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
455,778,601
4,100
alexey-milovidov
`test_insert_into_distributed/test.py::test_reconnect` - flappy test.
2019-01-19T13:03:24
2019-01-19T13:03:24
{}
MEMBER
ClickHouse
ClickHouse
456,879,375
4,133
abyss7
stateless.00411_long_accurate_number_comparison has failed because of timeout
2019-01-23T16:53:18
2019-01-23T16:53:18
{}
CONTRIBUTOR
ClickHouse
ClickHouse
457,153,886
4,108
4ertus2
According to MySQL behaviour we should result only one column from using statement for asterisks. It'll probably solve the main issue. ``` >select * from persons inner join children using(id) id name childName 1 John Robert 1 John Susan 3 Daniel Sarah 4 James David 4 James Jo...
2019-01-24T10:50:09
2019-01-24T10:50:09
{}
CONTRIBUTOR
ClickHouse
ClickHouse
457,409,761
1,614
alexey-milovidov
There is `hdfs` table function in recent ClickHouse versions. It allows to import/export data from HDFS in various formats.
2019-01-25T00:26:41
2019-01-25T00:26:41
{}
MEMBER
ClickHouse
ClickHouse
456,358,284
4,060
nvartolomei
```c stat("/proc/self/fd", {st_mode=S_IFDIR|0500, st_size=0, ...}) = 0 open("/proc/self/fd", O_RDONLY|O_NONBLOCK|O_DIRECTORY|O_CLOEXEC) = 12 fstat(12, {st_mode=S_IFDIR|0500, st_size=0, ...}) = 0 getdents(12, /* 287 entries */, 32768) = 6888 close(3) = 0 close(4) ...
2019-01-22T10:58:26
2019-01-22T10:58:26
{}
CONTRIBUTOR
ClickHouse
ClickHouse
455,705,711
4,097
alexey-milovidov
Yes, just a minimal test. I've added one.
2019-01-18T22:16:46
2019-01-18T22:16:46
{}
MEMBER
ClickHouse
ClickHouse
456,094,110
4,060
alesapin
This assumption is not correct: `The problem is that during vfork/execv dance, fd table of the parent process which has sockets for TCP and HTTP connections are copied to the odbc-bridge. However odbc-bridge is not aware of them, so they are never closed.` odbc-bridge closes all parents FDs on start https://github.co...
2019-01-21T14:34:01
2019-01-21T14:34:01
{}
MEMBER
ClickHouse
ClickHouse
457,145,224
4,108
4ertus2
Trying reproduce on master. The queries fail but without segfault: ``` 0. ./clickhouse-server(StackTrace::StackTrace()+0x28) [0xab42278] 1. ./clickhouse-server(DB::Exception::Exception(std::__1::basic_string<char, std::__1::char_traits<char>, std::__1::allocator<char> > const&, int)+0x4b) [0xab0e0eb] 2. ./clickho...
2019-01-24T10:22:02
2019-01-24T10:28:35
{}
CONTRIBUTOR
ClickHouse
ClickHouse
457,264,268
4,129
alexey-milovidov
In most cases, you should not type test results manually. Just run `clickhouse-client -n --testmode < test.sql > test.reference` and validate the result.
2019-01-24T16:35:20
2019-01-24T16:35:20
{ "+1": 1 }
MEMBER
ClickHouse
ClickHouse
455,670,053
4,097
bgranvea
what kind of test can I write? because I can't check that the returned value is correct (at least not in stateless sql scripts). So just a simple test which checks that the functions return something?
2019-01-18T20:00:53
2019-01-18T20:00:53
{}
CONTRIBUTOR
ClickHouse
ClickHouse
456,062,432
2,402
minaevmike
We have the same issue, any progress on this?
2019-01-21T12:46:49
2019-01-21T12:46:49
{}
NONE
ClickHouse
ClickHouse
456,525,111
4,117
alexey-milovidov
You own tests have not passed.
2019-01-22T19:11:51
2019-01-22T19:11:51
{}
MEMBER
ClickHouse
ClickHouse
457,567,321
4,080
4ertus2
Cannot reproduce on master ``` Connected to ClickHouse server version 19.1.6 revision 54413. chertus-nix :) select u, days, cumSum from ( :-] select u, min(d) mn, max(d) mx, groupArray(d) dg, groupArray(v) vg, :-] arrayMap(x -> x + mn, range(toUInt32(mx - mn + 1))) days, :-] toString(arrayCum...
