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 | 565,773,856 | 6,460 | alexey-milovidov | > have to build icudata from sources (from txt files with generator)
I prefer a compromise solution because build of data is too complicated. | 2019-12-15T03:26:20 | 2019-12-15T03:26:47 | {} | MEMBER |
ClickHouse | ClickHouse | 565,854,986 | 7,900 | alexey-milovidov | About the choice of default port in configuration example, 8001.
Is it reasonable?
We should default to some most widely used port for that type of handles or simply reuse http_port. | 2019-12-15T22:45:11 | 2019-12-15T22:48:21 | {} | MEMBER |
ClickHouse | ClickHouse | 565,864,097 | 8,218 | alexey-milovidov | ASM version of SHA256 is not compatible with query profiler (unwinder):
```
Thread 82 "ParalInputsProc" received signal SIGSEGV, Segmentation fault.
[Switching to Thread 0x7ffe0eb7c700 (LWP 984386)]
libunwind::DwarfInstructions<libunwind::LocalAddressSpace, libunwind::Registers_x86_64>::evaluateExpression (expres... | 2019-12-16T00:31:01 | 2019-12-16T00:31:01 | {} | MEMBER |
ClickHouse | ClickHouse | 565,884,540 | 2,399 | alexey-milovidov | Для шаблонов без `(?t)` исправлено давно, а в общем случае, исправить пока не получится. | 2019-12-16T02:53:43 | 2019-12-16T02:53:43 | {} | MEMBER |
ClickHouse | ClickHouse | 566,019,014 | 7,900 | azat | >Common Prometheus port is 9090
Hm, but this is prometheus server, not the exporter (while this is an exporter)
*P.S. this is just a note, no objections against 9090 port* | 2019-12-16T11:19:03 | 2019-12-16T11:19:03 | {} | MEMBER |
ClickHouse | ClickHouse | 566,194,376 | 8,034 | alexey-milovidov | https://github.com/ClickHouse/ClickHouse/commit/0042c8271e49f37256e2c577610685faefca9d38
https://github.com/ClickHouse/ClickHouse/commit/0bc1c6d91d966778f2443640282d79b7919242ad
1. Why I have removed warning suppression of `-Wdocumentation` for clang?
Because h3 headers are included as system headers:
https://g... | 2019-12-16T18:57:02 | 2019-12-16T18:57:20 | {} | MEMBER |
ClickHouse | ClickHouse | 566,342,928 | 8,014 | ghost | @volfco
I am fixing the perfect storm of pending merges at https://github.com/ClickHouse/ClickHouse/pull/8242
Thank you. | 2019-12-17T01:50:06 | 2019-12-17T01:50:06 | {} | NONE |
ClickHouse | ClickHouse | 566,393,598 | 8,027 | alexey-milovidov | @amosbird This is related to configuration of jemalloc.
It is trying to select a method to purge pages:
pages.c:
```
#elif defined(JEMALLOC_PURGE_MADVISE_FREE)
return (madvise(addr, size,
# ifdef MADV_FREE
MADV_FREE
# else
JEMALLOC_MADV_FREE
# endif
) != 0);
#elif defined(JEMALLOC_PURGE... | 2019-12-17T05:55:35 | 2019-12-17T05:55:35 | {} | MEMBER |
ClickHouse | ClickHouse | 566,454,514 | 8,127 | alesapin | failed test flap, so I'll merge it. | 2019-12-17T09:18:35 | 2019-12-17T09:18:35 | {} | MEMBER |
ClickHouse | ClickHouse | 566,477,545 | 8,209 | akuzm | Test failures are not related to this patch. | 2019-12-17T10:19:01 | 2019-12-17T10:19:01 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 565,705,347 | 5,749 | filimonov | We are.
