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 | 414,408,395 | 2,902 | coreground | version 18.6.0 pulled from docker.io:
REPOSITORY TAG IMAGE ID CREATED SIZE
docker.io/yandex/clickhouse-client latest 85ab63b6bb17 2 weeks ago 349 MB
docker.io/yandex/clickhouse-server latest e129e1cdb731 ... | 2018-08-20T18:01:42 | 2018-08-20T18:01:42 | {} | NONE |
ClickHouse | ClickHouse | 414,454,675 | 2,838 | alexey-milovidov | The best way is to enable synchronous inserts into Distributed table
by setting `insert_distributed_sync` to 1.
The setting is located at `users.xml` file and could be tuned on per user basis.
Probably we will eventually made this behaviour as default.
| 2018-08-20T20:37:30 | 2018-08-20T20:37:30 | {} | MEMBER |
ClickHouse | ClickHouse | 414,520,354 | 2,796 | alexey-milovidov | @vavrusa @bocharov | 2018-08-21T01:31:59 | 2018-08-21T01:31:59 | {} | MEMBER |
ClickHouse | ClickHouse | 414,531,414 | 2,447 | alexey-milovidov | It's in my queue.
We have to rewrite code to modern style
(remove malloc, free, new, delete, sprintf, strdup, strlen, std::thread::detach)
Fix memory leaks. Fix thread safety errors.
Remove ternary search tree as it is not needed.
Extract completion for keywords from the parser (similar to "expected one of ... "... | 2018-08-21T02:42:32 | 2018-08-21T02:46:26 | {} | MEMBER |
ClickHouse | ClickHouse | 414,541,612 | 2,833 | kshvakov | @sundarv85 Do you have an example that can reproduce the problem? | 2018-08-21T03:53:36 | 2018-08-21T03:53:36 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 415,315,572 | 2,926 | alexey-milovidov | I wonder, how it is resolved? | 2018-08-23T07:13:03 | 2018-08-23T07:13:03 | {} | MEMBER |
ClickHouse | ClickHouse | 414,461,705 | 2,833 | alexey-milovidov | It looks like the data for INSERT queries was not sent for more than 300 seconds.
(Probably there was some data, then a pause for more than 300 seconds.)
You should consult with author of the library: @kshvakov
Also you can increase timeouts: it is `send_timeout` and `receive_timeout`
- these settings should be ... | 2018-08-20T21:00:36 | 2018-08-20T21:00:36 | {} | MEMBER |
ClickHouse | ClickHouse | 414,464,589 | 2,811 | alexey-milovidov | "Too many parts" error usually means that the replica fails to process merges. There may be two reasons:
1. merges are not processing at all due to misconfiguration, network issues or logical error.
2. merges are processing not fast enough due to high load.
First reason is more common.
To figure out the reason,... | 2018-08-20T21:11:26 | 2018-08-20T21:12:56 | {} | MEMBER |
ClickHouse | ClickHouse | 414,691,068 | 2,896 | champtar | To be sure I understand, you want me to:
1) create a new table, do ` SET low_cardinality_use_single_dictionary_for_part = 1`, and copy the data from the old table (vme_low_sortmin_file_ip)
2) rebuild with the patch in PR #2911, tests if the query works on old and new table
is that exact ? | 2018-08-21T14:20:14 | 2018-08-21T14:20:14 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 414,830,967 | 2,156 | wake-up-neo | Sorry, the previous message is related to **Distributed** on **ReplicatedCollapsingMergeTree**. | 2018-08-21T21:41:09 | 2018-08-21T21:41:09 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 414,964,117 | 1,621 | blinkov | @Saturma have you ReplicatedMergeTree as suggested above? Non-replicated MergeTree do not have the deduplication feature indeed. | 2018-08-22T09:02:34 | 2018-08-22T09:02:34 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 414,967,443 | 1,568 | blinkov | To analyze logs with ClickHouse it is indeed recommended to parse them into individual columns before ingestion instead of storing arbitrary strings/blobs as is.
