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 | 814,731,563 | 22,708 | alesapin | Keeper segfault in flaky check https://github.com/ClickHouse/ClickHouse/pull/22743 | 2021-04-07T08:51:05 | 2021-04-07T08:51:05 | {} | MEMBER |
ClickHouse | ClickHouse | 814,731,750 | 22,708 | alesapin | `01358_union_threads_bug` -- not related to changes. | 2021-04-07T08:51:21 | 2021-04-07T08:51:21 | {} | MEMBER |
ClickHouse | ClickHouse | 814,746,359 | 22,618 | kitaisreal | Other necessary code for test were not backported. | 2021-04-07T09:08:57 | 2021-04-07T09:08:57 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 814,845,854 | 22,745 | vdimir | We will allow more cases in `JOIN ON` section, considering expressions from unknown table as from left table.
Previously it was forbidden in https://github.com/ClickHouse/ClickHouse/pull/15162, this PR relax some rules.
E.g. `JOIN ON 1 = t2.x` will be allowed | 2021-04-07T11:41:45 | 2021-04-07T11:41:45 | {} | MEMBER |
ClickHouse | ClickHouse | 814,217,869 | 22,217 | gyuton | Internal documentation ticket: DOCSUP-8324. | 2021-04-06T15:34:50 | 2021-04-06T15:34:50 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 814,688,323 | 13,594 | Essun013 | 大佬解决了吗？为啥关闭了？我同样碰到了这个问题。。。Who can tell me How to solve this problem? | 2021-04-07T07:49:34 | 2021-04-07T07:49:34 | {} | NONE |
ClickHouse | ClickHouse | 814,708,702 | 22,696 | filimonov | Maybe some issue with byte order (initialization vector / and arithmetic)? From what I know java is big-endian, while x86 is little-endian. | 2021-04-07T08:18:34 | 2021-04-07T08:18:55 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 814,747,938 | 22,743 | alesapin | Yarr
```
WARNING: ThreadSanitizer: lock-order-inversion (potential deadlock) (pid=52727)
Cycle in lock order graph: M3412 (0x7b4400002c68) => M3497 (0x7b6... | 2021-04-07T09:10:56 | 2021-04-07T09:10:56 | {} | MEMBER |
ClickHouse | ClickHouse | 814,910,702 | 22,518 | abyss7 | Yandex build looks like a fluke | 2021-04-07T13:22:57 | 2021-04-07T13:22:57 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 814,237,129 | 22,711 | stephane-moreau | As a side note if I run the query that timed out manually in the client It get the expected 1 output within a few seconds | 2021-04-06T15:59:26 | 2021-04-06T15:59:26 | {} | NONE |
ClickHouse | ClickHouse | 814,729,692 | 22,325 | vdimir | > is it really CI problem?
yes | 2021-04-07T08:48:28 | 2021-04-07T08:48:28 | {} | MEMBER |
ClickHouse | ClickHouse | 814,796,565 | 22,696 | EwainCai | Does clickhouse make special encodings for non-English characters? | 2021-04-07T10:22:15 | 2021-04-07T10:22:15 | {} | NONE |
ClickHouse | ClickHouse | 814,884,250 | 21,781 | bun4uk | @alexey-milovidov from which CH version these fields exist?
We have 20.10 and I don't see them | 2021-04-07T12:45:42 | 2021-04-07T12:45:42 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 814,376,984 | 21,943 | azat | >I'm also in doubt - isn't it an overkill?
Rewritten using `SHOW TABLES FROM system` | 2021-04-06T19:18:37 | 2021-04-06T19:18:37 | {} | MEMBER |
ClickHouse | ClickHouse | 814,373,435 | 22,321 | ceedok | This is how i count cumulative LTV with help of window function. And do PARTITION BY source_id (source of traffic) - it works fine.
