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 | 781,770,225 | 20,630 | amosbird | https://github.com/ClickHouse/ClickHouse/pull/20206 | 2021-02-19T02:27:52 | 2021-02-19T02:27:52 | {} | COLLABORATOR |
ClickHouse | ClickHouse | 781,887,240 | 20,585 | alesapin | ```
node2.query("INSERT INTO ordinary.t1 SELECT number FROM numbers(10)")
node1.query("SYSTEM SYNC REPLICA ordinary.t1", timeout=10)
node3.query("SYSTEM SYNC REPLICA ordinary.t1", timeout=10)
> assert node1.query("SELECT COUNT() FROM ordinary.t1") == "10\n"
E Asserti... | 2021-02-19T07:23:13 | 2021-02-19T07:23:13 | {} | MEMBER |
ClickHouse | ClickHouse | 781,356,896 | 20,868 | den-crane | Your cluster_live configuration has 3 shards. Each shard has only 1 replica.
Can you show `show create table mytable` ?
In general to remove replica you should:
1. edit remote_servers and remove replica
2. execute `drop database ...` for all databases at this replica. | 2021-02-18T13:50:18 | 2021-02-18T13:50:18 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 781,536,271 | 20,867 | iameugenejo | @den-crane Here is the new test without the TTL:
```sql
create table tmp.rep_repl
(
`timestamp` DateTime,
`account_number` UInt64,
`account_name` String
) Engine = ReplicatedReplacingMergeTree('/clickhouse/tables/{layer}-{shard}/tmp.rep_repl', '{replica}')
PARTITION BY formatDateTime(timestamp, '%... | 2021-02-18T18:10:06 | 2021-02-18T18:10:06 | {} | NONE |
ClickHouse | ClickHouse | 781,540,129 | 20,867 | iameugenejo | Here is the same test with `ReplacingMergeTree` - https://gist.github.com/iameugenejo/c3e3af1bdf11d9fe1c0112817864d8a0 (not posting the snippet here directly to keep the irrelevant info from polluting the discussion) | 2021-02-18T18:16:21 | 2021-02-18T18:16:21 | {} | NONE |
ClickHouse | ClickHouse | 781,561,782 | 20,867 | den-crane | ```
insert into tmp.rep_repl select toStartOfDay(now()), 1, 'test';
insert into tmp.rep_repl select toStartOfDay(now()), 1, 'test';
insert into tmp.rep_repl select toStartOfDay(now()), 1, 'test';
```
try ` set insert_deduplicate=0`
https://clickhouse.tech/docs/en/engines/table-engines/mergetree-family/replica... | 2021-02-18T18:52:14 | 2021-02-18T18:54:43 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 781,567,284 | 19,673 | azat | Rebased to fix conflicts with upstream | 2021-02-18T19:01:51 | 2021-02-18T19:01:51 | {} | MEMBER |
ClickHouse | ClickHouse | 781,339,934 | 10,637 | terox | Hello @George3d6 ! The variables aren't documented in the docker hub but works fine in my container. | 2021-02-18T13:22:41 | 2021-02-18T13:22:41 | {
"+1": 1
} | NONE |
ClickHouse | ClickHouse | 781,354,213 | 20,871 | den-crane | >visits_v1.tsv " for many times following the instructions
Why? Can you clarify what instructions did you follow?
tutorial.visits_v1 has the ENGINE = CollapsingMergeTree(Sign)
This engine removes duplicates https://clickhouse.tech/docs/en/engines/table-engines/mergetree-family/collapsingmergetree/ automatical... | 2021-02-18T13:46:00 | 2021-02-18T13:46:00 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 781,454,064 | 20,867 | iameugenejo | @den-crane I just tested without `timestamp` VER column, and it's behaving the same.
The documentation says ```The last in the selection, if ver not set. A selection is a set of rows in a set of parts participating in the merge. The most recently created part (the last insert) will be the last one in the selection. ... | 2021-02-18T16:09:52 | 2021-02-18T16:30:08 | {} | NONE |
ClickHouse | ClickHouse | 781,894,451 | 20,945 | filimonov | @oandrew are you interested in that? | 2021-02-19T07:38:19 | 2021-02-19T07:38:19 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 782,041,421 | 6,107 | Felixoid | > add a line `<allow_experimental_data_skipping_indices>1</allow_experimental_data_skipping_indices>` into users.xml in profiles / default subsection.
