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 | 327,263,463 | 1,197 | KochetovNicolai | @daledude,
As I can understand, your issue is not the same. You have degradation in performance because during reading from table with `Engine = Merge(...)` threads are doing different amount of work.
You can also try to exclude big table from merge and rewrite query with `UNION ALL`: https://clickhouse.yandex/docs... | 2017-09-05T18:31:45 | 2017-09-05T18:31:45 | {} | MEMBER |
ClickHouse | ClickHouse | 327,371,154 | 1,197 | fleapapa | The following tree view might be a clue about why system load was >24. No idea why "set max_threads=10" can result in 20+ threads....
2925 root 20 0 520M 129M 11592 S 0.0 13.1 0:41.22 ├─ clickhouse-server --config=/etc/clickhouse-server/config.xml --daemon
2960 root 20 0 520M 129M 11592 S 0... | 2017-09-06T04:27:50 | 2017-09-06T04:27:50 | {} | NONE |
ClickHouse | ClickHouse | 327,830,053 | 1,144 | hnakamur | Oh, I'm sorry.
I noticed you fixed my mistake at https://github.com/yandex/ClickHouse/commit/17edcdec96b01d988e749f44c4be0194f8546aee
Thanks!
| 2017-09-07T15:11:36 | 2017-09-07T15:11:36 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 330,260,610 | 1,241 | ludv1x | Also, there is a presentation about it https://events.yandex.com/events/meetings/05-10-2017/?talk=247
But it will be presented in upcoming ClickHouse meetup. | 2017-09-18T15:30:11 | 2017-09-18T15:30:11 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 330,271,720 | 1,243 | ludv1x | Confirmed.
v1.1.54289
```
SELECT
toUnixTimestamp('1999-12-31 22:00:01', 'UTC') AS ts,
toDateTime(ts) AS time,
toString(time, 'UTC') AS time_utc,
toString(time, 'Europe/Amsterdam') AS time_amst,
toStartOfYear(time, 'Europe/Amsterdam') AS year_amst,
toStartOfQuarter(time, 'Europe/... | 2017-09-18T16:06:13 | 2017-09-18T16:06:13 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 330,292,213 | 1,241 | silviucpp | But what the use case when are in different tables ? I don't think the use case with all events in the same table is so common. | 2017-09-18T17:19:35 | 2017-09-18T17:31:42 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 330,305,136 | 1,256 | proller | new freebsd port submitted. | 2017-09-18T17:55:53 | 2017-09-18T17:55:53 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 327,065,336 | 1,197 | daledude | I ran into same, or similar, issue. Whether using a single large table (3tb) or a merge of many tables my queries would start off using all CPU and most IO then dwindle down to a single thread and little IO. It's not an issue with the system as increasing max_streams_to_max_threads_ratio uses most, if not all, resource... | 2017-09-05T03:58:30 | 2017-09-05T03:58:30 | {} | NONE |
ClickHouse | ClickHouse | 329,170,496 | 360 | silviucpp | While compiling on OSX I get the following error:
```
Scanning dependencies of target mysqlxx
[ 45%] Building CXX object libs/libmysqlxx/CMakeFiles/mysqlxx.dir/src/Connection.cpp.o
/Users/silviu.caragea/Desktop/ClickHouse/libs/libmysqlxx/src/Connection.cpp:1:25: fatal error: mysql/mysql.h: No such file or directo... | 2017-09-13T13:38:28 | 2017-09-13T13:38:28 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 330,508,137 | 360 | sundy-li | `[100%] Linking CXX executable clickhouse
ld: archive has no table of contents file '../../../libs/libmysqlxx/libmysqlclient.a' for architecture x86_64
collect2: error: ld returned 1 exit status
make[3]: *** [dbms/src/Server/clickhouse] Error 1`
same error using the latest code | 2017-09-19T11:17:16 | 2017-09-19T11:17:16 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 327,025,320 | 1,196 | alexey-milovidov | This is too slow for ClickHouse:
```
SELECT avg(pointInPolygon((x, y), (0., 0.), (1., 100.), (1., 0.), (2., 100.), (2., 0.), (3., 100.), (3., 0.), (4., 100.), (4., 0.), (5., 100.), (5., 0.), (6., 100.), (6., 0.), (7., 100.), (7., 0.), (8., 100.), (8., 0.), (9., 100.), (9., 0.), (10., 100.), (10., 0.), (11., 100.), ... | 2017-09-04T20:36:38 | 2017-09-04T20:36:38 | {} | MEMBER |
ClickHouse | ClickHouse | 330,733,678 | 198 | wangyinsheng | is it solved? | 2017-09-20T03:16:01 | 2017-09-20T03:16:01 | {} | NONE |
