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 | 611,871,170 | 6,387 | Slach | @alexey-milovidov could you suggest SQL queries which can produce `memory profler` related data in readable format? i can quick add it to https://githhub.com/Slach/clickhouse-flamegraph | 2020-04-10T04:22:51 | 2020-04-10T04:24:59 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 612,049,523 | 10,161 | den-crane | > Перевод на английский язык планируется?
да, я не хочу тратить время на перевод неправильной доки и потом корректировать по сто раз оба языка.
И планирую описать все параметры. Я не хочу делать это в одном PR , потому что его согласование займет бесконечное время и в итоге он будет закрыт. | 2020-04-10T14:20:15 | 2020-04-10T14:21:53 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 612,078,832 | 10,130 | CurtizJ | I've added tests. Also now there is a check, that runs all tests with enabled compact parts. | 2020-04-10T15:28:14 | 2020-04-10T15:30:30 | {} | MEMBER |
ClickHouse | ClickHouse | 612,233,914 | 10,154 | filimonov | Test case:
```
set send_logs_level='trace',max_block_size=256, min_insert_block_size_rows=1, min_insert_block_size_bytes=1;
create table test_merge_oom engine=AggregatingMergeTree ORDER BY id as select number%1024 as id, arrayReduce( 'groupArrayState', arrayMap( x-> randomPrintableASCII(1024), range(8192) )) fat_sta... | 2020-04-10T21:51:53 | 2020-04-10T21:51:53 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 612,340,389 | 9,990 | Enmk | Currently, Hyperscan does not support that:
https://github.com/intel/hyperscan/issues/189
> * Hyperscan doesn't support _greedy_ or _ungreedy_ semantics but reports all matches instead. So in your case, `"2"`, `"20"`, `"201"` and `"2018"` are all valid matches for `[0-9]+`.
> You can refer to http://inte... | 2020-04-11T06:06:55 | 2020-04-11T06:06:55 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 612,407,321 | 10,179 | vzakaznikov | All new tests have passed so should be good to go. | 2020-04-11T12:15:01 | 2020-04-11T12:15:01 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 612,492,060 | 10,189 | alexey-milovidov | It reproduces even if I replace the arrows with ASCII characters. | 2020-04-11T19:04:57 | 2020-04-11T19:04:57 | {} | MEMBER |
ClickHouse | ClickHouse | 612,502,682 | 10,189 | trygveaa | According to the author of kitty, it is an error to send multiple restore cursor commands following each other without sending save cursor each time: https://github.com/kovidgoyal/kitty/issues/1264#issuecomment-449788068
It works in xterm though, so I'm not sure which terminal emulator has the correct behavior. | 2020-04-11T20:01:41 | 2020-04-11T20:01:41 | {} | NONE |
ClickHouse | ClickHouse | 611,828,193 | 10,155 | alexey-milovidov | Or: simply replace `countDistinct` to `uniq`. | 2020-04-10T01:18:14 | 2020-04-10T01:18:14 | {} | MEMBER |
ClickHouse | ClickHouse | 611,864,779 | 10,140 | bohutang | Comments:
can be tested | 2020-04-10T03:51:28 | 2020-04-10T03:51:28 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 611,996,014 | 10,168 | azat | >Add libmsgpack-dev into the image (for unbundled build)
This requires updating & push docker image manually, or I added in the wrong place?
>2020-04-10 09:04:14 -- Using msgpack: MSGPACK_INCLUDE_DIR-NOTFOUND | 2020-04-10T11:45:56 | 2020-04-10T11:45:56 | {} | MEMBER |
ClickHouse | ClickHouse | 612,036,766 | 10,173 | alexey-milovidov | Ok but...
1. let's add small sleep between retries.
2. let's add `.python` to the list of file extensions that will trigger tests run. | 2020-04-10T13:50:01 | 2020-04-10T13:50:01 | {} | MEMBER |
ClickHouse | ClickHouse | 612,045,157 | 6,464 | sfriquet | > @sfriquet Distributed table has to merge the intermediate results from remote servers.