@shalugin do you have any further questions? | 2018-08-22T09:14:36 | 2018-08-22T09:14:36 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 415,050,168 | 2,778 | dqminh | So, we ended up doing the following steps:
- disable 1/3 of the cluster from accepting queries ( removing them from shards config in zookeeper )
- upgrade 1/3 of the cluster to newer version
- do the same for another 1/3 of cluster
- redirect queries to 2/3 of cluster on newer version
- upgrade the last 1/3 of c... | 2018-08-22T14:24:22 | 2018-08-22T14:24:22 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 415,152,233 | 2,908 | pryg-skok | I don't agree - sometimes removing data silently is completely acceptable, because data has different nature and importance :) In my project it's OK that data will be dropped, because it's just become outdated.
Also I have different setups with different disk spaces - somewhere there will be more data, with higher r... | 2018-08-22T19:32:01 | 2018-08-22T19:32:01 | {} | NONE |
ClickHouse | ClickHouse | 414,378,250 | 2,897 | alexey-milovidov | You have to apply it to `.inner.` table (that you can find in `system.tables`) and mutations will work if the underlying storage is of MergeTree family. | 2018-08-20T16:26:44 | 2018-08-20T16:26:44 | {} | MEMBER |
ClickHouse | ClickHouse | 414,784,478 | 147 | pavel-odintsov | Hello!
Do you have any updates about it? | 2018-08-21T18:56:45 | 2018-08-21T18:57:29 | {} | NONE |
ClickHouse | ClickHouse | 414,957,117 | 2,015 | zhang2014 | @alexey-milovidov There are some problems with the current implementation.
### Can't work
```SQL
SELECT * FROM (SELECT * FROM test.test) WHERE id = 1;
```
### Unexpected results
``` SQL
SELECT * FROM (SELECT id, 'S' AS name FROM test.test_join_1) AS a ALL LEFT JOIN (SELECT id, 'S' AS name FROM test.test_join_... | 2018-08-22T08:37:55 | 2018-08-22T08:39:59 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 414,962,978 | 1,617 | fsitedev | Yes, we have already solved this issue. The old version of SSE was the reason. Thank you! | 2018-08-22T08:58:41 | 2018-08-22T08:58:41 | {} | NONE |
ClickHouse | ClickHouse | 415,151,143 | 2,919 | alexey-milovidov | Need to investigate failed test in "continuous-integration/travis-ci/push". | 2018-08-22T19:28:25 | 2018-08-22T19:28:25 | {} | MEMBER |
ClickHouse | ClickHouse | 414,966,646 | 1,581 | blinkov | @weishishuo haven't you figured out what happened by now?
how exactly you have checked the row count of distributed table? | 2018-08-22T09:11:40 | 2018-08-22T09:11:40 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 415,230,275 | 1,234 | aggarwaa | The most reliable way of reproducing this is by doing large streaming INSERTs using clickhouse-client. As soon as a single streaming insert exceeds 1G, tcmalloc spits out this error. | 2018-08-23T00:20:49 | 2018-08-23T00:20:49 | {} | NONE |
ClickHouse | ClickHouse | 414,466,371 | 2,826 | alexey-milovidov | Yes, you can insert data concurrently into Distributed tables on every server.
Exact way of choosing servers to send the data is the responsibility of your application.
PS. Note that the performance will be better if you split your data on application side and then insert into local tables on each server. (This is ... | 2018-08-20T21:17:48 | 2018-08-20T21:20:24 | {} | MEMBER |
ClickHouse | ClickHouse | 414,519,698 | 2,316 | alexey-milovidov | 3b39f6a4c36512aeb49b36d2d8d962698e8419de will help. | 2018-08-21T01:28:02 | 2018-08-21T01:28:02 | {} | MEMBER |
ClickHouse | ClickHouse | 414,546,194 | 2,904 | alexey-milovidov | > I though I needed -g to have proper debug packages but that but maybe it's a bad idea, and I should stick to Release?