ClickHouse | ClickHouse | 327,003,532 | 1,197 | KochetovNicolai | Hi,
You can increase performance by adding both index columns in condition. For example, if you know that `pickup_datetime < dropoff_datetime`, than instead of query `select ... from table where dropoff_datetime > '2013-01-01 00:00:00'` you can use `select ... from table where dropoff_datetime > '2013-01-01 00:00:... | 2017-09-04T17:08:43 | 2017-09-04T17:08:43 | {} | MEMBER |
ClickHouse | ClickHouse | 327,014,606 | 1,197 | fleapapa | The comparison is conducted on the same physical host same h/w and s/w environment, so i/o or any objective environment seems not the main factor of the performance difference. (my point here is, on my host CH should be able to read data from physical disk as fast as Spark does)
I raise max_threads to 4 on my dual c... | 2017-09-04T18:50:08 | 2017-09-05T01:23:17 | {} | NONE |
ClickHouse | ClickHouse | 327,184,455 | 1,197 | fleapapa | 1. the result is 10
2. io was close to 100% on both threads (by default max_threads)
3. when max_threads = 16, io is nearly 100% on 8 threads but very slow on the other 8 thread.
cpu is only ~3% on one thread and nearly 0% on all the other 15 thread.
iotop:
11857 be/4 root 2.16 M/s 0.00 B/s 0.00 % ... | 2017-09-05T14:01:11 | 2017-09-05T14:01:11 | {} | NONE |
ClickHouse | ClickHouse | 327,781,685 | 1,092 | proller | Should be fixed here https://github.com/yandex/ClickHouse/commit/d422777d9cf9a51908d10db11d05f961334b7294 | 2017-09-07T12:15:53 | 2017-09-07T12:15:53 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 327,818,841 | 829 | valyala | There is yet another workaround: just wrap the original `SELECT` query into another `SELECT` and move `ORDER BY` into outer select:
```sql
SELECT
Payload1,
Payload2
FROM (
SELECT
Payload1,
Payload2
FROM foobar_all
WHERE EventTime > now() - 60
)
ORDER BY EventTime
``` | 2017-09-07T14:35:20 | 2017-09-07T14:35:20 | {
"+1": 1
} | NONE |
ClickHouse | ClickHouse | 327,863,099 | 1,205 | KochetovNicolai | Thank you for report. It's true that ip_trie may cause segmentation fault if dictionary has more than 2^20 different ip addresses. Will fix that problem soon. | 2017-09-07T17:06:51 | 2017-09-07T17:06:51 | {} | MEMBER |
ClickHouse | ClickHouse | 328,299,992 | 167 | shahbour | @alexey-milovidov is it still on road map the delete | 2017-09-09T19:54:56 | 2017-09-09T19:54:56 | {} | NONE |
ClickHouse | ClickHouse | 329,053,581 | 1,044 | mkocikowski | I think 2 is cleaner. 3: read password from an environment variable? | 2017-09-13T04:10:44 | 2017-09-13T04:10:44 | {} | NONE |
ClickHouse | ClickHouse | 329,187,278 | 1,225 | silviucpp | Beside my previous questions there is any challenge to support fractional offsets like Asia/Kolkata which is UTC+05:30 ? Or just something that was missed ?
| 2017-09-13T14:33:54 | 2017-09-13T14:33:54 | {
"+1": 3
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 329,335,353 | 1,225 | alexey-milovidov | > for each timezone at this point you create an instance of that timezone (as a singleton of DateLUT) which means if we add seconds as second param we will have lot of singletons created for each offset (timezone names are limited - number of seconds somehow are not) you don't consider this as a problem ?
I think, w... | 2017-09-14T00:35:53 | 2017-09-14T00:35:53 | {} | MEMBER |
ClickHouse | ClickHouse | 330,147,144 | 1,250 | alexey-milovidov | Implementation is rather inefficient, also we lack limit on array sizes and memory tracking. This is Ok for first version (will merge after few modifications). Superior implementation can use something similar to `AggregateFunctionGroupUniqArray`. | 2017-09-18T07:34:43 | 2017-09-18T07:35:15 | {} | MEMBER |
ClickHouse | ClickHouse | 326,987,204 | 1,195 | proller | `Cannot specify include directories for target "mariadbclient" which is not
built by this project.` - delete build/CMakeCache.txt before cmake | 2017-09-04T15:23:32 | 2017-09-04T15:23:32 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 327,333,398 | 1,203 | alexey-milovidov | The test is still incorrect.
