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 | 416,212,260 | 2,963 | zhang2014 | > Also I have noticed that there is a little delay between inserting in a distributed table and the results appearing in it for selection. Is there a way to have no delay for testing purposes?
You can try the following two configurations:
- insert_distributed_sync --This type is bool
- insert_distributed_timeout ... | 2018-08-27T12:37:05 | 2018-08-27T12:39:05 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 416,481,207 | 2,934 | amosbird | `arrayEnumerateUniq` is not the right function to generate ranks, neither is `arrayEnumerate` (as you've noticed that it doesn't handle equal values which is just a row_number function). Traditional `RANK` window function returns `1,1,3` and `DENSE_RANK` returns `1,1,2` in your example. Suppose you really want a dense_... | 2018-08-28T07:33:37 | 2018-08-28T07:34:43 | {} | COLLABORATOR |
ClickHouse | ClickHouse | 417,505,883 | 2,427 | alexey-milovidov | @luc1ph3r will try to prepare this PR for merge in September. | 2018-08-31T00:00:23 | 2018-08-31T00:00:23 | {} | MEMBER |
ClickHouse | ClickHouse | 417,795,265 | 2,796 | kmatt | Similar issue. Attempting to shutdown the clickhouse-server service hangs, with last log entry:
<Trace> StorageKafka (logs_queue): Unsubscribing from assignments
| 2018-08-31T21:35:37 | 2018-08-31T21:35:37 | {} | NONE |
ClickHouse | ClickHouse | 417,827,862 | 2,808 | alexey-milovidov | Usages of aggregate states:
FunctionsMiscellaneous
FunctionsArray
GraphiteRollupSortedBlockInputStream
SummingSortedBlockInputStream
TotalsHavingBlockInputStream
were forgotten. | 2018-09-01T02:58:20 | 2018-09-01T02:58:20 | {} | MEMBER |
ClickHouse | ClickHouse | 417,874,342 | 3,008 | alexey-milovidov | `00411_long_accurate_number_comparison: [ FAIL ] - Timeout!`
In Sandbox. I retried it two times. | 2018-09-01T17:19:09 | 2018-09-01T17:19:09 | {} | MEMBER |
ClickHouse | ClickHouse | 417,883,325 | 3,010 | alexey-milovidov | I've added the info.
https://github.com/yandex/ClickHouse/pull/3016 | 2018-09-01T19:53:20 | 2018-09-01T19:53:20 | {} | MEMBER |
ClickHouse | ClickHouse | 415,649,701 | 2,929 | zhang2014 | All tests passed. need to turn off `enable_optimize_predicate_expression` by default ? | 2018-08-24T04:46:36 | 2018-08-24T04:47:22 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 415,894,222 | 2,946 | alexey-milovidov | But the tests have failed. | 2018-08-24T22:03:05 | 2018-08-24T22:03:05 | {} | MEMBER |
ClickHouse | ClickHouse | 416,187,784 | 2,577 | rpfennin | in my case the issue was haproxy making a tcp check. | 2018-08-27T10:45:42 | 2018-08-27T10:45:42 | {
"+1": 2
} | NONE |
ClickHouse | ClickHouse | 416,203,296 | 1,102 | henyihanwobushi | GOOD Suggestion!!!