Just a little clarification: the DDDL are executed as the default user, so it's necessary to add the setting to whatever profile related to the `default` | 2021-02-19T12:20:35 | 2021-02-19T12:20:35 | {} | MEMBER |
ClickHouse | ClickHouse | 782,064,224 | 20,550 | CurtizJ | Hung queries seem to be related to changes.
From https://clickhouse-test-reports.s3.yandex.net/20550/0b315fccc286adc7fa23d7849fdff5a8ca51c6a9/functional_stateless_tests_(release,_wide_parts_enabled)/test_run.txt.out.log
```sql
SELECT ( SELECT rank() ) , ( SELECT * FROM zeros(NULL, (CAST(('') AS Date))) ) AS u;
``... | 2021-02-19T13:09:48 | 2021-02-19T13:09:48 | {} | MEMBER |
ClickHouse | ClickHouse | 782,182,084 | 20,878 | tavplubix | Yandex third-party checks - flacky, passed after restart | 2021-02-19T16:24:28 | 2021-02-19T16:24:28 | {} | MEMBER |
ClickHouse | ClickHouse | 781,598,258 | 20,867 | iameugenejo | @den-crane Great, that's the exact information I needed. Thanks for clarifying it.
Revised the table to have the timestamp as is for the uniqueness of the insert, and it's working fine without setting off the flag. Either approach seems to serve the same purpose for my need though. In the end, I don't want insert de... | 2021-02-18T19:56:52 | 2021-02-18T20:21:39 | {
"+1": 1
} | NONE |
ClickHouse | ClickHouse | 781,730,322 | 20,608 | den-crane | What CH version do you use with old servers? Try that version with new servers. | 2021-02-19T00:35:16 | 2021-02-19T00:35:16 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 781,890,088 | 20,585 | alesapin | ```
2021.02.19 01:53:32.947332 [ 15 ] {12b069f3-4315-439c-8135-86ca0aba55da} <Debug> executeQuery: (from 172.19.0.1:33412, using production parser) SYSTEM SYNC REPLICA ordinary.t1
2021.02.19 01:53:32.947690 [ 15 ] {12b069f3-4315-439c-8135-86ca0aba55da} <Trace> ContextAccess (default): Access granted: SYSTEM SYNC REPL... | 2021-02-19T07:28:53 | 2021-02-19T07:42:15 | {} | MEMBER |
ClickHouse | ClickHouse | 781,926,667 | 19,644 | taiyang-li | Failed tests look like irrelevant to my changes. | 2021-02-19T08:44:02 | 2021-02-19T08:44:02 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 781,980,807 | 16,481 | amosbird | @azat Another uncaught exception from dtor. Please take a look. https://clickhouse-test-reports.s3.yandex.net/16481/5d150cce4778dd14f58dcff67435bdec1efa155b/stress_test_(thread).html#fail1 | 2021-02-19T10:19:37 | 2021-02-19T10:19:37 | {} | COLLABORATOR |
ClickHouse | ClickHouse | 782,090,210 | 20,462 | azat | >Marker check Pending — Starting checks (No special conditions apply)
Looks hanged
@alexey-milovidov can you merge this? (maybe it will reproduce the problem and will have full stacktrace, in the meantime I will look into this https://github.com/ClickHouse/ClickHouse/pull/16481#issuecomment-781980807) | 2021-02-19T13:55:45 | 2021-02-19T13:55:45 | {} | MEMBER |
ClickHouse | ClickHouse | 782,098,053 | 20,864 | alesapin | Ready https://clickhouse-test-reports.s3.yandex.net/0/ea27c3ca32bdf9a18e90d75bf38bbc725c6db4db/push_to_dockerhub.html, I'll rerun check. | 2021-02-19T14:09:00 | 2021-02-19T14:09:00 | {} | MEMBER |
ClickHouse | ClickHouse | 782,159,871 | 18,722 | oleg-savko | It would be great if in catboost configuration file we can specify for example s3 bucket, and clickhouse load models from it periodically. | 2021-02-19T15:47:44 | 2021-02-19T15:47:44 | {} | NONE |
ClickHouse | ClickHouse | 781,233,424 | 20,611 | kitaisreal | > Ok.
>
> BTW, why don't allow to reinterpret DateTime64 <-> Decimal64?