```
$ echo $?
28
```
Look at autotests results. | 2017-09-05T23:41:46 | 2017-09-05T23:41:46 | {} | MEMBER |
ClickHouse | ClickHouse | 328,066,414 | 1,216 | innerr | uniqExact, my fault... | 2017-09-08T10:19:10 | 2017-09-08T10:19:10 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 328,244,574 | 1,197 | fleapapa | Any update?
When i redid the test with almost the same environment as previously, except that this time i added 1GB more memory to my Ubuntu VM host, the result got worse! Immediately after i launched Clickhouse server and submitted the first (and the only one) query as below, my VM got completely frozen and no remo... | 2017-09-09T01:06:26 | 2017-09-09T01:22:12 | {} | NONE |
ClickHouse | ClickHouse | 328,245,829 | 1,197 | fleapapa | By the way, when the column _c6 in the WHERE condition was changed to _c5, the query took <1s, so index does help a lot. (Note: in my schema, _c5 is the first column in primary index, _c6 is the second one) The problem is CH doesn't support secondary index. Right? | 2017-09-09T01:28:07 | 2017-09-09T01:28:07 | {} | NONE |
ClickHouse | ClickHouse | 330,059,355 | 253 | arduanov | Decimal is needed to work with financial data, UInt64 is not enough for sum aggregations | 2017-09-17T16:13:13 | 2017-09-17T16:13:13 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 327,007,655 | 1,197 | KochetovNicolai | If CPU cores were under-utilized, there should be bottle neck in something else. It makes sense to measure performance of i/o operations, such as disk and network loading. You can use `iotop` to measure disk and `iftop` to measure network performance during query execution.
You can try to increase the number of thre... | 2017-09-04T17:44:35 | 2017-09-04T17:44:35 | {} | MEMBER |
ClickHouse | ClickHouse | 327,167,255 | 1,200 | proller | with swap your compile time will take days and you have chance to kill sd card.
and now you need ~3.4G ram
I have tried to compile on boards with 2g ram + swap and all attempts are unsuccessful | 2017-09-05T12:58:19 | 2017-09-05T12:58:19 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 328,166,832 | 1,217 | alexey-milovidov | It seems easy to implement. Do you want to try?
Similar pull request (as an example): https://github.com/yandex/ClickHouse/pull/983 | 2017-09-08T17:33:16 | 2017-09-08T17:33:16 | {} | MEMBER |
ClickHouse | ClickHouse | 330,136,974 | 1,253 | alexey-milovidov | Fixed.
Sorry for inconvenience. | 2017-09-18T06:23:34 | 2017-09-18T06:23:34 | {} | MEMBER |
ClickHouse | ClickHouse | 330,506,562 | 1,257 | ludv1x | Hi!
Could you show a backtrace in the saved core file (`gdb /usr/bin/clickhouse /var/lib/clickhouse/cores/core -batch -ex bt`)?
Also, you could recompile ClickHouse with `-DDEBUG_LIBTCMALLOC=1` CMake variable for better future diagnostics. | 2017-09-19T11:09:52 | 2017-09-19T11:09:52 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 330,766,021 | 1,231 | silviucpp | Seems this functionality is already implemented in clickhouse but I don't see any documentation excepting comments in code:
```c++
/**
* The function calculates distance in meters between two points on Earth specified by longitude and latitude in degrees.