SELECT
cohort,
month,
CAST(sum(revenue) AS Float64) AS total_revenue,
uniqExact(uid) AS total_users,
total_revenue / total_users AS ltv,
CAST(MAX(cum_revenue) AS Float64) AS cum_r... | 2021-04-06T19:12:29 | 2021-04-06T19:26:49 | {} | NONE |
ClickHouse | ClickHouse | 814,701,182 | 22,696 | alexey-milovidov | I'm not a Java developer, maybe someone else can help you. | 2021-04-07T08:08:04 | 2021-04-07T08:08:04 | {} | MEMBER |
ClickHouse | ClickHouse | 814,848,691 | 22,325 | azat | >Why wasn't it backported to 21.3 LTS?
In my opinion it is not that significant for backporting (since it just do one extra conversion in some cases), however since 21.3 is LTS maybe it worth doing, @vdimir thoughts? | 2021-04-07T11:47:06 | 2021-04-07T11:47:06 | {} | MEMBER |
ClickHouse | ClickHouse | 814,936,116 | 19,766 | vdimir | @achimbab
Second part of https://github.com/ClickHouse/ClickHouse/pull/19766#issuecomment-814019342
> Is it true that in most cases base_cond and first event will be same? Also order of arguments looks confusing, because we have base_cond between timestamp_column and event_column. Maybe (timestamp, event_column,... | 2021-04-07T13:56:49 | 2021-04-07T13:56:49 | {} | MEMBER |
ClickHouse | ClickHouse | 814,746,249 | 22,605 | kitaisreal | Other necessary code for test were not backported. | 2021-04-07T09:08:48 | 2021-04-07T09:08:48 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 814,431,881 | 22,707 | alesapin | A failed test is not related to changes, all sessions were stable since creation:
```
$ zgrep 'KeeperTCPHandler' clickhouse-server.log.gz
2021.04.06 21:17:37.608463 [ 193 ] {} <Trace> KeeperTCPHandlerFactory: Keeper request. Address: [::1]:54028
2021.04.06 21:17:37.608649 [ 193 ] {} <Information> KeeperTCPHandle... | 2021-04-06T20:50:44 | 2021-04-06T20:51:20 | {} | MEMBER |
ClickHouse | ClickHouse | 814,669,452 | 22,514 | alesapin | Yes, no related changes in performance. | 2021-04-07T07:19:39 | 2021-04-07T07:19:39 | {} | MEMBER |
ClickHouse | ClickHouse | 808,728,746 | 14,298 | mastertheknife | If it helps, i recently decided to share the RPMs i created for use at my workplace.
We're a 95% CentOS company, and i didn't like the deb->rpm approach, i preferred native RPMs, so i made the effort of getting it done. I saw that @filimonov started working on it here: https://github.com/ClickHouse/ClickHouse/pull/712... | 2021-03-27T12:56:02 | 2021-04-07T09:23:30 | {
"heart": 1
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 814,847,147 | 22,325 | millin | Why wasn't it backported to 21.3 LTS? | 2021-04-07T11:44:14 | 2021-04-07T11:44:14 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 815,036,332 | 21,781 | Slach | @bun4uk according to git blame
https://github.com/ClickHouse/ClickHouse/commit/272311f6bdccbc9832fd63b7f230136010d2547a
available from 20.13 | 2021-04-07T16:07:03 | 2021-04-07T16:07:03 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 814,419,344 | 22,321 | akuzm | A more precise proof that all users are seen in the month 0:
```
SELECT
distance_month,
length(users_cumulative) AS cumulative_users,
length(arrayIntersect(users_cumulative, users_prev)) AS already_seen_users
FROM
(
SELECT
distance_month,
users_cumulative,
lagInFrame(... | 2021-04-06T20:31:21 | 2021-04-06T20:31:21 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 814,438,451 | 22,414 | oandrew | I think it would be useful to be able to specify which fields should be exported as avro string. Maybe wildcard/regex pattern? Then you could use something like avro_string_column_pattern=* to set it for all fields if needed.
It was originally avro string when I implemented it but as @alexey-milovidov noted we have n... | 2021-04-06T21:02:47 | 2021-04-06T21:02:47 | {
"+1": 1
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 814,487,747 | 21,814 | den-crane | It depends on python driver. Each driver has own unique timeout settings. Check with a driver's vendor. | 2021-04-06T23:00:08 | 2021-04-06T23:00:08 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 814,591,785 | 14,113 | alexey-milovidov | ```
<interserver_http_credentials>
<user>admin</user>
<password>333</password>
<users>
<admin>222</admin>
</users>
</interserver_http_credentials>
```
This example has inconsistency: the user name is
- in the content of `<user>` tag;
- in tag name: `<admin>`.
