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
517,976,851
6,313
slimtom95
Try `showduplicates` to see all available versions. For example: ``` yum search clickhouse-server --showduplicates ``` There is version 19.7.3.9 available, if it suits you.
2019-08-04T06:07:22
2019-08-04T06:07:22
{ "+1": 1 }
NONE
ClickHouse
ClickHouse
518,180,905
6,330
abyss7
It's a revert of broken merge. I will fix it in a proper way - along with tests.
2019-08-05T10:32:24
2019-08-05T10:32:24
{}
CONTRIBUTOR
ClickHouse
ClickHouse
518,203,201
6,028
alexey-milovidov
> And why does IsEvent have the same data_uncompressed_bytes as SocialAction? All SocialAction are empty strings.
2019-08-05T11:52:42
2019-08-05T11:52:42
{}
MEMBER
ClickHouse
ClickHouse
518,182,082
6,308
abyss7
This bug emerged with #6075 - it shows the breakage in tests but nevertheless it was merged. I will do a proper fix ASAP.
2019-08-05T10:36:31
2019-08-05T10:36:31
{}
CONTRIBUTOR
ClickHouse
ClickHouse
518,184,936
6,337
Tri0L
+1. Need this functionality.
2019-08-05T10:46:35
2019-08-05T10:46:35
{ "+1": 1 }
CONTRIBUTOR
ClickHouse
ClickHouse
518,224,242
6,335
filimonov
Do all servers sharing that zookeeper ddl path has that server configured?
2019-08-05T12:59:58
2019-08-05T12:59:58
{}
CONTRIBUTOR
ClickHouse
ClickHouse
518,071,686
6,314
yuzhichang
@alexey-milovidov Renamed to bitmapSubsetInRange per your suggestion.
2019-08-05T03:21:12
2019-08-06T02:11:17
{}
CONTRIBUTOR
ClickHouse
ClickHouse
518,001,812
5,042
MahmoudGoda0
Can you please share how i can disable optimize_read_in_order manually ?
2019-08-04T12:57:25
2019-08-04T12:57:25
{}
NONE
ClickHouse
ClickHouse
518,004,919
6,195
alexey-milovidov
@4ertus2 Now we have an issue with `operator delete[]`. It doesn't correctly untrack memory (you may see it in rare test runs that happens to be on AVX2 machines). And we get false report `Memory limit exceeded` while parsing JSONs.
2019-08-04T13:43:46
2019-08-04T13:44:11
{}
MEMBER
ClickHouse
ClickHouse
518,005,190
6,285
alexey-milovidov
Step 2: https://github.com/yandex/ClickHouse/pull/6320
2019-08-04T13:47:21
2019-08-04T13:47:21
{}
MEMBER
ClickHouse
ClickHouse
518,017,515
6,313
filimonov
1. Check available versions, and pick the version to install, ``` yum --showduplicates list clickhouse-server Loaded plugins: fastestmirror, langpacks, priorities Loading mirror speeds from cached hostfile * epel: ftp.icm.edu.pl * ius: lon.mirror.rackspace.com * nux-dextop: mirror.li.nux.ro 244 packages ...
2019-08-04T16:31:15
2019-08-04T16:32:06
{ "+1": 1 }
CONTRIBUTOR
ClickHouse
ClickHouse
518,089,087
6,289
filimonov
> I have some projects which need to be migrated to Docker deployment, and that might cause this ownership problem. Official docker image run clickhouse process in non-root mode
2019-08-05T05:17:16
2019-08-05T05:17:16
{}
CONTRIBUTOR
ClickHouse
ClickHouse
518,124,315
5,846
Rostish
@abyss7 Hi! Could you comment on the new bug? I did rollback to 19.11.3.11
2019-08-05T07:44:40
2019-08-05T07:44:40
{}
NONE
ClickHouse
ClickHouse
518,186,694
6,338
zhang2014
@blinkov
2019-08-05T10:52:48
2019-08-05T10:52:48
{}
CONTRIBUTOR
ClickHouse
ClickHouse
518,203,119
6,028
alexey-milovidov
> You mean data_uncompressed_bytes or the type size? type size. UInt8 takes the same size on disk than empty String but it is much cheaper to read.