2019-01-25T13:07:47
2019-01-25T13:07:47
{}
CONTRIBUTOR
ClickHouse
ClickHouse
455,907,999
4,105
filimonov
Default config doesn't have any zookeeper configs, only reference substitutions file. Maybe you have one (check /etc/metrika.xml)?
2019-01-20T22:20:57
2019-01-20T22:20:57
{}
CONTRIBUTOR
ClickHouse
ClickHouse
455,916,557
4,019
den-crane
AFAIK Several recent CH versions have the same .deb for testing and stable as a result select value from system.build_options where name = 'VERSION_DESCRIBE' v18.14.18-testing v18.16.0-testing v18.16.1-testing
2019-01-21T00:21:51
2019-01-21T00:21:51
{}
CONTRIBUTOR
ClickHouse
ClickHouse
455,982,600
4,047
zzn01
what's the affection of this error? Is it safe to ignore?
2019-01-21T08:11:50
2019-01-21T08:11:50
{ "+1": 2 }
NONE
ClickHouse
ClickHouse
457,094,015
4,126
Miniwoffer
Is zstd optional? If it is this needs to be done differently.
2019-01-24T07:24:04
2019-01-24T07:24:04
{}
NONE
ClickHouse
ClickHouse
457,095,567
3,890
zhang2014
Ok, I'll try to mark them.
2019-01-24T07:31:03
2019-01-24T07:31:03
{}
CONTRIBUTOR
ClickHouse
ClickHouse
457,144,037
4,139
blinkov
It's possible, but less convenient than 2 shards with 2 replicas (on 4 nodes). Here's an article describing this: https://www.altinity.com/blog/2018/5/10/circular-replication-cluster-topology-in-clickhouse
2019-01-24T10:18:18
2019-01-24T10:18:18
{}
CONTRIBUTOR
ClickHouse
ClickHouse
457,245,415
1,567
alexey-milovidov
Solved.
2019-01-24T15:47:45
2019-01-24T15:47:45
{ "+1": 1, "hooray": 5 }
MEMBER
ClickHouse
ClickHouse
457,250,094
4,091
alexey-milovidov
Solved in https://github.com/yandex/ClickHouse/pull/4136
2019-01-24T15:59:22
2019-01-24T15:59:22
{ "+1": 1 }
MEMBER
ClickHouse
ClickHouse
457,316,638
4,137
healiseu
Thanks @alexey-milovidov, I tested your code on a small table of 16 rows by 15 columns and with missing data. It takes **TRIACLICK** about 40msec to do the transformation on my machine. This is better than my nested `ANY LEFT OUTER JOIN` approach that took about 56msec to return the result. Anyway I expect this benchma...
2019-01-24T19:02:37
2019-01-24T20:12:09
{}
NONE
ClickHouse
ClickHouse
457,559,927
4,146
KochetovNicolai
You just need to use different zookeeper's hosts in config.xml (zookeeper cluster must be the same for replicas in ReplicationMergeTree, but may be different for different shards). Actually, we use single zookeeper cluster for more than 500 nodes, and it works fine.
2019-01-25T12:36:04
2019-01-25T12:36:04
{}
MEMBER
ClickHouse
ClickHouse
457,715,082
3,764
reflection
Actually, we can get accurate results without changing _max_threads_ to 1 but rather disabling _group_by_two_level_ completely: ``` SET group_by_two_level_threshold = 0; SET group_by_two_level_threshold_bytes = 0; ``` Are there bad side effects to doing this?
2019-01-25T20:25:13
2019-01-25T20:25:13
{}
CONTRIBUTOR
ClickHouse
ClickHouse
456,137,013
4,060
nvartolomei
pid process name ``` 20577 /usr/bin/clickhouse odbc-bridge --http-port 9018 --listen-host localhost --http-timeout 1800000000 ``` some random lines that i think are related to the issue, stripped `ss` ``` LISTEN 0 64 127.0.0.1:9018 *:* users:(("clickhouse...
