column DBs. Thank you!
2018-09-12T15:52:56
2018-09-12T15:55:00
{}
NONE
ClickHouse
ClickHouse
420,870,300
3,115
alexey-milovidov
You can have a column like `uuid DEFAULT generateUUIDv4()`
2018-09-13T03:19:09
2018-09-13T03:19:09
{ "+1": 2 }
MEMBER
ClickHouse
ClickHouse
421,044,381
3,110
alex-zaitsev
I confirm the issue exist at latest 18.12.13, but with slightly different output: ``` select sum(number1) from merge('test','distributed_table_2|distributed_table_1'); SELECT sum(number1) FROM merge('test', 'distributed_table_2|distributed_table_1') ┌─sum(number1)─┐ │ 100 │ └──────────────┘ ┌─s...
2018-09-13T15:16:13
2018-09-13T15:16:13
{}
CONTRIBUTOR
ClickHouse
ClickHouse
419,886,255
1,118
akonyaev90
is the task dead? any news about explain query plan?
2018-09-10T11:52:41
2018-09-10T11:52:41
{}
CONTRIBUTOR
ClickHouse
ClickHouse
420,174,774
3,096
fizerkhan
Yes, we will narrow down from minutes/days/weeks/months.
2018-09-11T07:25:28
2018-09-11T07:25:28
{}
NONE
ClickHouse
ClickHouse
420,262,757
3,100
ztlpn
The comment in config.xml refers to the zstd compression algorithm which achieves better compression ratios than the default lz4 algorithm, but is also more CPU-intensive. You can experiment with it and see if you are ok with the increased CPU load.
2018-09-11T12:51:16
2018-09-11T12:51:16
{}
CONTRIBUTOR
ClickHouse
ClickHouse
420,279,203
3,096
fizerkhan
@filimonov - Thanks. One question - We would usually query only using `accountId`, `appId` and `time`. There would of course be times when we query along with hostname (+`account id`, `appId` and `time`). Since the former was the most often called query (without `hostname`), we assumed that if we added `hostname` befor...
2018-09-11T13:43:06
2018-09-11T13:49:37
{}
NONE
ClickHouse
ClickHouse
420,574,101
3,096
fizerkhan
@filimonov Thanks. We have two databases: 1. In one database, we have approx 700,000 rows in a month per host. We might have around 5,000 hosts. so it will be 3500,000,000 rows in a month. And almost every select call will either have hostname in `where` statement or in the `group by` statement. So in this case, I a...
2018-09-12T09:15:41
2018-09-12T10:04:02
{}
NONE
ClickHouse
ClickHouse
421,006,381
3,121
alexey-milovidov
Спасибо! Смешная опечатка.
2018-09-13T13:27:20
2018-09-13T13:27:20
{}
MEMBER
ClickHouse
ClickHouse
419,872,466
3,088
Chenjp
select now() as nowts,* from tst_date | nowts | ts | val | 3 | | 2018-09-10 18:52:26 | 2023-09-08 | 1 | |
2018-09-10T10:54:48
2018-09-10T10:54:48
{}
NONE
ClickHouse
ClickHouse
419,956,789
1,118
blinkov
@akonyaev90 not dead, just in backlog, not currently in progress.
2018-09-10T15:33:47
2018-09-10T15:33:47
{}
CONTRIBUTOR
ClickHouse
ClickHouse
420,640,412
3,108
alesapin
In ClickHouse 18.12.13 we have changed interaction with ODBC managers. Previously .so libraries were linked with clickhouse-server binary, now they link with clickhouse-odbc-bridge binary. `file not found` exception doesn't mean, that the file doesn't exist on fs. It may mean that shared object cannot be loaded. Pl...
2018-09-12T13:04:31
2018-09-12T13:04:31
{}
MEMBER
ClickHouse
ClickHouse
420,836,782
3,117
alexey-milovidov
Ok. > Hack in code :(. But changes are very small. The only good solution is to split MergeTreeBlockInputStream, as said in comment https://github.com/yandex/ClickHouse/blob/master/dbms/src/Storages/MergeTree/MergeTreeBlockInputStream.h#L15. Note that the methods `createReadBufferFromFileBase` and `createWriteBuf...