Let the rename statement support multi renames, and support transaction for the renames, like mysql, will solve my problem. | 2018-08-27T11:58:48 | 2018-08-27T11:58:48 | {} | NONE |
ClickHouse | ClickHouse | 416,210,924 | 2,963 | faucct | Also I have noticed that there is a little delay between inserting in a distributed table and the results appearing in it for selection. Is there a way to have no delay for testing purposes? | 2018-08-27T12:31:33 | 2018-08-27T12:31:42 | {} | NONE |
ClickHouse | ClickHouse | 417,474,913 | 2,942 | alexey-milovidov | Do you have a time to do the required modifications? | 2018-08-30T21:36:14 | 2018-08-30T21:36:14 | {} | MEMBER |
ClickHouse | ClickHouse | 417,505,604 | 2,464 | alexey-milovidov | Outdated. | 2018-08-30T23:58:52 | 2018-08-30T23:58:52 | {} | MEMBER |
ClickHouse | ClickHouse | 417,577,575 | 2,974 | zhang2014 | Done and update https://github.com/yandex/ClickHouse/pull/2974#issuecomment-416459103 | 2018-08-31T07:27:25 | 2018-08-31T07:27:25 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 417,773,879 | 3,007 | alexey-milovidov | Data in ClickHouse is processed by blocks and limits are checked after each block is processed.
For example, max_result_rows will stop query processing if the size of result appears to be larger than the limit after processing a block.
This should be clearly stated in docs.
The size of block may be arbitary an... | 2018-08-31T19:58:47 | 2018-08-31T19:59:05 | {
"+1": 1
} | MEMBER |
ClickHouse | ClickHouse | 417,774,998 | 3,007 | alexey-milovidov | Updated docs (ru) cfb448cce6656a81d8c30bd009fee9737a5b4cf7. | 2018-08-31T20:03:17 | 2018-08-31T20:03:17 | {} | MEMBER |
ClickHouse | ClickHouse | 415,937,593 | 2,427 | maddhuapplicate | Any idea when this would be released? | 2018-08-25T05:05:59 | 2018-08-25T05:05:59 | {} | NONE |
ClickHouse | ClickHouse | 416,030,580 | 2,447 | alexey-milovidov | The diff in this pull request is outdated. See master branch. | 2018-08-26T11:04:51 | 2018-08-26T11:04:51 | {} | MEMBER |
ClickHouse | ClickHouse | 417,592,323 | 2,914 | VladRassokhin | @Imaskar index_granularity is used for Primary Key, so if we're not skipping parts (based on Partitioning Key), but only blocks with same indexes there would be no missed values.
Unfortunately both Partitioning and Primary starts with P, so there's a confusion with what PK means :) | 2018-08-31T08:27:10 | 2018-08-31T08:27:10 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 417,876,362 | 2,907 | alexey-milovidov | Ok.
PS. Non-obvious usage of Context remains as technical debt. | 2018-09-01T17:50:42 | 2018-09-01T17:50:42 | {} | MEMBER |
ClickHouse | ClickHouse | 415,639,215 | 2,929 | zhang2014 | I turned on `enable_optimize_predicate_expression` and was checking each test.
https://github.com/yandex/ClickHouse/pull/2015#issuecomment-413029315 | 2018-08-24T03:20:26 | 2018-08-24T03:20:26 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 416,163,389 | 1,966 | alexey-milovidov | > Как я понимаю, речь идет только о том, чтобы учитывать порядок колонок в заголовке CSV. Сейчас заголовок просто пропускается. Возможно ли это поправить?
Да, совсем несложно добавить возможность учёта столбцов в заголовке. | 2018-08-27T09:03:33 | 2018-08-27T09:03:33 | {} | MEMBER |
ClickHouse | ClickHouse | 416,262,441 | 2,963 | alexey-milovidov | DDL operations for Distributed tables are not propagated to the cluster by default and this is done intentionally.
First, we don't propagate DROP TABLE. This allows to drop Distributed table itself without affecting the cluster. And this also helps for isolation when your users only have access to an empty server wi... | 2018-08-27T15:18:30 | 2018-08-27T15:18:30 | {} | MEMBER |
ClickHouse | ClickHouse | 417,666,021 | 2,942 | VadimPlh | #3009 Fix | 2018-08-31T13:35:32 | 2018-08-31T13:35:32 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 417,867,263 | 3,014 | alexey-milovidov | @kshvakov | 2018-09-01T15:33:06 | 2018-09-01T15:33:06 | {} | MEMBER |
ClickHouse | ClickHouse | 417,876,117 | 2,974 | alexey-milovidov | Ok.