> It is needed because data for the same GROUP BY key (in your case - `id`) may reside on multiple shards.
@alexey-milovidov Thanks for your reply.
In cases where there are few distinct keys to aggregate (here only 9), what ... | 2020-04-10T14:10:45 | 2020-04-10T14:17:50 | {} | NONE |
ClickHouse | ClickHouse | 612,324,970 | 10,015 | Slach | @blinkov I fix conflicts with master could you merge this PR? | 2020-04-11T04:38:45 | 2020-04-11T04:38:45 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 612,453,928 | 10,155 | amosbird | wait a sec. There is a problem. FixedHashSet takes quite some space. I just noticed this comment
` /// When creating, the hash table must be small.
`
Per group lookup table would be problematic if the actual load factor is low. | 2020-04-11T16:06:16 | 2020-04-11T16:07:23 | {} | COLLABORATOR |
ClickHouse | ClickHouse | 611,308,901 | 10,094 | alexey-milovidov | Ivan Blinkov has written a wrong script that assigns random people. This is by mistake and to be removed soon. Sorry for that. | 2020-04-09T03:32:40 | 2020-04-11T20:04:38 | {} | MEMBER |
ClickHouse | ClickHouse | 612,504,267 | 10,168 | azat | >Now it fails for different reason.
It was able to find msgpack:
>2020-04-10 16:15:04 -- Using msgpack: /usr/include
But failed to find lots of other stuff (cctz and so on), looks like the image was not completely correct.
P.S. I'm trying to pull the image and the layer ids are different | 2020-04-11T20:10:48 | 2020-04-11T20:10:48 | {} | MEMBER |
ClickHouse | ClickHouse | 612,504,994 | 10,085 | alexey-milovidov | > if there is NULL in IN operator values
Throw an exception if there is NULL in explicit list of values at the right hand of IN:
`x IN (1, NULL, 2)`
(but we cannot do that for subqueries or for Nullable columns at the left hand side)
Yes, that will be good feature (for separate pull request). | 2020-04-11T20:15:07 | 2020-04-11T20:15:27 | {} | MEMBER |
ClickHouse | ClickHouse | 611,786,080 | 1,604 | filimonov | I think it is about that
https://github.com/apache/thrift/blob/master/lib/java/src/org/apache/thrift/TSerializer.java
https://stackoverflow.com/questions/12328896/thrift-is-it-possible-to-do-only-serialization-with-c-thrift-library
https://www.programcreek.com/java-api-examples/?api=org.apache.thrift.TSerialize... | 2020-04-09T22:34:10 | 2020-04-09T22:35:55 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 612,045,261 | 10,009 | qoega | should be rebased. Can you make similar change for other languages also? | 2020-04-10T14:10:58 | 2020-04-10T14:10:58 | {} | MEMBER |
ClickHouse | ClickHouse | 612,051,806 | 10,161 | BayoNet | Спасибо большое :-)
Какая-то часть настроек в каком-то виде описана здесь https://clickhouse.tech/docs/en/engines/table_engines/mergetree_family/mergetree/#mergetree-query-clauses. Мы их потом оттуда выпилим и оставим только ссылку на новый раздел. | 2020-04-10T14:25:24 | 2020-04-10T14:25:24 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 612,164,941 | 10,053 | azat | > And it's strange if I cannot simply list all values that I want to see.
I can add parsing of a string, i.e. `log_queries_type="QUERY_FINISH EXCEPTION_WHILE_PROCESSING"`, but this also does not looks good (at least to me), and making [this](https://github.com/ClickHouse/ClickHouse/pull/10053#issuecomment-610674443)... | 2020-04-10T18:45:15 | 2020-04-10T18:45:15 | {} | MEMBER |
ClickHouse | ClickHouse | 612,346,926 | 9,990 | filimonov | That sound like more about captured groups in regexps (i.e. smth like `xyz=([0-9]+)([a-z]+)`).