* The function uses great circle distance formula http... | 2017-09-20T07:15:15 | 2017-09-20T07:19:57 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 327,002,136 | 1,196 | alexey-milovidov | Worth for testing. | 2017-09-04T16:57:36 | 2017-09-04T16:57:36 | {} | MEMBER |
ClickHouse | ClickHouse | 327,007,578 | 1,195 | proller | Now merged to master. Please update and try again | 2017-09-04T17:44:05 | 2017-09-04T17:44:05 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 327,022,035 | 1,196 | alexey-milovidov | You have to add performance test (Look at `tests/performance`). | 2017-09-04T20:04:06 | 2017-09-04T20:04:06 | {} | MEMBER |
ClickHouse | ClickHouse | 327,139,953 | 1,197 | KochetovNicolai | @fleapapa ,
Could you do the following:
1. run `select count() from system.parts where table = 'your_table_name'`
2. run `iotop` and look at IO column to see if clickhouse threads spent mots of the time on IO operations (I expect it could be close to 100%)
3. increase `max_threads` more (8, 12, 16) and look at ... | 2017-09-05T10:46:18 | 2017-09-05T10:46:18 | {} | MEMBER |
ClickHouse | ClickHouse | 327,692,096 | 1,205 | vavrusa | Hi John, this is a simple binary trie, so it branches on every bit. I have a note to reimplement it with poptries (since the original implementation is encumbered by a restrictive license), but haven't had a time for that yet. | 2017-09-07T05:36:46 | 2017-09-07T05:36:46 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 330,269,046 | 1,233 | alexey-milovidov | Thanks for fixing! I've edited the code a bit. | 2017-09-18T15:58:20 | 2017-09-18T15:58:20 | {} | MEMBER |
ClickHouse | ClickHouse | 330,285,347 | 1,256 | iron-udjin | Thank you for fast reply!
Works. | 2017-09-18T16:54:56 | 2017-09-18T16:54:56 | {} | NONE |
ClickHouse | ClickHouse | 330,373,097 | 1,256 | iron-udjin | Great, thank you! | 2017-09-18T22:23:06 | 2017-09-18T22:23:06 | {} | NONE |
ClickHouse | ClickHouse | 327,021,682 | 1,196 | alexey-milovidov | There is a trash in inner loop:
```
- 10.19% 10.17% clickhouse [.] DB::Block::rows
- DB::Block::rows
- 68.28% DB::FunctionPointInPolygon<DB::PointInPolygonCrossing>::executeImpl
- 100.00% DB::IFunction::execute
``` | 2017-09-04T20:00:25 | 2017-09-04T20:00:25 | {} | MEMBER |
ClickHouse | ClickHouse | 327,021,919 | 1,196 | alexey-milovidov | Can you explain what is the point of choosing "franklin" algrorithm? | 2017-09-04T20:02:42 | 2017-09-04T20:02:42 | {} | MEMBER |
ClickHouse | ClickHouse | 327,158,193 | 1,200 | KochetovNicolai | Hi,
arm32 and arm64 architectures are not currently supported. It was possible to build ClickHouse on arm64, but there might be problems though.
For 32 bit architectures it's more difficult. The ability to build on 32 bit was recently committed, but visible problems are:
- Possible lack of memory during compilat... | 2017-09-05T12:18:11 | 2017-09-05T12:18:11 | {} | MEMBER |
ClickHouse | ClickHouse | 327,179,813 | 1,200 | proller | No, we need to split/refactor some c++ template code.
Maybe for now easier/faster to use debian arm on qemu or cross compile | 2017-09-05T13:45:49 | 2017-09-05T13:45:49 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 327,257,682 | 1,197 | KochetovNicolai | 1. means that you have only 10 parts in table, so more than 10 threads won't help. Is it still works slower with 10 threads than spark?
Could you also show table structure (`show create table your_table_name`) and full query text? | 2017-09-05T18:10:22 | 2017-09-05T18:10:22 | {} | MEMBER |
ClickHouse | ClickHouse | 327,993,386 | 1,213 | alexey-milovidov | It doesn't fixed. | 2017-09-08T04:04:41 | 2017-09-08T04:04:41 | {} | MEMBER |
ClickHouse | ClickHouse | 327,993,671 | 1,210 | alexey-milovidov | The task is not done, because now code in ExecutableDictionarySource simply ignores exit code of command. | 2017-09-08T04:07:02 | 2017-09-08T04:07:02 | {} | MEMBER |
ClickHouse | ClickHouse | 328,147,726 | 1,220 | alexey-milovidov | Look at `ext::bit_cast`. | 2017-09-08T16:13:45 | 2017-09-08T16:13:45 | {} | MEMBER |
ClickHouse | ClickHouse | 328,456,776 | 1,217 | bgaifullin | patch is merged to master. | 2017-09-11T08:27:44 | 2017-09-11T08:27:44 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 328,693,113 | 1,231 | sirkon | https://clickhouse.yandex/docs/en/functions/math_functions.html#tgamma-x
 | 2017-09-12T00:08:42 | 2017-09-12T00:14:15 | {} | NONE |
ClickHouse | ClickHouse | 330,293,513 | 1,256 | proller | root of problem already fixed in lz4 master, but still outdated in freebsd ports:
https://github.com/lz4/lz4/blob/dev/lib/lz4.h#L432 | 2017-09-18T17:24:05 | 2017-09-18T17:24:05 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 327,006,084 | 1,197 | fleapapa | Hi, thanks a lot for the quick response.
