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 | 380,748,401 | 2,207 | filimonov | Didn't you see also the following message is server logs?
https://github.com/yandex/ClickHouse/blob/08170d0d779b12eff05777e12f3d6b086e54ae02/dbms/src/Interpreters/DDLWorker.cpp#L239
Check that part of your config.xml and your profile:
```xml
<distributed_ddl>
<!-- Path in ZooKeeper to queue with DD... | 2018-04-12T10:02:41 | 2018-04-12T10:07:05 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 381,039,614 | 1,661 | filimonov | Also you can take a look on that: https://www.percona.com/blog/2018/01/09/updating-deleting-rows-clickhouse-part-1/ | 2018-04-13T06:42:49 | 2018-04-13T06:42:49 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 381,227,547 | 2,208 | alexey-milovidov | Fixed in master: 669b0d4b | 2018-04-13T18:46:15 | 2018-04-13T18:46:15 | {
"heart": 1
} | MEMBER |
ClickHouse | ClickHouse | 381,233,289 | 2,201 | alexey-milovidov | The functions are capable to sort the array according to transformed values.
Please check the test: https://github.com/yandex/ClickHouse/blob/master/dbms/tests/queries/0_stateless/00390_array_sort.sql | 2018-04-13T19:06:44 | 2018-04-13T19:06:44 | {} | MEMBER |
ClickHouse | ClickHouse | 381,567,099 | 2,220 | meolu | there are two way to create distributed table.
# 1、Without Zookeeper
refer to [this test script](https://github.com/yandex/ClickHouse/blob/0114b69f6180740848b5d3acef4804fa5ac87134/dbms/tests/integration/test_distributed_ddl/test.py#L267)
## config.xml
```xml
<rrc_ck_cluster_without_zookeeper>
... | 2018-04-16T11:25:31 | 2018-04-16T11:26:19 | {} | NONE |
ClickHouse | ClickHouse | 381,936,342 | 2,187 | millin | @jonatasfreitasv
1.1.54370 has to many bugs...
The new version 1.1.5437**8** should be more stable, but I haven't tried it yet. | 2018-04-17T10:18:53 | 2018-04-17T10:18:53 | {
"+1": 1
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 380,639,156 | 2,206 | badvir | Unfortunately, there were no merge tasks and no queries in my case. | 2018-04-12T00:45:15 | 2018-04-12T00:46:35 | {} | NONE |
ClickHouse | ClickHouse | 380,731,373 | 2,207 | jpiper | @ludv1x Thanks for the reply! So now I get a different error, but only on some nodes...
> Code: 164, e.displayText() = DB::Exception: Received from XXXXXXXX:9000, 10.110.208.170. DB::Exception: Cannot execute query in readonly mode., e.what() = DB::Exception
Which is odd as I am logged in as a user with admin per... | 2018-04-12T09:03:07 | 2018-04-12T09:03:07 | {} | NONE |
ClickHouse | ClickHouse | 380,921,071 | 1,661 | Slind14 | > Yes, that's what i've tried to propose, may be my expkanations wasn't clear enought. So last day/or few from dictionary. Older data - materialized and stored in column.
You wrote:
> if in a meanwhile some background merges will happen they will materialize b again
for merged subrange, so you need to call CLEA... | 2018-04-12T19:39:02 | 2018-04-12T19:39:02 | {} | NONE |
ClickHouse | ClickHouse | 380,923,336 | 1,661 | Slind14 | > @Slind14 Dictionary lookups aren't done directly against your db, so it doesn't really matter what you use. ClickHouse polls your dictionary for changes and keeps the most recent data in RAM for query-time lookups. Just use whatever db is easiest for you.
ahh, I see. @derekperkins is there something I can read on ... | 2018-04-12T19:47:32 | 2018-04-12T19:47:32 | {} | NONE |
ClickHouse | ClickHouse | 381,736,180 | 2,202 | alexey-milovidov | > So generally it will use the same credential to connect to leader replica which were used by current user?