Here we need to return whole match (by several regexps) | 2020-04-11T06:43:57 | 2020-04-11T06:43:57 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 612,445,565 | 10,155 | amosbird | > Both of them are using the same data structure (hash table).
@alexey-milovidov UInt16 could use a lookup table which is faster. Though in this case it doesn't matter much :) | 2020-04-11T15:23:58 | 2020-04-11T15:25:14 | {} | COLLABORATOR |
ClickHouse | ClickHouse | 612,485,695 | 10,193 | alexey-milovidov | @blinkov What kind of increase? | 2020-04-11T18:34:48 | 2020-04-11T18:34:48 | {} | MEMBER |
ClickHouse | ClickHouse | 612,494,510 | 10,189 | alexey-milovidov | But it may flicker on slow ssh sessions. | 2020-04-11T19:17:04 | 2020-04-11T19:17:04 | {} | MEMBER |
ClickHouse | ClickHouse | 612,496,067 | 10,189 | alexey-milovidov | This task can be fixed together: https://github.com/ClickHouse/ClickHouse/issues/5908 | 2020-04-11T19:24:40 | 2020-04-11T19:24:40 | {} | MEMBER |
ClickHouse | ClickHouse | 612,503,525 | 10,168 | alexey-milovidov | But probably only because the CI was broken at that specific time... | 2020-04-11T20:06:24 | 2020-04-11T20:06:24 | {} | MEMBER |
ClickHouse | ClickHouse | 612,503,689 | 10,168 | alexey-milovidov | Let's test it again. | 2020-04-11T20:07:24 | 2020-04-11T20:07:24 | {} | MEMBER |
ClickHouse | ClickHouse | 612,508,111 | 10,168 | azat | FWIW I checked the image, and it does contains libmsgpack-dev now:
```
$ docker run --rm -it yandex/clickhouse-deb-builder apt-cache policy libmsgpack-dev
libmsgpack-dev:
Installed: 3.0.1-3
Candidate: 3.0.1-3
Version table:
*** 3.0.1-3 500
500 http://archive.ubuntu.com/ubuntu eoan/universe amd6... | 2020-04-11T20:33:28 | 2020-04-11T20:33:28 | {} | MEMBER |
ClickHouse | ClickHouse | 611,827,966 | 10,155 | alexey-milovidov | Yes, it can be accelerated.
```
SET group_by_two_level_threshold = 1, group_by_two_level_threshold_bytes = 1;
SELECT sum(c) FROM (SELECT count(DISTINCT profile_id_hash) AS c FROM clustered_face4 GROUP BY profile_id_hash % 256)
```
Todo: provide explanation. | 2020-04-10T01:17:25 | 2020-04-10T01:17:51 | {} | MEMBER |
ClickHouse | ClickHouse | 611,902,791 | 5,029 | filimonov | It will be fixed here #10163 | 2020-04-10T06:41:18 | 2020-04-10T06:41:18 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 612,221,207 | 10,141 | filimonov | May be related to https://github.com/ClickHouse/ClickHouse/issues/9965 (which was fixed here https://github.com/ClickHouse/ClickHouse/pull/9971 , but not in release yet) | 2020-04-10T21:15:31 | 2020-04-10T21:16:19 | {
"+1": 1
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 612,354,859 | 8,636 | DAOkiin | > There should be a query string in the trace log before the crash.
do you use ArgMax(column, version) to aggregate data from shards? if you select more than one column, try doing it like this:
argMax(tuple(col1,col2,col3), version) FROM table | 2020-04-11T07:27:53 | 2020-04-11T07:27:53 | {} | NONE |
ClickHouse | ClickHouse | 612,452,678 | 8,313 | blinkov | @svladykin, do you have any further questions? | 2020-04-11T15:58:01 | 2020-04-11T15:58:01 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 612,485,636 | 10,193 | blinkov | It'll lead to unnecessary latency increases during work time at Moscow. | 2020-04-11T18:34:29 | 2020-04-11T18:34:29 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 612,491,633 | 10,189 | alexey-milovidov | Actually the reason is different. | 2020-04-11T19:02:54 | 2020-04-11T19:02:54 | {} | MEMBER |
ClickHouse | ClickHouse | 611,835,700 | 3,822 | 631068264 | ClickHouse server version 20.3.5 revision also get this error how to fix it. I update from **version 18.16**
Even I del `/var/lib/clickhouse/data/loh/attack_event/tmp_clone_20190731_20190731_470338_470758_110_15067` when stop server. **It will create similar dir get the same error when it restart** .