Performance drop is significant, but it's ok as the cost for correctness of this rather specific function. | 2018-09-01T17:47:38 | 2018-09-01T17:47:38 | {} | MEMBER |
ClickHouse | ClickHouse | 416,236,154 | 2,932 | Ouriel | Looks a lots like #2156 | 2018-08-27T13:59:48 | 2018-08-27T13:59:48 | {} | NONE |
ClickHouse | ClickHouse | 416,773,780 | 2,982 | alexey-milovidov | Your own test has failed. | 2018-08-28T23:32:24 | 2018-08-28T23:32:24 | {} | MEMBER |
ClickHouse | ClickHouse | 417,294,503 | 2,994 | alexey-milovidov | :+1: | 2018-08-30T12:01:54 | 2018-08-30T12:01:54 | {} | MEMBER |
ClickHouse | ClickHouse | 417,829,802 | 2,808 | alexey-milovidov | ColumnAggregateFunction::getExtremes | 2018-09-01T03:39:14 | 2018-09-01T03:39:14 | {} | MEMBER |
ClickHouse | ClickHouse | 417,876,496 | 2,881 | alexey-milovidov | > 00646_url_engine is not a stable test
Yes, I confirm. | 2018-09-01T17:52:51 | 2018-09-01T17:52:51 | {} | MEMBER |
ClickHouse | ClickHouse | 417,878,038 | 3,015 | nvartolomei | This seems to fix it, at least for cases I was playing with.
However you should be careful not to include `#include <Interpreters/QueryThreadLog.h>` before declaring `thread_local`s, otherwise `thread_local` keyword is redefined by https://github.com/pocoproject/poco/blob/develop/Foundation/include/Poco/Thread_STD.h... | 2018-09-01T18:19:29 | 2018-09-01T18:19:29 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 415,613,954 | 2,261 | alexey-milovidov | The calculation in the PR looks incorrect:
```
$ docker run -m 1g --cpus="1.0" ubuntu:trusty cat /sys/fs/cgroup/cpu/cpu.cfs_quota_us
WARNING: Your kernel does not support swap limit capabilities or the cgroup is not mounted. Memory limited without swap.
100000
```
But I have no idea, why it is using the value... | 2018-08-24T00:29:45 | 2018-08-24T00:29:45 | {} | MEMBER |
ClickHouse | ClickHouse | 415,674,044 | 2,929 | zhang2014 | > Yes, just in case. We will first enable manually on our servers, then we will enable by default.
BTW, i wanna tell a digression, `compile` and `compile_expressions` open on your servers? How did it work? | 2018-08-24T07:17:50 | 2018-08-24T07:17:50 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 415,941,396 | 2,946 | amosbird | > Do I understand correctly the difference is that we avoid unlocking the mutex before flushBuffer and then locking it again?
Exactly. | 2018-08-25T05:56:43 | 2018-08-25T05:56:43 | {} | COLLABORATOR |
ClickHouse | ClickHouse | 416,344,318 | 2,929 | alexey-milovidov | We enable `compile` setting on our servers, but not `compile_expressions`.
`compile` is pretty mature and stable feature.
When this setting is enabled, ClickHouse will generate specialized code for first step of aggregation (GROUP BY). See `SpecializedAggregator.cpp`, `Compiler` and `Aggregator::compileIfPossible... | 2018-08-27T19:42:56 | 2018-08-27T19:45:23 | {} | MEMBER |
ClickHouse | ClickHouse | 417,454,351 | 2,542 | sethhochberg | Hi,
Confirming we are seeing the same issue using the `IN` keyword with the Set engine, `ClickHouse server version 18.6.0 revision 54401`
With a complete stack trace:
```
2018.08.30 10:24:37.028008 [ 354385 ] <Error> HTTPHandler: Code: 48, e.displayText() = DB::Exception: Method read is not supported by stora... | 2018-08-30T20:21:04 | 2018-08-30T20:21:45 | {} | NONE |
ClickHouse | ClickHouse | 417,866,273 | 2,754 | alexey-milovidov | > @alexey-milovidov Hmm, why do some structs need alignment on their own while the underlying architecture supports unaligned memory access?
