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 | 319,941,381 | 1,057 | KochetovNicolai | Can you show what is in clickhouse-server.err.log and clickhouse-server.log when it happens? | 2017-08-03T11:20:09 | 2017-08-03T11:20:09 | {} | MEMBER |
ClickHouse | ClickHouse | 319,957,466 | 1,045 | alexey-milovidov | Ok. If you are not going to use this compression method, it's not worth to implement. | 2017-08-03T12:40:52 | 2017-08-03T12:40:52 | {} | MEMBER |
ClickHouse | ClickHouse | 320,471,793 | 1,067 | ramazanpolat | @alexey-milovidov thank you for detailed answer.
Typically clients connect to db to insert data to only a handful of tables. Most of the time, it's just one table and so one insert statement with different values. Think it like 1000+ clients sending insert statements for an individual table with only one row but they ... | 2017-08-05T21:37:44 | 2017-08-05T21:41:06 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 320,560,200 | 972 | ipolevoy | hello? | 2017-08-07T03:27:14 | 2017-08-07T03:27:14 | {} | NONE |
ClickHouse | ClickHouse | 321,009,318 | 972 | vas-and-tor | We have similar problem.
Here is our cluster:
```
SELECT *
FROM system.clusters
WHERE cluster = 'logs'
┌─cluster─┬─shard_num─┬─shard_weight─┬─replica_num─┬─host_name───┬─host_address────┬─port─┬─is_local─┬─user────┬─default_database─┐
│ logs │ 1 │ 1 │ 1 │ clickhouse2 │ 127.0... | 2017-08-08T16:28:26 | 2017-08-08T16:28:26 | {} | NONE |
ClickHouse | ClickHouse | 321,530,056 | 1,052 | prog8 | I can declare my hands-on to help with this but I think it can be only **help** and I cannot lead this. | 2017-08-10T11:52:40 | 2017-08-10T11:52:40 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 320,416,442 | 1,062 | silviucpp | Hmm, and how do you solve the problem of adding later let's say in a table a row which is array and you want to select records from past where this property was not available ? I mean where new field it's NULL not empty array..
Silviu | 2017-08-05T06:29:01 | 2017-08-13T11:37:41 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 322,036,909 | 1,062 | silviucpp | Hy @alexey-milovidov do you have an opinion on my previous question ?
Silviu | 2017-08-13T11:38:24 | 2017-08-13T11:38:24 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 322,098,708 | 1,095 | alexey-milovidov | Missing test. | 2017-08-14T04:36:52 | 2017-08-14T04:36:52 | {} | MEMBER |
ClickHouse | ClickHouse | 322,618,776 | 1,103 | alexey-milovidov | Thanks! | 2017-08-15T23:41:58 | 2017-08-15T23:41:58 | {} | MEMBER |
ClickHouse | ClickHouse | 322,729,957 | 1,106 | robot-metrika-test | Can one of the admins verify this patch? | 2017-08-16T10:29:03 | 2017-08-16T10:29:03 | {} | NONE |
ClickHouse | ClickHouse | 320,383,498 | 1,067 | alexey-milovidov | This is expected.