```... | 2020-04-10T01:48:48 | 2020-04-10T02:01:19 | {
"+1": 1
} | NONE |
ClickHouse | ClickHouse | 612,025,421 | 10,121 | alesapin | backport only to 20.3 | 2020-04-10T13:21:08 | 2020-04-10T13:21:08 | {} | MEMBER |
ClickHouse | ClickHouse | 611,649,015 | 10,130 | CurtizJ | Will fix #10157 in new PR.
UPD: fixed in this PR. | 2020-04-09T17:21:58 | 2020-04-10T13:37:48 | {} | MEMBER |
ClickHouse | ClickHouse | 612,039,521 | 6,464 | alexey-milovidov | @sfriquet Distributed table has to merge the intermediate results from remote servers.
It is needed because data for the same GROUP BY key (in your case - `id`) may reside on multiple shards. | 2020-04-10T13:56:58 | 2020-04-10T13:56:58 | {} | MEMBER |
ClickHouse | ClickHouse | 612,040,461 | 6,464 | alexey-milovidov | @achimbab But your case is different. | 2020-04-10T13:59:21 | 2020-04-10T13:59:21 | {} | MEMBER |
ClickHouse | ClickHouse | 612,486,210 | 10,193 | blinkov | It obviously depends on website visitor location on the globe and network connection quality along the way. | 2020-04-11T18:37:23 | 2020-04-11T18:37:23 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 610,232,148 | 10,076 | toby080508050805 | We use version
> ClickHouse client version 19.17.5.18 (official build).
> Connected to ClickHouse server version 19.17.5 revision 54428.
The query try to select count from merge table. The first one is show create merge table and rest of them are member of merge tables create table stamen.
> CREATE TABLE gt_... | 2020-04-07T07:46:51 | 2020-04-10T04:44:51 | {} | NONE |
ClickHouse | ClickHouse | 612,076,233 | 9,972 | akuzm | (did not backport to 20.1 as requested above, the tag is just to get it out of the list) | 2020-04-10T15:22:14 | 2020-04-10T15:22:32 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 612,216,597 | 10,180 | filimonov | May be related: https://github.com/ClickHouse/ClickHouse/issues/10022 | 2020-04-10T21:03:00 | 2020-04-10T21:03:00 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 612,439,107 | 10,191 | alexey-milovidov | Fixed with the following commands:
```
find ru -name '*.md' | xargs grep -l -F 'machine_translated: true' > list.txt
cat list.txt | xargs rm
cat list.txt | while read f; do ln -s ${f/ru/en} $f; done
``` | 2020-04-11T14:53:52 | 2020-04-11T14:53:52 | {} | MEMBER |
ClickHouse | ClickHouse | 612,469,889 | 10,192 | vitlibar | Actually on master it already works well because of using `StorageID` when checking access.
This PR is to fix the checking in 20.3 too. | 2020-04-11T17:20:48 | 2020-04-11T17:21:17 | {} | MEMBER |
ClickHouse | ClickHouse | 611,801,333 | 10,164 | alexey-milovidov | This is not non-significant. | 2020-04-09T23:29:57 | 2020-04-09T23:29:57 | {
"+1": 1
} | MEMBER |
ClickHouse | ClickHouse | 611,849,923 | 10,162 | den-crane | try to disable experimental_use_processors
`set experimental_use_processors = 0;` before the query | 2020-04-10T02:44:52 | 2020-04-10T02:44:52 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 612,077,535 | 10,162 | cw9 | Hi the stack above is from the node issuing the distributed query.