2018-09-12T23:55:44
2018-09-12T23:55:44
{}
MEMBER
ClickHouse
ClickHouse
420,870,057
3,004
alexey-milovidov
LowCardinality data type is in active development right now. Both of the mentioned issues are in todo.
2018-09-13T03:17:24
2018-09-13T03:17:24
{ "+1": 1 }
MEMBER
ClickHouse
ClickHouse
419,680,271
3,075
alexey-milovidov
Имеет смысл выставить более сильные ограничения на потребление памяти запросами: max_memory_usage - например, в 1 GB max_memory_usage_for_user - например, в 4 GB max_memory_usage_for_all_queries - например, в 4 GB https://clickhouse.yandex/docs/ru/operations/settings/query_complexity/#max_memory_usage Ограни...
2018-09-08T23:43:51
2018-09-08T23:43:51
{}
MEMBER
ClickHouse
ClickHouse
419,879,172
3,088
Chenjp
@sarg Datatype of function now() result is `DateTime`. So we can just call like this: **insert into tst_datetime values (cast(now() as Date),now(), 1) ;**
2018-09-10T11:19:41
2018-09-10T11:19:41
{}
NONE
ClickHouse
ClickHouse
419,916,405
188
alexey-milovidov
Fixed in master. Ctrl+C started to work as expected.
2018-09-10T13:42:40
2018-09-10T13:42:54
{ "hooray": 1 }
MEMBER
ClickHouse
ClickHouse
420,229,759
2,770
alex-krash
Thanks, @alexey-milovidov , will apply changes.
2018-09-11T10:42:34
2018-09-11T10:42:34
{}
CONTRIBUTOR
ClickHouse
ClickHouse
420,268,370
3,096
filimonov
Did you check that: https://clickhouse.yandex/docs/en/operations/table_engines/mergetree/#primary-keys-and-indexes-in-queries ? 1) `(accountId, appId, hostname, dateTime)` that option will probably give a better compression ratio and would be effective if you mostly always use conditions for `accountId, appId, hos...
2018-09-11T13:09:18
2018-09-11T13:09:18
{}
CONTRIBUTOR
ClickHouse
ClickHouse
420,473,313
3,105
zhang2014
Two questions, I will fix them as soon as possible. - Should the ArrayJoin push down? - Support to https://github.com/yandex/ClickHouse/blob/master/dbms/tests/queries/0_stateless/00597_push_down_predicate.sql#L15
2018-09-12T00:44:41
2018-09-12T00:44:41
{}
CONTRIBUTOR
ClickHouse
ClickHouse
420,552,524
1,464
blinkov
@OzRamsay could you please create a new one describing what exactly you are doing and which version of Docker image is used?
2018-09-12T08:05:12
2018-09-12T08:05:12
{}
CONTRIBUTOR
ClickHouse
ClickHouse
421,011,120
1,821
alexey-milovidov
Limit on number of connections and throttling of recovery speed is actual task.
2018-09-13T13:41:51
2018-09-13T13:41:51
{}
MEMBER
ClickHouse
ClickHouse
419,503,087
3,054
ztlpn
You will have to add new aggregated columns by hand: * Execute `ALTER ADD COLUMN` * Do `DETACH TABLE Test` * Find the metadata file (usually its path is `/var/lib/clickhouse/metadata/<database>/<table>.sql`) and add the new column to the list of the aggregated columns. * Do `ATTACH TABLE Test`
2018-09-07T17:00:44
2018-09-07T17:00:44
{ "+1": 1, "heart": 2 }
CONTRIBUTOR
ClickHouse
ClickHouse
419,519,506
2,809
simPod
Yup
2018-09-07T18:01:03
2018-09-07T18:01:03
{}
CONTRIBUTOR
ClickHouse
ClickHouse
419,984,170
3,035
alesapin
lgtm
2018-09-10T16:56:07
2018-09-10T16:56:07
{}
MEMBER
ClickHouse
ClickHouse
420,281,584
3,096
blinkov
@fizerkhan https://clickhouse.yandex/docs/en/operations/table_engines/custom_partitioning_key/
2018-09-11T13:50:41
2018-09-11T13:50:41
{}
CONTRIBUTOR
ClickHouse
ClickHouse
419,762,689
2,770
alexey-milovidov
There are four reasons why the code is slow: 1. You append data to a temporary `std::string` object for each row. 2. You create extra temporary `std::string` objects in some cases. 3. `sprintf` function is used (it is extremely slow). 4. `std::function` is slower or about the same than virtual call that is subseque...