2019-01-21T16:48:18
2019-01-21T16:48:18
{}
CONTRIBUTOR
ClickHouse
ClickHouse
456,140,045
4,060
nvartolomei
Live from a machine: ``` $ ps ax | grep clickhouse 69950 ? Sl 128877:09 /usr/bin/clickhouse-server --config=/etc/clickhouse-server/config.xml 73474 ? Z 36:04 [clickhouse] <defunct> 99308 ? Sl 7:54 /usr/bin/clickhouse odbc-bridge --http-port 9018 --listen-host localhost --htt...
2019-01-21T16:57:57
2019-01-21T16:57:57
{}
CONTRIBUTOR
ClickHouse
ClickHouse
456,249,814
4,114
den-crane
I agree that ``` SELECT if(NULL, 'YES', 'NO') AS x ┌─x────┐ │ ᴺᵁᴸᴸ │ └──────┘ ``` is wrong behavior. The result have to be **NO**. ----------------------- psql ``` select case when null then 'YES' else 'NO' end as x; x ---- NO ``` ----------------------- mysql ``` mysql> select IF(null, "YES...
2019-01-22T02:29:06
2019-01-22T02:33:34
{}
CONTRIBUTOR
ClickHouse
ClickHouse
456,888,695
4,132
alexey-milovidov
Stress test: in `AlterCommands`: known issue.
2019-01-23T17:17:15
2019-01-23T17:17:15
{}
MEMBER
ClickHouse
ClickHouse
457,220,255
4,140
KochetovNicolai
Probably, it's a duplicate for #3783
2019-01-24T14:41:05
2019-01-24T14:41:05
{}
MEMBER
ClickHouse
ClickHouse
457,259,675
4,122
alexey-milovidov
A few notes: - you can also search for header `.h` files (most of the comments are there); - you can also look for multiline comments (we usually write them as ``` /** ... * ... */ ``` ) I wonder if we can integrate spell check to our CI process. But there are too many exceptions (spell checker don't lik...
2019-01-24T16:23:19
2019-01-24T16:23:59
{}
MEMBER
ClickHouse
ClickHouse
457,565,550
3,998
4ertus2
It looks like 'data.value' has been cropped to 'value' before aliases calculation. Correlated issue #4110
2019-01-25T13:00:22
2019-01-25T13:00:56
{}
CONTRIBUTOR
ClickHouse
ClickHouse
455,768,127
4,097
bgranvea
ok, thank you.
2019-01-19T10:28:42
2019-01-19T10:28:42
{}
CONTRIBUTOR
ClickHouse
ClickHouse
456,089,730
4,104
agend
`<compile_expressions>0</compile_expressions>` fix the issue with memory. Thank you for help.
2019-01-21T14:20:44
2019-01-21T14:20:44
{}
NONE
ClickHouse
ClickHouse
456,426,321
4,113
KochetovNicolai
In master the error is ``` Code: 47. DB::Exception: Received from localhost:9004, ::1. DB::Exception: Unknown identifier: d, context: required_names: 'd' source_tables: 't' table_aliases: private_aliases: column_aliases: 'x' public_columns: masked_columns: array_join_columns: source_columns: . ``` @4ertus2
2019-01-22T14:49:19
2019-01-22T14:49:19
{}
MEMBER
ClickHouse
ClickHouse
457,177,540
4,138
den-crane
1.1.54236 it's two years old CH version one of the early beta version. Try something newer https://clickhouse.yandex/#quick-start http://repo.yandex.ru/clickhouse/rpm/ https://www.altinity.com/download/
2019-01-24T12:20:01
2019-01-24T12:20:01
{ "+1": 1 }
CONTRIBUTOR
ClickHouse
ClickHouse
457,243,884
329
alexey-milovidov
https://github.com/yandex/ClickHouse/pull/4143
2019-01-24T15:43:46
2019-01-24T15:44:14
{}
MEMBER
ClickHouse
ClickHouse
457,304,252
4,126
alexey-milovidov
Zstd is mandatory. The only option is to switch between packages and bundled version.
2019-01-24T18:26:05
2019-01-24T18:26:05
{}
MEMBER
ClickHouse
ClickHouse
457,327,004
4,137
alexey-milovidov
You're right, the timings measured on just a few rows are misleading. You should measure on millions. For less amount of rows, the query speed is dominated by various preparations than the data processing itself. > I expect that the table can be reconstructed likewise the MergeTree() engine by reading and concatena...
