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 | 448,358,100 | 3,811 | alexey-milovidov | https://github.com/yandex/ClickHouse/commit/d5d1c34017729640fcbe72a2bbafaed2523de15d | 2018-12-18T20:25:52 | 2018-12-18T20:25:52 | {} | MEMBER |
ClickHouse | ClickHouse | 448,439,954 | 3,721 | alexey-milovidov | ```
mtlog-perftest03j.yandex.ru :) SELECT count() FROM system.numbers WHERE NOT ignore(roundAge(rand() % 100))
SELECT count()
FROM system.numbers
WHERE NOT ignore(roundAge(rand() % 100))
↙ Progress: 2.41 billion rows, 19.26 GB (109.80 million rows/s., 878.44 MB/s.) Cancelling query.
Ok.
Query was cancelled.... | 2018-12-19T01:49:42 | 2018-12-19T01:49:42 | {} | MEMBER |
ClickHouse | ClickHouse | 448,708,229 | 2,440 | knoguchi | I know this is a closed PR but I tested anyway. The biggest problem is it can't handle multiple PB messages. The Kafka storage polls Kafka, then the messages consumed are batched up and copied into a buffer. The delimiter option in Kafka storage won't work because PB is not a delimiter safe format unfortunately. ... | 2018-12-19T19:03:44 | 2018-12-19T19:21:44 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 448,738,247 | 3,869 | den-crane | CollapsingMergeTree(**Sign**)
ORDER BY **UserID**;
INSERT INTO test3 VALUES (1 .. 1);
INSERT INTO test3 VALUES (1 .. 1),(1 .. 1),(1 .. -1),(1 .. -1);
INSERT INTO test3 VALUES(1 .. -1)
1 + 1 + 1 - 1 - 1 -1
https://clickhouse.yandex/docs/en/operations/table_engines/versionedcollapsingmergetree/#algorithm
W... | 2018-12-19T20:47:22 | 2018-12-19T20:47:22 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 448,816,082 | 3,876 | qietingfengling | My ram has 16G, there are 100W rows of data per second. Every query has more than 1 hour of data. Do I need to increase the value of index_granularity? | 2018-12-20T01:03:56 | 2018-12-20T01:03:56 | {} | NONE |
ClickHouse | ClickHouse | 449,003,272 | 3,870 | zhang2014 | Old tests are keep in:
- https://github.com/yandex/ClickHouse/pull/3870/files#diff-a485106b9dfd012a710477225331f6daR27
- https://github.com/yandex/ClickHouse/pull/3870/files#diff-a485106b9dfd012a710477225331f6daR36
- https://github.com/yandex/ClickHouse/pull/3870/files#diff-a485106b9dfd012a710477225331f6daR49
Sho... | 2018-12-20T13:42:58 | 2018-12-20T13:44:23 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 449,162,729 | 3,851 | kanterov | @abyss7 @alexey-milovidov cool! thanks for taking care of this! | 2018-12-20T22:57:55 | 2018-12-20T22:57:55 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 449,358,954 | 3,817 | blinkov | @denisovser1406 I think we have discussed this on email, do you have any further questions? | 2018-12-21T11:02:58 | 2018-12-21T11:02:58 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 449,368,663 | 3,893 | blinkov | @sliontc it sounds like the timeout you see is not between ClickHouse and MySQL, but between your desktop client and ClickHouse.