It's true that more constrained condition can improve the "indexing" performance, but actually my point is the poor utilization of massive parallel processing (compare with Spark SQL) in cases of a useless index, which i guess can be improved by forking more threads to fill ... | 2017-09-04T17:30:36 | 2017-09-04T17:30:53 | {} | NONE |
ClickHouse | ClickHouse | 327,065,754 | 1,178 | daledude | I can't remember which version introduced the DEDUPLICATE to the OPTIMIZE query but it was effective for me in removing dupes. It does appear to "rewrite" the whole table. Also, I can't remember if I needed to use FINAL.
OPTIMIZE TABLE tablename [FINAL] DEDUPLICATE;
| 2017-09-05T04:02:44 | 2017-09-05T04:02:44 | {
"+1": 7,
"-1": 3
} | NONE |
ClickHouse | ClickHouse | 327,774,467 | 1,205 | johndickinson | I was more concerned with the fact that the server segfaulted rather than handling the error gracefully. | 2017-09-07T11:39:14 | 2017-09-07T11:39:14 | {} | NONE |
ClickHouse | ClickHouse | 328,439,390 | 1,152 | kszucs | @Slach How could I extract a list of clickhouse query identifiers? Something like [this](https://github.com/kszucs/ibis/blob/0c99a582ea967f97841d480c2ac2589db6c87535/ibis/clickhouse/identifiers.py). | 2017-09-11T07:17:53 | 2017-09-11T07:17:53 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 330,162,316 | 1,016 | jackpgao | I sloved the problem.
The reason is docker.
As I do not have a Ubuntu OS, I run clickhouse-client in docker.
The Docker Lang problem is the root.
| 2017-09-18T08:55:53 | 2017-09-18T08:55:53 | {} | NONE |
ClickHouse | ClickHouse | 330,526,354 | 1,257 | ludv1x | BTW I found the same errors in our production servers: `tcmalloc.cc:290] Attempt to realloc invalid pointer 0x3f000000`. But only two mistakes of all time in ~700 servers (Ubuntu build). | 2017-09-19T12:41:18 | 2017-09-19T12:42:14 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 327,014,475 | 1,092 | proller | problem:
https://github.com/yandex/ClickHouse/blob/master/dbms/src/Dictionaries/ExecutableDictionarySource.cpp#L71
sometimes thread is not joinable -> thread.join() shipped -> thread-zombie
| 2017-09-04T18:48:49 | 2017-09-04T18:48:49 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 327,025,617 | 1,196 | alexey-milovidov | ```
SELECT avg(pointInPolygon((x, y), (0., 0.), (1., 100.), (1., 0.), (2., 100.), (2., 0.), (3., 100.), (3., 0.), (4., 100.), (4., 0.), (5., 100.), (5., 0.), (6., 100.), (6., 0.), (7., 100.), (7., 0.), (8., 100.), (8., 0.), (9., 100.), (9., 0.), (10., 100.), (10., 0.), (11., 100.), (11., 0.), (12., 100.), (12., 0.), (... | 2017-09-04T20:39:47 | 2017-09-04T20:39:47 | {} | MEMBER |
ClickHouse | ClickHouse | 327,176,285 | 1,200 | marklit | I've been able to attach mechanical SATA drives via USB to avoid hitting the SD card too much.
Do you know of any parameters make or cmake can be given to try and lower that memory footprint? 3.4 GB doesn't feel like a world away from >3 GB.
> There are parts of the code where size_t assumed to be 64 bit
If th... | 2017-09-05T13:32:59 | 2017-09-05T13:37:02 | {} | NONE |
ClickHouse | ClickHouse | 327,462,020 | 1,110 | alexber220 | apt-get install libboost1.62-dev
solved this problem in debian 9
fatal error: 'boost/noncopyable.hpp' file not found
| 2017-09-06T12:02:39 | 2017-09-06T12:02:39 | {} | NONE |
ClickHouse | ClickHouse | 327,370,610 | 1,197 | fleapapa | 1. With 10 thread, it is even worse -- took 567.033 secs. Weird is during the query, top shows "Load average: 24.83 15.13 6.59" while cpu usage on both cores was very low <5%. Due to the high load, all terminal windows were locked up till the query ended.