There is no performance difference between RelWithDebInfo and Release. In fact we always use RelWithDebInfo. | 2018-08-21T04:28:28 | 2018-08-21T04:28:28 | {} | MEMBER |
ClickHouse | ClickHouse | 414,569,798 | 2,906 | zhang2014 | The advantage of doing so is that it can support:
``` SQL
SELECT * FROM table_name_a AS a ANY LEFT JOIN remote(...) AS b ON a.id = b.id
SELECT * FROM table_name_a AS a ANY LEFT JOIN mysql(...) AS b ON a.id = b.id
``` | 2018-08-21T06:57:26 | 2018-08-21T06:57:26 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 414,665,333 | 2,904 | champtar | Ok, problem is that `CMAKE_LIBRARY_ARCHITECTURE` is no defined or empty on CentOS7, we should use `CMAKE_SYSTEM_PROCESSOR` | 2018-08-21T13:01:14 | 2018-08-21T13:01:14 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 414,749,643 | 2,896 | KochetovNicolai | It would be helpful if you could `show create table` for table and send data form problem part (it's in `/var/lib/clickhouse/data/default/vme_low_sortmin_file_ip/20180710_2037_2293_3/`) | 2018-08-21T17:07:56 | 2018-08-21T17:07:56 | {} | MEMBER |
ClickHouse | ClickHouse | 414,920,406 | 147 | filimonov | In common case changing PK is not possible (data reordering will require the same amount of work as recreate the table with another PK and copy data to it).
There is support for some very special case (feature not documented because it's too specific). See https://github.com/yandex/ClickHouse/issues/2210#issuecomme... | 2018-08-22T05:56:43 | 2018-08-22T05:56:43 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 414,920,326 | 2,865 | qinzl1 | Any new for this? @alexey-milovidov @proller | 2018-08-22T05:56:13 | 2018-08-22T05:56:53 | {} | NONE |
ClickHouse | ClickHouse | 414,377,560 | 2,888 | alexey-milovidov | (if you want to be sure, I need detailed test case) | 2018-08-20T16:24:32 | 2018-08-20T16:24:32 | {} | MEMBER |
ClickHouse | ClickHouse | 414,407,068 | 2,888 | francisdb | Test case would look like this
```sql
CREATE TABLE test (
col1 String,
col2 String,
// below probably not needed
col3 Nullable(UInt64) materialized cityHash64(col1, col2)
) engine = MergeTree
INSERT INTO test (col1) VALUES ('bar')
```
Just omit some non-nullable culumns from the insert | 2018-08-20T17:57:19 | 2018-08-20T17:57:19 | {} | NONE |
ClickHouse | ClickHouse | 414,436,810 | 2,888 | francisdb | Excuses for the bad example, should have tried this.
So a String is always Nullable?
We also have `Array(String)` `FixedString(26)` `DateTime` `Nested(...)` and `UInt8`. Maybe it was one of those triggering the NULL conversion issue
Looks like I will have to learn Russian or wait for this to be fixed :-)
https://... | 2018-08-20T19:36:16 | 2018-08-20T19:42:03 | {} | NONE |
ClickHouse | ClickHouse | 414,442,800 | 2,903 | champtar | Asking on google groups instead of here
https://groups.google.com/forum/#!topic/clickhouse/izXdc52aF0k | 2018-08-20T19:56:58 | 2018-08-20T19:56:58 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 414,542,761 | 2,904 | champtar | Just tried with `Release`
```
# cat build/dbms/src/Common/config_build.cpp
// .cpp autogenerated by cmake
#include <Common/config_build.h>
const char * auto_config_build[]
{
"VERSION_FULL", "ClickHouse 18.11.0",
"VERSION_DESCRIBE", "v18.11.0-testing",
"VERSION_GITHASH", "76af46ed5d223b3a7af92e3... | 2018-08-21T04:02:16 | 2018-08-21T04:02:16 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 414,951,813 | 1,695 | blinkov | At the moment there are no plans to support old CPUs in official ClickHouse builds. | 2018-08-22T08:19:07 | 2018-08-22T08:19:07 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 414,958,904 | 1,666 | blinkov | As far as I know, most people are satisfied with UNION ALL performance of ClickHouse. | 2018-08-22T08:44:17 | 2018-08-22T08:44:17 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 415,045,970 | 2,896 | champtar | It's working with the repro table and the full data
Thanks! | 2018-08-22T14:12:27 | 2018-08-22T14:12:27 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 414,452,857 | 2,867 | alexey-milovidov | Known issue:
- materialized columns and alias columns are grouped together and cannot be interleaved in table definition with ordinary columns.