What happens if you open http://192.168.1.37:8123/ping in web browser? | 2018-12-21T11:49:23 | 2018-12-21T11:49:23 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 448,442,182 | 3,721 | alexey-milovidov | Note that there are many optimization possibilities:
- SIMD optimization of comparison;
- better layout for binary search;
- no binary search for small arrays. | 2018-12-19T02:01:41 | 2018-12-19T02:01:41 | {} | MEMBER |
ClickHouse | ClickHouse | 448,653,875 | 3,874 | alexey-milovidov | Already comitted. | 2018-12-19T16:18:26 | 2018-12-19T16:18:26 | {} | MEMBER |
ClickHouse | ClickHouse | 448,939,017 | 3,871 | DChistyakov | Yes, looks like it is! Thank you very much. Hope it will solve after hardware replace. | 2018-12-20T09:53:59 | 2018-12-20T09:53:59 | {} | NONE |
ClickHouse | ClickHouse | 449,047,559 | 3,885 | den-crane | > Maybe we don't push down in this case? or , do you have any good suggestions ? :
Of course you should not. The same as with order by limit ... and so on. | 2018-12-20T16:03:29 | 2018-12-20T17:00:12 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 449,085,101 | 3,826 | blinkov | @alexey-milovidov there's a test for it's implicit call from one of `trim` variants | 2018-12-20T18:02:45 | 2018-12-20T18:02:45 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 449,228,833 | 3,890 | zhang2014 | Maybe we should tag these functions in the function interface?
For example:
``` SQL
CREATE TABLE order_view AS SELECT * FROM table_xxx ORDER BY create_time;
SELECT * FROM order_view WHERE name = 'XXX' AND create_time = 'XXX';
```
In this case, optimization makes sense. | 2018-12-21T03:25:25 | 2018-12-21T03:25:25 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 449,350,992 | 3,714 | blinkov | Added in #3826, will be available in next release. | 2018-12-21T10:28:02 | 2018-12-21T10:28:02 | {
"+1": 1
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 449,392,967 | 3,893 | sliontc | It returns "Ok.". Seem no errors. | 2018-12-21T13:48:19 | 2018-12-21T13:48:19 | {} | NONE |
ClickHouse | ClickHouse | 448,304,838 | 3,849 | ztlpn | I have meant that the server must finish running destructors before responding to the Cancel packet, but the Cancel packet doesn't need a response! So the mystery remains. Also, if you wait minutes before issuing the second query and the slowdown still persists, destructors of aggregate functions can't be the reason.
... | 2018-12-18T17:36:02 | 2018-12-18T17:36:02 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 448,310,043 | 3,597 | ztlpn | Hi @5cmpersec, the problem arises because the `i` column is in the index. It is not specific to mutations (try executing `SELECT * FROM test WHERE i IN mset` - you'll get the same error). BTW it has already been reported: #2542 | 2018-12-18T17:52:04 | 2018-12-18T17:52:04 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 448,539,718 | 3,849 | crakjie | Ok, found the issue. Thank you for pointing me in the right direction. The problem came from the connection settings. The driver was not using the `tcp no delay` parameter, making it waiting a little time to see if the connection buffer is full. Sending the Cancel must have filled this buffer before the "buffer timeout... | 2018-12-19T10:07:27 | 2018-12-19T10:07:27 | {} | NONE |
ClickHouse | ClickHouse | 448,596,423 | 3,868 | Jack012a | Thank you! | 2018-12-19T13:27:43 | 2018-12-19T13:27:43 | {} | NONE |
ClickHouse | ClickHouse | 448,669,886 | 3,845 | f41gh7 | awesome, thx! | 2018-12-19T17:03:16 | 2018-12-19T17:03:16 | {} | NONE |
ClickHouse | ClickHouse | 449,000,240 | 3,877 | ztlpn | You can use the shard number as `replica_group` (`01`, `02`, etc.) and the hostname (`host-ch-0X`) as `replica`.
ZooKeeper is needed for Replicated tables. In your current configuration there is only one ZooKeeper node. This configuration is not fault-tolerant - if the ZooKeeper node goes down, the Replicated tables... | 2018-12-20T13:31:40 | 2018-12-20T13:31:40 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 449,082,584 | 3,826 | alexey-milovidov | No test for `regexpQuoteMeta`? | 2018-12-20T17:54:12 | 2018-12-20T17:54:12 | {} | MEMBER |
ClickHouse | ClickHouse | 449,122,129 | 3,876 | den-crane | you'd better leave it as default 8192. | 2018-12-20T20:13:49 | 2018-12-20T20:13:49 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 449,351,210 | 3,705 | blinkov | Added in #3826, will be available in next release. | 2018-12-21T10:28:43 | 2018-12-21T10:28:43 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 449,369,094 | 3,814 | blinkov | Let's keep only #2675 for this. | 2018-12-21T11:51:27 | 2018-12-21T11:51:27 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 449,397,886 | 3,893 | sliontc | Yes, it should be the db client issue. When I load data in local clickhouse-client, no such timeout present, as shown below:
| 2018-12-21T14:09:32 | 2018-12-21T14:09:32 | {} | NONE |
ClickHouse | ClickHouse | 448,246,516 | 3,858 | ztlpn | Hi, can you please state your question with more detail?