Hardware allows unaligned memory access in most cases, but C++ has its own rules.
In C++ compiler may assume that the object is aligned and generate code under that assumption... | 2018-09-01T15:15:38 | 2018-09-01T15:15:38 | {} | MEMBER |
ClickHouse | ClickHouse | 417,866,376 | 2,754 | alexey-milovidov | gcc-8 started to exploit this rule in Aggregator and AutoArray and this cause failed tests.
(We didn't notice it before because there was no automatic builds with gcc-8 neither with ubsan.)
The common part in this code is placement new. | 2018-09-01T15:17:29 | 2018-09-01T15:18:10 | {} | MEMBER |
ClickHouse | ClickHouse | 415,638,235 | 2,929 | alexey-milovidov | There are failed tests, example:
https://travis-ci.org/yandex/ClickHouse/jobs/419930263#L9155 | 2018-08-24T03:12:47 | 2018-08-24T03:12:47 | {} | MEMBER |
ClickHouse | ClickHouse | 416,123,947 | 2,904 | zhang2014 | I have a suggestion, if the data is below a certain threshold, can we use bitmap coding to optimize the data file ? this should be friendly to file io | 2018-08-27T06:07:31 | 2018-08-27T06:08:25 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 416,270,143 | 2,963 | faucct | I guess I didn’t consider TRUNCATE a DDL command. | 2018-08-27T15:42:26 | 2018-08-27T15:42:26 | {} | NONE |
ClickHouse | ClickHouse | 416,773,911 | 2,982 | alexey-milovidov | And a few more. | 2018-08-28T23:33:04 | 2018-08-28T23:33:04 | {} | MEMBER |
ClickHouse | ClickHouse | 415,650,398 | 2,929 | alexey-milovidov | Yes, just in case. We will first enable manually on our servers, then we will enable by default. | 2018-08-24T04:52:44 | 2018-08-24T04:52:44 | {} | MEMBER |
ClickHouse | ClickHouse | 416,143,920 | 2,577 | onine007 | my logs have the same error, How to solve ? | 2018-08-27T07:45:46 | 2018-08-27T07:45:46 | {} | NONE |
ClickHouse | ClickHouse | 416,152,645 | 2,629 | alexey-milovidov | Fixed in master by @wawaka. | 2018-08-27T08:22:18 | 2018-08-27T08:22:18 | {} | MEMBER |
ClickHouse | ClickHouse | 416,256,267 | 2,958 | alexey-milovidov | :+1: | 2018-08-27T15:00:41 | 2018-08-27T15:00:41 | {} | MEMBER |
ClickHouse | ClickHouse | 416,288,660 | 2,961 | WouldYouKindly | @ztlpn Thank you! We found that in testing, so we just dropped the table (:
But good to know that it's fixed. | 2018-08-27T16:42:32 | 2018-08-27T16:42:32 | {} | NONE |
ClickHouse | ClickHouse | 416,449,514 | 2,295 | derekperkins | This used to happen on every upgrade, but it hasn't happened in the last 3-5 stable versions, so it appears to be resolved. | 2018-08-28T04:43:05 | 2018-08-28T04:43:05 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 416,694,860 | 2,976 | alexey-milovidov | Yes, `charset=UTF-8` is technically wrong.
Strings in ClickHouse can contain arbitary binary data,
that is typically assumed to be UTF-8 for text strings,
but the data type and the server itself is charset-agnostic.
We definitely should not recode this binary data in neither way. If we will write `C3 B8` instead ... | 2018-08-28T18:35:59 | 2018-08-28T18:35:59 | {} | MEMBER |
ClickHouse | ClickHouse | 416,459,103 | 2,974 | zhang2014 | Take the best result of the three runs.