INSERT statements with single rows are slower not only because it need SQL parsing for each statement, but also, because each INSERT into MergeTree goes directly into filesystem without any buffering and forms new "data part" with files per each column. (These data parts will be merged in larger pa... | 2017-08-05T01:18:40 | 2017-08-05T01:18:40 | {
"+1": 23
} | MEMBER |
ClickHouse | ClickHouse | 322,680,720 | 1,104 | robot-metrika-test | Can one of the admins verify this patch? | 2017-08-16T06:42:01 | 2017-08-16T06:42:01 | {} | NONE |
ClickHouse | ClickHouse | 322,798,081 | 1,107 | robot-metrika-test | Can one of the admins verify this patch? | 2017-08-16T14:54:19 | 2017-08-16T14:54:19 | {} | NONE |
ClickHouse | ClickHouse | 322,921,163 | 1,110 | vavrusa | Perhaps you have mysqlxx includes somewhere in the include path already, which is why it succeeds? | 2017-08-16T22:50:54 | 2017-08-16T22:50:54 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 323,148,272 | 1,110 | vavrusa | @proller builds okay with MySQL disabled 🙌 | 2017-08-17T17:58:44 | 2017-08-17T17:58:44 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 320,636,265 | 1,002 | PolPavel | Solution found аt: [Here](https://stackoverflow.com/questions/39188919/different-exception-specifier-with-g-6-2)
I just deleted ' throw (std::bad_alloc)' on line 60 and 64 in file: /opt/src/ClickHouse/contrib/libtcmalloc/src/libc_override_gcc_and_weak.h
| 2017-08-07T11:11:18 | 2017-08-07T11:11:18 | {} | NONE |
ClickHouse | ClickHouse | 320,814,841 | 1,075 | alexey-milovidov | In fact,
`uniq` uses hash table with adaptive sampling,
`uniqHLL12` uses HyperLogLog,
`uniqCombined` uses array + hash table + HyperLogLog
and `uniqExact` uses hash table. | 2017-08-08T00:16:56 | 2017-08-08T00:17:13 | {} | MEMBER |
ClickHouse | ClickHouse | 320,815,187 | 1,051 | alexey-milovidov | Thanks! | 2017-08-08T00:19:47 | 2017-08-08T00:19:47 | {} | MEMBER |
ClickHouse | ClickHouse | 321,797,865 | 1,091 | robot-metrika-test | Can one of the admins verify this patch? | 2017-08-11T12:17:03 | 2017-08-11T12:17:03 | {} | NONE |
ClickHouse | ClickHouse | 322,729,537 | 1,106 | robot-metrika-test | Can one of the admins verify this patch? | 2017-08-16T10:27:03 | 2017-08-16T10:27:03 | {} | NONE |
ClickHouse | ClickHouse | 322,776,498 | 58 | egorlitvinenko | @alexey-milovidov when? :) | 2017-08-16T13:48:04 | 2017-08-16T13:48:04 | {} | NONE |
ClickHouse | ClickHouse | 320,413,938 | 1,067 | daledude | This goes for just about any database. Bulk inserts, as Alexey showed, are faster. | 2017-08-05T05:31:29 | 2017-08-05T05:31:29 | {} | NONE |
ClickHouse | ClickHouse | 320,609,262 | 972 | alex-zaitsev | Please share your cluster configuration. You seem to have one shard and two replicas. The data has to be replicated, so it looks like your Distributed table is not properly configured and thinks there are two shards. | 2017-08-07T08:59:44 | 2017-08-07T08:59:44 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 320,692,542 | 972 | slyderby | 'show create table' output for one of the tables:
`CREATE TABLE test.campaign_events ( campaign_id UInt64, contact_id UInt64, type UInt8, email String, user_id UInt64, user_agent String, url String, latitude Float32, longitude Float32, ip String, created_at DateTime, event_date Date, merged_url String) EN... | 2017-08-07T15:15:58 | 2017-08-07T15:15:58 | {} | NONE |
ClickHouse | ClickHouse | 321,358,620 | 1,080 | alexey-milovidov | As you have mentioned in chat, the filesystem doesn't look good: https://la.wentropy.com/nXPX | 2017-08-09T19:33:31 | 2017-08-09T19:33:31 | {} | MEMBER |
ClickHouse | ClickHouse | 322,018,618 | 1,067 | ramazanpolat | I'm thinking something like this:
Clients connect to CH with a hint in JDBC URL:
jdbc:clickhouse://<host>:<port>[/<database>][?format=table]
This connection style hints CH server to indicate that client will send a text with a supported format to be directly inserted into a table.
With hinted JDBC connecti... | 2017-08-13T02:57:55 | 2017-08-13T02:57:55 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 322,923,393 | 1,077 | alexey-milovidov | 1. You can do manual intervention to data parts only when server is not running. So, you should stop the server, then modify parts and then start the server. This is because the server maintains set of parts in memory.
2. For Replicated tables, you cannot change set of parts, because the server will validate this s... | 2017-08-16T23:04:21 | 2017-08-16T23:04:21 | {} | MEMBER |
ClickHouse | ClickHouse | 320,387,772 | 1,066 | vavrusa | You're right, it'd be nice to show it in the regular logs, but query log is good enough I think. | 2017-08-05T01:53:12 | 2017-08-05T01:53:12 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 321,056,991 | 1,078 | alexey-milovidov | About warnings:
```
Include not found: clickhouse_compression
```
These are totally fine.
About
```
Couldn't save preprocessed config
```
This is not harmful, but also isn't good.
Preprocessed config is dumpred mostly for informational purposes.