You can create views (as well as ordinary tables) on a cluster that you configured in the `<remote_servers>` section of `config.xml` with the [ON CLUSTER command](https://clickhouse.yandex/docs/en/query_language/create/#distributed-ddl-queries-on-cluster-claus... | 2018-12-18T14:52:03 | 2018-12-18T14:52:03 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 448,415,078 | 3,868 | den-crane | >Based on my understanding, ...
it's not quite right. There is a merge scheduler which plans merges for a table to reduce number of parts without causing I/O starvation. All merges will be finished eventually = never.
You should not bother about unmerged data. 99.99% poor performance cases related to other issues... | 2018-12-18T23:47:53 | 2018-12-18T23:49:45 | {
"+1": 6
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 449,351,570 | 3,704 | blinkov | Case insensitivity for some additional functions (including listed here) was introduced in #3826, will be available in next release. | 2018-12-21T10:30:11 | 2018-12-21T10:30:11 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 448,320,343 | 3,590 | abyss7 | We decided to forbid such kind of subqueries with AggregateFunction states - as a workaround your query may be simply transformed to another one:
SELECT * FROM ( SELECT uniqState('') ) | 2018-12-18T18:24:58 | 2018-12-18T18:24:58 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 448,437,393 | 3,721 | alexey-milovidov | Escpecially performance comparison against `roundAge` and `roundDuration`. | 2018-12-19T01:36:49 | 2018-12-19T01:36:49 | {} | MEMBER |
ClickHouse | ClickHouse | 448,531,640 | 3,846 | AlexanderJLiu | @ztlpn Got it, Thanks very much 👍 | 2018-12-19T09:40:07 | 2018-12-19T09:40:07 | {} | NONE |
ClickHouse | ClickHouse | 448,618,054 | 3,591 | abyss7 | PR #3875 | 2018-12-19T14:38:48 | 2018-12-19T14:38:48 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 448,998,755 | 3,870 | zhang2014 | How to build `Functional stateless tests (debug)`
In my environment, -DCMAKE_BUILD_TYPE=Release and -DCMAKE_BUILD_TYPE=Debug both pass | 2018-12-20T13:25:51 | 2018-12-20T13:25:51 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 449,001,524 | 3,870 | alexey-milovidov | Let's also retain the old test,
and add your test in a new file. | 2018-12-20T13:36:28 | 2018-12-20T13:36:28 | {} | MEMBER |
ClickHouse | ClickHouse | 449,007,554 | 3,870 | alexey-milovidov | Yes. Just in case, it's reasonable to also keep the existing case separate.
Because if both tests will be broken, the old test may be simpler for investigation. | 2018-12-20T13:58:08 | 2018-12-20T13:58:08 | {} | MEMBER |
ClickHouse | ClickHouse | 449,351,065 | 3,712 | blinkov | Added in #3826, will be available in next release. | 2018-12-21T10:28:13 | 2018-12-21T10:28:13 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 448,239,972 | 3,846 | ztlpn | You need to stop insertions into the tables, copy ZooKeeper data to the new cluster, change the `<zookeeper>` section in the config and restart ClickHouse.
The most tedious part is copying the data from the old ZooKeeper cluster to the new one. You should NOT use tools such as https://github.com/ksprojects/zkcopy be... | 2018-12-18T14:32:13 | 2018-12-18T14:32:13 | {
"+1": 3
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 448,441,101 | 3,721 | alexey-milovidov | For extra large arrays the performance is down due to array preparation on each block. That's expected (we may use PreparedFunction to make it better).