2019-01-24T19:33:10
2019-01-24T19:33:10
{}
MEMBER
ClickHouse
ClickHouse
457,577,371
4,151
alexey-milovidov
> As of today canBeWiden returns true (and getWidenDataType doesn't throw) only for DataTypeDecimal and DataTypeNumber. I am still unsure whether it should also be implemented for all the types If we name the method `promoteNumericType` instead of `getWidenDataType`, no question exists. ("type promotion" is a widel...
2019-01-25T13:46:25
2019-01-25T13:46:25
{}
MEMBER
ClickHouse
ClickHouse
457,719,572
4,139
filimonov
> I can’t normally use queries with ON CLUSTER statement Unfortunately currently ON CLUSTER doesn't support that scenario (AFAIK)
2019-01-25T20:39:13
2019-01-25T20:39:13
{}
CONTRIBUTOR
ClickHouse
ClickHouse
456,309,709
4,060
alesapin
hmm... looks strange. Which version of clickhouse do you use? You use linux or some other OS (maybe run it in docker)? Seems like `closeFDs` doesn't work correctly with your system. Actually, it was tested on linux only. P.S. Why do you have so much Zombie clickhouse processes (`[clickhouse] <defunct>`)?
2019-01-22T08:20:47
2019-01-22T08:25:42
{}
MEMBER
ClickHouse
ClickHouse
456,565,907
4,122
maiha
@alexey-milovidov 1. **detecting** is automated by using **aspell** like this. ``` cat **/*.cpp | grep '///' | aspell -a | cut -d ' ' -f 2 | grep -v '*' | sort | uniq > typo.txt ``` 2. **proofread** is manual. So it was limited to `dbms` for this time. I'll try to check whole src when I have a time 😸
2019-01-22T21:14:52
2019-01-22T21:14:52
{}
CONTRIBUTOR
ClickHouse
ClickHouse
457,013,010
4,119
alexey-milovidov
This task is assigned to https://github.com/CurtizJ. He will implement custom TTL expressions for columns and tables.
2019-01-23T23:55:16
2019-01-23T23:56:37
{}
MEMBER
ClickHouse
ClickHouse
457,181,454
4,129
LeoErcolanelli
@alexey-milovidov I addressed all of your concerns. I am not too sure about the specialization of `std::hash<Decimal128>` if you want to double check! Let me know!
2019-01-24T12:35:24
2019-01-24T12:35:24
{}
CONTRIBUTOR
ClickHouse
ClickHouse
457,440,913
4,091
tungdam
Thanks for your quick feedback. I have a question related to taskstats info ( Hope you don't mind me asking it here ) As i understand, enable cap_net_admin capabilities will provide us more info about taskstats, basically it's ProfileEvents info in query_log/query_thread_log/processes table. But I noticed no signifi...
2019-01-25T03:14:35
2019-01-25T03:14:35
{}
NONE
ClickHouse
ClickHouse
457,607,802
4,146
iameugenejo
Just having different zookeer hosts in config makes the different shards to use different zookeeper clusters? How does that work? I already have multiple zookeepers in the config but they are clustered. First, how do I configure multiple zookeeper clusters, and second how do I point a shard to one zookeeper cluster an...
2019-01-25T15:23:14
2019-01-25T15:23:14
{}
NONE
ClickHouse
ClickHouse
455,647,006
4,097
alexey-milovidov
Missing test.
2019-01-18T18:43:08
2019-01-18T18:43:08
{}
MEMBER
ClickHouse
ClickHouse
456,700,133
3,960
blinkov
@liuyimin-bytedance @chenxing-xc maybe this feature will be relevant for you.
2019-01-23T07:36:47
2019-01-23T07:36:47
{ "+1": 1 }
CONTRIBUTOR
ClickHouse
ClickHouse
456,760,402
4,063
Miniwoffer
I'm sorry this is a little late. It looks good, i also did some manual testing and wrote two additional tests that takes care of edge cases if there are changes in the future. The tests are not needed.
2019-01-23T11:01:59
2019-01-23T11:01:59
{}
NONE
ClickHouse
ClickHouse
456,917,314
4,135
4ertus2
By default sum uses Decimal128 and it's affected by #3783. Decimal128 + groupArray leads to crash in some cases. It's not fixed yet. Probably cause of Aggregator returns unaligned data and compiler makes SSE aligned instructions to load 128-bit types.