Added support for Decimals, DateTime64. | 2021-02-18T10:09:41 | 2021-02-18T10:09:41 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 781,341,483 | 20,209 | Qsimple | > > We are developing some Table parts monitor
>
> Can you clarify ? What for do you need `SELECT xxx FROM system.parts` ?
>
> ```
> SELECT COUNT(active)
> FROM system.parts
>
> ┌─COUNT(active)─┐
> │ 11155 │
> └───────────────┘
>
> 1 rows in set. Elapsed: 0.827 sec. Processed 11.15 thousand rows... | 2021-02-18T13:25:19 | 2021-02-18T13:25:19 | {} | NONE |
ClickHouse | ClickHouse | 781,391,126 | 20,089 | bharatnc | @alexey-milovidov made the necessary changes, does it look good to you ? | 2021-02-18T14:42:48 | 2021-02-18T14:42:48 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 781,616,743 | 6,349 | azat | @alexey-milovidov here is an RFC #20477 to address this | 2021-02-18T20:31:45 | 2021-02-18T20:31:45 | {} | MEMBER |
ClickHouse | ClickHouse | 781,917,951 | 20,955 | filimonov | Also - maybe a possibility to cross-check the source and target at the end (by rows number and some checksums?). | 2021-02-19T08:26:10 | 2021-02-19T08:26:10 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 781,934,614 | 20,168 | alexey-milovidov | This PR looks ready but I cannot enable compression by default:
- sometimes it improves performance sometimes not; it depends on queries, on the dataset and on specific server configurations. | 2021-02-19T08:58:54 | 2021-02-19T08:58:54 | {} | MEMBER |
ClickHouse | ClickHouse | 781,997,391 | 7,752 | PHameete | This would be a very useful feature for filling gaps in time series data. My use case looks a lot like @zhang2014 's use case:
```
WITH
INTERVAL 5 minute as granule,
toDateTime('2021-02-19 09:25:00') as from_date,
toDateTime('2021-02-19 09:50:00') as to_date
SELECT subject_id, toStartOfInterval(start_d... | 2021-02-19T10:50:35 | 2021-02-19T10:53:13 | {
"+1": 4
} | NONE |
ClickHouse | ClickHouse | 782,019,795 | 19,885 | vdimir | I've added implementation of type conversion for `JOIN ON` syntax. It don't change types in result table. For that conversion do not replace original columns, but and new ones and join performed by this new columns.
| 2021-02-19T11:34:39 | 2021-02-19T11:34:39 | {} | MEMBER |
ClickHouse | ClickHouse | 782,037,807 | 20,382 | stigsb | @alexey-milovidov could you approve this for testing please? | 2021-02-19T12:13:15 | 2021-02-19T12:13:15 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 782,065,121 | 20,550 | CurtizJ | From https://clickhouse-test-reports.s3.yandex.net/20550/e5cef576e589f4307f35074cf45e8dbb08801c65/functional_stateless_tests_(memory)/test_run.txt.out.log
```sql
SELECT (CAST((( SELECT row_number() ) , ( SELECT * FROM merge('\0') ) AS mdjw) AS Date)), NULL, (CAST((( SELECT avgWeighted() ) , ( SELECT * FROM jdbc() )... | 2021-02-19T13:11:37 | 2021-02-19T13:11:37 | {} | MEMBER |
ClickHouse | ClickHouse | 780,512,587 | 19,766 | ghost | @vdimir
It's okay.
I implemented `sequenceNextNode` for [flow analysis](https://mixpanel.com/topics/user-flow-analysis/)
> event
It refers to activities of users of websites.
> chain
The `chain` consists of the events of users.
> data in table
In the below tables schema, the action column refer... | 2021-02-17T12:06:57 | 2021-02-18T11:25:55 | {} | NONE |
ClickHouse | ClickHouse | 781,406,342 | 17,623 | tisonkun | @KochetovNicolai what do you mean by "Remove DataStreams"? Is there an isolated issue or explain document? | 2021-02-18T15:04:01 | 2021-02-18T15:04:01 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 781,592,171 | 20,867 | den-crane | Do you realize that it's about **insert** (all rows in the same order) not about **rows** in insert? Deduplication skips repeated inserts not rows.
>that's our real life production situation we have that real inserts do repeat naturally.