We have discussed, that this is Ok. But few downsides exist. | 2018-04-16T20:20:44 | 2018-04-16T20:20:44 | {} | MEMBER |
ClickHouse | ClickHouse | 381,958,355 | 2,156 | lamberken | @KochetovNicolai , it seems dosn't been fixed, version 1.1.54378
```
SELECT time
FROM centerlog
WHERE date = '2018-01-21'
LIMIT 1
Received exception from server (version 1.1.54378):
Code: 171. DB::Exception: DB::Exception: Block structure mismatch in UNION stream: different number of columns:
time DateTime U... | 2018-04-17T11:34:00 | 2018-04-17T11:34:00 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 382,492,449 | 2,240 | alex-zaitsev | This is yet experimental. Please try using Altinity RPMs: https://packagecloud.io/altinity/clickhouse | 2018-04-18T18:55:02 | 2018-04-18T18:55:02 | {
"+1": 1
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 382,522,163 | 2,238 | alexey-milovidov | Thank you! | 2018-04-18T20:40:43 | 2018-04-18T20:40:43 | {} | MEMBER |
ClickHouse | ClickHouse | 382,605,715 | 1,925 | alexey-milovidov | Performance difference is neglible. Perfect. | 2018-04-19T04:19:17 | 2018-04-19T04:19:17 | {} | MEMBER |
ClickHouse | ClickHouse | 380,904,099 | 1,661 | filimonov | Updates in that scenario cost mostly nothing (as they happen outside of clickhouse). Selects will use external dict before 'fixing' the final value (by OPMIZIE FINAL), so it's more about performance of external dictionary during that time. But that allows you to store outside only limited amount of data (for last day o... | 2018-04-12T18:37:45 | 2018-04-12T18:40:41 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 380,912,818 | 1,661 | filimonov | > or could we keep the dictionary lookup active for the last n hours/days and do the optimize partition final for older records? So old old partitions would be a single table read and new partitions with the dictionary lookup. I don't know how performance will with this. I have up to 100 million new records per day.
... | 2018-04-12T19:07:19 | 2018-04-12T19:07:19 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 381,356,262 | 2,070 | k0st1an | Demonstration of bug. OS: Ubuntu 16.04 - https://gist.github.com/k0st1an/782576761f4fce1a752ba7aea4991ec6 | 2018-04-14T19:59:14 | 2018-04-14T19:59:14 | {
"+1": 1
} | NONE |
ClickHouse | ClickHouse | 382,258,868 | 2,218 | jonatasfreitasv | Last commits:
- Database, Tables, Columns spec in SideBar
- Split panels
- Database settings | 2018-04-18T04:40:43 | 2018-04-18T04:40:43 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 380,754,932 | 2,207 | jpiper | My config.xml looks like that already.
I did not see that message - some nodes gave the "Cannot execute query in readonly mode" - others did not (it looked successful).
My default user is read-only and I set the admin username and password in `<remote_servers>` I am logged in via the admin user when I run the com... | 2018-04-12T10:24:52 | 2018-04-12T10:24:52 | {} | NONE |
ClickHouse | ClickHouse | 380,881,950 | 1,661 | filimonov | Ok. So you need do one-time change the status (one column) of records which were inserted in last day. That status updates will happen in 'bulk' mode, like some ETL at the evening will fill the data for whole day. Did I understand that correctly?
You can try the following trick:
1) use usual MergeTree / ReplicatedM... | 2018-04-12T17:22:25 | 2018-04-12T17:33:08 | {
"+1": 5
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 381,388,701 | 2,218 | sundy-li | Sounds great! | 2018-04-15T08:15:07 | 2018-04-15T08:15:07 | {
"+1": 2
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 381,465,207 | 1,661 | Slind14 | I'll try it with a union query.
1. table with +2 days old data, with the seen bool as column
2. table with the live data partitioned by day and a join to the second table with seen records
This is the best I could come up with in terms of keeping it simple, on one database and fast.