2019-01-23T18:38:13
2019-01-23T18:38:13
{}
CONTRIBUTOR
ClickHouse
ClickHouse
457,089,956
4,138
kiragoo
Sorry for adding the info about the clickhouse: ``` [root@k8sdev ~]# rpm -qa | grep clickhouse clickhouse-server-common-1.1.54236-4.el7.x86_64 clickhouse-debuginfo-1.1.54236-4.el7.x86_64 clickhouse-server-1.1.54236-4.el7.x86_64 clickhouse-compressor-1.1.54236-4.el7.x86_64 clickhouse-client-1.1.54236-4.el7.x86_64...
2019-01-24T07:05:28
2019-01-24T07:05:28
{ "+1": 1 }
NONE
ClickHouse
ClickHouse
457,624,619
4,146
iameugenejo
We see `fsync-ing the write ahead log in SyncThread:* took *ms which will adversely effect operation latency` warning frequently atm. It's not causing any issues but I'd need to prepare to scale the zookeeper if it does in the future when we increase the inserts by 20 fold (650billion rows per day). And also we're plan...
2019-01-25T16:10:57
2019-01-25T16:10:57
{}
NONE
ClickHouse
ClickHouse
457,650,699
4,151
alexey-milovidov
> Are you aware of this issue or should I dig in? No need to worry, it is just an unstable test.
2019-01-25T17:29:36
2019-01-25T17:29:36
{}
MEMBER
ClickHouse
ClickHouse
457,705,232
4,151
LeoErcolanelli
@alexey-milovidov I hesitated implementing `sumMapWithOverflow` but as I didn't know at all about `SummingSortedBlockInputStream` I had no idea it was so crucial! Everything should be implemented now. Thank you very much for your explanations and reactivity, it is a pleasure contributing to the project!
2019-01-25T19:55:09
2019-01-25T19:55:09
{}
CONTRIBUTOR
ClickHouse
ClickHouse
455,827,419
3,988
alexey-milovidov
@proller How the user can disable SSE4 from the command line?
2019-01-20T00:41:26
2019-01-20T00:41:26
{}
MEMBER
ClickHouse
ClickHouse
456,908,976
4,135
xicmiah
After adding an extra cast around sum, the query succeeds: ``` select groupArray(s) from (select toDecimal32(sum(n), 2) s from (select toDecimal32(1, 2) as n)) ```
2019-01-23T18:14:46
2019-01-23T18:14:46
{}
NONE
ClickHouse
ClickHouse
456,938,012
4,095
alexey-milovidov
#4136
2019-01-23T19:39:06
2019-01-23T19:39:06
{}
MEMBER
ClickHouse
ClickHouse
457,246,805
3,605
alexey-milovidov
This task is assigned to @igron99 (external contributor).
2019-01-24T15:51:20
2019-01-24T15:51:20
{}
MEMBER
ClickHouse
ClickHouse
455,912,426
4,019
pdeva
details here: https://groups.google.com/forum/?utm_medium=email&utm_source=footer#!msg/clickhouse/WrEbn8O2Bbk/pWzncnQHEgAJ
2019-01-20T23:23:13
2019-01-20T23:23:13
{}
NONE
ClickHouse
ClickHouse
455,915,445
4,104
den-crane
What CH version do you use? How much memory does your servers have? Please provide table DDL. max_memory_usage* -- these settings are limits. So your query will fail if it tries to use more than this limit. To reduce insert memory usage try to lower max_insert_block_size (1mil rows by default). Try to disable <c...
2019-01-21T00:07:29
2019-01-21T00:09:09
{}
CONTRIBUTOR
ClickHouse
ClickHouse
456,006,150
4,107
KochetovNicolai
That error may happen if insertion into `Log` table failed. [`Log` engine](https://clickhouse.yandex/docs/en/operations/table_engines/log/) is only for small data, testing or demonstration purposes. Generally, it's better to use `MergeTree` family engines.
2019-01-21T09:33:24
2019-01-21T09:33:24
{}
MEMBER
ClickHouse
ClickHouse
456,430,543
4,112
nvartolomei
Interesting, thanks. I'll close it for now and get back if I see it again.