Writing user names i... | 2021-04-07T04:31:20 | 2021-04-07T04:31:20 | {
"eyes": 1
} | MEMBER |
ClickHouse | ClickHouse | 814,789,174 | 22,399 | vdimir | Query without alias for ARRAY JOIN works. I'm working on fix for aliased version too.
```
SELECT
id,
name,
sum(m1)
FROM f
LEFT ARRAY JOIN d_ids
LEFT JOIN d ON d.id = d_ids
GROUP BY
id,
name
``` | 2021-04-07T10:10:01 | 2021-04-07T10:10:01 | {} | MEMBER |
ClickHouse | ClickHouse | 814,791,458 | 21,888 | Avogar | > Is there no way to check if two rows are actually 'next' to each other according to ordering key?
In general, no way.
> If there was a way to not use parallel aggregation for one field, that would work.
We can disable parallel aggregation for certain aggregate function (I hope so). I will try do it for delt... | 2021-04-07T10:13:54 | 2021-04-07T12:48:14 | {} | MEMBER |
ClickHouse | ClickHouse | 814,892,065 | 22,299 | nikitamikhaylov | I tried to add a performance test, but selecting only one or several columns is extremely fast (and short queries should last at least 0.2s). So let's merge as is. BTW,
<img width="1698" alt="image" src="https://user-images.githubusercontent.com/25705399/113869867-e0ba9400-97b9-11eb-874d-a19d3255cdf1.png">
Seems lik... | 2021-04-07T12:57:51 | 2021-04-07T12:57:51 | {
"+1": 1,
"hooray": 1
} | MEMBER |
ClickHouse | ClickHouse | 814,904,978 | 22,754 | den-crane | Check result of
```
select i_action, count(*) as kol
from coem.events_month
where identity(i_action)='DownloadStart'
group by i_action
```
Can you share `show create table coem.events_month` ?
Upgrade at least to 20.9.7.11 | 2021-04-07T13:14:43 | 2021-04-07T13:15:55 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 814,911,120 | 22,742 | den-crane | You can limit only number of parallel fetches. Currently it is 3 (background_fetches_pool_size)
```
SELECT *
FROM system.merge_tree_settings
WHERE name LIKE '%replicated_max_parallel%'
Query id: f1e814d0-3fa6-472a-a463-a19ba84b74af
┌─name──────────────────────────────────────┬─value─┬─changed─┬─description─... | 2021-04-07T13:23:32 | 2021-04-07T13:23:32 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 814,923,211 | 22,696 | den-crane | ```sql
SELECT hex('æ')
┌─hex('æ')─┐
│ C3A6 │
└──────────┘
```
```java
public static String print(byte[] bytes) {
StringBuilder sb = new StringBuilder();
sb.append("[ ");
for (byte b : bytes) {
sb.append(String.format("0x%02X ", b));
}
sb.append(... | 2021-04-07T13:40:46 | 2021-04-07T13:43:50 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 814,941,799 | 22,754 | den-crane | What about the result with ` identity(i_action)`?
```
Check result of
select i_action, count(*) as kol
from coem.events_month
where identity(i_action)='DownloadStart'
group by i_action
``` | 2021-04-07T14:03:52 | 2021-04-07T14:04:07 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 814,960,176 | 22,754 | den-crane | Upgrade at least to 20.9.7.11 | 2021-04-07T14:27:13 | 2021-04-07T14:27:13 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 814,960,913 | 22,754 | zav379 | thanks | 2021-04-07T14:28:05 | 2021-04-07T14:28:05 | {} | NONE |
ClickHouse | ClickHouse | 814,203,354 | 22,706 | abyss7 | We put the work-in-progress pull-requests into draft state - for convenience. Feel free to move it back to regular PR when it's ready. | 2021-04-06T15:15:38 | 2021-04-06T15:15:38 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 814,615,957 | 21,639 | mishfish | @changvvb I've been trying, but got only
`CMake Error at cmake/arch.cmake:14 (message):
32bit platforms are not supported
Call Stack (most recent call first):
CMakeLists.txt:44 (include)` | 2021-04-07T05:36:52 | 2021-04-07T05:36:52 | {} | NONE |
ClickHouse | ClickHouse | 814,753,864 | 22,744 | filimonov | yep
```
Thread 22 "rdk:b/kafka" received signal SIGILL, Illegal instruction.