But error handling is still very poor in Kafka (in case of error it just leave the consumer group to join it back after 5 sec, creating 2 rebalances per one error :) ) , i think i know how to improve that, i will try to do it next week and also create some test for that issue. | 2019-12-14T10:34:20 | 2019-12-14T10:34:20 | {
"+1": 1
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 565,748,657 | 8,172 | alexey-milovidov | Fixed in master. | 2019-12-14T20:02:45 | 2019-12-14T20:02:45 | {} | MEMBER |
ClickHouse | ClickHouse | 565,849,797 | 7,517 | alexey-milovidov | > As far as I know, modifying the order by clause is not currently supported. I remember there was an issue about it.
There was support for extending ORDER BY key and adding a new column simultaneously. As far as I remember, it was implemented only for non-replicated MergeTree tables by feature request of one intern... | 2019-12-15T21:42:49 | 2019-12-15T21:42:49 | {} | MEMBER |
ClickHouse | ClickHouse | 565,859,672 | 8,059 | alexey-milovidov | Ok.
PS. It's unclear that the part will be moved only if all rows satisfy the condition. | 2019-12-15T23:45:09 | 2019-12-15T23:45:09 | {} | MEMBER |
ClickHouse | ClickHouse | 565,882,221 | 7,898 | alexey-milovidov | 00463_long_sessions_in_http_interface - flacky test. | 2019-12-16T02:41:02 | 2019-12-16T02:41:02 | {} | MEMBER |
ClickHouse | ClickHouse | 565,882,689 | 8,223 | krnaveen14 | @den-crane Going through the Feature request you've mentioned, it seems that it is stale right now.
I was thinking that second proposed solution might be straightforward approach. Probably by always sorting the data blocks by primary key (after predicate pushdown) and removing the older version of rows while doing t... | 2019-12-16T02:43:19 | 2019-12-16T02:47:21 | {} | NONE |
ClickHouse | ClickHouse | 565,885,403 | 1,168 | alexey-milovidov | > Perhaps it should accept connection first, and allow certain queries despite the query limit (SYSTEM, KILL)?
Now it's possible to execute KILL query regardless to the number of simultaneous queries. | 2019-12-16T02:58:22 | 2019-12-16T02:58:22 | {} | MEMBER |
ClickHouse | ClickHouse | 565,945,997 | 7,900 | vdimir | > About the choice of default port in configuration example, 8001.
> Is it reasonable?
>
> We should default to some most widely used port for that type of handles or simply reuse http_port.
Common Prometheus port is 9090 ([link](https://github.com/prometheus/prometheus/wiki/Default-port-allocations)), I had to ... | 2019-12-16T07:53:24 | 2019-12-16T07:54:35 | {} | MEMBER |
ClickHouse | ClickHouse | 566,006,462 | 8,201 | hczhcz | I updated it to the latest git version. It does not reproduce.
I have no clue on what happened. | 2019-12-16T10:43:58 | 2019-12-16T10:43:58 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 566,199,334 | 6,847 | alesapin | Failed test depend on unimplemented feature. | 2019-12-16T19:09:59 | 2019-12-16T19:09:59 | {} | MEMBER |
ClickHouse | ClickHouse | 566,230,509 | 8,034 | alexey-milovidov | > Why don't we want to prefix C function from h3?