2. show create table trip:
CREATE TABLE default.trip ( _cx D... | 2017-09-06T04:23:08 | 2017-09-07T01:29:03 | {} | NONE |
ClickHouse | ClickHouse | 328,572,324 | 863 | tobad357 | Is there any update on this development.
As currently to get BI tools to work with AggregatingMergeTree we would need a view on top that does the sumMerge, countMerge etc from the sumState.
But it isn't workable as long as the view reads the whole table when querying | 2017-09-11T15:49:38 | 2017-09-11T15:49:38 | {} | NONE |
ClickHouse | ClickHouse | 328,684,190 | 1,205 | vavrusa | @KochetovNicolai are you refactoring the dictionary? | 2017-09-11T23:10:07 | 2017-09-11T23:10:31 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 330,257,570 | 1,236 | ludv1x | Could you provide a minimal reproducible example, please?
I can't reproduce it.
```
:) create table dummy (p Date, k UInt64, d String) ENGINE = MergeTree(p, k, 8192)
CREATE TABLE dummy
(
p Date,
k UInt64,
d String
) ENGINE = MergeTree(p, k, 8192)
Ok.
0 rows in set. Elapsed: 0.005 sec.
... | 2017-09-18T15:20:25 | 2017-09-18T15:20:25 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 330,259,601 | 1,241 | ludv1x | You could try `sequenceMatch` aggregate function.
https://clickhouse.yandex/docs/en/single/#sequencematch-pattern-time-cond1-cond2
But it is suitable when all events are stored in one table.
| 2017-09-18T15:26:52 | 2017-09-18T15:27:04 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 330,317,605 | 1,252 | ludv1x | You could use `ALTER TABLE [db.]table DROP PARTITION 'month'` command to clear all data from a specified month (for example '201709') (according to partitioning key). | 2017-09-18T18:40:03 | 2017-09-18T18:40:03 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 330,517,761 | 1,231 | IanCal | @silviucpp Perhaps they just were linking to the page on mathematical operators?
the equation posted looks like it's for the great circle distance : https://en.wikipedia.org/wiki/Great-circle_distance and it's composed entirely of operators that are already supported | 2017-09-19T12:03:50 | 2017-09-19T12:03:50 | {} | NONE |
ClickHouse | ClickHouse | 327,160,117 | 1,200 | marklit | Before trying to compile I did increase my swapfile and rebooted my Raspberry Pi:
```
$ grep CONF_SWAPSIZE /etc/dphys-swapfile
CONF_SWAPSIZE=2048
```
It allowed for 1 GB of RAM and 2 GB of Swap. It's slow but I was hoping it would alleviate any memory issues. So far I've made it to about 54% of the way through... | 2017-09-05T12:27:18 | 2017-09-05T12:27:18 | {} | NONE |
ClickHouse | ClickHouse | 329,304,641 | 1,225 | silviucpp | Hy Alexey can this become a priority for you ? I checked a bit your code and I found that:
- It is not hard to add an integer as offset as the second parameter (my opinion the highest flexibility when working with ISO 8601)
- for each timezone at this point you create an instance of that timezone (as a singleton of ... | 2017-09-13T21:38:51 | 2017-09-13T21:38:51 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 330,314,891 | 1,241 | ludv1x | You could write subqueries for each table, concatenate them with UNION ALL and use `sequenceMatch`.
Also, the common ClickHouse's case is vast denormalized tables (500 columns is Ok). | 2017-09-18T18:30:30 | 2017-09-18T18:30:30 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 330,525,747 | 1,250 | bocharov | I changed return type to `Tuple(Array(K), Array(V))` and did other requested changes.
Please review again and I hope we could merge then. | 2017-09-19T12:38:48 | 2017-09-19T12:38:48 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 326,991,898 | 1,195 | marklit | Still seeing the issue.
```
$ pwd
/home/pi/ClickHouse/build
$ rm CMakeCache.txt
$ cmake ..
```
```
CMake Error at contrib/CMakeLists.txt:69 (target_include_directories):
Cannot specify include directories for target "mariadbclient" which is not
built by this project.