I also have a click data (no... | 2018-04-16T02:47:41 | 2018-04-16T02:49:45 | {} | NONE |
ClickHouse | ClickHouse | 382,119,146 | 2,230 | alexey-milovidov | Thank you! | 2018-04-17T19:49:22 | 2018-04-17T19:49:22 | {} | MEMBER |
ClickHouse | ClickHouse | 382,119,564 | 2,223 | alexey-milovidov | Ok. | 2018-04-17T19:50:34 | 2018-04-17T19:50:34 | {} | MEMBER |
ClickHouse | ClickHouse | 382,156,598 | 2,215 | alexey-milovidov | Спасибо - действительно, пропало при редактировании. | 2018-04-17T21:39:00 | 2018-04-17T21:39:00 | {} | MEMBER |
ClickHouse | ClickHouse | 382,205,667 | 2,233 | duaneya | ```bash
$ sudo dpkg-reconfigure tzdata
[sudo] duan 的密码：
Current default time zone: 'Asia/Shanghai'
Local time is now: Wed Apr 18 08:39:04 CST 2018.
Universal Time is now: Wed Apr 18 00:39:04 UTC 2018.
```
@filimonov That don't work,but I pull latest image and figure it out.Thanks. | 2018-04-18T00:45:20 | 2018-04-18T00:45:20 | {} | NONE |
ClickHouse | ClickHouse | 382,392,168 | 2,237 | ludv1x | It works fine in the latest 54378.
```
CREATE TABLE nullable
(
k UInt32,
v1 Nullable(UInt32),
v2 Nullable(UInt32)
)
ENGINE = MergeTree
ORDER BY k
Ok.
0 rows in set. Elapsed: 0.007 sec.
vludv-ub :) insert into nullable VALUES (1, 1, 1)
INSERT INTO nullable VALUES
Ok.
1 rows in... | 2018-04-18T13:47:15 | 2018-04-18T13:47:15 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 382,596,148 | 2,134 | alexey-milovidov | Thank you! | 2018-04-19T03:11:19 | 2018-04-19T03:11:19 | {} | MEMBER |
ClickHouse | ClickHouse | 382,597,213 | 1,925 | alexey-milovidov | Yes, this is very useful.
The only missing part is performance tests. | 2018-04-19T03:17:40 | 2018-04-19T03:17:40 | {} | MEMBER |
ClickHouse | ClickHouse | 380,786,305 | 2,199 | fahrettinf89 | @filimonov What is the best way to insert data both servers ? Are there any other functions like rand()? | 2018-04-12T12:23:21 | 2018-04-12T12:23:21 | {} | NONE |
ClickHouse | ClickHouse | 382,383,633 | 2,239 | jpiper | You're not using a distributed table are you? You'd need to run alter on all copies of the table.
i.e.
alter table X on cluster Y add column Z String | 2018-04-18T13:20:02 | 2018-04-18T13:20:02 | {} | NONE |
ClickHouse | ClickHouse | 382,488,886 | 2,169 | vavrusa | It's probably not supported by the SQL parser for anything else than MergeTree. You can configure it for the default user profile in [users.xml](https://github.com/yandex/ClickHouse/blob/master/dbms/src/Server/users.xml) like:
```
<yandex>
<profiles>
<default>
<max_block_size>100</max_bloc... | 2018-04-18T18:43:22 | 2018-04-18T18:43:22 | {
"+1": 4
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 382,612,686 | 777 | Slach | Latest Clickhouse release completelly broke zetcd compatibility
previous releases created table successfully
```
2018.04.19 05:00:53.427019 [ 3 ] <Trace> TCPHandlerFactory: TCP Request. Address: 172.18.0.10:53392
2018.04.19 05:00:53.427255 [ 3 ] <Debug> TCPHandler: Connected ClickHouse client version 1.1.54378, u... | 2018-04-19T05:14:01 | 2018-04-19T05:14:01 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 380,424,030 | 2,206 | den-crane | I experience the same issue. 50/50 need to send "kill -9" because stop works too long (once I waited for 1 hour and gave up).
It seems related to long merges (some of merges take 10 / 12 hours on my env.). So I am checking system.merges before initiating a stop. | 2018-04-11T11:48:37 | 2018-04-11T11:50:23 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 380,818,510 | 2,180 | proller | now tested on latest master:
```
mkdir -p build_client
cd build_client
cmake .. -DENABLE_CLICKHOUSE_ALL=0 -DENABLE_CLICKHOUSE_CLIENT=1 -DCMAKE_CXX_COMPILER=`which g++-7` -DCMAKE_C_COMPILER=`which gcc-7`
make -j clickhouse-client
dbms/src/Server/clickhouse-client --port 19000
СlickHouse client version 1.1.54377.