We are going to fix it. | 2018-08-20T20:31:09 | 2018-08-20T20:31:30 | {
"+1": 1
} | MEMBER |
ClickHouse | ClickHouse | 414,463,621 | 2,827 | alexey-milovidov | Ready test case:
```
drop table if exists test.t;
create table test.t
(
a Int32,
b Int32
) engine = MergeTree partition by (a,b) order by (a);
insert into test.t values (1, 1);
alter table test.t add column c Int32;
select b from test.t prewhere a < 1000;
select c from test.t where a < 1000;
select c ... | 2018-08-20T21:07:38 | 2018-08-20T21:07:38 | {} | MEMBER |
ClickHouse | ClickHouse | 414,469,617 | 2,806 | alexey-milovidov | Version 18.10.3 should have this issue fixed. | 2018-08-20T21:29:20 | 2018-08-20T21:29:20 | {
"hooray": 1
} | MEMBER |
ClickHouse | ClickHouse | 414,539,934 | 2,904 | champtar | Thanks a lot @alexey-milovidov
I'm using CentOS 7 / GCC 7 building with `-DCMAKE_BUILD_TYPE:STRING=RelWithDebInfo`
```
-- Will build ClickHouse 18.11.0
-- Using internal=OFF compiler=0: headers=/usr/share/clickhouse/headers/18.11.0 : clickhouse-clang -pipe -msse4.1 -msse4.2 -mpopcnt -fno-omit-frame-pointer -W... | 2018-08-21T03:41:10 | 2018-08-21T03:41:10 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 414,688,608 | 2,896 | KochetovNicolai | I've found a bug while dictionary deserialization and fixed it. Can you test to run your query from pr above?
Also, it's interesting if the same bug happens when you set setting `low_cardinality_use_single_dictionary_for_part = 1 ` and reinsert data. | 2018-08-21T14:12:56 | 2018-08-21T14:14:30 | {} | MEMBER |
ClickHouse | ClickHouse | 414,952,338 | 1,464 | blinkov | If someone has similar issues with recent ClickHouse releases, please reopen this issue or better create a new one. | 2018-08-22T08:21:03 | 2018-08-22T08:21:03 | {
"+1": 1
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 414,959,156 | 1,946 | blinkov | @bytecodedan could you provide a way to reproduce this? | 2018-08-22T08:45:12 | 2018-08-22T08:45:12 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 415,223,402 | 2,922 | alexey-milovidov | The release job is supposed to do it automatically but most of the time it just doesn't work. We will make efforts to fix the infrastructure behind the release job.
You can build a package for yourself by running single command on Ubuntu or Debian:
https://clickhouse.yandex/docs/en/development/build/
(The build is... | 2018-08-22T23:43:35 | 2018-08-22T23:45:15 | {} | MEMBER |
ClickHouse | ClickHouse | 414,434,224 | 2,895 | alexey-milovidov | For Replicated tables, you can also look at `system.replication_queue`. | 2018-08-20T19:28:05 | 2018-08-20T19:28:05 | {} | MEMBER |
ClickHouse | ClickHouse | 414,436,483 | 2,895 | simPod | Thank you!
For those wondering, also check server error log. | 2018-08-20T19:35:14 | 2018-08-20T19:35:14 | {
"+1": 1
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 414,534,929 | 2,904 | alexey-milovidov | PS. About your flame graph:
In production build you should have `memcpy` instead of `__memcpy_ssse3_back`, because we use internal variant of memcpy instead of libc.