2019-08-05T11:52:22
2019-08-05T11:52:22
{}
MEMBER
ClickHouse
ClickHouse
518,222,272
6,331
filimonov
Sounds reasonable, supported by most of popular DBMS (including MySQL, Postgres, MsSQL etc)
2019-08-05T12:54:26
2019-08-05T12:54:26
{}
CONTRIBUTOR
ClickHouse
ClickHouse
518,544,290
6,353
oflasch
Many thanks for the detailed solution, looks very good. I did not know about the effect of GROUP BY on memory usage. Neat optimization! Your observation that the series must be sorted anyway brought another possible solution to mind that I wanted to document for others perhaps facing the same task: We can simply so...
2019-08-06T07:28:27
2019-08-06T07:30:09
{}
NONE
ClickHouse
ClickHouse
517,927,999
6,308
ivanprostran
> This must be a noobish question but how to you downgrade clickhouse version safely from the production server without turning off the while cluster? Downtime! I turned off the cluster, backed up the data directory, metrika.xml and /etc/clickhouse-server. Another solution is rolling restart/upgrade/downgrade ...
2019-08-03T14:19:59
2019-08-03T14:19:59
{ "+1": 1 }
NONE
ClickHouse
ClickHouse
518,009,187
6,318
alexey-milovidov
Yes, it is very easy to implement with table settings. (we should also do it for StorageSet)
2019-08-04T14:41:57
2019-08-04T14:42:15
{}
MEMBER
ClickHouse
ClickHouse
518,213,396
6,334
4ertus2
Can you reproduce the same locally with a MergeTree engine?
2019-08-05T12:26:43
2019-08-05T12:27:06
{}
CONTRIBUTOR
ClickHouse
ClickHouse
518,225,706
6,335
astudnev
There are servers connected to ZK, that do not have this cluster written in remote_servers.xml But they do not use this cluster.
2019-08-05T13:04:15
2019-08-05T13:04:15
{}
CONTRIBUTOR
ClickHouse
ClickHouse
518,226,824
6,339
filimonov
Please check: https://github.com/yandex/ClickHouse/issues/5256 https://github.com/yandex/ClickHouse/issues/3526 https://github.com/yandex/ClickHouse/issues/3388
2019-08-05T13:07:48
2019-08-05T13:07:48
{}
CONTRIBUTOR
ClickHouse
ClickHouse
518,516,921
6,318
zombiemonkey
Thanks Alexey, Thinking about this more it would probably have to be Memory vs temporary as I would need to access the same data from multiple sessions (as well as perform parallel operations). Thanks again!
2019-08-06T06:03:39
2019-08-06T06:03:39
{}
NONE
ClickHouse
ClickHouse
518,556,333
5,859
fancyqlx
@4ertus2 Hi, I found there was a segmentation fault coredump after this fix. It can be reproduced as follows. create table test_join (date Date, id Int32, name Nullable(String)) engine = MergeTree partition by date order by id; insert into test_join values ('2019-01-01', 1, 'a'); insert into test_join values (...
2019-08-06T08:00:10
2019-08-06T08:00:10
{}
CONTRIBUTOR
ClickHouse
ClickHouse
518,181,120
6,332
amosbird
CI somehow gets confused when force pushing along with pr creation at the same time.
2019-08-05T10:33:10
2019-08-05T10:33:10
{}
COLLABORATOR
ClickHouse
ClickHouse
518,228,773
6,343
dimarub2000
Was thinking about > You can also implement some statistical tests for "is median (or mean) of one distribution is less than another" and show the results of these tests on the fly. Looks like `query_time_percentiles` provide enough information for comparison. But not sure.
2019-08-05T13:13:26
2019-08-05T13:14:18
{}
CONTRIBUTOR
ClickHouse
ClickHouse
518,340,924
6,348
alexey-milovidov
In what build configuration you've encountered this issue?
2019-08-05T18:12:49
2019-08-05T18:12:49
{}
MEMBER
ClickHouse
ClickHouse
518,349,854
3,776
alexey-milovidov
Yes, it is still relevant. > I think I can try to implement this feature. That will be very appreciated!
2019-08-05T18:38:12
2019-08-05T18:38:35
{}
MEMBER
ClickHouse
ClickHouse
518,501,702
6,353
zombiemonkey
You can use arrays to get a series per key, sort by timestamp then calculate pretty much anything on the series. Performance on array vectors is certainly not the same as column vectors however if you consider the fact you have to sort to get things in order you end up with performance that roughly matches the new opti...
2019-08-06T05:06:20
2019-08-06T05:06:20
{}
NONE
ClickHouse
ClickHouse
517,941,221
6,285
lemire
Let us know if there is anything we can do to help. I have added documentation to our README.