```
SELECT count() FROM system.numbers_mt WHERE NOT ignore(roundDown(rand() % 65536, (SELECT groupArray(number) FROM numbers(65536))))
``` | 2018-12-19T01:55:53 | 2018-12-19T01:55:53 | {} | MEMBER |
ClickHouse | ClickHouse | 449,045,087 | 3,884 | den-crane | hm, the issue seems in Nullable
workaround:
INSERT INTO temp VALUES (5), (10);
SELECT * FROM temp WHERE name1 IN ( cast(5 as Nullable(Decimal(38,12))) )
5.000000000000
| 2018-12-20T15:55:57 | 2018-12-20T15:56:06 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 449,087,031 | 3,826 | alexey-milovidov | ```
mtlog-perftest03j.yandex.ru :) SELECT URL FROM hits_1000m_transformed WHERE URL != regexpQuoteMeta(URL) LIMIT 100
SELECT URL
FROM hits_1000m_transformed
WHERE URL != regexpQuoteMeta(URL)
LIMIT 100
Received exception from server (version 18.14.17):
Code: 241. DB::Exception: Received from localhost:9000, ... | 2018-12-20T18:09:47 | 2018-12-20T18:09:47 | {} | MEMBER |
ClickHouse | ClickHouse | 449,389,653 | 3,891 | ztlpn | Can you look up log messages with `ReplicatedMergeTreeAlterThread` in them on replicas that couldn't apply ALTER? | 2018-12-21T13:33:46 | 2018-12-21T13:33:46 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 448,310,442 | 3,597 | ztlpn | I'm going to close this issue because the fix for the default database problem has been released in 18.16.0. | 2018-12-18T17:53:28 | 2018-12-18T17:53:28 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 448,669,781 | 3,829 | alesapin | v18.14.19-stable released, please update and try to disable settings. This fix is almost the only in this version. | 2018-12-19T17:02:58 | 2018-12-19T17:03:35 | {} | MEMBER |
ClickHouse | ClickHouse | 448,908,007 | 3,830 | blinkov | @TH-HA in short term (4) sounds like the easiest solution, I've asked some people who use ClickHouse+Tableau to check it out.
As far as I can see from Tableau documentation, the config might have looked like this (not tested myself though): https://github.com/yandex/clickhouse-odbc/blob/clickhouse-tbc/clickhouse.tdc
... | 2018-12-20T08:02:57 | 2018-12-20T08:02:57 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 449,086,568 | 3,866 | heyciao | Replaced the big SummingMergeTree materialized view with an AggregatingMergeTree and ran the usual workload for a day => the average hourly amount of inserted records and queries that ran against the matview is the same as with SummingMergeTree => no difference => closing this ticket.
Cheers | 2018-12-20T18:08:08 | 2018-12-20T18:08:08 | {} | NONE |
ClickHouse | ClickHouse | 449,332,053 | 3,891 | vasyaabr | After downgrade to 18.14.11 problem disapeared. | 2018-12-21T09:33:13 | 2018-12-21T09:33:13 | {} | NONE |
ClickHouse | ClickHouse | 449,382,353 | 3,573 | abyss7 | We're thinking about secondary indices, and for now the most probable solution is an auxiliary minmax index with the same granularity as a primary index. | 2018-12-21T13:00:00 | 2018-12-21T13:00:00 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 448,289,482 | 3,245 | dqminh | This is too old. We dont see such errors now on 18.14.17. | 2018-12-18T16:50:35 | 2018-12-18T16:50:35 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 448,424,217 | 3,721 | alexey-milovidov | Some minor details remain. I will merge it myself. | 2018-12-19T00:30:54 | 2018-12-19T00:30:54 | {} | MEMBER |
ClickHouse | ClickHouse | 448,440,666 | 3,721 | alexey-milovidov | ```
mtlog-perftest03j.yandex.ru :) SELECT count() FROM system.numbers WHERE NOT ignore(roundDuration(rand() % 65536))
SELECT count()
FROM system.numbers
WHERE NOT ignore(roundDuration(rand() % 65536))
↙ Progress: 6.22 billion rows, 49.76 GB (121.83 million rows/s., 974.66 MB/s.) Cancelling query.