... | 2018-04-12T14:09:32 | 2018-04-12T14:09:32 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 381,254,736 | 2,216 | filimonov | Please check assumption #4 about accepted JSON here: https://clickhouse.yandex/docs/en/functions/json_functions/
> The JSON doesn't have space characters outside of string literals.
| 2018-04-13T20:39:10 | 2018-04-13T20:39:10 | {
"heart": 1
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 381,575,581 | 2,218 | jonatasfreitasv | Wow! Thx a lot @tannerlinsley ! | 2018-04-16T12:02:34 | 2018-04-16T12:02:34 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 381,614,946 | 2,211 | jonatasfreitasv | Why do not you run with POST in HTTP Interface? | 2018-04-16T14:17:27 | 2018-04-16T14:17:27 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 381,764,512 | 2,228 | derekperkins | When I rolled back to `1.1.54370`, everything initialized properly.
cc @alex-zaitsev | 2018-04-16T22:02:42 | 2018-04-16T22:02:42 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 382,390,874 | 2,237 | mighTY94 | How recent?
Clickhouse version - 1.1.54310 | 2018-04-18T13:43:24 | 2018-04-18T13:43:24 | {} | NONE |
ClickHouse | ClickHouse | 382,596,445 | 1,722 | alexey-milovidov | This is almost ready for merge. Alexey @ztlpn is doing final reviews. | 2018-04-19T03:13:22 | 2018-04-19T03:13:22 | {} | MEMBER |
ClickHouse | ClickHouse | 380,482,697 | 2,210 | filimonov | Generally it's something very specific. You can't really change primary key columns with that command. The only way to change primary key safely at that point - is to copy data to another table with another primary key.
`ALTER TABLE xxx MODIFY PRIMARY KEY (...)` allows you only to add _new_ (and empty) columns at ... | 2018-04-11T14:56:43 | 2018-04-11T18:48:25 | {
"+1": 2
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 381,226,537 | 2,206 | alexey-milovidov | It shouldn't be related to long merges or any other background operations as we are calcelling them on shutdown.
Could you please provide a stack trace of your server when it hangs on shutdown?
Just type `sudo gdb -batch -ex 't apply all bt' -p $(pidof clickhouse-server)` | 2018-04-13T18:42:23 | 2018-04-13T18:42:23 | {} | MEMBER |
ClickHouse | ClickHouse | 381,509,981 | 2,220 | xinchen0104 | # My config:
```xml
<rrc_ck_cluster>
<shard>
<internal_replication>false</internal_replication>
<replica>
<host>ch-00a</host>
<port>9000</port>
</replica>
<replica>
<... | 2018-04-16T07:47:57 | 2018-04-16T09:17:00 | {} | NONE |
ClickHouse | ClickHouse | 381,807,758 | 2,086 | meolu | @filimonov what filimonov say is right.
```sql
:) drop table pk_local;
:) CREATE TABLE default.pk_local ( id Int64, name String, age Int64, created_at Date, updated_at Date) ENGINE = ReplacingMergeTree(updated_at) partition by created_at order by id settings enable_vertical_merge_algorithm = 1, vertical_merge_a... | 2018-04-17T02:14:01 | 2018-04-17T02:14:01 | {} | NONE |
ClickHouse | ClickHouse | 382,383,719 | 2,169 | ennio1991 | @vavrusa any news? | 2018-04-18T13:20:18 | 2018-04-18T13:20:18 | {} | NONE |
ClickHouse | ClickHouse | 382,531,002 | 2,141 | alexey-milovidov | Pending:
- using hostnames in directory names of Distributed tables;
- using hostnames for names of nodes in ZK for distributed DDL. | 2018-04-18T21:12:18 | 2018-04-18T21:12:18 | {} | MEMBER |
ClickHouse | ClickHouse | 382,305,850 | 1,469 | tuedelue | I am also looking for a solution to this very problem: I would like to get running accumulates grouped by a certain key.