### Compile Parameters:
``` Shell
cmake .. -DCMAKE_CXX_COMPILER=`which g++-8 ` -DCMAKE_C_COMPILER="`which gcc-8`" -DENABLE_TESTS=0 -DUSE_UNWIND=0 -DENABLE_CLICKHOUSE_COMPRESSOR=0 -DENABLE_MONGODB=1 -DENABLE_JEMALLOC=0 -DCMAKE_BUILD_TYPE=Release
```
### Before
``` Json
[... | 2018-08-28T05:47:59 | 2018-08-31T07:24:44 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 417,867,344 | 3,005 | alexey-milovidov | @serebrserg | 2018-09-01T15:34:29 | 2018-09-01T15:34:29 | {} | MEMBER |
ClickHouse | ClickHouse | 417,891,537 | 3,015 | nvartolomei | A minimal example with thread_local failing under gcc 8.2 on macOS, however works with llvm.
`main.cpp`
```c++
#include "lib.h"
int main() {
getProfileEvents();
return 0;
}
```
`lib.h`
```c++
#pragma once
#include <memory>
#include <iostream>
class ThreadStatus {
public:
int x = 0;... | 2018-09-01T22:26:59 | 2018-09-01T22:26:59 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 415,606,594 | 2,261 | reflection | Clickhouse still doesn't respect cgroup limits, unfortunately. The merged pull request incorrectly calculates physical cpus and doesn't address memory limits (docker image below modified to log to console at debug level):
```
# Start with memory limited to 1gb and cpus limited to 1 core
# https://docs.docker.com/... | 2018-08-23T23:39:57 | 2018-08-23T23:39:57 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 416,064,729 | 2,261 | bobrik | I'll take this chance to also mention that if you're using CFS quotas, there's a high chance you're throttling when you shouldn't: https://github.com/kubernetes/kubernetes/issues/67577. | 2018-08-26T19:40:57 | 2018-08-26T19:40:57 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 417,866,913 | 2,754 | amosbird | Oh, I misunderstood the term `on their own`. Thanks | 2018-09-01T15:27:18 | 2018-09-01T15:27:18 | {} | COLLABORATOR |
ClickHouse | ClickHouse | 415,406,502 | 1,344 | nvartolomei | Hello, does the ClickHouse team have any plans regarding on improving this query pattern?
ps. Also, I'm interested in this optimisation to work together with `where` conditions. | 2018-08-23T13:03:57 | 2018-08-23T14:26:06 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 415,623,106 | 2,261 | reflection | Thanks for reverting. In the #2325 pull request @bobrik linked to [openjdk's cgroup work](http://hg.openjdk.java.net/jdk/jdk/rev/7f22774a5f42).
For cpu: `int OSContainer::active_processor_count()`
For memory: `long OSContainer::memory_limit_in_bytes()` | 2018-08-24T01:30:02 | 2018-08-24T01:30:02 | {
"+1": 1
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 417,452,455 | 2,980 | alexey-milovidov | Ok.
But the code in `Memory` is a little bit tangled (from my point of view).
Because capacity is first initialized to value without padding and then padded. | 2018-08-30T20:14:22 | 2018-08-30T20:14:22 | {} | MEMBER |
ClickHouse | ClickHouse | 417,468,373 | 2,948 | alexey-milovidov | Submodules are updated accidentially.
| 2018-08-30T21:10:43 | 2018-08-30T21:10:43 | {} | MEMBER |
ClickHouse | ClickHouse | 417,878,845 | 3,015 | alexey-milovidov | I will just remove these lines from our Poco fork.
| 2018-09-01T18:34:37 | 2018-09-01T18:34:37 | {} | MEMBER |
ClickHouse | ClickHouse | 415,605,279 | 1,344 | alexey-milovidov | This is planned to do after more fundamental modifications in query pipeline.