2019-08-03T17:24:50
2019-08-03T17:24:50
{}
NONE
ClickHouse
ClickHouse
518,001,324
5,042
MahmoudGoda0
OK, you mean if i tried after some time from the import time it should be faster ?
2019-08-04T12:50:26
2019-08-04T12:50:26
{}
NONE
ClickHouse
ClickHouse
518,002,411
5,042
alexey-milovidov
`SET optimize_read_in_order = 0` in clickhouse-client interactively or add `optimize_read_in_order=0` URL parameter in HTTP interface or add this setting to the user profile.
2019-08-04T13:05:37
2019-08-04T13:05:37
{}
MEMBER
ClickHouse
ClickHouse
518,014,770
6,285
lemire
You are right.
2019-08-04T15:54:30
2019-08-04T15:54:30
{}
NONE
ClickHouse
ClickHouse
518,019,155
6,313
luck143
Thankyou very much
2019-08-04T16:56:25
2019-08-04T16:56:25
{}
NONE
ClickHouse
ClickHouse
518,021,453
6,248
alexey-milovidov
@dimarub2000 Please check if the monotonic functions processing for index works correctly for this case.
2019-08-04T17:29:40
2019-08-04T17:29:40
{}
MEMBER
ClickHouse
ClickHouse
518,062,435
6,287
dkhodorchenko
![identity](https://user-images.githubusercontent.com/51990161/62434417-645ceb80-b762-11e9-9416-550efc9212ba.png) With identity it works
2019-08-05T02:22:14
2019-08-05T02:22:14
{}
NONE
ClickHouse
ClickHouse
518,304,206
6,149
4ertus2
Continuation #6347
2019-08-05T16:27:39
2019-08-05T16:27:39
{}
CONTRIBUTOR
ClickHouse
ClickHouse
518,306,706
5,846
AntonLiashenka
Got this bug on 19.10.1.5 as well
2019-08-05T16:35:03
2019-08-05T16:35:03
{}
NONE
ClickHouse
ClickHouse
518,585,547
6,335
astudnev
This functionality is very confusing and wrong For example, i want to have a separate cluster of servers that get all replicated data, but do not know about any other clusters. It means i still have to configure all clusters on that machines as well, even if they do NOT need it? Also it breaks incapsulation of my conf...
2019-08-06T09:16:49
2019-08-06T09:16:49
{}
CONTRIBUTOR
ClickHouse
ClickHouse
517,933,815
6,191
alexey-milovidov
I will look at query_profiler test.
2019-08-03T15:40:43
2019-08-03T15:40:43
{}
MEMBER
ClickHouse
ClickHouse
517,970,722
6,285
alexey-milovidov
https://github.com/ClickHouse-Extras/simdjson/commit/e3f6322af762213ff2087ce3366bf9541c7fd355 I can send a pull request if you are Ok with this solution.
2019-08-04T03:30:40
2019-08-04T03:30:40
{}
MEMBER
ClickHouse
ClickHouse
517,970,862
6,282
alexey-milovidov
Let's use @filimonov to prove that this code is well understandable. You can write comments explaining this code until @filimonov will say that it's clear.
2019-08-04T03:33:30
2019-08-04T03:33:30
{}
MEMBER
ClickHouse
ClickHouse
518,001,463
5,042
alexey-milovidov
We continue optimizing "in order" read: https://github.com/yandex/ClickHouse/pull/6299 This is in active development. But some rare queries will remain slightly slower with `optimize_read_in_order` enabled. They are the queries that has ORDER BY prefix of primary key, large LIMIT and WHERE condition that will requi...
2019-08-04T12:52:32
2019-08-04T12:52:57
{}
MEMBER
ClickHouse
ClickHouse
518,024,861
6,166
kreuzerkrieg
Calamity! It was more like 5 hours... I want to see if it passes CI and then I will additional changes, mostly to bring headers in line with CPPs https://github.com/yandex/ClickHouse/pull/6325
2019-08-04T18:20:38
2019-08-04T18:20:55
{}
CONTRIBUTOR
ClickHouse
ClickHouse
518,058,902
6,289
ddliu
> Currently it is not possible. When you run it as root it will finish with exception. See: [#3785 (comment)](https://github.com/yandex/ClickHouse/pull/3785#discussion_r240163579) > > Can i ask why do you need to run clickhouse as root? I have some projects which need to be migrated to Docker deployment, and that...