We don't need that - there is no name collisions. | 2019-12-16T20:31:43 | 2019-12-16T20:31:49 | {} | MEMBER |
ClickHouse | ClickHouse | 565,680,522 | 8,212 | Slach | there is for queries which used UNION ALL
```
SELECT 'undefined' AS field FROM table AS t
UNION ALL
SELECT field FROM table AS t
``` | 2019-12-14T04:11:24 | 2019-12-14T04:11:24 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 565,783,118 | 8,141 | liukun4515 | My clang version is
```
apple LLVM version 10.0.1 (clang-1001.0.46.4)
Target: x86_64-apple-darwin18.6.0
Thread model: posix
InstalledDir: /Library/Developer/CommandLineTools/usr/bin
``` | 2019-12-15T06:48:08 | 2019-12-15T06:48:18 | {} | NONE |
ClickHouse | ClickHouse | 565,850,024 | 7,327 | alexey-milovidov | Let's continue? | 2019-12-15T21:45:21 | 2019-12-15T21:45:21 | {} | MEMBER |
ClickHouse | ClickHouse | 565,850,349 | 7,327 | stavrolia | Yep, I would soon | 2019-12-15T21:49:12 | 2019-12-15T21:49:12 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 565,855,231 | 7,900 | alexey-milovidov | `2019.12.16 01:47:49.789705 [ 1 ] {} <Information> Application: Listening http://[::1]:8001`
not informative enough. | 2019-12-15T22:48:14 | 2019-12-15T22:48:14 | {} | MEMBER |
ClickHouse | ClickHouse | 566,051,615 | 8,200 | hombit | @alesapin done | 2019-12-16T13:02:41 | 2019-12-16T13:02:41 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 566,186,925 | 8,236 | alexey-milovidov | `00800_low_cardinality_join` has failed. | 2019-12-16T18:37:25 | 2019-12-16T18:37:25 | {} | MEMBER |
ClickHouse | ClickHouse | 565,830,764 | 8,171 | alexey-milovidov | Note that there is no integration test for TLS 1.3. | 2019-12-15T17:48:27 | 2019-12-15T17:48:41 | {} | MEMBER |
ClickHouse | ClickHouse | 565,882,954 | 8,151 | alexey-milovidov | Please investigate... | 2019-12-16T02:44:36 | 2019-12-16T02:44:36 | {} | MEMBER |
ClickHouse | ClickHouse | 565,940,668 | 8,223 | krnaveen14 | > Isn't it the same as FINAL? `select ... from table FINAL` ?
On the query output, yes it's. But usage of FINAL limits the query being single threaded. I've tested it with our dataset and found the performance hit to be pretty huge.
One thing I realised right now is, the limitation of single threaded which we hav... | 2019-12-16T07:35:52 | 2019-12-16T07:47:59 | {} | NONE |
ClickHouse | ClickHouse | 565,955,951 | 8,200 | alesapin | Maybe add a reference to https://clickhouse.yandex/docs/en/query_language/create/#column-compression-codecs? | 2019-12-16T08:25:04 | 2019-12-16T08:25:04 | {} | MEMBER |
ClickHouse | ClickHouse | 566,161,880 | 8,237 | alesapin | Own tests failed :( | 2019-12-16T17:31:33 | 2019-12-16T17:31:33 | {} | MEMBER |
ClickHouse | ClickHouse | 566,202,349 | 8,228 | alexey-milovidov | > Unknown MySQL server host 'host' (-2) ((nullptr):0)
Probably you have erroneously specified `host` as hostname for MySQL server, like this:
`<host>host</host>`
The `(nullptr):0` part is Ok - it's what we have as the error message from the library. | 2019-12-16T19:17:50 | 2019-12-16T19:19:36 | {} | MEMBER |
ClickHouse | ClickHouse | 566,392,125 | 8,242 | alexey-milovidov | > I could improve performance after fixing this bug.
Could you please add this into `dbms/tests/performance`? | 2019-12-17T05:49:27 | 2019-12-17T05:49:27 | {} | MEMBER |
ClickHouse | ClickHouse | 566,477,870 | 8,235 | akuzm | Test failures are not related to this patch. | 2019-12-17T10:19:55 | 2019-12-17T10:19:55 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 565,783,400 | 7,970 | liukun4515 | But I'm very confused, why not support the gcc compiler on the mac os?