Here is a stack trace from the node with the underlying replicated merge tree table that had the seg fault:
```
{"log":"2020.04.10 15:03:04.808968 [ 399 ] {} \u003cFatal\u003e BaseDaemon: ########################################\n"}
{"log":"2020... | 2020-04-10T15:25:22 | 2020-04-10T15:25:22 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 612,453,291 | 10,155 | amosbird | yeah, I think it's worth doing. Lemme add it along with a perftest | 2020-04-11T16:01:58 | 2020-04-11T16:01:58 | {} | COLLABORATOR |
ClickHouse | ClickHouse | 612,500,479 | 10,189 | alexey-milovidov | Slow connections can be tested with https://github.com/sjmulder/trickle | 2020-04-11T19:48:59 | 2020-04-11T19:48:59 | {} | MEMBER |
ClickHouse | ClickHouse | 611,811,751 | 9,808 | azat | Actually this is not only fixes DISTINCT, but also makes **auto**-`distributed_group_by_no_merge`, which should improve the query processing | 2020-04-10T00:11:05 | 2020-04-10T00:11:31 | {} | MEMBER |
ClickHouse | ClickHouse | 611,872,068 | 6,387 | Slach | found https://github.com/ClickHouse/ClickHouse/blob/97f2a2213e754ba25dabba4bc8ddf507cd67660c/tests/queries/0_stateless/01092_memory_profiler.sql
thanks | 2020-04-10T04:27:45 | 2020-04-10T04:27:45 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 611,903,808 | 1,604 | filimonov | > As MsgPack format was merged to master, this issue is fixed.
:+1:
If somebody will need Thrift, he should open a separate issue with motivation and it can be linked to this one.
https://github.com/ClickHouse/ClickHouse/labels/comp-formats | 2020-04-10T06:45:07 | 2020-04-10T06:45:07 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 612,025,289 | 10,095 | alesapin | Backport only to 20.3 | 2020-04-10T13:20:44 | 2020-04-10T13:20:44 | {} | MEMBER |
ClickHouse | ClickHouse | 612,036,159 | 10,161 | BayoNet | Перевод на английский язык планируется? | 2020-04-10T13:48:28 | 2020-04-10T13:48:28 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 612,247,938 | 10,141 | genzgd | That appears to be the case for our issue -- turning off `experimental_use_processors` fixed the problem. Interestingly, the query ran fine on the "last" shard of the cluster, since it started all the remote queries before processing the local one. Hopefully the #9971 PR fixes that behavior. | 2020-04-10T22:34:12 | 2020-04-10T22:34:12 | {
"+1": 1
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 612,460,534 | 10,155 | alexey-milovidov | Yes, you're right. It will be worse. | 2020-04-11T16:38:34 | 2020-04-11T16:38:40 | {} | MEMBER |
ClickHouse | ClickHouse | 612,505,996 | 10,168 | alexey-milovidov | The image is Ok for commits on master. Merge... | 2020-04-11T20:20:41 | 2020-04-11T20:20:41 | {} | MEMBER |
ClickHouse | ClickHouse | 612,028,925 | 10,168 | alexey-milovidov | @alesapin will update the image before merging. | 2020-04-10T13:29:58 | 2020-04-10T13:29:58 | {} | MEMBER |
ClickHouse | ClickHouse | 612,453,036 | 10,155 | alexey-milovidov | @amosbird We still don't have such optimization for AggregateFunctionUniqExactData. | 2020-04-11T16:00:21 | 2020-04-11T16:00:21 | {} | MEMBER |
ClickHouse | ClickHouse | 612,492,587 | 10,189 | alexey-milovidov | The reason is in terminal interpretation of
```
#define SAVE_CURSOR_POSITION "\033""7"
#define RESTORE_CURSOR_POSITION "\033""8"
```
escape sequences.