Then you should add unique insert identifier generated at the source (additi... | 2021-02-18T19:45:39 | 2021-02-18T19:45:39 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 781,875,808 | 17,623 | alexey-milovidov | @TisonKun We have tech dept - we started to rewrite DataStreams to Processors but only completed it for 90%. | 2021-02-19T07:00:55 | 2021-02-19T07:00:55 | {} | MEMBER |
ClickHouse | ClickHouse | 781,876,753 | 17,623 | alexey-milovidov | @itzikiusa We did not find any ways to do it efficiently* - after some solution will be proposed we can add it to roadmap.
\* there are ways to implement it if table size is less than a few billion records or server has fast SSD and table is not replicated. | 2021-02-19T07:02:48 | 2021-02-19T07:02:48 | {} | MEMBER |
ClickHouse | ClickHouse | 781,934,086 | 20,168 | alexey-milovidov | @amosbird If we lower block size to 8192 for this Memory table, query will be faster - about 30 ms if I remember correctly.
| 2021-02-19T08:57:55 | 2021-02-19T08:57:55 | {} | MEMBER |
ClickHouse | ClickHouse | 781,951,324 | 20,959 | filimonov | P.S. usually it's handier to create a storage table for MV explicitly and use TO syntax for materialized view creation:
```
CREATE MATERIALIZED VIEW mv TO mv_storage AS SELECT ...
```
| 2021-02-19T09:26:11 | 2021-02-19T09:26:11 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 782,035,930 | 9,510 | PHameete | We too are running into this issue. It would be nice if toStartOfInterval will work in the correct time zone (aligning with 00:00:00 for that day in the correct time zone). It seems this is only an issue with time, not with date.
For example:
```
e5ad12e1236a :) SELECT toStartOfInterval(toDateTime('2021-02-19 01... | 2021-02-19T12:09:23 | 2021-02-19T12:09:23 | {
"+1": 1
} | NONE |
ClickHouse | ClickHouse | 781,203,704 | 20,864 | azat | >Fasttest doesn't use an updated image from PR
@alesapin maybe you can update it manually? | 2021-02-18T09:19:48 | 2021-02-18T09:19:48 | {} | MEMBER |
ClickHouse | ClickHouse | 781,878,076 | 20,089 | alexey-milovidov | Almost. But the changes in this PR are "orphane" - they have no usages and cannot be tested.
I will take this PR and integrate the changes into another PR. | 2021-02-19T07:05:28 | 2021-02-19T07:05:28 | {
"+1": 1
} | MEMBER |
ClickHouse | ClickHouse | 782,061,373 | 20,679 | kitaisreal | accurateCast function was not back ported to 20.8. | 2021-02-19T13:04:09 | 2021-02-19T13:04:09 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 782,063,643 | 20,658 | kitaisreal | Previous changes necessary to resolve this conflicts without regressions were not back ported to 20.8. | 2021-02-19T13:08:35 | 2021-02-19T13:08:35 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 782,145,392 | 14,197 | bobelev | Got this errors dozens of times while replicating huge cluster to new hosts | 2021-02-19T15:24:15 | 2021-02-19T15:24:15 | {} | NONE |
ClickHouse | ClickHouse | 782,155,984 | 20,864 | azat | >New report https://clickhouse-test-reports.s3.yandex.net/20864/ee18f6a7ec23304c7ebc5128882d163d510525e0/fast_test.html
Yes, it works, thank you! | 2021-02-19T15:41:24 | 2021-02-19T15:41:24 | {} | MEMBER |
ClickHouse | ClickHouse | 781,203,321 | 20,612 | zav379 | thanks | 2021-02-18T09:19:11 | 2021-02-18T09:19:11 | {} | NONE |
ClickHouse | ClickHouse | 781,312,408 | 19,766 | ghost | @vdimir
I added comments to the source code.
Feel free to comment if there are anything I have to do in order to merge this PR. Have a nice day. | 2021-02-18T12:34:57 | 2021-02-18T12:34:57 | {} | NONE |
ClickHouse | ClickHouse | 781,315,807 | 11,876 | xwydq | > it is impossible with AggregatingMergeTree. AggregatingMergeTree does not know that you need arrayReduce('sumMap' and just rewrites a value in _discrete column by function ANY on each insert.