@nikitamikhaylov | 2019-12-15T06:54:40 | 2019-12-15T06:54:40 | {} | NONE |
ClickHouse | ClickHouse | 565,835,427 | 2,890 | alexey-milovidov | List of third-party libraries:
```
milovidov@milovidov-Latitude-E7470:~/work/ClickHouse/build$ cat licenses.sh
#!/bin/bash
jq --raw-output '.[].file' compile_commands.json | xargs dirname | sort | uniq | while read file
do
echo -n '.' 1>&2
DIR=$file
while [[ "$DIR" != '/' ]]
do
li... | 2019-12-15T18:42:19 | 2019-12-15T18:42:19 | {} | MEMBER |
ClickHouse | ClickHouse | 565,877,755 | 8,223 | den-crane | the same as https://github.com/ClickHouse/ClickHouse/issues/5125
>Don't use the unmerged data for select queries until the actual merges / deduplication happens
| 2019-12-16T02:15:33 | 2019-12-16T02:15:33 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 566,260,063 | 7,676 | kmatt | @alexey-milovidov Related to Parquet partitions, it appears the current S3 table function (v19.19.1) does not support path style access nor wildcards? Segmented files in the root of a bucket don't seem accessable:
```
SELECT * FROM s3('http://host:port/test/part-00000-1f19d701-3aab-4398-98d4-07aec0a99402.c000.snap... | 2019-12-16T21:50:50 | 2019-12-16T22:15:15 | {
"+1": 1
} | NONE |
ClickHouse | ClickHouse | 565,549,889 | 8,168 | alexey-milovidov | @zhang2014 I have created this issue according to one user's complaint. Maybe there is no such issue :) I will check more carefully... | 2019-12-13T18:16:26 | 2019-12-13T18:16:26 | {} | MEMBER |
ClickHouse | ClickHouse | 565,702,059 | 5,653 | ishirav | I got the same problem when using default values with CSV format. I'm using ClickHouse 19.17.5 and input_format_defaults_for_omitted_fields is on by default. For example:
```
~ clickhouse-client -q "CREATE TABLE foo (id UInt32 DEFAULT 1, date Date DEFAULT today(), random UInt32 DEFAULT rand()) ENGINE = Log"
~ echo -... | 2019-12-14T09:40:02 | 2019-12-14T09:40:02 | {
"confused": 1
} | NONE |
ClickHouse | ClickHouse | 565,857,232 | 8,226 | alexey-milovidov | OPTIMIZE query should flush the Buffer table. | 2019-12-15T23:13:22 | 2019-12-15T23:13:22 | {
"+1": 3
} | MEMBER |
ClickHouse | ClickHouse | 565,965,396 | 8,227 | onine007 | @alesapin thank you. | 2019-12-16T08:54:54 | 2019-12-16T08:54:54 | {} | NONE |
ClickHouse | ClickHouse | 566,101,972 | 8,236 | 4ertus2 | #8154 | 2019-12-16T15:11:11 | 2019-12-16T15:11:11 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 566,166,367 | 5,335 | alexey-milovidov | @Melancholic It was really easy! Thank you! | 2019-12-16T17:43:39 | 2019-12-16T17:43:39 | {} | MEMBER |
ClickHouse | ClickHouse | 566,390,876 | 8,228 | ashkank66 | I actually haven't configured MySQL on my clickhouse, and to be honest, I have to admit I tried to find a configuration for MySQL but I couldn't.
Could you tell me where should I configure it? | 2019-12-17T05:43:54 | 2019-12-17T05:43:54 | {} | NONE |
ClickHouse | ClickHouse | 565,727,786 | 8,215 | filimonov | Only in Russian for now: https://clickhouse.yandex/docs/ru/extended_roadmap/
Will be translated soon. | 2019-12-14T15:35:04 | 2019-12-14T15:35:04 | {
"+1": 2
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 565,744,602 | 7,477 | Vinatorul | Hi! I want to work on this as my first issue.
Am I correct that I should replace `thread_number` assignment logic with `os_thread_id` one (for example [here](https://github.com/ClickHouse/ClickHouse/blob/master/dbms/src/Common/ThreadStatus.cpp#L37))?
Firstly I thought I can change `getThreadNumber`, but have found th... | 2019-12-14T19:21:25 | 2019-12-14T19:23:12 | {} | NONE |
ClickHouse | ClickHouse | 565,771,392 | 8,159 | aa8808021 | > Currently HA mode is not supported.