Any ideas? | 2018-04-18T08:22:40 | 2018-04-18T08:22:40 | {} | NONE |
ClickHouse | ClickHouse | 382,481,126 | 2,146 | ludv1x | It has been fixed in 1.1.54378. | 2018-04-18T18:18:23 | 2018-04-18T18:18:23 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 382,594,027 | 2,134 | alexey-milovidov | As I understand correctly, it just introduce optional parentheses around SELECT.
```
(SELECT 1) UNION ALL (SELECT 1)
```
This works.
```
(SELECT 1)
```
This also works.
(I wonder why these obvious cases are not covered in tests.)
Parentheses have no meaning and totally ignored.
So, why we don't imp... | 2018-04-19T02:56:54 | 2018-04-19T02:56:54 | {} | MEMBER |
ClickHouse | ClickHouse | 380,927,283 | 1,661 | Slind14 | > Clickhouse is not good/optimal for single row lookup, here a lot of such lookups will be needed.
> most recent data in RAM
hmm, when Dictionaries are cached in memory for quick access, why would it do one lookup for every key/value?
I have a few tables with ~100 million new rows per day, if we work with dai... | 2018-04-12T20:02:01 | 2018-04-12T20:02:01 | {} | NONE |
ClickHouse | ClickHouse | 380,930,087 | 1,661 | filimonov | Check that https://clickhouse.yandex/docs/en/dicts/external_dicts_dict_layout/
If flat or hashed will siute your needs - you don't need key/value lookups. For cache (which is slowest, but most universal) - key lookups needed.
| 2018-04-12T20:12:28 | 2018-04-12T20:12:28 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 381,865,358 | 2,228 | filimonov | Check the changelog ( #2215 ) regarding port reuse and listen configuration. | 2018-04-17T06:40:24 | 2018-04-17T06:45:25 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 381,970,211 | 2,231 | the-tinderbox | There is data sample in CSV format: [https://yadi.sk/d/FvSTeuE53UVHTw](https://yadi.sk/d/FvSTeuE53UVHTw) | 2018-04-17T12:14:07 | 2018-04-17T12:14:07 | {} | NONE |
ClickHouse | ClickHouse | 382,208,847 | 2,222 | sundy-li | #2223 | 2018-04-18T01:01:57 | 2018-04-18T01:01:57 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 382,596,741 | 1,886 | alexey-milovidov | Should we continue? | 2018-04-19T03:14:59 | 2018-04-19T03:14:59 | {} | MEMBER |
ClickHouse | ClickHouse | 380,921,528 | 1,661 | derekperkins | > For the external data source I'm probably good off with mongodb (if the look ups are fast enough, which they should), cassandra or redis
@Slind14 Dictionary lookups aren't done directly against your db, so it doesn't really matter what you use. ClickHouse polls your dictionary for changes and keeps the most recent... | 2018-04-12T19:40:46 | 2018-04-12T19:40:46 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 380,923,497 | 1,661 | derekperkins | https://clickhouse.yandex/docs/en/dicts/external_dicts_dict_lifetime/ | 2018-04-12T19:48:08 | 2018-04-12T19:48:08 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 380,925,606 | 1,661 | filimonov | > Doesn't this mean, that the latest records are not being read from the dictionary if they haven been merged which happens automatically at random times?
Yes. And It is not true realtime updates. It is some workaround for _some_ updates
Everything depends on concrete usage scenario.
> Dictionary lookups aren'... | 2018-04-12T19:55:42 | 2018-04-12T19:55:42 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 381,228,456 | 2,208 | alexey-milovidov | Thank you for the report! | 2018-04-13T18:49:55 | 2018-04-13T18:49:55 | {} | MEMBER |
ClickHouse | ClickHouse | 381,507,014 | 2,221 | meolu | # Question about ReplacingMergeTree
Optimize table will work while there 3 rows with same pk in table, but it fails while 2 rows with same pk. **Is there any params to set, so clickhouse will replacing merge when there 2 rows with same pk in table**? Thank you : )
## table
```sql
CREATE TABLE default.pk_local (... | 2018-04-16T07:36:26 | 2018-04-16T09:15:15 | {} | NONE |
ClickHouse | ClickHouse | 382,152,013 | 2,215 | filimonov | Кажется потерялась информация про port reuse.