>
> You can use SummingMT and nested https://clickhouse.tech/docs/en/engines/table-engines/mergetree-family/summingmergetre... | 2021-02-18T12:41:38 | 2021-02-18T12:41:38 | {} | NONE |
ClickHouse | ClickHouse | 781,536,909 | 20,490 | traceon | > It's limitation of headers itself, all of them sent before data
Right, we can't do better, when it comes to Summary. But then, each Progress should still contain updated elapsed time. | 2021-02-18T18:11:10 | 2021-02-18T23:19:30 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 781,899,576 | 20,167 | yiguolei | @ianton-ru
**part_size <= s3_max_single_part_upload_size** - use simple upload.
**s3_max_single_part_upload_size < part_size <= s3_min_upload_part_size** - use multipart upload for entry part.
What's the difference between them since the multipart upload in the second case only contains just one part ?
Is the pe... | 2021-02-19T07:49:20 | 2021-02-19T07:54:35 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 781,967,747 | 20,254 | siradjev | Thanks @amosbird , your proposal is obviously more generic and covers abovementioned problem.
It seems there is a need for ruleset to identify "proper stage" correctly.
E.g. query
01-grouping() -> 02-filtering() -> 03-grouping()
shall filter on initiator and continue on initiator. So basically after the first g... | 2021-02-19T09:56:19 | 2021-02-19T09:56:19 | {} | NONE |
ClickHouse | ClickHouse | 782,005,714 | 20,548 | tobzsc | In this scenario it was a single Kafka broker. A single Clickhouse instance is the only consumer. | 2021-02-19T11:06:52 | 2021-02-19T11:06:52 | {} | NONE |
ClickHouse | ClickHouse | 782,023,805 | 20,585 | alesapin | Actually, timeouts caused by the bad implementation of NuKeeperTCPHandler. | 2021-02-19T11:42:52 | 2021-02-19T11:42:52 | {} | MEMBER |
ClickHouse | ClickHouse | 782,133,862 | 17,761 | kitaisreal | @den-crane could you please specify version. Could not reproduce on current master. | 2021-02-19T15:09:21 | 2021-02-19T15:09:21 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 781,932,782 | 20,949 | alexey-milovidov | Obscure function. | 2021-02-19T08:55:34 | 2021-02-19T08:55:34 | {} | MEMBER |
ClickHouse | ClickHouse | 781,949,577 | 20,959 | filimonov | for database engine ordinary (pre 20.10)
```
table = '.inner.mv_name'
```
for database engine Atomic (since 20.10)
```
table = '.inner.UUID_OF_MV' /* you can find uuid in system.tables */
```
| 2021-02-19T09:23:51 | 2021-02-19T09:23:51 | {
"+1": 1
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 782,050,328 | 18,185 | alexey-milovidov | As I have difficulties understanding this code I need someone else to help with review.
@amosbird Maybe you can help? | 2021-02-19T12:40:26 | 2021-02-19T12:40:26 | {} | MEMBER |
ClickHouse | ClickHouse | 782,052,093 | 18,185 | amosbird | > As I have difficulties understanding this code I need someone else to help with review.
> @amosbird Maybe you can help?
my pleasure. | 2021-02-19T12:44:14 | 2021-02-19T12:44:14 | {
"+1": 1
} | COLLABORATOR |
ClickHouse | ClickHouse | 782,069,848 | 20,929 | CurtizJ | Already in 20.8 | 2021-02-19T13:20:37 | 2021-02-19T13:20:37 | {} | MEMBER |
ClickHouse | ClickHouse | 782,092,930 | 20,965 | alexey-milovidov | It's not an error.
1. `Address family for hostname not supported` for IPv6 address means that IPv6 is disabled on your server.
The server will print warning and continue to work if it is able to listen at least one of the addresses.
No actions required. | 2021-02-19T14:00:24 | 2021-02-19T14:02:28 | {
"+1": 2
} | MEMBER |
ClickHouse | ClickHouse | 771,709,958 | 19,805 | vitlibar | 1. [test_materialize_mysql_database/test.py::test_select_without_columns_8_0](https://clickhouse-test-reports.s3.yandex.net/19805/cbd4bace6f265561902455aa18a9a894c43e6acc/integration_tests_(release).html#fail1) will be fixed in https://github.com/ClickHouse/ClickHouse/pull/19971
2. [AST fuzzer (UBSan)](https://click... | 2021-02-02T15:20:42 | 2021-02-18T13:46:07 | {} | MEMBER |
ClickHouse | ClickHouse | 781,359,959 | 20,867 | den-crane | You defined table as
```
ReplicatedReplacingMergeTree('/clickhouse/tables/{layer}-{shard}/tmp.rep_repl', '{replica}', timestamp)
```
1. This means that `timestamp` column will be used as VER column. It will be compared during replacing to make decision witch row to remain.