These modifications are planned for Q4. You may look at the prototype here: https://github.com/yandex/ClickHouse/compare/processors#diff-d84e5dc28812accede7d6803ffb73be9R13 | 2018-08-23T23:32:04 | 2018-08-23T23:32:04 | {
"hooray": 3
} | MEMBER |
ClickHouse | ClickHouse | 416,209,688 | 2,963 | faucct | > You can use TRUNCATE TABLE local_table_name ON CLUSTER cluster_name;
Yes, I have figured this out.
> BTW, TRUNCATE a distributed table only removes temporary files
Don't you think this seems a little counterintuitive? | 2018-08-27T12:26:25 | 2018-08-27T12:26:25 | {} | NONE |
ClickHouse | ClickHouse | 416,265,913 | 2,961 | ztlpn | The bug (vertical merge fails for empty parts that are created as a result of DELETE) was fixed in #2930.
To fix the problem before this fix is released, you can temporarily turn off vertical merge (set `enable_vertical_merge_algorithm` to 0 in [MergeTree settings](https://clickhouse.yandex/docs/en/operations/server... | 2018-08-27T15:29:19 | 2018-08-27T15:29:19 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 416,292,629 | 2,966 | alexey-milovidov | :+1: | 2018-08-27T16:56:12 | 2018-08-27T16:56:12 | {} | MEMBER |
ClickHouse | ClickHouse | 417,461,819 | 2,974 | alexey-milovidov | Thank you!
PS. The performance test you quote is incorrect because
`visitParamExtractRaw('{"myparam":"test_string"}', 'myparam')`
is a constant expression and it is calculated just once.
(it's obvious because otherwise you cannot get 2+ billion rows/sec. on single core.)
To be sure, you can transform a stri... | 2018-08-30T20:47:33 | 2018-08-30T21:07:26 | {} | MEMBER |
ClickHouse | ClickHouse | 417,560,186 | 580 | sukantamaikap | For connecting to the clickhouse-server container as part of docker-compose
docker-compose exec -u 0 << service name or container id >> bash
run clickhouse client inside the container | 2018-08-31T05:57:32 | 2018-08-31T05:59:06 | {} | NONE |
ClickHouse | ClickHouse | 417,824,463 | 2,999 | champtar | Maybe you could use http trailers? | 2018-09-01T01:48:37 | 2018-09-01T01:48:37 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 417,832,644 | 2,754 | alexey-milovidov | Thank you! | 2018-09-01T04:45:24 | 2018-09-01T04:45:24 | {} | MEMBER |
ClickHouse | ClickHouse | 415,952,738 | 2,951 | blinkov | There are a bunch of different options on how to integrate ClickHouse and MySQL for various use cases:
* https://clickhouse.yandex/docs/en/operations/table_engines/mysql/
* https://clickhouse.yandex/docs/en/query_language/dicts/external_dicts_dict_sources/#mysql
* https://github.com/sysown/proxysql/wiki/ClickHouse-S... | 2018-08-25T08:20:53 | 2018-08-25T08:20:53 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 416,003,665 | 2,955 | alexey-milovidov | Great, thank you! | 2018-08-25T23:47:19 | 2018-08-25T23:47:19 | {} | MEMBER |
ClickHouse | ClickHouse | 416,223,963 | 2,833 | sundarv85 | @kshvakov I don't have an example right now, except the issue we are seeing in our staging server. I will work on an example and get back to you. | 2018-08-27T13:20:00 | 2018-08-27T13:20:00 | {} | NONE |
ClickHouse | ClickHouse | 416,278,192 | 1,102 | filimonov | > Let the rename statement support multi renames, and support transaction for the renames
Rename support renaming multiple tables at once. Syntax is usual.