[Switching to Thread 0x7fffe9dfc000 (LWP 130373)]
0x0000000009ff936d in rand () at ../base/harmful/harmful.c:211
211 TRAP(rand)
(gdb) bt
#0 0x0000000009ff936d in rand () at ../base/harmful/harmful.c:211
#1 0x00000000135be573 in rd_j... | 2021-04-07T09:18:31 | 2021-04-07T09:18:31 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 814,894,510 | 22,221 | Blackmorse | @traceon, any updates? | 2021-04-07T13:01:41 | 2021-04-07T13:01:41 | {} | NONE |
ClickHouse | ClickHouse | 814,214,274 | 21,888 | rf | Is there no way to check if two rows are actually 'next' to eachother according to ordering key?
If there was a way to not use parallel aggregation for one field, that would work. I guess I could also have a second MV that only has `deltaSum` and has parallel agg disabled, if that was available at the table level.
... | 2021-04-06T15:30:08 | 2021-04-06T15:30:08 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 814,323,211 | 20,964 | EvgenyVinogradov | I have exactly same issue with 21.4.3.25. ZK is away from Clickhouse nodes.
Any suggestions? | 2021-04-06T17:59:04 | 2021-04-06T18:00:15 | {} | NONE |
ClickHouse | ClickHouse | 814,369,884 | 21,948 | azat | >Example benchmark with our HashTable
But I don't see where filling with preallocation is slower.
>If you can implement preallocation without breaking IDictionarySource abstraction, we can put it under some option.
Ok, thanks | 2021-04-06T19:06:28 | 2021-04-06T19:06:28 | {} | MEMBER |
ClickHouse | ClickHouse | 814,390,390 | 21,948 | kitaisreal | @azat
When we reserve with before inserting block (variant1) we get
Inserted in 53860 milliseconds
If we reserve whole size before all insertions (variant2) we get
Inserted in 105929 milliseconds
Without any resize it means hash table will resize itself where load_factor will be 0.5 we get
Inserted in 12424... | 2021-04-06T19:40:45 | 2021-04-06T19:40:45 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 814,432,062 | 22,707 | alesapin | Jepsen also looks ok https://clickhouse-test-reports.s3.yandex.net/22707/80d441ec9536dd93b3db3d03434dfb684f3d45a2/clickhouse_keeper_jepsen.html | 2021-04-06T20:51:05 | 2021-04-06T20:51:05 | {} | MEMBER |
ClickHouse | ClickHouse | 814,471,051 | 22,728 | vzakaznikov | Similar issue with `rankCorr`
```
Apr 06,2021 18:12:07 ⟥ Example func='rankCorr(salary, 0.5)'
Checking aggregate funcs over rows frame.