Please file a report back to Kitty. | 2020-04-11T19:07:35 | 2020-04-11T19:07:35 | {} | MEMBER |
ClickHouse | ClickHouse | 612,150,673 | 10,174 | alexey-milovidov | The task is being developed by @xzkostyan (Telegram login: @tkostyan) | 2020-04-10T18:09:21 | 2020-04-10T18:33:47 | {} | MEMBER |
ClickHouse | ClickHouse | 612,398,867 | 10,149 | vitlibar | @nvartolomei It seems the problem what you're solving here is already solved in https://github.com/ClickHouse/ClickHouse/pull/9447 and https://github.com/ClickHouse/ClickHouse/pull/9852.
> Consider a case where initial query is sent by a user which has
readonly=1 on the profile and the query is sent to a node under... | 2020-04-11T11:29:34 | 2020-04-11T11:35:20 | {} | MEMBER |
ClickHouse | ClickHouse | 612,502,804 | 10,149 | alexey-milovidov | @vitlibar Thank you! | 2020-04-11T20:02:25 | 2020-04-11T20:02:25 | {} | MEMBER |
ClickHouse | ClickHouse | 612,038,502 | 10,170 | den-crane | you can `select hostName(), .... from distribute_query_log `
or `select FQDN(), .... from distribute_query_log` | 2020-04-10T13:54:12 | 2020-04-10T13:54:12 | {
"+1": 1
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 612,193,611 | 10,141 | genzgd | Does this issue exist on older ClickHouse versions? I'm seeing what looks like similar behavior on 20.3.5 -- my distributed query is not sending all the shard queries at once. Our cluster is 10 shards on 10 different physical servers. Running a trace log, I see the distributed query sent to only 3 other shards initi... | 2020-04-10T20:00:19 | 2020-04-10T20:03:33 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 612,211,455 | 9,945 | den-crane | Did you tune `compile_expressions=0`?
Do you have some graphics with memory usage?
Do you know SQL query which causes the issue?
Need at least /var/log/clickhouse-server/clickhouse-server.log with logging level trace to find out what causes it. | 2020-04-10T20:51:42 | 2020-04-10T20:51:42 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 612,213,212 | 10,186 | alexey-milovidov | @filimonov FYI @blinkov is actually a robot. | 2020-04-10T20:56:14 | 2020-04-10T20:56:14 | {
"laugh": 2
} | MEMBER |
ClickHouse | ClickHouse | 611,062,493 | 9,088 | filimonov | Reproduces on 20.4.1 (master).
```
SELECT arrayJoin([0, 1, 3, NULL]) AS x, x = 0, if(x = 0, 'x=0', 'x<>0');
```
Related https://github.com/ClickHouse/ClickHouse/pull/5238
/cc @janplus | 2020-04-08T16:35:34 | 2020-04-10T22:26:56 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 612,440,428 | 10,191 | alexey-milovidov | Slightly incorrect, we need relative symlinks.
https://unix.stackexchange.com/questions/85060/getting-relative-links-between-two-paths
Will do... | 2020-04-11T15:00:18 | 2020-04-11T15:00:18 | {} | MEMBER |
ClickHouse | ClickHouse | 612,441,280 | 10,191 | alexey-milovidov | The correct list of commands:
```
find ru -name '*.md' | xargs grep -l -F 'machine_translated: true' > list.txt
cat list.txt | xargs rm
cat list.txt | while read f; do ln -s $(readlink -f ${f/ru/en}) $f; done
symlinks -cdr ru
``` | 2020-04-11T15:04:25 | 2020-04-11T15:04:25 | {} | MEMBER |
ClickHouse | ClickHouse | 612,453,522 | 10,155 | alexey-milovidov | Yes, it will be great! | 2020-04-11T16:03:29 | 2020-04-11T16:03:29 | {} | MEMBER |
ClickHouse | ClickHouse | 612,050,875 | 10,085 | qoega | > NULL is not equals to NULL
OK, is it more user-friendly if we return warning or exception if there is NULL in IN operator values, but this setting is disabled? | 2020-04-10T14:23:20 | 2020-04-11T20:13:21 | {} | MEMBER |
ClickHouse | ClickHouse | 612,506,363 | 10,189 | alexey-milovidov | I'm happy that the root cause is found!