2019-01-22T15:00:39
2019-01-22T15:00:47
{}
CONTRIBUTOR
ClickHouse
ClickHouse
457,012,504
4,137
alexey-milovidov
You can write it with GROUP BY: ``` SELECT hb1, any(supCountry), any(catTotal), any(supName) FROM ( SELECT hb1, supCountry, NULL AS catTotal, NULL AS supName FROM VW1 UNION ALL SELECT hb1, NULL AS supCountry, catTotal, NULL AS supName FROM VW3 UNION ALL SELECT hb1, NULL AS supCountry, NUL...
2019-01-23T23:52:52
2019-01-23T23:53:21
{}
MEMBER
ClickHouse
ClickHouse
457,244,341
530
alexey-milovidov
https://github.com/yandex/ClickHouse/pull/4143
2019-01-24T15:44:57
2019-01-24T15:44:57
{}
MEMBER
ClickHouse
ClickHouse
457,249,766
4,136
alexey-milovidov
Running `docker` with `--cap-add` is more obscure and non-obvious for end users. As these capabilities and corresponding features are optional, this solution is considered Ok. We can re-work init script in Docker image to enable capabilities if available on startup.
2019-01-24T15:58:34
2019-01-24T15:58:34
{}
MEMBER
ClickHouse
ClickHouse
457,577,825
4,151
LeoErcolanelli
@alexey-milovidov Looks good to me ! I'll rename `canBeWiden` to `canBePromoted` and `getWidenType` to `promoteNumericType`!
2019-01-25T13:47:57
2019-01-25T13:47:57
{}
CONTRIBUTOR
ClickHouse
ClickHouse
457,632,082
4,028
chipitsine
sorry, I'm somewhat busy. will address this PR later
2019-01-25T16:32:22
2019-01-25T16:32:22
{}
CONTRIBUTOR
ClickHouse
ClickHouse
457,632,458
4,151
LeoErcolanelli
@alexey-milovidov I did the renaming but I am not sure as to why the test `00446_clear_column_in_partition_zookeeper` is failing. It appears that there is no use of `sumMap` in this test that could explain it failing. Are you aware of this issue or should I dig in?
2019-01-25T16:33:38
2019-01-25T16:37:47
{}
CONTRIBUTOR
ClickHouse
ClickHouse
455,826,828
4,090
alexey-milovidov
It does not work this way. Either write ``` insert into xxx select * from xxx limit 10; ``` or ``` insert into xxx (column1, column2) select * from xxx limit 10; ```
2019-01-20T00:28:39
2019-01-20T00:28:39
{}
MEMBER
ClickHouse
ClickHouse
455,988,596
4,105
aido93
I took config from there: https://github.com/yandex/ClickHouse/blob/master/dbms/programs/server/config.xml And it uses zookeeper config from there: https://github.com/yandex/ClickHouse/blob/master/dbms/programs/server/config.d/zookeeper.xml So you see that in zookeeper.xml hardcoded three zookeeper servers that...
2019-01-21T08:34:39
2019-01-21T08:34:39
{}
NONE
ClickHouse
ClickHouse
456,061,394
4,015
lapkofear
This bug is blocker for our CH update =(
2019-01-21T12:42:49
2019-01-21T12:42:49
{ "+1": 2 }
CONTRIBUTOR
ClickHouse
ClickHouse
456,762,628
4,117
abyss7
> You own tests have not passed. I know, will fix - still haven't figured out how to run them locally before pushing to PR.
2019-01-23T11:09:40
2019-01-23T11:09:40
{}
CONTRIBUTOR
ClickHouse
ClickHouse
456,861,927
4,133
abyss7
All changes can be divided in three categories: - merge of the source code from two classes - actual rename `IProfilingBlockInputStream` → `IBlockInputStream` - removal of constructions like `if (auto * stream = dynamic_cast<IProfilingBlockInputStream*>(…))`
2019-01-23T16:10:05
2019-01-23T16:10:05
{}
CONTRIBUTOR
ClickHouse
ClickHouse
457,308,870
4,129
alexey-milovidov
Failed UBSan tests: #4145
2019-01-24T18:39:37
2019-01-24T18:39:37
{}
MEMBER
ClickHouse
ClickHouse
457,506,551
3,852
bgranvea
I performed a benchmark with different storage solutions: - an AggregatingMergeTree with anyLast and with the flag enable_optimize_predicate_expression - a patched SummingMergeTree which uses anyLast as aggregating function - a ReplacingMergeTree which is our current solution but this forces us to reload old valu...