Arguments
func
... | 2021-04-06T22:14:42 | 2021-04-06T22:14:42 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 814,615,023 | 22,696 | EwainCai | @alexey-milovidov I have been troubled for a long time, could you please help me solve it? | 2021-04-07T05:34:31 | 2021-04-07T05:34:31 | {} | NONE |
ClickHouse | ClickHouse | 814,642,819 | 22,730 | filimonov | > And also it should be possible to publish custom source code tarball on GitHub Releases page
That is a nice practice in general and requirement for reproducable builds. | 2021-04-07T06:33:14 | 2021-04-07T06:33:30 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 814,746,404 | 22,604 | kitaisreal | Other necessary code for test were not backported. | 2021-04-07T09:09:01 | 2021-04-07T09:09:01 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 814,906,917 | 22,760 | cfroystad | See [issue#20205](https://github.com/ClickHouse/ClickHouse/issues/20205) | 2021-04-07T13:17:25 | 2021-04-07T13:17:25 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 814,951,468 | 22,754 | zav379 | with identity it's ok
result:
DownloadStart 47683
what's reason? | 2021-04-07T14:16:29 | 2021-04-07T14:16:29 | {} | NONE |
ClickHouse | ClickHouse | 815,023,425 | 19,766 | ghost | I fixed source codes https://github.com/ClickHouse/ClickHouse/pull/19766/commits/d13d69e246ec5d2fe26f80cb67fd8afdd27c77d2
- Add allow_experimental_funnel_functions
- Modify the order of arguments | 2021-04-07T15:49:47 | 2021-04-07T15:50:00 | {
"+1": 1
} | NONE |
ClickHouse | ClickHouse | 814,263,685 | 18,695 | olegkv | Just my 2 cents:
It seems much easier to understand and likely, implement such an idea, if partition expressions just put limits on number of partitions of each type.
So, instead of:
alter table x add partition expr hourly toStartOfHour(dt) where toDate(dt) >= today() - 3 days
this seems much easier:
create table ... | 2021-04-06T16:36:13 | 2021-04-06T16:36:13 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 814,422,045 | 22,321 | akuzm | Or even simper:
```
SELECT
first_month,
count(*) AS users
FROM
(
SELECT min(distance_month) AS first_month
FROM table
GROUP BY uid
)
GROUP BY first_month
ORDER BY users DESC
┌─first_month─┬──users─┐
│ 0 │ 380121 │
└─────────────┴────────┘
``` | 2021-04-06T20:36:08 | 2021-04-06T20:36:08 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 814,447,305 | 21,948 | kitaisreal | @azat I rebenchmarked with O3 and updated code, it seems that reserve whole size can make sense ~20% on my machine, and we can make that work with modification of IDictionarySource interface
```
virtual BlockInputStreamPtr loadAll(size_t result_size_hint * = nullptr) = 0;
virtual BlockInputStreamPtr loadUpdatedAll(s... | 2021-04-06T21:20:29 | 2021-04-06T21:20:50 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 814,452,187 | 22,414 | ilejn | I like the idea.
Although this way seems slightly less elegant than _avro_type_for_string_, it is more useful from practical standpoint. | 2021-04-06T21:31:10 | 2021-04-06T21:31:10 | {
"+1": 1
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 814,582,618 | 22,730 | alexey-milovidov | Source code tarballs are generated automatically by GitHub:
https://github.com/ClickHouse/ClickHouse/releases/tag/v21.5.1.6463-testing
and GitHub does not include submodules for some reason.
Nevertheless, we don't pay any attention to these tarballs and don't use them.
I have found a ton of feature requests for... | 2021-04-07T04:00:25 | 2021-04-07T04:00:25 | {} | MEMBER |
ClickHouse | ClickHouse | 814,637,606 | 22,490 | Sallery-X | > Normally all replicas do merges
I know that! that's not my question | 2021-04-07T06:23:13 | 2021-04-07T06:23:13 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 814,141,228 | 22,696 | EwainCai | I tried to use UTF-8 encoding last week, but did not get the expected results. I think this is not a problem caused by the encoding. | 2021-04-06T13:55:52 | 2021-04-06T13:55:52 | {} | NONE |
ClickHouse | ClickHouse | 814,408,134 | 20,882 | azat | >@azat The test is flaky: https://clickhouse-test-reports.s3.yandex.net/22561/36b03b27401afcf0b0ee704c827a1f89d377e50e/functional_stateless_tests_(memory).html#fail1