`CRC32Hash::operator()`, `memEqualSSE2Wide` and `Aggregator...::getKey` methods should probably better to be inlined (note that clang inlines slight... | 2018-08-21T03:05:40 | 2018-08-21T03:05:40 | {} | MEMBER |
ClickHouse | ClickHouse | 414,605,343 | 2,908 | blinkov | Usually people just do this externally by dropping old partitions whenever it seems suitable. | 2018-08-21T09:06:43 | 2018-08-21T09:06:43 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 414,692,563 | 2,896 | KochetovNicolai | Yes. You can actually rebuild with patch and test if query already works on your table. If it is, that was probably the same bug. Otherwise test both builds with `low_cardinality_use_single_dictionary_for_part`. | 2018-08-21T14:24:24 | 2018-08-21T14:24:24 | {} | MEMBER |
ClickHouse | ClickHouse | 414,734,719 | 2,909 | alexey-milovidov | Ok. | 2018-08-21T16:22:53 | 2018-08-21T16:22:53 | {} | MEMBER |
ClickHouse | ClickHouse | 414,954,184 | 1,659 | blinkov | @tacyuuhon have you finally fixed these issues on your own? | 2018-08-22T08:27:45 | 2018-08-22T08:27:45 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 414,961,081 | 1,617 | blinkov | @fsitedev you definitely had some other circumstances that have lead to this, what are they?
Can you reproduce this on recent ClickHouse releases? | 2018-08-22T08:51:53 | 2018-08-22T08:51:53 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 415,257,189 | 1,234 | alexey-milovidov | Fixed in 18.10.3. | 2018-08-23T02:04:26 | 2018-08-23T02:04:26 | {} | MEMBER |
ClickHouse | ClickHouse | 414,459,682 | 2,838 | alexey-milovidov | It can be solved even in async mode, but it is little bit more difficult.
Possible implementation:
- add a settings 'native_insert_adapt_block_structure' with default 0;
- `RemoteBlockOutputStream` should set it to 1;
- when this setting is 1, `InterpreterInsertQuery` should apply the same transform as `Convertin... | 2018-08-20T20:54:04 | 2018-08-20T20:54:20 | {} | MEMBER |
ClickHouse | ClickHouse | 414,463,438 | 2,827 | alexey-milovidov | Not quite.
The last query returns empty set while it should return the same result as the previous query. | 2018-08-20T21:06:56 | 2018-08-20T21:06:56 | {} | MEMBER |
ClickHouse | ClickHouse | 414,509,065 | 2,316 | alexey-milovidov | But the issue still exists. There are more exceptions on the graph than in logs. | 2018-08-21T00:34:21 | 2018-08-21T01:29:21 | {} | MEMBER |
ClickHouse | ClickHouse | 414,967,881 | 1,573 | blinkov | @vavrusa @bobrik have you figured out how this happened by now? | 2018-08-22T09:16:15 | 2018-08-22T09:16:15 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 415,035,690 | 2,900 | ztlpn | fixed in master | 2018-08-22T13:41:42 | 2018-08-22T13:41:42 | {
"hooray": 1
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 414,450,986 | 2,667 | alexey-milovidov | This feature was implemented 1.5 years ago in https://github.com/yandex/ClickHouse/commit/9918965c | 2018-08-20T20:24:50 | 2018-08-20T20:24:50 | {} | MEMBER |
ClickHouse | ClickHouse | 414,533,751 | 2,904 | alexey-milovidov | GROUP by is not optimized yet for LowCardinality data type. It is still to do.