```
ORDER BY (account_number, timest... | 2021-02-18T13:55:30 | 2021-02-18T13:56:40 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 781,450,293 | 20,585 | alesapin | Reproduced on network HDD. | 2021-02-18T16:04:20 | 2021-02-18T16:04:20 | {} | MEMBER |
ClickHouse | ClickHouse | 781,520,164 | 17,623 | itzikiusa | Any plans of supporting INSERT IGNORE and REPLACE INTO
I understand mergetree tables are bit problematic for it, but i think it will be pretty helpfull | 2021-02-18T17:43:59 | 2021-02-18T17:43:59 | {} | NONE |
ClickHouse | ClickHouse | 781,933,300 | 20,942 | alexey-milovidov | Wrong PR. | 2021-02-19T08:56:31 | 2021-02-19T08:56:31 | {} | MEMBER |
ClickHouse | ClickHouse | 782,198,798 | 20,971 | alexey-milovidov | Part is removed if its level is zero and if sanity threshold is passed. | 2021-02-19T16:51:18 | 2021-02-19T16:52:02 | {} | MEMBER |
ClickHouse | ClickHouse | 781,030,405 | 20,608 | filimonov | Can you compare with 'baremetal' setup? I.e. install clickhouse on the same node directly w/o lxc and check if problem is still there. | 2021-02-18T03:54:14 | 2021-02-18T07:42:37 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 781,179,618 | 20,627 | alesapin | Reported https://github.com/ClickHouse/ClickHouse/issues/20870 | 2021-02-18T08:41:18 | 2021-02-18T08:41:18 | {} | MEMBER |
ClickHouse | ClickHouse | 781,252,455 | 20,712 | Avogar | Failed tests are not related. | 2021-02-18T10:43:12 | 2021-02-18T10:43:12 | {} | MEMBER |
ClickHouse | ClickHouse | 781,581,777 | 20,867 | iameugenejo | And I do understand what insert deduplication means for the replicating engines. I understand why it's needed, but I think ReplacingMergeTree engine itself performs the deduplication that serves the purpose of that as well, and it has an additional characteristic of keeping the last inserted row. But the additional cha... | 2021-02-18T19:28:13 | 2021-02-18T19:28:35 | {} | NONE |
ClickHouse | ClickHouse | 781,859,349 | 13,564 | chu1070y | Hi, do you have any updates on support for this issue? | 2021-02-19T06:26:30 | 2021-02-19T06:26:30 | {} | NONE |
ClickHouse | ClickHouse | 782,000,715 | 20,625 | kitaisreal | Hello, `argMax`, `argMin` function require two arguments, but if only one argument is passed and this argument is tuple of two elements special case is applied and first tuple columns used and first argument and second tuple column used as second argument.
```
CREATE TABLE test_argmax (user String, salary UInt64) ENG... | 2021-02-19T10:57:14 | 2021-02-19T10:57:14 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 782,039,691 | 20,392 | vdimir | For now rules fort `==` operator in `JOIN ON` more strict than in `WHERE` section. `JOIN ON` requires exact type match, but `WHERE` allows implicit type conversion. So, some queries may do not work after rewrite. But after changes from PR #19885 would be merged we can move expression safely. | 2021-02-19T12:17:00 | 2021-02-19T12:17:00 | {} | MEMBER |
ClickHouse | ClickHouse | 782,076,160 | 20,864 | alesapin | > > Fasttest doesn't use an updated image from PR
>
> @alesapin maybe you can update it manually?
Unfortunately, it's not easy to do. Let me add gdb in fasttest image latest version and after that, your script should work well. | 2021-02-19T13:29:57 | 2021-02-19T13:29:57 | {} | MEMBER |
ClickHouse | ClickHouse | 782,085,693 | 20,864 | alesapin | Done, let's wait until it will be pushed https://github.com/ClickHouse/ClickHouse/commit/ea27c3ca32bdf9a18e90d75bf38bbc725c6db4db. | 2021-02-19T13:47:37 | 2021-02-19T13:47:37 | {
"+1": 1
} | MEMBER |
ClickHouse | ClickHouse | 781,289,104 | 20,176 | amosbird | What happens if you do the following:
```
SELECT dictGetString('geoip_asn_blocks_ipv4', 'autonomous_system_organization', tuple(toUInt32(value)))
FROM fdns
``` | 2021-02-18T11:50:33 | 2021-02-18T11:50:33 | {
"+1": 1,
"hooray": 1
} | COLLABORATOR |
ClickHouse | ClickHouse | 781,318,136 | 20,306 | alesapin | Looks ok, but ALTER documentation outdated:
```
The `ALTER` query blocks all reads and writes for the table. ...