> -* Сервер больше не использует по умолчанию опцию сокета SO_REUSEPORT. Теперь многократный listen одного адреса например `<listen_host>::</listen_host><listen_host>0.0.0.0</listen_host>` выдаст ошибку. Для включения этой опции добавьте в конфиг `<listen_reuse_port>1</l... | 2018-04-17T21:24:24 | 2018-04-17T21:24:24 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 382,427,521 | 2,232 | arduanov | Last stable build for homebrew
https://github.com/arduanov/homebrew-clickhouse | 2018-04-18T15:28:06 | 2018-04-18T15:28:06 | {
"+1": 4,
"-1": 2
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 382,594,278 | 2,134 | alexey-milovidov | Even this works:
```
((SELECT (((1)))))
``` | 2018-04-19T02:58:22 | 2018-04-19T02:58:22 | {} | MEMBER |
ClickHouse | ClickHouse | 381,255,156 | 2,216 | jonatasfreitasv | wow, nice, thx a lot. | 2018-04-13T20:41:05 | 2018-04-13T20:41:05 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 381,831,939 | 2,187 | millin | From now you can checkout official repo:
https://repo.yandex.ru/clickhouse/rpm/stable/x86_64/ | 2018-04-17T04:08:26 | 2018-04-17T04:08:26 | {
"+1": 1
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 381,947,723 | 2,228 | derekperkins | Thanks @filimonov! For future reference, here is the Google Translate version of the changelog. Adding `<listen_reuse_port>1</listen_reuse_port>` to my config.xml solved the problem immediately.
> The server no longer uses the SO_REUSEPORT socket option by default. Now a multiple listen of one address, for example <... | 2018-04-17T10:56:46 | 2018-04-17T10:56:46 | {
"+1": 5,
"heart": 2,
"rocket": 3
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 380,800,938 | 2,199 | filimonov | If you need to have the same data on 2 servers - you need a replication, not a shardind.
Please read answers here: https://github.com/yandex/ClickHouse/issues/2161 | 2018-04-12T13:15:36 | 2018-04-12T13:15:36 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 380,908,569 | 1,661 | Slind14 | > Updates in that scenario cost mostly nothing (as they happen outside of clickhouse). Selects will use external dict before 'fixing' the final value (by OPMIZIE FINAL), so it's more about performance of external dictionary during that time. But that allows you to store outside only limited amount of data (for last day... | 2018-04-12T18:52:52 | 2018-04-12T18:52:52 | {} | NONE |
ClickHouse | ClickHouse | 381,037,250 | 1,661 | filimonov | You said that you can have about 100 mln 'seen' records dayly - that means that your dict will have ~16Gb size overall for one day (100 mln * 16 chars for your uuid + some overhaed). `hashed` or `flat` need to reread *whole* table after each change. So in that case after every update of underlying table (and you want t... | 2018-04-13T06:29:33 | 2018-04-13T06:30:53 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 381,039,294 | 1,661 | filimonov | Another variant - if that "seen" event can happen only once in short timeframe - may be it would be optimal to you just to store one day of data in some other DB outside of clickhouse, and just copy it to clickchouse after that 'mutable' period? So some external DB like mysql/postgres/whatever will store data for one d... | 2018-04-13T06:41:00 | 2018-04-13T06:41:00 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 381,478,513 | 2,218 | tannerlinsley | I can be of assistance. | 2018-04-16T04:44:36 | 2018-04-16T04:44:36 | {
"hooray": 1
} | NONE |
ClickHouse | ClickHouse | 381,850,605 | 2,187 | jonatasfreitasv | @alex-zaitsev and @millin
The 1.1.54370 version is stable for now ? | 2018-04-17T05:40:16 | 2018-04-17T05:40:16 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 382,605,670 | 1,925 | alexey-milovidov | How to use performance test:
- place a table with name `hits_100m_single` containing `URL` column with real dataset on your server;
- run `clickhouse performance-test --lite .` in `performance/url` directory and compare results before and after your changes.