@alexey-milovidov Not really, the problem is in settings for default user, anyway had been already fixed in 278ad443e4aeae800e960d83c811c3d7568efbc3 (b... | 2021-04-06T20:11:07 | 2021-04-06T20:11:07 | {} | MEMBER |
ClickHouse | ClickHouse | 814,141,922 | 22,696 | EwainCai | https://github.com/ClickHouse/clickhouse-jdbc/issues/610 | 2021-04-06T13:56:43 | 2021-04-07T08:13:42 | {} | NONE |
ClickHouse | ClickHouse | 814,800,880 | 22,743 | alesapin | Now fix is more correct, but required changes in NuRaft :( | 2021-04-07T10:29:21 | 2021-04-07T10:29:21 | {} | MEMBER |
ClickHouse | ClickHouse | 814,805,853 | 22,700 | alesapin | Need to remove `test_testkeeper_back_to_back` from this branch. | 2021-04-07T10:37:59 | 2021-04-07T10:37:59 | {} | MEMBER |
ClickHouse | ClickHouse | 814,938,319 | 22,754 | zav379 | CREATE TABLE coem.events_month
(
`hash_ud` UInt64,
`i_device` UInt64,
`i_platform` String,
`i_user_guid` UInt64,
`i_os_version` String,
`i_time` DateTime,
`i_time_zone` Int8,
`i_insert_date` DateTime,
`i_patch_group_id` String,
`i_id` UInt32,
`i_fb_id` UInt64,
... | 2021-04-07T13:59:35 | 2021-04-07T13:59:35 | {} | NONE |
ClickHouse | ClickHouse | 814,137,212 | 21,639 | changvvb | I think it's ready to review this PR. | 2021-04-06T13:51:24 | 2021-04-06T13:51:24 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 814,471,203 | 21,639 | mishfish | hi @changvvb
how do you build ClickHouse?
I try with
`cmake -DCMAKE_C_COMPILER=$(brew --prefix gcc)/bin/gcc-10 -DCMAKE_CXX_COMPILER=$(brew --prefix gcc)/bin/g++-10 -DCMAKE_BUILD_TYPE=RelWithDebInfo -DENABLE_JEMALLOC=OFF ..`
but it doesn't build
Have I use some toolchain for build? | 2021-04-06T22:15:02 | 2021-04-06T22:15:02 | {} | NONE |
ClickHouse | ClickHouse | 814,592,649 | 22,710 | alexey-milovidov | The behaviour is correct:
`SELECT 'N' > 0.0` is not allowed to run. | 2021-04-07T04:34:13 | 2021-04-07T04:34:13 | {} | MEMBER |
ClickHouse | ClickHouse | 814,641,625 | 22,490 | filimonov | If you explicitly forbid to some replicas to do merges - they will not do it, and task will stay in queue. It is not related to leaders etc. | 2021-04-07T06:30:55 | 2021-04-07T06:30:55 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 814,710,293 | 22,696 | alexey-milovidov | Please also ensure that UTF-8 encoded string does not have BOM. | 2021-04-07T08:20:45 | 2021-04-07T08:20:45 | {} | MEMBER |
ClickHouse | ClickHouse | 814,715,803 | 22,420 | lingnancfy | > lineorder_demo -- Distributed?
>
> Because stop merges is not applicable to Distributed Engine
I have understand it, stop merges is ok, This question is relevant to the old_part_time. | 2021-04-07T08:28:01 | 2021-04-07T08:28:01 | {} | NONE |
ClickHouse | ClickHouse | 814,881,352 | 22,325 | vdimir | > Why wasn't it backported to 21.3 LTS?
I also think that if it is not bug fix, doesn't affect to user, but removes overhead in some cases, so backporting is not necessary. | 2021-04-07T12:40:42 | 2021-04-07T12:40:48 | {} | MEMBER |
ClickHouse | ClickHouse | 814,914,818 | 22,420 | den-crane | SYSTEM STOP MERGES is not related to inactive parts | 2021-04-07T13:28:44 | 2021-04-07T13:28:44 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 814,945,365 | 22,697 | TCeason | The test is passed. Request review. | 2021-04-07T14:08:23 | 2021-04-07T14:08:23 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 814,081,883 | 22,707 | alesapin | From logs:
```
2021.04.04 14:46:23.148835 [ 210 ] RaftInstance: Receive a client_request message from 0 with LastLogIndex=0, LastLogTerm=0, EntriesLength=1, CommitIndex=0 and Term=0
2021.04.04 14:46:29.851160 [ 210 ] KeeperLogStore: Starting new changelog /var/lib/clickhouse/coordination/logs/changelog_100001_110000... | 2021-04-06T12:31:17 | 2021-04-06T12:33:46 | {} | MEMBER |
ClickHouse | ClickHouse | 813,921,324 | 21,888 | Avogar | @rf unfortunately, it seems that deltaSumTimestamp and deltaSum won't work properly in materialized view with AggregatingMergeTree because parallel aggregation is used. I think I can make deltaSumTimestamp work using my algorithm and using some other tricks that rely on some unreliable invariants, but it would be a dir... | 2021-04-06T08:09:03 | 2021-04-06T13:13:07 | {} | MEMBER |
ClickHouse | ClickHouse | 814,332,356 | 22,592 | azat | >Could you please merge with master?