Currently the most performance benefit happens:
- when you apply filtering (WHERE), by LowCardinality column or by any other column (because filtering itself is much faster);
- when you apply functions on LowCardinality column (though t... | 2018-08-21T02:57:48 | 2018-08-21T02:58:27 | {} | MEMBER |
ClickHouse | ClickHouse | 414,643,295 | 2,888 | francisdb | here you have a reproducer, it's related to the non-nullable materialized from null values
```sql
drop table test;
create table test (
col1 String,
validation_error Nullable(String),
error Nullable(String),
error_fingerprint UInt64 materialized cityHash64(validation_error, error)
) ENGINE = MergeTre... | 2018-08-21T11:33:11 | 2018-08-21T11:33:11 | {} | NONE |
ClickHouse | ClickHouse | 414,754,185 | 2,914 | VladRassokhin | Here's script to reproduce:
```
-- Assume uniq(S) = 3000 in origin data
CREATE DATABASE IF NOT EXISTS test;
CREATE TABLE IF NOT EXISTS test.data (
S UInt32,
T UInt32,
C UInt32,
M UInt32
)
ENGINE = MergeTree() PARTITION BY toUInt32(S / 128) ORDER BY (S, C, M);
INSERT INTO test.data (S, T, C, M) SE... | 2018-08-21T17:22:57 | 2018-08-21T17:22:57 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 415,318,909 | 2,926 | thanhthai3010 | Thank you guys, in my case, the problem is ClickHouse cannot connect to MongoDB using the default `admin` user, with `root` role. So I try with the solution by creating a new MongoDB with reading role only, for example:
`
use test;
db.createUser(
{
user: "userReadTest",
pwd: "read#test#2018",
roles... | 2018-08-23T07:27:25 | 2018-08-23T07:27:25 | {} | NONE |
ClickHouse | ClickHouse | 414,686,919 | 2,896 | KochetovNicolai | #2911 | 2018-08-21T14:07:38 | 2018-08-21T14:07:38 | {} | MEMBER |
ClickHouse | ClickHouse | 414,745,482 | 2,896 | champtar | This doesn't fix my bug (currently running https://github.com/champtar/ClickHouse/commit/cccae457b8198edb34bf7fdee81b92902df50f51)
```
new stack strace in later messages
``` | 2018-08-21T16:55:48 | 2018-08-21T17:58:41 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 414,924,156 | 2,914 | filimonov | In other words you're proposing to assume that the whole data block have the same value of column (which is a part of PK) if consequent values of that column in index are equal.
I.e. if 2 values of PK in sparse index are A and next is also A, just assume that whole block have value A repeated 8192 times (or index_gr... | 2018-08-22T06:18:52 | 2018-08-22T06:18:52 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 414,952,812 | 1,552 | blinkov | Let's close this in favor of #863 | 2018-08-22T08:22:46 | 2018-08-22T08:22:46 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 415,316,378 | 2,926 | blinkov | @thanhthai3010 yep, please, post the reason/solution in case someone will google this up. | 2018-08-23T07:16:43 | 2018-08-23T07:16:43 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 414,393,475 | 2,866 | alexey-milovidov | Yes, it should be inside:
https://jsfiddle.net/w6uqfytj/
Need to investigate... | 2018-08-20T17:14:27 | 2018-08-20T17:14:27 | {} | MEMBER |
ClickHouse | ClickHouse | 414,421,418 | 2,888 | alexey-milovidov | I cannot reproduce the issue with your example:
```
milovidov-Pro-P30 :) CREATE TABLE test.test
:-] (
:-] col1 String,
:-] col2 String,
:-] // below probably not needed
:-] col3 Nullable(UInt64) materialized cityHash64(col1, col2)
:-] ) engine = MergeTree ORDER BY tuple();
CREATE TABLE test.test
(
c... | 2018-08-20T18:44:28 | 2018-08-20T18:44:28 | {} | MEMBER |
ClickHouse | ClickHouse | 414,435,084 | 2,667 | champtar | Hi @TH-HA I just tested with latest master and it works
```
CREATE TABLE testenum
(
date Date,
datetime DateTime,
a Enum8('GET' = 1, 'POST' = 2, 'HEAD' = 3, 'PUT' = 4, 'DELETE' = 5, 'OPTIONS' = 6, 'TRACE' = 7, 'CONNECT' = 8)
)
ENGINE = MergeTree
PARTITION BY date
ORDER BY (date, toStartOfMinute(... | 2018-08-20T19:31:00 | 2018-08-20T19:31:00 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 414,469,257 | 2,825 | alexey-milovidov | Care should be taken, because data in response body is considered as full featured resultset with its own data structure, tunable data format, etc.
Possible solutions:
- add it as a separate HTTP header;
- add it in the same way as `send_progress_in_http_headers`; | 2018-08-20T21:28:02 | 2018-08-20T21:28:02 | {} | MEMBER |
ClickHouse | ClickHouse | 414,616,561 | 2,908 | pryg-skok | The problem with such approach that I need to externally watch on database size and size of partitions, and also If in a moment many metrics rush into the base I can encounter overflow by size. Or I need to externally check DB size before insert every N metrics, which can be CPU intensive, isn't it?