2019-08-05T01:57:45
2019-08-05T01:57:45
{}
NONE
ClickHouse
ClickHouse
518,169,791
6,332
alesapin
Merge with master seems like flap.
2019-08-05T09:58:52
2019-08-05T09:58:52
{}
MEMBER
ClickHouse
ClickHouse
518,176,098
6,205
zhang2014
@alexey-milovidov The tests was wrong. Can you help me see what happened? I can't open these links.
2019-08-05T10:16:30
2019-08-05T10:16:30
{}
CONTRIBUTOR
ClickHouse
ClickHouse
518,227,968
6,341
filimonov
You need to use url encoding (aka [Percent-encoding](https://en.wikipedia.org/wiki/Percent-encoding)) when sending queries in GET requests Please try: ``` http://ip:10024/?user=xx&password=xx&query=select%20event%2Cvalue%20from%20system.events%20where%20event%20%3D%20%27MergedRows%27 ```
2019-08-05T13:11:08
2019-08-05T13:11:08
{}
CONTRIBUTOR
ClickHouse
ClickHouse
517,971,322
6,194
alexey-milovidov
@yurriy You can switch off MySQL protocol support for "unbundled" build.
2019-08-04T03:46:32
2019-08-04T03:46:32
{}
MEMBER
ClickHouse
ClickHouse
517,979,571
6,127
Enmk
> I noticed that we get an empty array for point (45, 45, 45, 45). Is this expected? > > ``` > ┌────n─┬─geohashesInBox(n, n, n, n, 1)─┐ > │ 44.9 │ ['s'] │ > │ 45 │ [] │ > │ 45.1 │ ['v'] │ > └──────┴─────────────────────────────...
2019-08-04T07:02:54
2019-08-04T07:02:54
{}
CONTRIBUTOR
ClickHouse
ClickHouse
518,021,280
5,734
alexey-milovidov
@filimonov The trivial query `select * from xxx limit 1` will also work fine. ClickHouse will automatically lower number of threads (to one) and block size (also to one).
2019-08-04T17:27:11
2019-08-04T17:27:11
{}
MEMBER
ClickHouse
ClickHouse
518,025,095
6,295
alexey-milovidov
Superseded by https://github.com/yandex/ClickHouse/pull/6326
2019-08-04T18:23:46
2019-08-04T18:23:46
{}
MEMBER
ClickHouse
ClickHouse
518,208,524
6,342
filimonov
bad syntax :) should be either `where 1=0` either `CREATE TABLE test AS system.one ENGINE = ...`
2019-08-05T12:10:42
2019-08-05T12:12:54
{}
CONTRIBUTOR
ClickHouse
ClickHouse
518,215,411
6,246
banburybill
I suspect all the faults I am seeing have the same cause. I managed to reproduce #6245 with a small schema this morning - I just added a report to #6245. I don't have `topKWeighted`, but I do have `topK()` in `AggregatingMergeTree`.
2019-08-05T12:33:17
2019-08-05T12:33:58
{}
NONE
ClickHouse
ClickHouse
517,948,784
6,166
kreuzerkrieg
> I hit the :a(a) some time ago > > Let's replace all a(a) to a(a_). I appreciate if you can help! Will do
2019-08-03T19:26:52
2019-08-03T19:26:52
{}
CONTRIBUTOR
ClickHouse
ClickHouse
517,996,535
5,042
alexey-milovidov
@MahmoudGoda0 Do you mind converting `imsi` to UInt64?
2019-08-04T11:47:30
2019-08-04T11:47:30
{}
MEMBER
ClickHouse
ClickHouse
518,349,486
6,349
alexey-milovidov
Settings and ClientInfo issues look to be easy to fix. Do you see the possibility to make it?
2019-08-05T18:37:12
2019-08-05T18:37:12
{}
MEMBER
ClickHouse
ClickHouse
517,930,943
6,285
lemire
> Does it require to compile all translation units with -mavx2? No! Of course not. See simdjson/src/jsonparser.cpp for the dispatching logic. Of course, if you compile with -mavx2, then you will defeat the purpose of runtime dispatching. We recommend compiling with ' -msse4.2 -mpclmul' if you use GNU GCC, LL...