Done | 2021-04-06T18:12:04 | 2021-04-06T18:12:04 | {} | MEMBER |
ClickHouse | ClickHouse | 814,894,799 | 21,302 | nikitamikhaylov | This PR is merged in neighboring PR. All commits are saved. | 2021-04-07T13:02:08 | 2021-04-07T13:02:08 | {
"+1": 1
} | MEMBER |
ClickHouse | ClickHouse | 814,944,502 | 22,763 | CurtizJ | This pr should fix Sqlancer check. | 2021-04-07T14:07:16 | 2021-04-07T14:07:16 | {} | MEMBER |
ClickHouse | ClickHouse | 814,198,407 | 21,912 | KochetovNicolai | @nikitamikhaylov you forgot to change description | 2021-04-06T15:09:04 | 2021-04-06T15:09:04 | {} | MEMBER |
ClickHouse | ClickHouse | 814,330,504 | 22,490 | filimonov | Normally all replicas do merges | 2021-04-06T18:09:05 | 2021-04-06T18:09:05 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 814,396,290 | 21,948 | azat | Oh, you mean that reserving by small blocks is faster?
But variant1 does not fill other cells AFAICS, only first block | 2021-04-06T19:51:12 | 2021-04-06T19:51:12 | {} | MEMBER |
ClickHouse | ClickHouse | 814,542,767 | 21,639 | changvvb | > how do you build ClickHouse?
>
Hi @michon470, you should use `clang` to build the project.
Try this
```sh
cmake .. -DCMAKE_C_COMPILER=`brew --prefix llvm`/bin/clang -DCMAKE_CXX_COMPILER=`brew --prefix llvm`/bin/clang++ -DCMAKE_PREFIX_PATH=`brew --prefix llvm` -DENABLE_JEMALLOC=OFF
```
| 2021-04-07T01:52:17 | 2021-04-07T01:52:17 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 814,642,738 | 22,490 | Sallery-X | > If you explicitly forbid to some replicas to do merges - they will not do it, and task will stay in queue. It is not related to leaders etc.
some replicas not do merges, but they will fetch, I mean the replica is down that is doing merge
| 2021-04-07T06:33:05 | 2021-04-07T06:33:05 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 814,809,510 | 22,688 | hexiaoting | In version 21.5.1.1, I try it, it seems run OK.
Can you run it on the fresh master, or give us a minimal example that we can reproduce this error. | 2021-04-07T10:44:47 | 2021-04-07T10:44:47 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 814,895,738 | 20,129 | nikitamikhaylov | Done here #22299 | 2021-04-07T13:03:33 | 2021-04-07T13:03:33 | {} | MEMBER |
ClickHouse | ClickHouse | 814,358,402 | 22,325 | azat | >[ClickHouse build check — 14/15 builds are OK](https://clickhouse-builds.s3.yandex.net/22325/27d4fbd13b7c6eeade74a84ab0813c2fd1962eaf/clickhouse_build_check/report.html)
`Build failed without build log. It's inner CI problem. Maybe something wrong with docker or host. Just wait for build restart.` looks unrelated, ... | 2021-04-06T18:47:17 | 2021-04-06T18:47:17 | {} | MEMBER |
ClickHouse | ClickHouse | 814,412,608 | 22,321 | akuzm | You have the maximum number of users in month 0, so it might be correct that the cumulative number of users doesn't change:
```
$clickhouse-local --structure "uid UInt64, device String, source_id UInt64, start_dt DateTime, start_ts DateTime, end_ts DateTime, revenue Decimal(38, 2), birthday DateTime, distance_month I... | 2021-04-06T20:19:11 | 2021-04-06T20:19:38 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 814,487,067 | 22,724 | den-crane | CH will allocate memory of the dictionary for each replica | 2021-04-06T22:58:07 | 2021-04-06T22:58:07 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 814,632,215 | 21,639 | changvvb | > @changvvb I've been trying, but got only
>
> `CMake Error at cmake/arch.cmake:14 (message): 32bit platforms are not supported Call Stack (most recent call first): CMakeLists.txt:44 (include)`
According to https://github.com/ClickHouse/ClickHouse/pull/21639/files#diff-d5d8fc115656969af2961dbe1b6bf7f8a722df4998b... | 2021-04-07T06:12:17 | 2021-04-07T06:12:17 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 811,520,182 | 9,589 | mastertheknife | > Thanks a lot, is there any way to get a Build ClickHouse binary for PowerPC without building it at the moment?