```
RENAME TABLE a1 TO b1, a2 TO b2, a3 TO b3
```
`RENAME` operation can be considered atomic but "but only in cases when there was no exceptions during t... | 2018-08-27T16:07:54 | 2018-08-27T16:09:05 | {
"+1": 6
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 416,559,274 | 2,971 | urykhy | It's workaround for https://github.com/yandex/ClickHouse/issues/2621
(value 30000 makes the trick) | 2018-08-28T12:05:04 | 2018-08-28T12:05:29 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 416,920,966 | 2,770 | alexey-milovidov | We've just planned to polish this PR (or #2818),
and the way of implementation in this PR theoretically should be fine. | 2018-08-29T11:35:04 | 2018-08-29T11:35:04 | {} | MEMBER |
ClickHouse | ClickHouse | 417,834,116 | 2,754 | amosbird | > Nevertheless we should support alignment of aggregation states if your structs do require alignment.
@alexey-milovidov Hmm, why do some structs need alignment on their own while the underlying architecture supports unaligned memory access? | 2018-09-01T05:19:23 | 2018-09-01T05:19:34 | {} | COLLABORATOR |
ClickHouse | ClickHouse | 416,208,728 | 2,963 | zhang2014 | You can use `TRUNCATE TABLE local_table_name ON CLUSTER cluster_name;`
BTW, `TRUNCATE` a distributed table only removes temporary files | 2018-08-27T12:22:06 | 2018-08-27T12:22:06 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 416,209,067 | 2,960 | zhang2014 | Existing integration tests should be able to reproduce this problem | 2018-08-27T12:23:47 | 2018-08-27T12:24:26 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 416,216,055 | 2,963 | faucct | Regarding the configurations: neither `yandex > insert_distributed_sync = 1` nor `yandex > settings > insert_distributed_sync = 1` put in `/etc/clickhouse-server/conf.d/*.xml` file gets reflected in `SELECT * FROM system.settings`. Am I doing everything right?
Regarding `TRUNCATE` behaviour: I can't imagine anyone wan... | 2018-08-27T12:51:22 | 2018-08-27T12:55:31 | {} | NONE |
ClickHouse | ClickHouse | 416,520,347 | 1,851 | huangyaofu2 | i met this , when i run `clickehouse-client` , i got:
```
ClickHouse client version 1.1.54336.
Connecting to localhost:9000.
Code: 32. DB::Exception: Attempt to read after eof
```
and there are 65 'CLOSE_WAIT' connections, i think this matters.
after i restart the service ,problem resolved. | 2018-08-28T09:46:07 | 2018-08-28T09:51:07 | {} | NONE |
ClickHouse | ClickHouse | 416,549,062 | 2,971 | alexey-milovidov | I wonder, what are the use cases - how exactly you are going to change this parameter and why? | 2018-08-28T11:23:46 | 2018-08-28T11:23:46 | {} | MEMBER |
ClickHouse | ClickHouse | 417,241,916 | 1,568 | shalugin | @blinkov Thanks! | 2018-08-30T08:51:36 | 2018-08-30T08:51:36 | {} | NONE |
ClickHouse | ClickHouse | 415,010,125 | 2,914 | VladRassokhin | >I.e. if 2 values of PK in sparse index are A and next is also A, just assume that whole block have value A repeated 8192 times (or index_granularity times) without reading the block from the disk?
Correct, sounds reasonable.