Yes, it also works fine in Konsole.
But I have removed these sequences at all in my PR: #10194 | 2020-04-11T20:22:40 | 2020-04-11T20:22:40 | {
"+1": 1
} | MEMBER |
ClickHouse | ClickHouse | 611,817,733 | 9,808 | alexey-milovidov | That's awesome! | 2020-04-10T00:34:48 | 2020-04-10T00:34:48 | {} | MEMBER |
ClickHouse | ClickHouse | 612,155,091 | 10,002 | azat | Rebased against master and fixed some links (that has been broken due to massive rename) | 2020-04-10T18:20:21 | 2020-04-10T18:20:21 | {} | MEMBER |
ClickHouse | ClickHouse | 612,212,588 | 10,183 | filimonov | But what an action do you expect on failure?
We can't blindly run the script once again.
imagine it inserts 10K rows at the beginning successfully and fails later on line 10001.
If we will repeat you will get 20K rows instead of 10K.
Init scripts can be shell script too (just use .sh extension).
You can s... | 2020-04-10T20:54:42 | 2020-04-10T20:54:42 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 612,216,072 | 10,178 | filimonov | Can't reproduce. Checked 20.3.3.6 and 20.3.5.21.
Your sample just works.
| 2020-04-10T21:01:36 | 2020-04-10T21:01:36 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 612,501,326 | 10,189 | trygveaa | It seems like kitty doesn't use the same cursor position if you send the restore cursor sequence twice after each other, e.g. like this:
```
echo -e '\x1B7 hello \x1B8 world \x1B8 world'
```
My guess is that's what causes the issue. In e.g. xterm the two restore cursor position sequences will move the cursor to... | 2020-04-11T19:53:45 | 2020-04-11T20:33:47 | {} | NONE |
ClickHouse | ClickHouse | 611,826,185 | 10,155 | ZonghaoWang | thanks.
it work with 900millin rows on nvme with 30Gbps
when I process distinct(UInt32), data read in 7s, the progressbar goes to 98%. the last 2% cost 23s. `system.column` show as follow.
... | 2020-04-10T01:10:49 | 2020-04-10T01:10:49 | {} | NONE |
ClickHouse | ClickHouse | 611,873,525 | 10,155 | ZonghaoWang | thank you.
| 2020-04-10T04:35:15 | 2020-04-10T04:35:15 | {} | NONE |
ClickHouse | ClickHouse | 611,961,276 | 10,085 | ghost | @alexey-milovidov @blinkov
I fixed the failed case.
And I fixed some codes and testcases for Tuple and StorageSet.
Thank you. | 2020-04-10T09:41:19 | 2020-04-10T09:44:23 | {
"+1": 1
} | NONE |
ClickHouse | ClickHouse | 612,009,916 | 6,464 | sfriquet | Hi, I might have a similar question, so adding this here in case it's related.
I also see a case where a query is significantly slower on a Distributed table than on the union of all its shards.