``` | 2017-09-04T15:49:52 | 2017-09-04T15:49:52 | {} | NONE |
ClickHouse | ClickHouse | 327,015,789 | 1,196 | alexey-milovidov | Missing modifications:
https://github.com/yandex/ClickHouse/commit/6901e80c4cbc1ed56e4be770e72db187697c8540 | 2017-09-04T19:01:21 | 2017-09-04T19:01:21 | {
"+1": 1
} | MEMBER |
ClickHouse | ClickHouse | 329,257,946 | 1,225 | alexey-milovidov | > Would be nice if all datetime functions which now supports as second parameter a string representing the timezone will support also an integer that specify the timezone offset in seconds.
This is reasonable, and it seems relatively easy to implement.
> Beside my previous questions there is any challenge to supp... | 2017-09-13T18:35:08 | 2017-09-13T18:35:08 | {} | MEMBER |
ClickHouse | ClickHouse | 330,573,955 | 1,246 | ludv1x | Confirmed
```select subsite_id, case when u=0 then CAST(NULL AS Nullable(float)) else cast(v as float)/u end from (select 'test' as subsite_id, 100 as v, 33 as u from system.numbers)```
```select subsite_id, case when u=0 then cast(NULL AS Nullable(Float32)) else cast(v as float)/u end from (select 'test' as subsit... | 2017-09-19T15:18:58 | 2017-09-19T15:19:20 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 327,016,228 | 1,196 | alexey-milovidov | Need to support both Float32 and Float64 directly. | 2017-09-04T19:05:41 | 2017-09-04T19:05:41 | {} | MEMBER |
ClickHouse | ClickHouse | 330,428,499 | 1,236 | jaykelin | 6 nodes cluster, each shard has 2 replica
source table : distributed.t1 look at MergeTree, 1500万/day
tarket table : distributed.t2 look at AggregatingMergeTree
```shell
#!/bin/bash
start_date="$1"
end_date="$2"
while [[ $start_date < $end_date ]]
clickhouse-client --query "insert into distributed.t2 sele... | 2017-09-19T04:49:01 | 2017-09-19T04:49:01 | {} | NONE |
ClickHouse | ClickHouse | 330,655,106 | 1,257 | iron-udjin | I've recompilled clickhouse with -DDEBUG_LIBTCMALLOC=1 opition. A new message has been appeared in stderr log file:
`memory allocation bug: object at 0x80b000000 has never been allocated`
Here is backtrace of core file:
```
# gdb `which clickhouse` /var/tmp/800.clickhouse.core -batch -ex bt
(no debugging symbols f... | 2017-09-19T19:55:47 | 2017-09-19T19:55:47 | {} | NONE |
ClickHouse | ClickHouse | 326,997,813 | 1,195 | proller | try apply patch from #1185 .
Also you need ~3gb memory to compile clickhouse, now only one? cheap arm board have 4G (rock64) | 2017-09-04T16:24:46 | 2017-09-04T16:24:46 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 327,016,713 | 1,196 | alexey-milovidov | Constant arguments for polygon vertices should cast to floating point type implicitly. | 2017-09-04T19:10:59 | 2017-09-04T19:10:59 | {} | MEMBER |
ClickHouse | ClickHouse | 328,171,207 | 1,217 | bgaifullin | Yes. | 2017-09-08T17:51:42 | 2017-09-08T17:51:42 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 328,747,176 | 1,231 | silviucpp | @sirkon can you please elaborate. I don't see the link between havershine and tgamma.
Silviu | 2017-09-12T06:06:14 | 2017-09-12T06:06:14 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 329,223,641 | 360 | proller | pass incliude path to cmake:
cmake .. -DMYSQL_INCLUDE_DIR=/usr/local/Cellar/mysql/5.7.18_1/include/
but also you need find correct path to binary lib like
-DSTATIC_MYSQLCLIENT_LIB=/usr/local/Cellar/mysql/5.7.18_1/libmysqlclient.a
or -DMYSQLCLIENT_LIBRARIES=/usr/local/Cellar/mysql/5.7.18_1/libmysqlclient.so | 2017-09-13T16:30:58 | 2017-09-13T16:30:58 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 329,474,808 | 1,232 | artemkamolodets | Hello, i've solved this issue by myself. The problem is, that i'm specifying wrong path to zknode.