AFAIK for now it doesn't work even for simple one-column PrimaryKey | 2018-08-22T12:09:42 | 2018-08-31T08:27:18 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 417,828,917 | 2,808 | alexey-milovidov | AggregateFunctionForEach | 2018-09-01T03:20:24 | 2018-09-01T03:20:24 | {} | MEMBER |
ClickHouse | ClickHouse | 417,828,743 | 2,808 | alexey-milovidov | AggregateFunctionNull
... | 2018-09-01T03:17:11 | 2018-09-01T03:25:35 | {} | MEMBER |
ClickHouse | ClickHouse | 417,892,141 | 2,801 | VadimPlh | @amosbird Hi! Thanks for your PR! Your code has problems with test(00678_shard_funnel_window). I try to fix it. | 2018-09-01T22:37:39 | 2018-09-02T12:36:53 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 415,659,720 | 2,913 | alexey-milovidov | Fixed in master. | 2018-08-24T05:59:50 | 2018-08-24T05:59:50 | {} | MEMBER |
ClickHouse | ClickHouse | 415,894,042 | 2,946 | alexey-milovidov | Do I understand correctly the difference is that we avoid unlocking the mutex before flushBuffer and then locking it again? Looks reasonable. | 2018-08-24T22:01:59 | 2018-08-24T22:01:59 | {} | MEMBER |
ClickHouse | ClickHouse | 416,198,351 | 1,531 | badvir | @blinkov No, I don't have any, thank you.
| 2018-08-27T11:36:37 | 2018-08-27T11:36:37 | {} | NONE |
ClickHouse | ClickHouse | 416,211,217 | 2,963 | zhang2014 | You're right, but the current implementation is more consistent with ClickHouse's other DDL. Like `DROP` : ) | 2018-08-27T12:32:55 | 2018-08-27T12:32:55 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 416,644,900 | 2,976 | faucct | I think the only way to make header consistent with response encoding is to change header, as making response body UTF-8 will most likely double encode it. | 2018-08-28T16:06:28 | 2018-08-28T16:06:28 | {} | NONE |
ClickHouse | ClickHouse | 417,219,205 | 2,560 | extead | @alexey-milovidov, I found release 1.1.54343, which works correctly in this case. I hope this helps. | 2018-08-30T07:32:20 | 2018-08-30T07:32:20 | {} | NONE |
ClickHouse | ClickHouse | 417,589,416 | 2,914 | Imaskar | This could be reasoned about partitioning key, but not primary key. In PK it would lead to missing values (block was skipped, but shouldn't been). | 2018-08-31T08:15:31 | 2018-08-31T08:15:31 | {} | NONE |
ClickHouse | ClickHouse | 417,871,348 | 3,005 | serebrserg | There is no comprehensive balancing utilities implemented inside
clickhouse-jdbc driver itself. You can try to use
https://github.com/yandex/clickhouse-jdbc/blob/master/src/main/java/ru/yandex/clickhouse/BalancedClickhouseDataSource.java
but I suggest you read the javadoc carefully and review the code to decide
if it f... | 2018-09-01T16:37:33 | 2018-09-01T16:37:33 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 416,149,173 | 1,659 | tacyuuhon | @blinkov
Yes, it looks like I just made a wrong config.
| 2018-08-27T08:07:41 | 2018-08-27T08:07:41 | {} | NONE |
ClickHouse | ClickHouse | 417,457,035 | 2,999 | alexey-milovidov | The difficulty is that the server can start to send result before query is finished. | 2018-08-30T20:30:27 | 2018-08-30T20:30:27 | {} | MEMBER |
ClickHouse | ClickHouse | 417,460,086 | 2,975 | alexey-milovidov | Great! I always wanted a function that is "dual" to arrayEnumerateUniq.
| 2018-08-30T20:41:37 | 2018-08-30T20:41:37 | {} | MEMBER |
ClickHouse | ClickHouse | 417,867,184 | 3,015 | alexey-milovidov | This may happen due to a usage of thread_local variable in different translation units.
Does the issue reproduce if we merge ThreadStatus.cpp, ThreadStatusExt.cpp, CurrentThread.cpp to a single translation unit?
Also we need to check that non-trivial initialization of thread_local is not broken at all on this platf... | 2018-09-01T15:31:53 | 2018-09-01T15:31:53 | {} | MEMBER |
ClickHouse | ClickHouse | 417,879,154 | 3,010 | alexey-milovidov | There is no detailed information. | 2018-09-01T18:40:39 | 2018-09-01T18:40:39 | {} | MEMBER |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.