2019-08-03T15:01:43
2019-08-03T15:01:43
{}
NONE
ClickHouse
ClickHouse
517,935,074
6,166
alexey-milovidov
For clang we also have `-Wno-shadow-field-in-constructor`
2019-08-03T15:59:24
2019-08-03T15:59:24
{}
MEMBER
ClickHouse
ClickHouse
517,946,325
6,166
alexey-milovidov
> but what you gonna do with types declaration in Types.h? It is inside `enum class`: ``` using Float32 = ...; ... enum class TypeIndex { Float32 = ... }; ``` - looks perfectly Ok. But look at the first answer here: https://stackoverflow.com/questions/53841625/wshadow-global-considers-enum-class-...
2019-08-03T18:46:41
2019-08-03T18:46:41
{}
MEMBER
ClickHouse
ClickHouse
517,963,454
6,299
alexey-milovidov
TSan report looks relevant.
2019-08-04T00:13:04
2019-08-04T00:13:04
{}
MEMBER
ClickHouse
ClickHouse
518,174,816
6,240
zhang2014
@akuzm Done.
2019-08-05T10:12:41
2019-08-05T10:12:41
{}
CONTRIBUTOR
ClickHouse
ClickHouse
518,200,652
6,340
oflasch
Never mind, this bug seems to fixed in the ClickHouse client version 19.11.2.7 (official build).
2019-08-05T11:43:22
2019-08-05T11:43:22
{}
NONE
ClickHouse
ClickHouse
518,225,428
6,211
filimonov
Related: #6337
2019-08-05T13:03:23
2019-08-05T13:03:23
{}
CONTRIBUTOR
ClickHouse
ClickHouse
518,580,090
6,358
filimonov
Related: https://github.com/yandex/ClickHouse/issues/3311 https://github.com/yandex/ClickHouse/issues/5527 https://github.com/yandex/ClickHouse/pull/6240
2019-08-06T09:01:41
2019-08-06T09:01:41
{}
CONTRIBUTOR
ClickHouse
ClickHouse
517,938,903
6,285
alexey-milovidov
Another example: https://github.com/yandex/ClickHouse/pull/6141
2019-08-03T16:51:12
2019-08-03T16:51:12
{}
MEMBER
ClickHouse
ClickHouse
517,996,665
5,042
MahmoudGoda0
I cant as I have values ends with "b" Like this one "543036746630589b"
2019-08-04T11:49:24
2019-08-04T11:49:58
{}
NONE
ClickHouse
ClickHouse
518,004,775
6,191
alexey-milovidov
query_profiler test has been fixed.
2019-08-04T13:41:53
2019-08-04T13:41:53
{}
MEMBER
ClickHouse
ClickHouse
518,068,150
6,289
ddliu
Seems ok to just `chown` the top level directory. https://github.com/yandex/ClickHouse/blob/8dfe294fc4da29deeee0a46ddce381faabce8c65/dbms/programs/server/Server.cpp#L238
2019-08-05T02:58:45
2019-08-05T02:58:45
{}
NONE
ClickHouse
ClickHouse
518,181,863
5,846
abyss7
The new bug emerged with #6075 - it shows the breakage in tests but nevertheless it was merged. I will do a proper fix ASAP.
2019-08-05T10:35:46
2019-08-05T10:35:46
{}
CONTRIBUTOR
ClickHouse
ClickHouse
517,934,604
6,285
alexey-milovidov
Ok. Another question: is it tested with TSan (or is there something that may affect it in newer version)? We have one TSan report for pull request with upgrading but unfortunately without enough context. PS. Maybe dynamic dispatch use non atomic variable? You can switch to atomic variable with memory_order_relaxed...
2019-08-03T15:52:21
2019-08-03T15:52:21
{}
MEMBER
ClickHouse
ClickHouse
517,935,126
6,166
alexey-milovidov
But we can rewrite all the code and enable it. It's not hard (a few hours work).
2019-08-03T16:00:12
2019-08-03T16:00:12
{}
MEMBER
ClickHouse
ClickHouse
517,942,225
6,166
kreuzerkrieg
I hit the `:a(a)` some time ago https://stackoverflow.com/questions/48888177/shadowing-and-gcc-wshadow and the explanation why it is not valid code is quite simple I didnt find the `no-shadow-field-in-constructor` equivalent for GCC, so no luck here I can change the filed initialization myself, not too complica...