> I tried to follow the guide here from here #4132 , but it didnt work.
Hi, i now have unofficial RPMs and SRPMs for ppc64le here:
https://github.com/mastertheknife/clickhouse-rpms
https://repo.kfir.te... | 2021-03-31T23:01:38 | 2021-04-07T09:21:17 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 814,844,768 | 9,953 | vdimir | Was fixed by https://github.com/ClickHouse/ClickHouse/pull/9950 | 2021-04-07T11:39:50 | 2021-04-07T11:39:50 | {} | MEMBER |
ClickHouse | ClickHouse | 814,932,200 | 22,696 | den-crane | Though
```
String str3 = "aaabbbbbcccc";
System.out.println(str3.hashCode());
--
1478625409
┌─javaHash('aaabbbbbcccc')─┐
│ 1478625409 │
└──────────────────────────┘
String str3 = "hűtő";
System.out.println(str3.hashCode());
--
3456806
┌─javaHash('hűtő')─┐
│ -1374267284 │
└───... | 2021-04-07T13:52:03 | 2021-04-07T13:52:18 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 815,050,272 | 22,584 | olgarev | @kitaisreal Hi, will you review this PR, please? | 2021-04-07T16:26:07 | 2021-04-07T16:26:07 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 814,526,980 | 22,588 | foolchi | @CurtizJ Thanks for adding test case | 2021-04-07T01:00:33 | 2021-04-07T01:00:33 | {} | NONE |
ClickHouse | ClickHouse | 814,794,646 | 22,696 | EwainCai | I tried to call the method to remove the BOM, and the result did not change.
Regarding'little-endian', I used the corresponding encoding format.All the keyword results in java hash were affected, and they all became wrong data.The byte order does not seem to be the expected factor.
There are tens of millions of keywo... | 2021-04-07T10:19:05 | 2021-04-07T10:19:54 | {} | NONE |
ClickHouse | ClickHouse | 817,200,126 | 20,102 | alexey-milovidov | Ok. Let's wait until next try to reproduce.
Network issues should never lead to crash. | 2021-04-10T20:47:01 | 2021-04-10T20:47:01 | {} | MEMBER |
ClickHouse | ClickHouse | 817,254,856 | 21,850 | filimonov | @fastio please check https://clickhouse-test-reports.s3.yandex.net/21639/03a2ba315b6b75c9264a89f712adeb61db19ce3d/integration_tests_(release).html#fail1 | 2021-04-11T06:04:09 | 2021-04-11T06:04:09 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 817,574,216 | 22,945 | alesapin | I'll fix the broken test. | 2021-04-12T07:48:45 | 2021-04-12T07:48:45 | {
"+1": 1
} | MEMBER |
ClickHouse | ClickHouse | 817,713,994 | 21,511 | azat | >Ok. The only concern is that it will be very slow for long lists of values.
Right now there is `optimize_skip_unused_shards_limit` that will disable this optimization (so at least the server will not spend too much for one query on the initiator with this option).
>Maybe collect all of them into a column and pro... | 2021-04-12T10:59:47 | 2021-04-12T10:59:47 | {} | MEMBER |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.