Ok.
Query ... | 2018-12-19T01:53:31 | 2018-12-19T01:53:31 | {} | MEMBER |
ClickHouse | ClickHouse | 448,509,874 | 3,829 | alesapin | @angeldestroyer
odbc-bridge created pipe (stdout) for logging, which nobody read from. Logging messages were written to this pipe until it reaches 64K (default linux pipe size) and everything works perfectly. After that attempt to log any message led to hang up on write syscal (pipe were overflown). | 2018-12-19T08:27:31 | 2018-12-19T08:27:44 | {} | MEMBER |
ClickHouse | ClickHouse | 448,540,447 | 3,829 | k0st1an | @angeldestroyer After the changes, everything works well. | 2018-12-19T10:09:57 | 2018-12-19T10:09:57 | {} | NONE |
ClickHouse | ClickHouse | 448,830,849 | 3,847 | Crazylearner30 | I was not entering it in the shell instead I was doing it in clickhouse-client prompt.
Corrected it
Thanks for the suggestions. | 2018-12-20T02:00:16 | 2018-12-20T02:00:16 | {} | NONE |
ClickHouse | ClickHouse | 449,016,574 | 3,858 | ztlpn | Just as with ordinary tables:
```sql
CREATE MATERIALIZED VIEW default.vm_test_table ON CLUSTER ck_cluster_162
(applicationId Int32, ...)
ENGINE ...
AS SELECT ...
``` | 2018-12-20T14:28:07 | 2018-12-20T14:28:07 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 449,371,824 | 3,593 | blinkov | The roadmap in docs has been recently updated: https://clickhouse.yandex/docs/en/roadmap/
There's also an alternative more verbose blog post by Altinity: https://www.altinity.com/blog/clickhouse-roadmap-2019 | 2018-12-21T12:05:14 | 2018-12-21T12:05:14 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 448,300,503 | 3,845 | ztlpn | Hi,
The problem in your case arises because ClickHouse caches the resolved IP addresses of replicas. The cache should be flushed when there is a surge of network errors, but this functionality is broken now. A workaround is to manually issue the `SYSTEM DROP DNS CACHE` command.
Also note that the IP addresses th... | 2018-12-18T17:22:56 | 2018-12-18T17:22:56 | {
"+1": 1
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 448,489,334 | 3,829 | snvad | I don't know why, but after adding the specified logging settings, the error is not reproduced. Dictionaries reload in 2-3 minutes. Previously, you could wait more than 10 minutes and nothing happened or see timeout error.
@k0st1an, what about you? | 2018-12-19T06:55:20 | 2018-12-19T06:55:20 | {} | NONE |
ClickHouse | ClickHouse | 448,212,985 | 3,849 | ztlpn | Hi @crakjie, the issue would be easier to debug if you provide a self-contained test case. I assume that you are using one of the native drivers - can you write a script using this driver + a database schema that clearly reproduces the problem? Thanks. | 2018-12-18T13:01:19 | 2018-12-18T13:01:19 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 448,243,594 | 3,847 | ztlpn | The error
```
Syntax error: failed at position 1:
clickhouse-client -h1.0.1.2.3 -u default --password xxxx --query="INSERT INTO insightbasev2.company_temp FORMAT CSV" </temp/company_1/4.csv
Expected one of:
```
will appear if you enter the `clickhouse-client -h...` command not in the shell prompt but in the ... | 2018-12-18T14:43:15 | 2018-12-18T14:43:15 | {
"+1": 1
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 448,322,941 | 3,590 | code-of-kpp | I wanned to use this subquery in ALTER TABLE statement to add a new column to AggregatingMergeTree table.
@alexey-milovidov showed me another way to achieve this via `arrayReduce('uniqState', CAST([] AS Array(String)))`
but in the end I still had to recreate table and populate it with `INSERT ... SELECT *, new_f... | 2018-12-18T18:33:25 | 2018-12-18T18:33:25 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 448,329,337 | 3,864 | alexey-milovidov | Спасибо! | 2018-12-18T18:52:54 | 2018-12-18T18:52:54 | {} | MEMBER |
ClickHouse | ClickHouse | 448,611,520 | 3,871 | den-crane | Maybe a some clickhouse-client issue when it runs on machine without some ch configs or different ch-cli vs cl-ser versions.