(It could be used by server in rare cases: if configuratio... | 2017-08-08T19:27:17 | 2017-08-08T19:28:09 | {} | MEMBER |
ClickHouse | ClickHouse | 321,947,175 | 1,067 | ramazanpolat | I was thinking the same, just wanted to be sure if there is another way.
@alexey-milovidov 's comment "Probably we could detect multiple INSERT statements with similar structure to optimize this automatically" should be considered for client inserts. This will save users lots of time and optimization. Should I open an... | 2017-08-12T00:59:58 | 2017-08-12T00:59:58 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 322,004,130 | 1,067 | alexey-milovidov | Yes, it's better to open issue for enhancement.
I already thinking about it. It is also relevant for bulk INSERTs where VALUES contain expressions. | 2017-08-12T20:30:55 | 2017-08-12T20:30:55 | {} | MEMBER |
ClickHouse | ClickHouse | 322,099,842 | 1,095 | alexey-milovidov | https://github.com/yandex/ClickHouse/commit/09d83eecd0d12391c81b6e142c72904515d185fb
https://github.com/yandex/ClickHouse/commit/957fb9f2e5a477e3fa39583df5e28cad3f6f5a7d | 2017-08-14T04:50:14 | 2017-08-14T04:50:14 | {} | MEMBER |
ClickHouse | ClickHouse | 322,234,977 | 1,078 | amosbird | via `--net=host`. In fact, i did exactly what you mentioned `- 9000:9000` but still couldn't connect. | 2017-08-14T16:14:09 | 2017-08-14T16:14:09 | {} | COLLABORATOR |
ClickHouse | ClickHouse | 322,426,553 | 1,078 | bamx23 | I've just created a droplet with:
* CentOS 7.3.1611
* Docker 17.06.0-ce, build 02c1d87
* Docker Compose 1.15.0, build e12f3b9
After adding 9000 port as exposed and starting server with `docker-compose up -d server` I can see that 9000 port is opened:
```
[root@temp-docker ClickHouse]# telnet localhost 9000
Try... | 2017-08-15T09:47:27 | 2017-08-15T09:47:27 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 322,799,105 | 1,107 | robot-metrika-test | Can one of the admins verify this patch? | 2017-08-16T14:57:16 | 2017-08-16T14:57:16 | {} | NONE |
ClickHouse | ClickHouse | 323,066,517 | 1,110 | proller | Can you provide full cmake console output ? | 2017-08-17T13:01:10 | 2017-08-17T13:01:10 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 320,040,054 | 127 | erebus1 | same problem | 2017-08-03T17:42:24 | 2017-08-03T17:42:24 | {} | NONE |
ClickHouse | ClickHouse | 320,678,175 | 972 | slyderby | This is how our cluster configuration looks:
```
<remote_servers>
<test>
<shard>
<replica>
<host>10.1.1.10</host>
<port>9000</port>
</replica>
<replica>
<host>10.1.1.11</host>
... | 2017-08-07T14:25:03 | 2017-08-07T14:25:03 | {} | NONE |
ClickHouse | ClickHouse | 322,929,767 | 1,110 | vavrusa | Seems like it concerns other builtin libraries as well:
```
[23:15:50][build] # In case building clickhouse-server, adding to package binary of clang, ld and header files - for dynamic compilation.
[23:15:50][build] mkdir -p debian/tmp/usr/share/clickhouse/bin debian/tmp/usr/share/clickhouse/headers
[23:15:50][bu... | 2017-08-16T23:45:47 | 2017-08-16T23:45:47 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 323,190,806 | 1,118 | alexey-milovidov | This is the part of a larger task and we have some related tasks planned:
First, we are planning to implement rich profiling info for a query, to know,
how much time is spent in each server node, in each thread for each subquery, on
CPU, disk IO, network and synchronization.
Second, we are going to implement a ... | 2017-08-17T20:51:27 | 2017-08-17T20:51:27 | {
"+1": 24
} | MEMBER |
ClickHouse | ClickHouse | 320,373,842 | 1,065 | bobrik | `system.cluster_parts` is also missing:
```
:) show tables
SHOW TABLES
┌─name──────────────────────────────────┐
│ .inner.clickhouse_query_stats_by_user │
│ asynchronous_metrics │
│ build_options │
│ clickhouse_query_stats_by_user │
│ clusters ... | 2017-08-04T23:26:29 | 2017-08-04T23:26:29 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 321,415,713 | 745 | vavrusa | I just have a similar problem, it only happens to some tables / partitions it seems. It looks like dropping or detaching succeeds, but it doesn't seem to be doing anything. These partitions were converted from MergeTree to ReplicatedMergeTree if that helps.