2019-01-25T09:14:14
2019-01-25T09:14:14
{}
CONTRIBUTOR
ClickHouse
ClickHouse
457,530,817
4,139
igor-sh8
Great, thanks. It works well. Considering that we now have all the databases different, I can’t normally use queries with `ON CLUSTER` statement ? Or maybe `ON CLUSTER` has also some additional properties like the "default database" ?
2019-01-25T10:37:54
2019-01-25T10:37:54
{}
NONE
ClickHouse
ClickHouse
457,591,964
2,443
filimonov
Small issue with that: when starting clickhouse-server like that ```clickhouse-server --config /etc/clickhouse-server/config.xml -- --path=/tmp/bbb --http_port=30000 --tcp_port=30001 --interserver_http_port=30002 --logger.log=/dev/null --logger.errorlog=/dev/null``` it still tries to access `/var/lib/clickhouse//...
2019-01-25T14:36:23
2019-01-25T14:39:51
{}
CONTRIBUTOR
ClickHouse
ClickHouse
457,628,968
4,028
alexey-milovidov
No difference in our CI, because we have no bulds with gcc8.
2019-01-25T16:23:23
2019-01-25T16:23:23
{}
MEMBER
ClickHouse
ClickHouse
457,647,123
4,146
filimonov
Looks like I/O issues from zookeeper (slow write on disk). Stupid questions: 1) do you have zookeeper on separate servers (Installing zookeeper on the same servers as ClickHouse is usually a very bad idea)? 2) does zookeeper's `dataDir` points to some fast / not overused disk (ideally: separate SSD)? If you want...
2019-01-25T17:18:15
2019-01-25T17:20:31
{}
CONTRIBUTOR
ClickHouse
ClickHouse
457,654,724
4,151
alexey-milovidov
I need to clarify something. We have aggregate functions `sum` and `sumWithOverflow`. `sum` does summation in promoted type. `sumWithOverflow` does summation with overflow within the range of its argument. Also we have aggregate function `sumMap` but not `sumMapWithOverflow`. Before your modifications, `sumMap...
2019-01-25T17:42:54
2019-01-25T17:44:09
{}
MEMBER
ClickHouse
ClickHouse
456,423,112
4,060
alesapin
https://github.com/yandex/ClickHouse/pull/4120
2019-01-22T14:40:40
2019-01-22T14:40:40
{}
MEMBER
ClickHouse
ClickHouse
456,558,515
4,122
alexey-milovidov
Thank you! I wonder did you perform manual proofread or have used some automated spell check to find notable mistakes?
2019-01-22T20:51:29
2019-01-22T20:51:29
{}
MEMBER
ClickHouse
ClickHouse
456,889,102
4,132
alexey-milovidov
`clear_column_in_partition_zookeeper` - unrelated, probably flappy test.
2019-01-23T17:18:27
2019-01-23T17:18:27
{}
MEMBER
ClickHouse
ClickHouse
456,914,335
4,135
KochetovNicolai
Query works fine on master. @4ertus2 probably it's already fixed?
2019-01-23T18:29:22
2019-01-23T18:29:22
{}
MEMBER
ClickHouse
ClickHouse
456,947,411
4,129
alexey-milovidov
The name is Ok.
2019-01-23T20:07:12
2019-01-23T20:07:12
{}
MEMBER
ClickHouse
ClickHouse
457,215,953
4,108
4ertus2
It will hide an issue #4141 The main problem is not fixed: ``` select children.* from persons all inner join children using id; DB::Exception: Sizes of columns doesn't match: default.children.id: 5, childName: 6 select * from persons all inner join children as ch on id = ch.id; DB::Exception: Sizes of columns d...
2019-01-24T14:28:47
2019-01-24T19:08:56
{}
CONTRIBUTOR
ClickHouse
ClickHouse
456,420,608
4,112
KochetovNicolai
Replica is marked as lost (in zookeeper, there is is_lost flag for each replica) if it isn't active and it's replication queue has more than max_replicated_logs_to_keep (= 10000) items. If, for example, you delete information from zookeeper for all replicas which are not lost, you'll get this message.
2019-01-22T14:33:44
2019-01-22T14:33:44
{}
MEMBER
ClickHouse
ClickHouse
456,994,499
3,890
alexey-milovidov
Yes, we should mark them as `isStateful` (suggested property name) in `IFunctionBase`.