Also is there an... | 2018-08-21T09:44:48 | 2018-08-21T09:44:48 | {} | NONE |
ClickHouse | ClickHouse | 415,006,890 | 2,810 | BigBlueWhale0 | Upgrading to version 18.6.0 has solved my problem. It was so trivial.
Thank you
| 2018-08-22T11:56:13 | 2018-08-22T11:56:13 | {} | NONE |
ClickHouse | ClickHouse | 415,336,343 | 2,690 | prompt-bot | Show your docker logs, Check your configuration file, host selinux, and host folder permissions, because this container run as clickhouse user. | 2018-08-23T08:32:02 | 2018-08-23T08:32:28 | {} | NONE |
ClickHouse | ClickHouse | 414,451,965 | 2,903 | alexey-milovidov | Answered in groups. | 2018-08-20T20:28:16 | 2018-08-20T20:28:16 | {} | MEMBER |
ClickHouse | ClickHouse | 414,546,395 | 2,904 | alexey-milovidov | We use
```
USE_INTERNAL_MEMCPY
USE_GLIBC_COMPATIBILITY
```
for our build.
Though performance difference could be hardly noticed. | 2018-08-21T04:30:07 | 2018-08-21T04:30:07 | {} | MEMBER |
ClickHouse | ClickHouse | 414,682,824 | 2,609 | silviucpp | Hello,
Any reason why this is not merged ?
Silviu | 2018-08-21T13:56:14 | 2018-08-21T13:56:14 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 414,689,475 | 2,904 | champtar | https://github.com/yandex/ClickHouse/pull/2912 | 2018-08-21T14:15:31 | 2018-08-21T14:15:31 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 414,923,731 | 2,609 | alexey-milovidov | Tests have failed. | 2018-08-22T06:16:27 | 2018-08-22T06:16:27 | {} | MEMBER |
ClickHouse | ClickHouse | 414,379,922 | 2,897 | simPod | Thank's for the reply! I also see you implememented query forwarding to the underlying table 👍 | 2018-08-20T16:32:06 | 2018-08-20T16:32:06 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 414,387,882 | 2,902 | alexey-milovidov | Please provide more details on this issue:
- what version of ClickHouse do you run;
- do you use Yandex packages, Altinity packages or built ClickHouse by yourself;
- install `clickhouse-common-dbg` package and run `addr2line -iCe /usr/bin/clickhouse 0x4f82dcb 0x4f7f93b ...` for all addresses from the stack trace;
... | 2018-08-20T16:56:43 | 2018-08-20T16:57:16 | {} | MEMBER |
ClickHouse | ClickHouse | 414,470,780 | 2,888 | alexey-milovidov | > So a String is always Nullable?
No. Nullable types can be specified only explicitly.
Probably you can construct another test case to figure out the root cause. | 2018-08-20T21:33:52 | 2018-08-20T21:33:52 | {} | MEMBER |
ClickHouse | ClickHouse | 414,514,852 | 2,316 | alexey-milovidov | It is normal.
It happens in two locations:
```
#0 ZooKeeperImpl::Exception::Exception (this=0x452b9f280, msg=..., code=-110) at ../dbms/src/Common/ZooKeeper/ZooKeeperImpl.cpp:274
#1 0x0000000005443423 in zkutil::KeeperMultiException::KeeperMultiException (this=0x452b9f280, code=-110, requests=..., responses=.... | 2018-08-21T01:01:35 | 2018-08-21T01:02:19 | {} | MEMBER |
ClickHouse | ClickHouse | 414,434,090 | 2,895 | alexey-milovidov | > In mutations table I can only see parts_to_do=1 but nothing else.
There is nothing else to track mutations: only system.mutations table. | 2018-08-20T19:27:38 | 2018-08-20T19:27:38 | {
"+1": 1
} | MEMBER |
ClickHouse | ClickHouse | 414,462,692 | 2,827 | alexey-milovidov | It is started to work in master.