```
:) SELECT formatReadableSize(sum(bytes)) as sz FROM s... | 2017-08-10T00:16:52 | 2017-08-10T06:26:16 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 321,466,253 | 972 | vas-and-tor | I reproduced this problem even with direct inserts to a shard, like this:
```
INSERT INTO metrika_new.api_logs_local SELECT
Timestamp,
HttpMethod,
HostName,
UriPath,
UriParams,
HttpStatus,
RequestTime,
RequestLength,
BytesSent,
UpstreamResponseTime,
PortalUserI... | 2017-08-10T06:47:57 | 2017-08-10T06:47:57 | {} | NONE |
ClickHouse | ClickHouse | 321,706,770 | 745 | vavrusa | I talked with @alexey-milovidov, the issue is that partitions were converted from MergeTree to ReplicatedMergeTree in an older version of ClickHouse that attached the parts in a private block number space [0, 200), without increasing the actual block number. Sine the part hasn't been written into, the block number is 0... | 2017-08-11T00:18:55 | 2017-08-11T00:18:55 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 322,612,553 | 1,103 | robot-metrika-test | Can one of the admins verify this patch? | 2017-08-15T23:02:01 | 2017-08-15T23:02:01 | {} | NONE |
ClickHouse | ClickHouse | 322,920,503 | 1,102 | alexey-milovidov | We are planning to implement a set of related operations:
```
TRUNCATE TABLE
CREATE OR REPLACE TABLE
REPLACE TABLE
```
The main downside is that in first implementation they will be supported for all table engines except `Replicated-` tables. This is because for Replicated tables, CREATE and DROP just add or re... | 2017-08-16T22:47:06 | 2017-08-16T22:47:06 | {
"+1": 11
} | MEMBER |
ClickHouse | ClickHouse | 322,922,675 | 1,102 | ramazanpolat | Looking forward to it.
I would like to contribute to this project, unfortunately I'm far from being good at C++. I'll do my best if there is any other way to help. | 2017-08-16T22:59:45 | 2017-08-16T22:59:45 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 323,046,880 | 1,112 | ludv1x | Similar problem https://github.com/yandex/ClickHouse/issues/598 | 2017-08-17T11:42:29 | 2017-08-17T11:42:29 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 323,152,861 | 1,110 | proller | no, mysqlxx haeaders not needed for copy_headers.sh | 2017-08-17T18:16:52 | 2017-08-17T18:16:52 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 322,417,393 | 1,078 | amosbird | ```
Docker version 1.12.6,
docker-compose version 1.13.0, build 1719ceb
docker-py version: 2.3.0
CPython version: 2.7.5
OpenSSL version: OpenSSL 1.0.1e-fips 11 Feb 2013
centos 7.3
``` | 2017-08-15T09:01:05 | 2017-08-15T09:01:05 | {} | COLLABORATOR |
ClickHouse | ClickHouse | 322,576,894 | 1,096 | alexey-milovidov | Thanks!
Added in separate commits:
https://github.com/yandex/ClickHouse/commit/4272265128e12bfdfd57ae0169e76b3ecd81f426
https://github.com/yandex/ClickHouse/commit/dc29ae7e730c3bc0518d6b5250d1b6dc999c897a | 2017-08-15T20:18:31 | 2017-08-15T20:18:31 | {} | MEMBER |
ClickHouse | ClickHouse | 322,912,578 | 1,111 | robot-metrika-test | Can one of the admins verify this patch? | 2017-08-16T22:04:02 | 2017-08-16T22:04:02 | {} | NONE |
ClickHouse | ClickHouse | 322,919,602 | 1,110 | alexey-milovidov | > I'm not sure if you're deprecating MySQL support or is it just an omission?
This is just the ommission.
This is strange, because we have this version installed on half of our clusters.
We have MySQL dictionaries in that clusters and they work fine.