2019-01-23T22:39:19
2019-01-23T22:39:19
{}
MEMBER
ClickHouse
ClickHouse
457,128,765
4,136
filimonov
But postinst is called one time during docker image preparation. And you can't run build with `--privileged`, so the resulting image will always have that capability disabled, and nobody except people building docker image (generally: you) will see that warning 'To enable taskstats accounting you may add the required ...
2019-01-24T09:32:40
2019-01-24T10:16:06
{}
CONTRIBUTOR
ClickHouse
ClickHouse
457,420,561
3,764
reflection
@blinkov @KochetovNicolai this should be very high priority, please. We have to run `SET max_threads=1;` before *all* of our `WITH ROLLUP` queries because otherwise ClickHouse returns unusable data. Setting max_threads to 1 adversely affects ClickHouse performance.
2019-01-25T01:18:54
2019-01-25T01:18:54
{}
CONTRIBUTOR
ClickHouse
ClickHouse
457,562,431
4,126
alexey-milovidov
Merged after a few additional changes.
2019-01-25T12:46:53
2019-01-25T12:46:53
{}
MEMBER
ClickHouse
ClickHouse
457,569,562
4,080
4ertus2
test appended to 00819_ast_refactoring_bugs
2019-01-25T13:16:50
2019-01-25T13:16:50
{}
CONTRIBUTOR
ClickHouse
ClickHouse
455,988,342
4,104
agend
Here you go: `Connected to ClickHouse server version 18.16.1 revision 54412.` RAM: 8 GB We send data by 10000 rows. We already had decreased mark_cache_size `<mark_cache_size>1368709120</mark_cache_size>` Пробуем с опцией <compile_expressions>0</compile_expressions> Thx
2019-01-21T08:33:43
2019-01-21T08:33:43
{}
NONE
ClickHouse
ClickHouse
456,750,593
4,116
filimonov
you can try to set that in default profile (may be server restart / or Kafka table recreate can be needed). BTW: May be user_profile should be added to Kafka settings?
2019-01-23T10:30:06
2019-01-23T10:30:06
{}
CONTRIBUTOR
ClickHouse
ClickHouse
457,563,493
4,113
4ertus2
The last query fails. The first two are OK. ``` WITH (d < '2018-01-01') AND (d < '2018-01-02') AS x SELECT 1 FROM t WHERE x SETTINGS optimize_move_to_prewhere = 0; WITH (d < '2018-01-01') AS x SELECT 1 FROM t WHERE x; WITH (d < '2018-01-01') AND (d < '2018-01-02') AS x SELECT 1 FROM t WHERE x; ```` The m...
2019-01-25T12:51:36
2019-01-25T12:51:36
{}
CONTRIBUTOR
ClickHouse
ClickHouse
457,618,823
4,146
filimonov
Hm. Most probably in that case ON CLUSTER DDL/mutations will work only inside shard scope. Do you have some issues with single zookeeper cluster?
2019-01-25T15:54:15
2019-01-25T15:56:31
{}
CONTRIBUTOR
ClickHouse
ClickHouse
455,769,546
4,019
alexey-milovidov
Don't understand, why it is closed? If it was your mistake - what exact?
2019-01-19T10:49:32
2019-01-19T10:49:32
{}
MEMBER
ClickHouse
ClickHouse
456,023,430
4,108
KochetovNicolai
Have reproduced on master.
2019-01-21T10:25:19
2019-01-21T10:25:19
{}
MEMBER
ClickHouse
ClickHouse
456,926,916
4,091
alexey-milovidov
ClickHouse doesn't run at all in Docker with enabled capabilities. ``` docker run -it ubuntu:precise /bin/bash apt-get update apt-get install gnupg2 apt-key adv --keyserver keyserver.ubuntu.com --recv E0C56BD4 echo "deb http://repo.yandex.ru/clickhouse/deb/stable/ main/" | tee /etc/apt/sources.list.d/clic...
2019-01-23T19:06:30
2019-01-23T19:06:30
{}
MEMBER
ClickHouse
ClickHouse
457,571,108
4,028
alexey-milovidov
It barely makes any sense for our codebase, because we are using variable length stack allocations only in few places in code (example: `AggregateFunctionArray.h`) where it cannot overflow the stack. But it can make sense for third-party libraries.
2019-01-25T13:23:14
2019-01-25T13:23:46
{}
MEMBER