This is related to changes in #2226 | 2018-08-20T21:04:03 | 2018-08-20T21:04:03 | {} | MEMBER |
ClickHouse | ClickHouse | 414,507,924 | 2,316 | alexey-milovidov | The issue is clearly visible on graphs (from the random server):
Looks like it is better after upgrade to the latest revision. | 2018-08-21T00:28:08 | 2018-08-21T00:28:08 | {} | MEMBER |
ClickHouse | ClickHouse | 414,495,966 | 2,877 | alexey-milovidov | I've edited your implementation. Now it's much more simple.
I don't understand why resetting the signal handler was required. | 2018-08-20T23:24:33 | 2018-08-21T01:33:32 | {} | MEMBER |
ClickHouse | ClickHouse | 414,685,985 | 2,866 | KochetovNicolai | #2910 | 2018-08-21T14:05:00 | 2018-08-21T14:05:00 | {} | MEMBER |
ClickHouse | ClickHouse | 414,791,138 | 2,896 | champtar | I've haven't managed to do a simple repro but i've reduced it
```
SET allow_experimental_low_cardinality_type = 1;
CREATE TABLE invaliddict (
date Date,
datetime DateTime,
a LowCardinality(String) DEFAULT '',
b LowCardinality(String) DEFAULT ''
)
ENGINE = MergeTree
PARTITION BY date
ORDER BY (date, toStart... | 2018-08-21T19:19:14 | 2018-08-21T19:19:14 | {
"+1": 1
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 414,964,587 | 1,582 | blinkov | @badvir do you have any further questions? | 2018-08-22T09:04:02 | 2018-08-22T09:04:02 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 414,709,372 | 2,868 | vmanyushin | ```
SELECT
user_id,
if(user_id NOT IN
(
SELECT user_id
FROM events.finance
WHERE project_id = 282
GROUP BY user_id
), 'yes', 'no') AS project282
FROM events.login
LIMIT 100
2018.08.21 15:53:49.196117 [ 79 ] <Error> executeQuery: Code: 10, e.displayText() = D... | 2018-08-21T15:09:55 | 2018-08-22T11:07:58 | {} | NONE |
ClickHouse | ClickHouse | 414,451,772 | 2,667 | alexey-milovidov | And here is the test: https://github.com/yandex/ClickHouse/blob/master/dbms/tests/queries/0_stateless/00427_alter_primary_key.sh | 2018-08-20T20:27:34 | 2018-08-20T20:27:34 | {} | MEMBER |
ClickHouse | ClickHouse | 414,533,627 | 2,883 | DeanThompson | @blinkov Thanks for the suggestion, I would use CollapsingMergeTree in the future. | 2018-08-21T02:56:56 | 2018-08-21T02:56:56 | {} | NONE |
ClickHouse | ClickHouse | 414,659,983 | 2,908 | filimonov | I don't think that removing some data silently based on size is brilliant idea.
Generally **you** should know how much data you need to store - it depends on scenario, sometimes 1 day is enough, sometimes 50 years is not enough.
After that you should estimate how much servers and how big disks you need. And how... | 2018-08-21T12:41:47 | 2018-08-21T12:41:47 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 414,950,663 | 1,725 | blinkov | @zhaoplayboy sounds weird, ClickHouse don't perform any requests on it's own either. Have you figured out what exactly was happening? | 2018-08-22T08:14:45 | 2018-08-22T08:14:45 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 414,968,478 | 1,443 | blinkov | @qietingfengling do you have any further questions? | 2018-08-22T09:18:24 | 2018-08-22T09:18:24 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 414,970,503 | 1,536 | blinkov | @theseusyang there are pretty detailed answers above, do you have any further questions? | 2018-08-22T09:25:15 | 2018-08-22T09:25:15 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 415,023,152 | 2,896 | KochetovNicolai | Thank you for reproducing the bug. I think it's fixed in #2919 now. Could you test it on full data? | 2018-08-22T13:00:23 | 2018-08-22T13:00:23 | {} | MEMBER |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.