Regards | 2017-09-14T13:02:39 | 2017-09-14T13:02:39 | {} | NONE |
ClickHouse | ClickHouse | 330,277,256 | 1,256 | proller | Latest fixed port version is here https://github.com/proller/freebsd-ports
pr to port update here - https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=221690
Fix of your problem : DCOMPILER_FLAGS="-DLZ4_DISABLE_DEPRECATE_WARNINGS=1" | 2017-09-18T16:26:15 | 2017-09-18T16:26:15 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 330,521,160 | 1,231 | silviucpp | @IanCal yes I figured out this after looking closer. I left this open maybe as an enhancement for clickhouse where we can have a function exported that takes as pramaters the longitude, latitude and a distance, and match the rows in that inverval.
Something like:
```
geoWithin(latitude1, longitude1, latitude2, ... | 2017-09-19T12:19:44 | 2017-09-19T12:19:44 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 327,026,304 | 1,196 | alexey-milovidov | (the list is generated by
```
perl -e 'print map { "($_., 0.), (" . ($_ + 1) . "., 100.), " } 0..100'
```
and one additional point is added) | 2017-09-04T20:47:14 | 2017-09-04T20:47:14 | {} | MEMBER |
ClickHouse | ClickHouse | 327,047,245 | 1,120 | alexey-milovidov | The data on the large badge:
https://app.fossa.io/api/projects/git%2Bhttps%3A%2F%2Fgithub.com%2Fyandex%2FClickHouse.svg?type=large
is incorrect.
All code except third-party libraries is under Apache 2 license.
This could be figured out from root LICENSE file:
https://github.com/yandex/ClickHouse/blob/master... | 2017-09-05T01:19:15 | 2017-09-05T01:19:24 | {} | MEMBER |
ClickHouse | ClickHouse | 328,458,748 | 1,217 | alexey-milovidov | Thanks! | 2017-09-11T08:34:48 | 2017-09-11T08:34:48 | {} | MEMBER |
ClickHouse | ClickHouse | 329,296,179 | 1,232 | ludv1x | Hi!
I can't reproduce the problem.
Also, I can't find a similar exception message in ClickHouse's code base.
Could you provide the full exception message with a stacktrace (just run `clickhouse-client` with `--stacktrace` parameter and repeat the query), please?
And a quick question:
> The path: /clickhouse/tabl... | 2017-09-13T21:02:59 | 2017-09-13T21:04:00 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 330,658,651 | 1,257 | proller | you can try build without tcmalloc: -DENABLE_LIBTCMALLOC=0 | 2017-09-19T20:09:07 | 2017-09-19T20:09:07 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 327,009,323 | 1,195 | marklit | Great, thank you for that patch. I've got the code compiling now. | 2017-09-04T17:59:53 | 2017-09-04T17:59:53 | {} | NONE |
ClickHouse | ClickHouse | 327,675,553 | 1,144 | alexey-milovidov | Actually you updated to version 1.3.0, not 1.3.1.
https://github.com/yandex/ClickHouse/commit/17edcdec96b01d988e749f44c4be0194f8546aee | 2017-09-07T03:43:13 | 2017-09-07T03:43:13 | {} | MEMBER |
ClickHouse | ClickHouse | 330,275,976 | 1,243 | ludv1x | Sorry, but it is correct behaviour -)
Here you could read an explanantion https://github.com/yandex/ClickHouse/issues/505#issuecomment-280681017
So, you should convert a DateTime variable returned by `toStartOfHour` to String explicitly:
```
SELECT
toUnixTimestamp('1999-12-31 22:00:01', 'UTC') AS ts,
... | 2017-09-18T16:21:28 | 2017-09-18T16:21:28 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 330,498,989 | 1,258 | proller | libc++ in old freebsd (before 11) does not support c++17 features.
Easiest way to compile - upgrade your freebsd first ;) | 2017-09-19T10:34:06 | 2017-09-19T10:34:06 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 330,663,444 | 1,257 | iron-udjin | Done. There are no errors in stderr log.
In clickhouse-server.err.log:
```
2017.09.19 23:25:01.343512 [ 7 ] <Error> BaseDaemon: ########################################
2017.09.19 23:25:01.343707 [ 7 ] <Error> BaseDaemon: (from thread 3) Received signal Segmentation fault (11).
2017.09.19 23:25:01.343724 [ 7 ] <Er... | 2017-09-19T20:27:58 | 2017-09-19T20:27:58 | {} | NONE |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.