e.g:
```
SELECT count() FROM test.foo GROUP BY toDate(timestamp)
┌───count()─┐
│ 260533159 │
└───────────┘
```
... | 2020-04-10T12:33:46 | 2020-04-10T12:38:09 | {} | NONE |
ClickHouse | ClickHouse | 612,049,558 | 6,464 | sfriquet | @alexey-milovidov I think it might still be related to this issue:
Here's another example without `group by`:
```
:) select sum(id), count(*) from test.foo;
┌────────────sum(id)─┬───count()─┐
│ 385975829488090336 │ 260533159 │
└────────────────────┴───────────┘
1 rows in set. Elapsed: 0.023 sec. Proces... | 2020-04-10T14:20:19 | 2020-04-10T14:25:05 | {} | NONE |
ClickHouse | ClickHouse | 612,092,266 | 10,162 | cw9 | Found something new that is interesting
This cases panic in subqueries:
SELECT countIf(level = 'AAA')
FROM default."Distributed.table"
PREWHERE (level != 'AAA') AND (_service = '_service') AND ((_ts >= 1586369299944) AND (_ts < 1586455699944))
This does not cause panic in subqueries:
SELECT countIf(level = '... | 2020-04-10T15:53:06 | 2020-04-10T15:54:45 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 612,391,205 | 10,189 | qoega | Thank you. Actually you can try to fix it by yourself. It looks it should be easy even if you are not a C++ programmer. | 2020-04-11T10:47:07 | 2020-04-11T10:47:07 | {} | MEMBER |
ClickHouse | ClickHouse | 612,442,604 | 10,191 | alexey-milovidov | Fixed. Now we have to redeploy the website. | 2020-04-11T15:10:47 | 2020-04-11T15:10:47 | {} | MEMBER |
ClickHouse | ClickHouse | 612,460,924 | 10,155 | alexey-milovidov | > Todo: provide explanation.
If we have just a single aggregate function value, the merge operation is not parallelized.
(It's only parallelized across multiple different aggregation states.)
It is possible area for improvement. | 2020-04-11T16:40:08 | 2020-04-11T16:40:08 | {} | MEMBER |
ClickHouse | ClickHouse | 612,493,018 | 10,189 | alexey-milovidov | But at least this simple test passed:
`echo -e '\x1B7 hello \x1B8 world'` | 2020-04-11T19:09:33 | 2020-04-11T19:09:39 | {} | MEMBER |
ClickHouse | ClickHouse | 611,902,622 | 8,160 | filimonov | It will be fixed here: #10163 | 2020-04-10T06:40:36 | 2020-04-10T06:40:36 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 612,187,312 | 9,945 | FedericoCeratto | I reproduced the issue on a host with 32GB of RAM, kernel 4.19.0-8-cloud-amd64, same version of ClickHouse:
```
[ 9380.054067] Out of memory: Kill process 1779 (clickhouse-serv) score 818 or sacrifice child
[ 9380.062029] Killed process 1779 (clickhouse-serv) total-vm:55547844kB, anon-rss:26644232kB, file-rss:0kB, s... | 2020-04-10T19:41:45 | 2020-04-10T19:41:45 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 612,493,571 | 10,189 | alexey-milovidov | We can get the same effect by simply sending "carriage return". | 2020-04-11T19:12:07 | 2020-04-11T19:17:33 | {} | MEMBER |
ClickHouse | ClickHouse | 612,020,828 | 10,162 | salamanderrex | > try to disable experimental_use_processors
> `set experimental_use_processors = 0;` before the query
I just tried that. It still does not work. | 2020-04-10T13:08:13 | 2020-04-10T13:08:13 | {} | NONE |
ClickHouse | ClickHouse | 612,395,934 | 10,169 | abyss7 | > Ok. Let's wait for performance tests just in case...
@akuzm says that all failed performance tests are really ok. | 2020-04-11T11:13:46 | 2020-04-11T11:13:46 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 612,502,109 | 10,168 | alexey-milovidov | Now it fails for different reason. | 2020-04-11T19:58:22 | 2020-04-11T19:58:22 | {} | MEMBER |
ClickHouse | ClickHouse | 612,510,648 | 9,990 | alexey-milovidov | @filimonov
For single regular expression we can add `extractGroups` that will return an array of captured groups starting from 1 in the first match. Also we can add `extractAllGroups` that will return two-dimensional array of groups from all matches. | 2020-04-11T20:48:22 | 2020-04-11T20:48:42 | {} | MEMBER |
ClickHouse | ClickHouse | 612,535,644 | 9,741 | alexey-milovidov | It's also worth to note that replication is implement on per-table basis.
For example, you can have one table replicated three times, other table replicated ten times and other table not replicated at all. | 2020-04-11T23:35:41 | 2020-04-11T23:35:41 | {} | MEMBER |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.