Try curl
curl -s -w "%{time_total}\n" http://user:password@serverA:8123/?query=select%201 | 2018-12-19T14:18:50 | 2018-12-19T14:23:05 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 448,667,533 | 3,876 | den-crane | The right value of index_granularity depends of queries and number of rows in a table and RAM of your server.
So if you need to search rows by exact primary key value
(OLTP query -- which is completely against CH nature as OLAP database)
select ... from T where PK_field(s) in (1,2,3)
you can decrease index_granul... | 2018-12-19T16:56:51 | 2018-12-19T16:56:51 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 448,777,657 | 3,877 | den-crane | you are doing something confusing.
2 shard * 2 repica :
```
<yandex>
<remote_servers>
<cluster2shard2replica>
<shard><internal_replication>true</internal_replication>
<replica><host>clickhouse-primary1</host><port>9000</port></replica>
... | 2018-12-19T23:01:57 | 2018-12-19T23:07:41 | {
"+1": 1
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 449,045,758 | 3,885 | zhang2014 | @den-crane Because the calculation of the function depends on the order of the data.
@alexey-milovidov Maybe we don't push down in this case? or , do you have any good suggestions ? : ) | 2018-12-20T15:57:59 | 2018-12-20T15:57:59 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 449,083,673 | 3,884 | ztlpn | @4ertus2 can you take a look?
It seems that [this function](https://github.com/yandex/ClickHouse/blob/c77930fe30824ff09a005b11ebd23d4d74c6d6ad/dbms/src/Interpreters/convertFieldToType.cpp#L106) should support conversions from other Decimal types.
E.g. this works (because types are the same, but anyway, the `conve... | 2018-12-20T17:57:57 | 2018-12-20T17:59:09 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 449,363,009 | 3,817 | denisovser1406 | Yes,thank you for your help!
Sent from my Samsung Galaxy smartphone.
-------- Original message --------From: Ivan Blinkov <notifications@github.com> Date: 12/21/18 20:03 (GMT+09:00) To: yandex/ClickHouse <ClickHouse@noreply.github.com> Cc: denisovser1406 <denisovser1406@gmail.com>, Mention <mention@noreply.github.c... | 2018-12-21T11:21:17 | 2018-12-21T11:21:17 | {} | NONE |
ClickHouse | ClickHouse | 449,395,241 | 3,891 | ztlpn | Is this the first line with the `session expired` exception? I need the first similar error and some lines preceding it. If it is possible, just upload the logs somewhere so that I can look at them, that will be easier. | 2018-12-21T13:58:08 | 2018-12-21T13:58:08 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 448,330,258 | 3,730 | alexey-milovidov | @4ertus2
Implementation of `quantile`, `quantileWeighted` should be straightforward (decimals are processed as scaled integers).
`quantileTDigest` implies conversion to Float32.
`quantileTiming` implies conversion to UInt16.
| 2018-12-18T18:55:40 | 2018-12-18T18:55:40 | {} | MEMBER |
ClickHouse | ClickHouse | 448,818,901 | 3,877 | qietingfengling | Hello, I have tried this configuration, but I am not very clear about the role of zookeeper in this configuration? The statement I created for the ReplicatedMergeTree table is ENGINE = ReplicatedMergeTree('/ch/replica_{replica_group}database_name/table_name', '{replica}'..., how should I change the parameters replica_g... | 2018-12-20T01:12:21 | 2018-12-20T01:12:21 | {} | NONE |
ClickHouse | ClickHouse | 448,283,663 | 3,849 | crakjie | @ztlpn I'm using my own driver so I can't really send you snippet now.
The query does not use any aggregation. The query is pretty simple :
`select toString(Currency) from (select distinct(Currency) from Values limit 2000)`
The table contains 91054 lines an there is just 30 distinct currency.