2019-08-03T17:40:37
2019-08-03T17:40:37
{}
CONTRIBUTOR
ClickHouse
ClickHouse
518,001,080
5,042
alexey-milovidov
Reading "in order" that is implemented in this task, can be slower than usual read, due to: - merging of sorted data; - additional thread synchronization (to read sorted data in parallel with O(1) memory and maintain the order, the threads need to wait while another thread has finished reading some previous chunk of ...
2019-08-04T12:47:06
2019-08-04T12:47:06
{}
MEMBER
ClickHouse
ClickHouse
518,001,564
5,042
alexey-milovidov
> OK, you mean if i tried after some time from the import time it should be faster ? Queries may become slightly faster due to less number of data parts, but not very significantly, because expensive merging step is still required.
2019-08-04T12:54:13
2019-08-04T12:54:13
{}
MEMBER
ClickHouse
ClickHouse
518,029,323
6,314
alexey-milovidov
Ok. But we need to choose appropriate name to avoid confusion. As far as I understand, this function will filter bitmap by selecting values only in specified range. But the name can be confused with existing `range` function that is used to generate a range of values. Maybe we can name it `bitmapSubsetInRange` or `...
2019-08-04T19:23:23
2019-08-04T19:23:41
{}
MEMBER
ClickHouse
ClickHouse
518,039,747
6,324
alexey-milovidov
Test failures look relevant.
2019-08-04T22:00:16
2019-08-04T22:00:16
{}
MEMBER
ClickHouse
ClickHouse
518,120,761
6,028
amosbird
> select a column with minimum size of value in memory: You mean `data_uncompressed_bytes` or the type size? And why does `IsEvent` have the same `data_uncompressed_bytes` as `SocialAction`?
2019-08-05T07:32:40
2019-08-05T07:32:40
{}
COLLABORATOR
ClickHouse
ClickHouse
518,166,253
5,042
MahmoudGoda0
@alexey-milovidov By using optimize_read_in_order=0, the time now = 22 second. I believe it should be smaller than this time, Agree ?
2019-08-05T09:49:10
2019-08-05T09:49:10
{}
NONE
ClickHouse
ClickHouse
518,211,475
6,246
CurtizJ
I found that segfault is appeared in SpaceSaving class, which is used only in topK and topKWeighted aggregate functions. And I could reproduce it with query `select topKWeighted(10)(toString(number), number) from numbers(10000000)`. So, I guess that you have column with `topKWeighted` function in `AggregatingMergeTree`...
2019-08-05T12:20:37
2019-08-05T12:24:07
{ "+1": 1 }
MEMBER
ClickHouse
ClickHouse
518,224,957
6,336
filimonov
There are several important bugs in Kafka engine currently. :\
2019-08-05T13:01:58
2019-08-05T13:01:58
{}
CONTRIBUTOR
ClickHouse
ClickHouse
517,938,816
6,285
alexey-milovidov
That's right and it is very easy to fix.
2019-08-03T16:49:44
2019-08-03T16:49:44
{}
MEMBER
ClickHouse
ClickHouse
517,995,596
5,042
MahmoudGoda0
@den-crane I think this is not the case, becasue the same query but the where is where start_time > 971128806382 and start_time <971222387788 and imsi = '938036746436052' the time = 8 second, and this is very good. But by adding more condition in the where part to be where start_time > 971128806382 and start_tim...
2019-08-04T11:31:54
2019-08-04T11:31:54
{}
NONE
ClickHouse
ClickHouse
518,000,638
5,042
alexey-milovidov
@MahmoudGoda0 Ok. @den-crane > In case of order by(time) database does millions (index->column) reads on search rows with imsi = '938036746436052' and skipping most of them. These reads are slower than full scan of imsi column. In ClickHouse, indexed read is never slower than full scan. (A query can be slower d...
2019-08-04T12:41:13
2019-08-04T12:41:56
{}
MEMBER
ClickHouse
ClickHouse
518,008,337
6,285
lemire
Please do open a pull request upstream. I'll create an issue upstream to match. Regarding the proposed solution, the pointer can be assigned only in one function (json_parse_dispatch). Instead of changing the pointer type to an atomic type, can't we just create an exclusive lock in json_parse_dispatch with a test ea...
2019-08-04T14:31:24
2019-08-04T14:31:24
{}
NONE
ClickHouse
ClickHouse
518,008,954
6,285
alexey-milovidov
We can serialize `json_parse_dispatch` with lock (or with function static variable). This will prevent concurrent execution of CPU detection (though it's not a bad thing). But if the variable `json_parse_ptr` is assigned under lock but read without a lock, it doesn't make sense (TSan will report data race).