- compare specific queries with `clickhouse-benchmark`; lo... | 2018-04-19T04:19:02 | 2018-04-19T04:20:20 | {} | MEMBER |
ClickHouse | ClickHouse | 382,609,398 | 2,164 | alexey-milovidov | Security issue: the user can exploit storage `File` and table function `file` to read arbitary data from databases without access granted. | 2018-04-19T04:48:48 | 2018-04-19T04:48:48 | {} | MEMBER |
ClickHouse | ClickHouse | 380,893,057 | 1,661 | Slind14 | Wow, thank you. This sounds like a great work around.
I get the seen records whenever the event triggers, meaning at any given time of the day, just delayed to the impressions.
Do you think it is doable - performance wise - to update the data more frequently e.g. every minute or at least every 5 minutes?
Could I us... | 2018-04-12T18:00:35 | 2018-04-12T18:09:04 | {} | NONE |
ClickHouse | ClickHouse | 381,188,364 | 1,661 | Slind14 | hmm, since 100mln records daily is by table, it is a lot more if we look at all the tables. So yeah, its not going to work memory wise.
I'm still having trouble believing how even a fast storage like redis would be able to deliver tens of millions of single lookups in a matter of milliseconds.
The issue with anot... | 2018-04-13T16:23:48 | 2018-04-13T16:23:48 | {} | NONE |
ClickHouse | ClickHouse | 381,276,589 | 2,070 | k0st1an | Сonfirm the problem | 2018-04-13T22:35:20 | 2018-04-13T22:35:20 | {} | NONE |
ClickHouse | ClickHouse | 381,461,905 | 2,206 | badvir |
```
[root@localhost ~]# service clickhouse-server stop
Stop clickhouse-server service:
```
> Results from another terminal
```
[root@localhost ~]# gdb -batch -ex 't apply all bt' -p $(pidof clickhouse-server)
[New LWP 7536]
[New LWP 7537]
[New LWP 7539]
[New LWP 7549]
[New LWP 7550]
[New LWP 7551]
... | 2018-04-16T02:22:15 | 2018-04-16T02:22:15 | {} | NONE |
ClickHouse | ClickHouse | 382,386,815 | 2,239 | sundy-li | closed，this problem was caused by the distribute table create sql，the order of the column must be same with local table | 2018-04-18T13:30:27 | 2018-04-18T13:30:27 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 382,590,951 | 2,201 | alexey-milovidov | No mention about behaviour on NULLs and NaNs. | 2018-04-19T02:36:25 | 2018-04-19T02:36:25 | {} | MEMBER |
ClickHouse | ClickHouse | 380,816,434 | 1,966 | chipitsine | "clickhouse-local в основном используется чтоб какие-то простые вещи в command line делать без запуска сервера" - задача вполне попадала под это условие.
если вы с какой-то целью включаете clickhouse-local в дистрибутив, то его могут запускать. если вы не хотите, чтобы его запускали - не включайте в дистрибутив | 2018-04-12T14:03:52 | 2018-04-12T14:03:52 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 381,972,731 | 2,231 | kshvakov | Reproduced (version 1.1.54343):
```sql
CREATE TABLE Issue_2231_Invalid_Nested_Columns_Size (
Date Date,
NestedColumn Nested(
ID Int32,
Count Int64
)
) Engine = MergeTree
PARTITION BY tuple()
ORDER BY Date;
INSERT INTO Issue_2231_Invalid_Nested_Columns_Size VALUES ... | 2018-04-17T12:22:26 | 2018-04-17T12:22:26 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 382,389,380 | 2,237 | ludv1x | What ClickHouse version do you use?
We have fixed several bugs related to Nullable columns and logical operators in recent releases. | 2018-04-18T13:38:34 | 2018-04-18T13:38:34 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 382,591,351 | 2,201 | alexey-milovidov | These functions should be in "higher order functions" as they accept a function as an optional first argument. "Schwartzsort" algorithm should be mentioned with a link to Wikipedia article. | 2018-04-19T02:39:15 | 2018-04-19T02:39:15 | {} | MEMBER |
ClickHouse | ClickHouse | 380,935,572 | 1,661 | Slind14 | My keys are binary UUIDs. So I should probably go with hashed?