``` | 2021-02-18T12:46:25 | 2021-02-18T12:46:25 | {} | MEMBER |
ClickHouse | ClickHouse | 781,319,341 | 20,254 | amosbird | I have a plan to provide a different mode to execute distributed queries. It will work as follows:
1. Introduce a new bool setting called `execute_query_on_common_cluster` to enable this mode.
2. Traverse the ast and collect the common cluster name of encountered distributed tables. Throw if there is a difference.
... | 2021-02-18T12:48:55 | 2021-02-18T12:48:55 | {} | COLLABORATOR |
ClickHouse | ClickHouse | 781,350,797 | 20,877 | den-crane | You cannot intervene in a table on a directory level. I
t will break a table because a list of directories with parts is stored in CH-server memory.
Moreover your last insert can be merged right after insert.
You need to use atomic inserts. Use search in this repo to get the ideas how to insert atomically. | 2021-02-18T13:40:22 | 2021-02-18T13:40:22 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 781,574,684 | 20,867 | den-crane | >I'll make sure to use this flag for insert to any ReplicatedReplacingMergeTree.
You should not do this.
In real life in a production system `insert_deduplicate` does not intervene because REAL inserts do not repeat naturally.
> (Maybe `insert_deduplicate=0` should be default for `ReplicatedReplacingMergeTree... | 2021-02-18T19:15:25 | 2021-02-18T19:15:25 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 781,917,342 | 20,955 | filimonov | Also those 2 issues are often lead to confusions.
* https://github.com/ClickHouse/ClickHouse/issues/20957
* https://github.com/ClickHouse/ClickHouse/issues/17966
And in general - refreshing docs about copier is needed, to describe how it work. | 2021-02-19T08:24:55 | 2021-02-19T08:24:55 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 782,047,840 | 19,516 | alexey-milovidov | Ok.
1. Finally we can increase the 16KB limit on URL size. We can make it in next PR after merge.
2. The code in HTTP directory is based on Poco. We should put README.md file for attribution.
3. There can be dead code (unused methods) there (like sendFile). We can review and remove these methods.
4. `while (in.re... | 2021-02-19T12:35:01 | 2021-02-19T12:36:25 | {} | MEMBER |
ClickHouse | ClickHouse | 782,072,925 | 19,580 | alesapin | > commit [57c9b6c](https://github.com/ClickHouse/ClickHouse/commit/57c9b6c864d10909e74cd02e9ba4c90cd4487339) belongs to this pull request excludes darwin(macOS) from running ClickHouse with
>
> ```
> <Error> Application: DB::Exception: ClickHouse server built without NuRaft library. Cannot use internal coordination... | 2021-02-19T13:26:33 | 2021-02-19T13:26:33 | {} | MEMBER |
ClickHouse | ClickHouse | 781,169,575 | 20,864 | alesapin | > Actually docker image was updated, but it seems that fast test check does not use per-PR docker images
>
> ```
> $ docker run --rm -it yandex/clickhouse-fasttest:20864 which gdb
> /usr/bin/gdb
> ```
Yes, you right. Fasttest doesn't use an updated image from PR. Two reasons:
1) Fasttest should start as fast ... | 2021-02-18T08:35:32 | 2021-02-18T08:35:32 | {} | MEMBER |
ClickHouse | ClickHouse | 781,347,357 | 20,873 | den-crane | You need to use one of aggregateFunction (SimpleAggregateFunction or AggregateFunction)
Database engine in Materialized view table knows nothing about select in materialized view.
SummingMergeTree sums automatically. In your case it does not know that it should calc min/max.
https://den-crane.github.io/Everything_... | 2021-02-18T13:34:39 | 2021-02-18T13:36:10 | {
"+1": 9,
"heart": 3
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 781,577,836 | 20,867 | iameugenejo | @den-crane The use case for `ReplicatedReplacingMergeTree` I have is that I need to keep track of changes to this account information daily.