"immediately" in my... | 2018-12-18T16:34:28 | 2018-12-18T16:35:16 | {} | NONE |
ClickHouse | ClickHouse | 448,723,016 | 2,440 | alexey-milovidov | We have started another task as a replacement for this PR.
Protobuf messages will be delimited by size, as in https://github.com/protocolbuffers/protobuf/blob/master/src/google/protobuf/util/delimited_message_util.h | 2018-12-19T19:53:09 | 2018-12-19T19:53:09 | {} | MEMBER |
ClickHouse | ClickHouse | 448,320,634 | 3,839 | ztlpn | This is a duplicate of #3779 (the error is different because in your case the query is INSERT SELECT) | 2018-12-18T18:25:50 | 2018-12-18T18:25:50 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 448,433,077 | 3,721 | alexey-milovidov | You've edited the test reference file manually and tabs are accidentially replaced by spaces. That's why your test has failed. I will fix it. | 2018-12-19T01:15:23 | 2018-12-19T01:15:23 | {} | MEMBER |
ClickHouse | ClickHouse | 448,436,831 | 3,721 | alexey-milovidov | Performance evaluation remains... | 2018-12-19T01:33:47 | 2018-12-19T01:33:47 | {} | MEMBER |
ClickHouse | ClickHouse | 448,450,966 | 3,858 | MrWhiteL | @ztlpn
There is two nodes in my cluster , I want to create view by my program on one node and need not to connect to another one to execute it once more. create table can use " on cluster ck_cluster_162 " but create view how?
![image](https://user-images.githubusercontent.com/32286846/50195893-ec30e980-037b-11e9-8... | 2018-12-19T02:52:12 | 2018-12-19T02:56:26 | {} | NONE |
ClickHouse | ClickHouse | 448,618,204 | 3,871 | DChistyakov | dc@severA:~$ curl -s -w "%{time_total}\n" http://serverA:8123/?query=select%201
1
0.001034
dc@serverB:~$ curl -s -w "%{time_total}\n" http://serverA:8123/?query=select%201
1
0.180937 | 2018-12-19T14:39:16 | 2018-12-19T14:39:32 | {} | NONE |
ClickHouse | ClickHouse | 448,672,664 | 3,109 | ztlpn | quoting removed in v18.16.0 | 2018-12-19T17:11:35 | 2018-12-19T17:11:35 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 448,723,291 | 2,440 | alexey-milovidov | If your messages in Kafka will be prefixed by size in appropriate format (varint), the Kafka storage will be able to read them from Protobuf format. | 2018-12-19T19:54:06 | 2018-12-19T19:54:06 | {} | MEMBER |
ClickHouse | ClickHouse | 448,723,501 | 2,440 | alexey-milovidov | Short task description in russian: https://gist.githubusercontent.com/alexey-milovidov/5551a7c7b7aba731a0a74f01972ddc69/raw/054237b5165a26a6482329c1416ef153bec37b8e/protobuf_task_ru.txt | 2018-12-19T19:54:53 | 2018-12-19T19:54:53 | {} | MEMBER |
ClickHouse | ClickHouse | 448,800,996 | 3,869 | luongdieu86 | Thank you, I got it.
> PRIMARY KEY - The primary key if it differs from the sorting key.
> By default the primary key is the same as the sorting key (which is specified by the ORDER BY clause). Thus in most cases it is unnecessary to specify a separate PRIMARY KEY clause.