Now since it states that it is "completely" stored in memory, doesn't this mean that I can use a clickhouse table where I don't have issues with write and read performance as long as writes are batched and the read being one query for all the key/values?
... | 2018-04-12T20:32:58 | 2018-04-12T20:32:58 | {} | NONE |
ClickHouse | ClickHouse | 381,642,004 | 2,225 | sundy-li | You could build a restful web server to handle the front queries, and send the query to ClickHouse by http or native tcp client. In this way, the web server is just a query proxy, and you can do anything to have security control of queries.
May be you could try the client https://github.com/kshvakov/clickhouse. ... | 2018-04-16T15:18:15 | 2018-04-16T15:18:31 | {
"+1": 2
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 381,852,257 | 2,225 | jpiper | Try https://github.com/Vertamedia/chproxy | 2018-04-17T05:47:17 | 2018-04-17T05:47:17 | {
"+1": 2
} | NONE |
ClickHouse | ClickHouse | 382,380,500 | 2,147 | den-crane | I realized now that I did typo "SELECT (a, b) " it's a tuple so the right behaviour is error "Bad type of Field"
But still v1.1.54378 returns no error in 10% of executions.
```
:) SELECT * FROM trololo WHERE (a, b) IN (SELECT (a, b) FROM trololo);
SELECT *
FROM trololo
WHERE (a, b) IN
(
SELECT (a, b)
... | 2018-04-18T13:09:24 | 2018-04-18T13:09:24 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 382,595,079 | 2,134 | alexey-milovidov | > So, why we don't implement this in more natural way - just allow optional parentheses?
Current solution is Ok. | 2018-04-19T03:03:54 | 2018-04-19T03:03:54 | {} | MEMBER |
ClickHouse | ClickHouse | 382,609,595 | 2,164 | alexey-milovidov | You can add integration test for this feature. Look at `tests/integration`. | 2018-04-19T04:50:28 | 2018-04-19T04:50:28 | {} | MEMBER |
ClickHouse | ClickHouse | 380,856,560 | 1,661 | Slind14 | Unfortunately I still haven't found a way which performs with hundreds of million of rows for real time data. If only the data would be available at the same time and could be written with one query.
I'm seeing upwards 80 million row/s when selecting data from a single table. As soon as I join or or make use of Coll... | 2018-04-12T15:59:43 | 2018-04-12T15:59:43 | {} | NONE |
ClickHouse | ClickHouse | 381,477,151 | 2,218 | jonatasfreitasv | *Last commits:*
- Implement query post and ace editor
- Implement React Table
- Added basic layout with React Ant Design
| 2018-04-16T04:30:54 | 2018-04-16T04:30:54 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 381,808,701 | 2,221 | meolu | Ok, I figure out what's the right way to use ReplacingMergeTree, It just a skill `optimize table xx` after insert **every time**.
```sql
:) drop table pk_local;
:) CREATE TABLE default.pk_local ( id Int64, name String, age Int64, created_at Date, updated_at Date) ENGINE = ReplacingMergeTree(updated_at) partit... | 2018-04-17T02:17:49 | 2018-04-17T02:17:49 | {} | NONE |
ClickHouse | ClickHouse | 382,050,145 | 2,233 | filimonov | Check your system timezone settings.
Try:
```bash
sudo dpkg-reconfigure tzdata
``` | 2018-04-17T16:09:24 | 2018-04-17T16:09:24 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 382,204,714 | 2,234 | jonatasfreitasv | @arduanov For dev environment, Docker is best option. | 2018-04-18T00:40:24 | 2018-04-18T00:40:24 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 382,205,485 | 2,232 | jonatasfreitasv | For dev env, Docker is best option. | 2018-04-18T00:44:19 | 2018-04-18T00:44:19 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 382,384,321 | 2,239 | sundy-li | @jpiper surly I did that. see the last line of my question you will find that | 2018-04-18T13:22:26 | 2018-04-18T13:22:26 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 382,609,913 | 2,164 | alexey-milovidov | > Security issue
Possible solution: allow to read files only from inside `path/files` directory.
(`files` subdirectory will be reserved for user provided files for tables) | 2018-04-19T04:52:56 | 2018-04-19T04:52:56 | {} | MEMBER |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.