2018-09-10T01:32:20
2018-09-10T01:32:51
{}
MEMBER
ClickHouse
ClickHouse
419,876,324
3,088
Chenjp
We need cast now() to Date or DateTime!!! CREATE TABLE tst_datetime ( ts Date, ts2 DateTime, val UInt16 ) ENGINE = MergeTree(ts, val, 8192); insert into tst_datetime values (cast(now() as Date),cast(now() as DateTime), 1) ; select * from tst_datetime; | ts | ts2 | val | 3 | | 2018-09-10 |...
2018-09-10T11:08:38
2018-09-10T11:08:38
{}
NONE
ClickHouse
ClickHouse
419,927,810
3,092
alexey-milovidov
Does not build.
2018-09-10T14:14:21
2018-09-10T14:14:21
{}
MEMBER
ClickHouse
ClickHouse
420,589,319
3,096
fizerkhan
@champtar - you have mentioned about `toStartOfMinute(datetime) in primary key` -> Is there an advantage in doing this? Does it play a role
2018-09-12T10:04:13
2018-09-12T10:04:13
{}
NONE
ClickHouse
ClickHouse
419,878,901
3,082
heyciao
(yes, if I compress manually using level 1 then the sizes match. Cool, I don't have to do the whole export-thing anymore)
2018-09-10T11:18:28
2018-09-10T11:18:28
{}
NONE
ClickHouse
ClickHouse
420,475,395
3,105
alexey-milovidov
> Should the ArrayJoin push down? No. For example, in `ExpressionActions::optimizeArrayJoin` we try to "push" it "up". The reason: arrayJoin usually make data size larger and it becomes more difficult to process. So, we move it as far as possible to the end of query execution pipeline. Another reason: condition p...
2018-09-12T00:57:34
2018-09-12T00:57:50
{}
MEMBER
ClickHouse
ClickHouse
420,557,162
3,109
crakjie
After more checking i see it's possible to query this table like this ``` SELECT partition, database, table FROM system.parts WHERE partition = '\'b710e103-3d07-481e-afc0-1687a8782b73\'' group by partition, database, table ``` So the partition contain quotes while it's do not contains quote in t...
2018-09-12T08:20:53
2018-09-12T08:20:53
{}
NONE
ClickHouse
ClickHouse
420,675,368
167
simPod
I have a question. Shouldn't the UPDATE and DELETE propagate to materialized views as well?
2018-09-12T14:45:01
2018-09-12T14:45:01
{}
CONTRIBUTOR
ClickHouse
ClickHouse
420,708,334
2,908
filimonov
> In my project it's OK that data will be dropped, because it's just become outdated. I'm just telling that defining data lifetime in terms of diskspace can't guarantee how long your data will be in place. I.e. lets assume that you give 100 Gb of disk space for your logs, and currently it's enough to store 7 day...
2018-09-12T16:15:36
2018-09-13T08:12:15
{}
CONTRIBUTOR
ClickHouse
ClickHouse
421,114,747
3,125
alexey-milovidov
Ok.
2018-09-13T18:56:27
2018-09-13T18:56:27
{}
MEMBER
ClickHouse
ClickHouse
420,397,516
2,015
alexey-milovidov
FYI, there is an issue with a query that `clickhouse-client` send to obtain info for autocomplete: ``` SELECT DISTINCT arrayJoin(extractAll(name, '[\\w_]{2,}')) AS res FROM ( SELECT name FROM system.functions UNION ALL SELECT name FROM system.table_engines UNION ALL SELECT na...
2018-09-11T19:38:07
2018-09-11T19:38:07
{}
MEMBER
ClickHouse
ClickHouse
420,690,067
873
blinkov
@sartor There is some progress like basic ON support, removal of most limitations for right side, better asterisk behaviour, ANY/ALL can be configured to be optional. Multiple JOINs per SELECT are still not implemented yet, but they are next in queue of SQL compatibility tasks.
2018-09-12T15:24:35
2018-09-12T15:24:35
{ "+1": 1 }
CONTRIBUTOR