We have a separate source database that's not clickhouse for the account information,
and I will have a cron task that would insert the entire set of accounts a few times ... | 2021-02-18T19:21:01 | 2021-02-18T19:21:13 | {} | NONE |
ClickHouse | ClickHouse | 781,953,408 | 20,954 | filimonov | There is data for that table on default disk. And default disk is not a part of used policy.
You need either to move that data (manually) to one of disks which are the part of that policy, either add the default disk to that policy. | 2021-02-19T09:29:33 | 2021-02-19T09:29:33 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 782,131,840 | 20,864 | alesapin | New report https://clickhouse-test-reports.s3.yandex.net/20864/ee18f6a7ec23304c7ebc5128882d163d510525e0/fast_test.html | 2021-02-19T15:05:49 | 2021-02-19T15:05:49 | {} | MEMBER |
ClickHouse | ClickHouse | 782,144,243 | 11,305 | abyss7 | Closed in favor of #19516 | 2021-02-19T15:22:21 | 2021-02-19T15:22:21 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 781,114,715 | 20,710 | hexiaoting | one solution: sort these two map data in ascending order using key, then compare them one by one. Is this OK? | 2021-02-18T07:20:06 | 2021-02-18T07:20:06 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 781,982,022 | 20,254 | amosbird | > It seems there is a need for ruleset to identify "proper stage" correctly.
Only the outmost `group by` will trigger WithMergeableState. All subqueries only deal with local tables (proxy). | 2021-02-19T10:21:42 | 2021-02-19T10:21:53 | {} | COLLABORATOR |
ClickHouse | ClickHouse | 781,429,113 | 20,585 | alesapin | ```
2021-02-18 17:26:57 00992_system_parts_race_condition_zookeeper: [ FAIL ] 11.28 sec. - having stderror:
2021-02-18 17:26:57 [96825df77e83] 2021.02.18 17:26:57.042652 [ 113751 ] {1464e8b1-93ac-44b7-8a31-fcce5e5a34e4} <Error> zkutil::EphemeralNodeHolder::~EphemeralNodeHolder(): Cannot rem... | 2021-02-18T15:35:22 | 2021-02-18T15:35:22 | {} | MEMBER |
ClickHouse | ClickHouse | 781,452,677 | 20,867 | iameugenejo | I see, I should remove `timestamp` from the engine declaration.
@den-crane thank you for the clarification.
| 2021-02-18T16:07:52 | 2021-02-18T16:07:52 | {} | NONE |
ClickHouse | ClickHouse | 781,516,241 | 20,867 | den-crane | Can you provide a new example?
Try without `TTL timestamp + toIntervalDay(1)`. It could be that the recent rows has date older than 1 day and are deleted by TTL. | 2021-02-18T17:38:14 | 2021-02-18T17:38:14 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 781,947,352 | 19,580 | tisonkun | commit 57c9b6c864d10909e74cd02e9ba4c90cd4487339 belongs to this pull request excludes darwin(macOS) from running ClickHouse with
```
<Error> Application: DB::Exception: ClickHouse server built without NuRaft library. Cannot use internal coordination.
```
it is intended? If so, is there a plan to fix? | 2021-02-19T09:20:02 | 2021-02-19T09:20:02 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 782,103,927 | 17,966 | filimonov | Make it switchable maybe? | 2021-02-19T14:19:28 | 2021-02-19T14:19:28 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 773,254,686 | 20,064 | filimonov | Updated clickhouse, works correctly now. :+1:
Installed version had few days old :) Didn't expect it's too old already :) | 2021-02-04T11:58:54 | 2021-02-04T12:00:49 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 773,362,340 | 19,943 | den-crane | >So if I don't want to use the cache for testing disk io,
>can I use it --optimize_trivial_count_query=0 --min_bytes_to_use_direct_io=1
Yes.
>and why it is faster?
It's only for my test.
Because of my hardware + table table10kRows is too small. | 2021-02-04T14:44:36 | 2021-02-04T14:44:36 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 773,515,378 | 20,086 | simpl1g | It's a strange design, I would expect such behavior if mv2 was connected to kafka_queue directly. However in this case mv2 doesn't care how data is written to mergetree1, why should queue fail. Also in case of problem it is easier to write data from mergetree1 to mergetree2 from scratch, than to think how to deduplicat... | 2021-02-04T18:28:36 | 2021-02-04T18:28:36 | {} | CONTRIBUTOR |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.