And also we have tests for all the dictionary sources, includ... | 2017-08-16T22:41:56 | 2017-08-16T22:41:56 | {} | MEMBER |
ClickHouse | ClickHouse | 321,235,270 | 1,081 | robot-metrika-test | Can one of the admins verify this patch? | 2017-08-09T12:02:02 | 2017-08-09T12:02:02 | {} | NONE |
ClickHouse | ClickHouse | 321,473,723 | 1,028 | silviucpp | Is this fix a priority ? | 2017-08-10T07:27:55 | 2017-08-10T07:27:55 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 321,478,027 | 1,080 | amosbird | Ok. I'll avoid using nfs. | 2017-08-10T07:49:18 | 2017-08-10T07:49:18 | {} | COLLABORATOR |
ClickHouse | ClickHouse | 322,751,398 | 1,105 | blinkov | Could you please apply the same changes to english version too? | 2017-08-16T12:13:39 | 2017-08-16T12:13:39 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 322,899,979 | 1,062 | alexey-milovidov | This will be more difficult. You could add another field as a flag to indicate older data.
For example, you could have two fields `Items Array(String)` and `HasItems UInt8`.
But sometimes it is enough to not differentiate between empty array and undefined array value. | 2017-08-16T21:08:51 | 2017-08-16T21:08:51 | {} | MEMBER |
ClickHouse | ClickHouse | 323,207,943 | 1,120 | robot-metrika-test | Can one of the admins verify this patch? | 2017-08-17T22:09:01 | 2017-08-17T22:09:01 | {} | NONE |
ClickHouse | ClickHouse | 319,723,692 | 1,003 | proller | Strange problem with mysql<->openssl lib. lib linking order seems good.
You can build without mysql: cmake -DENABLE_MYSQL=0
Or try buils with shared libs: cmake -DUSE_STATIC_LIBRARIES=0 | 2017-08-02T16:21:00 | 2017-08-02T16:21:00 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 319,616,921 | 1,045 | prog8 | @alexey-milovidov I just reminded myself why I didn't drop `memcpy` for decompression. It is not a low hanging fruit task because `CompressedReadBuffer` allocates memory so I'd have to refactor `CompressedReadBuffer`. In addition to that I will have to change `CompressedReadBufferBase::decompress` to accept pointer to ... | 2017-08-02T09:16:54 | 2017-08-03T10:57:18 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 320,918,993 | 1,025 | amosbird | thanks! | 2017-08-08T10:41:44 | 2017-08-08T10:41:44 | {} | COLLABORATOR |
ClickHouse | ClickHouse | 322,397,974 | 1,078 | bamx23 | What environment do you have? Like versions of Docker, Docker Compose and OS. | 2017-08-15T07:38:29 | 2017-08-15T07:38:29 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 322,618,845 | 1,103 | alexey-milovidov | And what about english file? | 2017-08-15T23:42:33 | 2017-08-15T23:42:33 | {} | MEMBER |
ClickHouse | ClickHouse | 323,122,609 | 1,110 | vavrusa | Sure, this is the tag + patch mentioned in this issue applied + MySQL disabled:
https://gist.github.com/vavrusa/25382af5a61584dcc7df01a5d5cc1734 | 2017-08-17T16:19:47 | 2017-08-17T16:19:47 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 319,978,235 | 1,002 | YiuRULE | Yep, no problem on the master branch for me. | 2017-08-03T14:03:14 | 2017-08-03T14:03:14 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 320,002,305 | 1,053 | prog8 | I've just realized that this is already possible to attach some plugins (`Compiler.h`) but this is just not documented in https://clickhouse.yandex/docs/en/. Is it only used to automatically compile queries so they can be inlined and speeded up? As far as I understand at the moment only aggregating functions can be com... | 2017-08-03T15:23:38 | 2017-08-03T15:29:30 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 320,681,534 | 1,075 | robot-metrika-test | Can one of the admins verify this patch? | 2017-08-07T14:37:03 | 2017-08-07T14:37:03 | {} | NONE |
ClickHouse | ClickHouse | 320,682,102 | 1,075 | robot-metrika-test | Can one of the admins verify this patch? | 2017-08-07T14:39:03 | 2017-08-07T14:39:03 | {} | NONE |
ClickHouse | ClickHouse | 321,075,643 | 1,076 | alexey-milovidov | Confirmed. | 2017-08-08T20:40:57 | 2017-08-08T20:40:57 | {} | MEMBER |
ClickHouse | ClickHouse | 321,739,507 | 1,057 | jackpgao | Forget that.