| 2018-12-20T00:13:09 | 2018-12-20T00:13:09 | {} | NONE |
ClickHouse | ClickHouse | 449,112,676 | 3,888 | alexey-milovidov | https://github.com/yandex/ClickHouse/commit/6c84afeb0f08166657770e12e06363a901863c61
Поправил: SQL запрос - это query, а не request. | 2018-12-20T19:38:35 | 2018-12-20T19:38:35 | {} | MEMBER |
ClickHouse | ClickHouse | 449,392,707 | 3,891 | vasyaabr | On fail we got 1 line for each table:
> 2018.12.20 20:43:17.023006 [ 33 ] {} <Error> funprodwh.Turnovers (ReplicatedMergeTreeAlterThread): void DB::ReplicatedMergeTreeAlterThread::run(): Code: 999, e.displayText() = Coordination::Exception: Session expired (Session expired), e.what() = Coordination::Exception, Stack... | 2018-12-21T13:47:29 | 2018-12-21T13:47:29 | {} | NONE |
ClickHouse | ClickHouse | 448,199,475 | 3,859 | ztlpn | Closing as a duplicate of #3782, please see https://github.com/yandex/ClickHouse/issues/3782#issuecomment-445228934 | 2018-12-18T12:09:56 | 2018-12-18T12:09:56 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 448,202,483 | 3,842 | ztlpn | What is your use case for this? The native TCP interface has transparent session support, and for the HTTP interface it is possible to maintain sessions with the `session_id` parameter. See [docs](https://clickhouse.yandex/docs/en/interfaces/http/) and a [test](https://github.com/yandex/ClickHouse/blob/master/dbms/test... | 2018-12-18T12:22:08 | 2018-12-18T12:22:08 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 448,327,128 | 3,590 | abyss7 | PR #3865 | 2018-12-18T18:46:27 | 2018-12-18T18:46:27 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 448,527,291 | 3,849 | crakjie | It there a bad usage possible of the connection, that can make Clickhouse take more time to answer after an EndOfStream? | 2018-12-19T09:26:09 | 2018-12-19T09:26:09 | {} | NONE |
ClickHouse | ClickHouse | 448,619,667 | 3,871 | den-crane | Looks like a network issue.
try
ping -s 1500 serverA | 2018-12-19T14:43:43 | 2018-12-19T14:43:43 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 448,669,736 | 3,825 | alesapin | v18.14.19-stable released, please update and try to disable settings. | 2018-12-19T17:02:50 | 2018-12-19T17:02:50 | {} | MEMBER |
ClickHouse | ClickHouse | 448,818,961 | 2,440 | knoguchi | Thanks for the description (I've read it using Google Translator). It seems Protobuf v3.3.0 added the delimited message utilities for C++ (Protobuf for Java supports delimiter for a long time).
I did a bit of experiment. Unfortunately `ParseDelimitedFromZeroCopyStream` takes `MessageLite` instead of `Message`. ... | 2018-12-20T01:12:34 | 2018-12-20T01:18:39 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 449,407,821 | 3,893 | sliontc | I have set the connect_timeout and keep_alive_timeout to:
<keep_alive_timeout>30</keep_alive_timeout>
<connect_timeout>60</connect_timeout>
But no lucky. | 2018-12-21T14:50:25 | 2018-12-21T14:50:25 | {} | NONE |
ClickHouse | ClickHouse | 448,249,086 | 3,849 | ztlpn | A suggestion by @alexey-milovidov: if the query contains some heavy aggregations the slowdown may be due to destructors of aggregate function states. The client thinks that the query execution is over immediately after it receives the EndOfStream packet, but destructors will continue to execute in the background. So if... | 2018-12-18T14:59:36 | 2018-12-18T14:59:36 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 449,021,769 | 3,885 | den-crane | @zhang2014 | 2018-12-20T14:44:47 | 2018-12-20T14:44:47 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 449,358,607 | 3,818 | blinkov | @lamber-ken you can run `SET optimize_throw_if_noop = 1;` and then https://clickhouse.yandex/docs/en/query_language/misc/#optimize will throw an exception with some explanation of why it doesn't want to merge anything (or will just do it). | 2018-12-21T11:01:21 | 2018-12-21T11:01:21 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 449,393,156 | 3,893 | sliontc | 
| 2018-12-21T13:49:01 | 2018-12-21T13:49:01 | {} | NONE |
ClickHouse | ClickHouse | 449,398,182 | 3,893 | sliontc | Then is it possible to fix it? Or are there related configuration in clickhouse? | 2018-12-21T14:10:46 | 2018-12-21T14:10:46 | {} | NONE |
ClickHouse | ClickHouse | 449,032,189 | 3,139 | alexey-milovidov | @tsl-karlp Merged to master. | 2018-12-20T15:17:37 | 2018-12-20T15:17:37 | {} | MEMBER |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.