2019-08-04T14:38:40
2019-08-04T14:38:40
{}
MEMBER
ClickHouse
ClickHouse
518,123,063
6,329
vdimir
I made one excess commit and decide to squash it to one to keep history clean. But changes have been already pushed and it requires force-push. First commit that you reviewed unchanged. I hope it is not a big problem.
2019-08-05T07:40:46
2019-08-05T08:55:53
{}
MEMBER
ClickHouse
ClickHouse
518,248,247
5,565
4ertus2
There's less confusing error message in master now.
2019-08-05T14:05:20
2019-08-05T14:05:20
{}
CONTRIBUTOR
ClickHouse
ClickHouse
518,370,531
5,126
4ertus2
Related #6351
2019-08-05T19:38:40
2019-08-05T19:38:40
{}
CONTRIBUTOR
ClickHouse
ClickHouse
518,019,872
6,103
nikitamikhaylov
If the code is really awful, maybe I should rewrite It with separate complex singleton?
2019-08-04T17:06:29
2019-08-04T17:06:29
{}
MEMBER
ClickHouse
ClickHouse
518,208,517
6,245
banburybill
I have managed to reproduce this problem with a minimal schema. Schema: ``` CREATE TABLE RawDataShard ( Date Date, DateTime DateTime, Name String ) ENGINE = MergeTree() ORDER BY (Date, DateTime); CREATE TABLE RawData ( Date Date, DateTime DateTime, Name String ) ENGINE = Dis...
2019-08-05T12:10:42
2019-08-05T12:10:42
{}
NONE
ClickHouse
ClickHouse
517,946,427
6,166
alexey-milovidov
> I hit the :a(a) some time ago Let's replace all `a(a)` to `a(a_)`. I appreciate if you can help!
2019-08-03T18:48:20
2019-08-03T18:48:20
{}
MEMBER
ClickHouse
ClickHouse
518,582,804
6,335
filimonov
ON CLUSTER commands are distributed between the nodes which share same zookeeper path. See https://github.com/yandex/ClickHouse/blob/6685ccba651606c32d8f84dd6fed9895a1b4c6d6/dbms/programs/server/config.xml#L368 It's a bit obscure...
2019-08-06T09:09:07
2019-08-06T09:09:07
{}
CONTRIBUTOR
ClickHouse
ClickHouse
517,936,954
6,285
lemire
The simdjson library is single-threaded. The runtime dispatch assigns the right function the first time the function is called. This is not thread-safe: so the first function call must be single-threaded.
2019-08-03T16:25:09
2019-08-03T16:25:09
{}
NONE
ClickHouse
ClickHouse
518,001,000
5,042
MahmoudGoda0
@alexey-milovidov Thanks for your explanation, so in my case, what is your recommendation ? I'm really counting on your support, let me know if you need any other info from my side.
2019-08-04T12:46:01
2019-08-04T12:46:01
{}
NONE
ClickHouse
ClickHouse
518,169,997
6,330
alesapin
Any explanations?
2019-08-05T09:59:22
2019-08-05T09:59:22
{}
MEMBER
ClickHouse
ClickHouse
518,286,003
3,776
vdimir
Is it still relevant? I think I can try to implement this feature. Cc @alexey-milovidov
2019-08-05T15:38:41
2019-08-05T15:38:41
{}
MEMBER
ClickHouse
ClickHouse
518,352,099
5,042
alexey-milovidov
Sorry, now I'm out of context of your queries, and cannot be sure.
2019-08-05T18:44:37
2019-08-05T18:44:37
{}
MEMBER
ClickHouse
ClickHouse
518,483,233
5,710
zhang2014
@filimonov Can this achieve #6240 ? Maybe I should turn off #6240 If this can, because this is more generic : ) BTW: If so, how does it match the password? Do I need to configure the password in the configuration file ?
2019-08-06T03:29:51
2019-08-06T03:31:49
{}
CONTRIBUTOR
ClickHouse
ClickHouse
507,164,128
4,498
Vxider
> Your version of ClickHouse was built without support for Kafka. This may happen if you are using unofficial packages or built ClickHouse by yourself. I'm build from source with tag `v19.8.3.8-stable` and encountered the same problem. How can I build a version that supports kafka?
2019-07-01T08:15:24
2019-07-01T08:15:24
{}
CONTRIBUTOR