I change the data dir in clickhouse-server.sh and it works fine.
| 2017-08-11T06:25:13 | 2017-08-11T06:25:13 | {} | NONE |
ClickHouse | ClickHouse | 322,201,413 | 1,078 | amosbird | well, it doesn't work either after i expose 9000 port. in fact i've exposed the whole network. | 2017-08-14T14:13:43 | 2017-08-14T14:13:43 | {} | COLLABORATOR |
ClickHouse | ClickHouse | 323,142,109 | 1,110 | vavrusa | @proller Rerunning the CI build with the patch, do we need to do the same thing for mysqlxx includes? (I disabled it to let the build pass) | 2017-08-17T17:34:53 | 2017-08-17T17:34:53 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 321,516,659 | 1,052 | otisg | This would be great and make CH more appealing to organizations that need really production-ready, mature tech. Is this something that could be added to CH roadmap for early 2018? | 2017-08-10T10:42:47 | 2017-08-10T10:42:47 | {
"+1": 1
} | NONE |
ClickHouse | ClickHouse | 321,798,220 | 1,091 | robot-metrika-test | Can one of the admins verify this patch? | 2017-08-11T12:19:02 | 2017-08-11T12:19:02 | {} | NONE |
ClickHouse | ClickHouse | 322,205,279 | 1,078 | bamx23 | How do you do that? I've added ` - 9000:9000` at the end of the `docker-compose.yml` and now `docker-compose up -d server` gives me an access to 9000 port of CH in the container. | 2017-08-14T14:28:15 | 2017-08-14T14:28:15 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 322,921,166 | 1,092 | alexey-milovidov | Interesting. We should look further.
Few questions:
1. Do you use cache or fully-reloaded dictionaries?
2. Do you have zombie process for every data reload (in case of fully-reloaded dictionaries) or for every processed block of data (in case of cache dictionaries), or zombies are left just occasionally? | 2017-08-16T22:50:55 | 2017-08-16T22:50:55 | {} | MEMBER |
ClickHouse | ClickHouse | 320,815,437 | 1,051 | alexey-milovidov | > Is the test suite going to be run by CI on this PR?
Yes. In fact, our CI is:
- internal (results are not published);
- broken at half the time; | 2017-08-08T00:21:38 | 2017-08-08T00:21:38 | {} | MEMBER |
ClickHouse | ClickHouse | 323,135,857 | 1,110 | proller | https://github.com/yandex/ClickHouse/pull/1115 | 2017-08-17T17:10:34 | 2017-08-17T17:10:34 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 323,207,569 | 1,120 | robot-metrika-test | Can one of the admins verify this patch? | 2017-08-17T22:07:02 | 2017-08-17T22:07:02 | {} | NONE |
ClickHouse | ClickHouse | 320,375,842 | 1,065 | bobrik | Turns out our automation just silently nuked `config.xml` for dictionaries after upgrade. | 2017-08-04T23:46:50 | 2017-08-04T23:46:50 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 320,652,680 | 1,002 | proller | https://github.com/yandex/ClickHouse/commit/465d3054790715586204775f65dd6803452f6585
Here was many problems with gcc7, not only in tcmalloc. But now anything fixed.
Also tcmalloc was fixed in upstream -
https://github.com/gperftools/gperftools/commit/c4de73c0e69b9a75b6795fdd4598234baed8496d
https://github.com/... | 2017-08-07T12:41:02 | 2017-08-07T12:44:40 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 321,235,666 | 1,081 | robot-metrika-test | Can one of the admins verify this patch? | 2017-08-09T12:04:02 | 2017-08-09T12:04:02 | {} | NONE |
ClickHouse | ClickHouse | 322,157,814 | 1,078 | bamx23 | `docker-compose up -d server` exposes only 8123 port. To use client run command `docker-compose run client` instead of `clickhouse-client`. Or you can modify `docker-compose.yml` file to expose 9000 port. | 2017-08-14T10:37:57 | 2017-08-14T10:37:57 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 322,680,993 | 1,103 | xzkostyan | Sorry, forgot about it. Here is PR with english docs: #1104. | 2017-08-16T06:43:37 | 2017-08-16T06:43:37 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 323,081,237 | 1,112 | inkrement | ok, then I'll close this one. | 2017-08-17T13:56:46 | 2017-08-17T13:56:46 | {} | NONE |
ClickHouse | ClickHouse | 319,725,071 | 962 | proller | Can you try now with fresh master ? | 2017-08-02T16:25:58 | 2017-08-02T16:25:58 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 320,365,004 | 956 | vavrusa | Has this been fixed in the recent commits by @KochetovNicolai ? | 2017-08-04T22:16:01 | 2017-08-04T22:16:01 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 320,384,333 | 1,066 | alexey-milovidov | In fact, we log these queries in `query_log`:
```
milovidov@milovidov-Latitude-E7470:~/work/ClickHouse$ curl 'http://localhost:8123/?session_id=test' -d 'SET log_queries = 1'
milovidov@milovidov-Latitude-E7470:~/work/ClickHouse$ curl 'http://localhost:8123/?session_id=test' -d 'SELECT @'
Code: 62, e.displayText()... | 2017-08-05T01:25:58 | 2017-08-05T01:33:37 | {} | MEMBER |
ClickHouse | ClickHouse | 320,930,770 | 1,067 | shuttie | @ramazanpolat I guess if you have 1000+ clients doing a lot of single-row inserts, you can re-architecture your app in a way like:
* a front-facing API to receive messages from thousands of clients
* a queue accumulating these messages (Kafka/RabbitMQ/whatever)
* a batch writer which dumps batches of messages ever... | 2017-08-08T11:41:50 | 2017-08-08T11:41:50 | {
"+1": 8
} | NONE |
ClickHouse | ClickHouse | 321,668,723 | 1,028 | alexey-milovidov | We should fix `groupUniqArray`.
This is not in first priority. Probably 4th or 5th in the list. | 2017-08-10T20:47:18 | 2017-08-10T20:47:18 | {} | MEMBER |
ClickHouse | ClickHouse | 322,912,191 | 1,111 | robot-metrika-test | Can one of the admins verify this patch? | 2017-08-16T22:02:02 | 2017-08-16T22:02:02 | {} | NONE |
ClickHouse | ClickHouse | 320,382,945 | 1,062 | alexey-milovidov | We decided, that for simplicity reasons we need to support either arrays with nullable values or nullable arrays, but not both. And arrays with nullable values seems more useful. Nullable arrays are less needed, because often empty arrays are enough.
This is still subject to change. | 2017-08-05T01:11:53 | 2017-08-05T01:11:53 | {} | MEMBER |
ClickHouse | ClickHouse | 320,687,223 | 972 | alex-zaitsev | Please also provide an output of 'show create table' for shard tables and distributed table.
But I guess the problem is that you do not have internal_replication for your shard tables:
<internal_replication>true</internal_replication> | 2017-08-07T14:57:02 | 2017-08-07T14:57:02 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 320,899,220 | 1,023 | hagen1778 | Can anyone take a look at this? Because we stuck with scheme updates | 2017-08-08T09:14:50 | 2017-08-08T09:14:50 | {} | NONE |
ClickHouse | ClickHouse | 321,379,026 | 1,083 | alexey-milovidov | Why we have `_LIBRARY` and `_LIBRARIES`?
Why in `/cmake` we have `find_zlib.cmake` and in `/cmake/Modules` we have `Findbtrie.cmake` or `FindGPerftools.cmake`?
This is neither snake_case or CamelCase, but something especially idiosyncratic.
I have merged this PR only because all CMake files considered as a trash a... | 2017-08-09T20:56:29 | 2017-08-09T20:56:29 | {} | MEMBER |
ClickHouse | ClickHouse | 322,612,123 | 1,103 | robot-metrika-test | Can one of the admins verify this patch? | 2017-08-15T22:59:01 | 2017-08-15T22:59:01 | {} | NONE |
ClickHouse | ClickHouse | 322,681,070 | 1,104 | robot-metrika-test | Can one of the admins verify this patch? | 2017-08-16T06:44:01 | 2017-08-16T06:44:01 | {} | NONE |
ClickHouse | ClickHouse | 322,924,794 | 1,073 | alexey-milovidov | Сейчас условные операторы вычисляются "жадно" - независимо от условия, вычисляются все аргументы.
Мы как раз хотели сделать, чтобы был отдельный режим вычисления:
- отфильтровать блоки данных для вычисления аргументов (then, else) по соответствующему условию;
- вычислить выражения then else для отфильтрованных бло... | 2017-08-16T23:13:30 | 2017-08-16T23:15:03 | {} | MEMBER |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.