Is there any plan to support ha? thanks | 2019-12-15T02:31:55 | 2019-12-15T02:31:55 | {} | NONE |
ClickHouse | ClickHouse | 565,791,696 | 7,684 | maqroll | Fixed in #8216 | 2019-12-15T09:21:14 | 2019-12-15T09:21:14 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 565,859,837 | 8,189 | alexey-milovidov | This is very old code that is needed only for some internal customers.
It's better to avoid usage of embedded dictionary for region names because the same can be achieved with external dictionaries in more flexible way. | 2019-12-15T23:47:07 | 2019-12-15T23:47:07 | {} | MEMBER |
ClickHouse | ClickHouse | 565,954,496 | 8,227 | alesapin | Seems like you dealing with tables, not dictionaries? Directories in `detached` directory called parts (parts of your table). If they start with `ignored_` or `broken_` and you sure, that all data is present in your table then you can remove them manually (`rm -r`) or use `DROP DETACHED` query https://clickhouse.yandex... | 2019-12-16T08:20:42 | 2019-12-16T08:20:42 | {
"+1": 2
} | MEMBER |
ClickHouse | ClickHouse | 566,035,223 | 7,900 | alexey-milovidov | Ok. Waiting for a PR!
> response more verbose
I already made it. | 2019-12-16T12:11:36 | 2019-12-16T12:11:36 | {} | MEMBER |
ClickHouse | ClickHouse | 565,843,748 | 7,900 | azat | >Is it acceptable to keep metric names in CamelCase without any changes?
Another reason for keeping the metric names - it will allow use various dashboards without any changes | 2019-12-15T20:28:45 | 2019-12-15T20:28:45 | {
"+1": 1
} | MEMBER |
ClickHouse | ClickHouse | 566,218,125 | 8,237 | alesapin | According to logs 10 seconds is not enough for mutation of two parts.
```
2019.12.16 20:20:58.616343 [ 36 ] {862e949e-b28f-4864-abfb-58f07ecd0141} <Debug> executeQuery: (from [::1]:54892) ALTER TABLE table_for_synchronous_mutations_no_replication UPDATE v1 = v1 + 1 WHERE 1 SETTINGS mutation_synchronous_wait_timeout ... | 2019-12-16T19:58:19 | 2019-12-16T19:58:19 | {} | MEMBER |
ClickHouse | ClickHouse | 566,414,403 | 7,195 | juansc | I tried various docker images and I found that this bug starts closer to `clickhouse-server:19.11.12.69`. I'll work on creating a minimal schema and then post it here. | 2019-12-17T07:16:50 | 2019-12-17T07:16:50 | {} | NONE |
ClickHouse | ClickHouse | 565,596,626 | 8,213 | alexey-milovidov | @alesapin You'll find something interesting in the test. | 2019-12-13T20:26:26 | 2019-12-13T20:26:26 | {} | MEMBER |
ClickHouse | ClickHouse | 566,099,673 | 8,231 | millin | Please update docs | 2019-12-16T15:05:33 | 2019-12-16T15:05:33 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 566,340,601 | 8,027 | jiugem | @alexey-milovidov
Use the following command to solve, I wonder if there is a problem：
cmake .. -DENABLE_TCMALLOC=OFF -DENABLE_JEMALLOC=OFF -DCMAKE_INSTALL_PREFIX=/opt/clickhouse | 2019-12-17T01:38:57 | 2019-12-17T01:38:57 | {} | NONE |
ClickHouse | ClickHouse | 565,566,405 | 5,735 | alexey-milovidov | `no space left on device` has one of the possible reasons:
- there is literally no space left on device;
- the filesystem cannot allocate more inodes (probably not the reason, because it should be thrown on `open` not `write`);
- some other limits on space have been configured and exceeded. | 2019-12-13T19:00:45 | 2019-12-13T19:00:45 | {} | MEMBER |
ClickHouse | ClickHouse | 565,571,209 | 5,749 | alexey-milovidov | @filimonov Please validate that now we are doing commit only after successful write. | 2019-12-13T19:13:35 | 2019-12-13T19:13:35 | {} | MEMBER |
ClickHouse | ClickHouse | 565,782,323 | 8,218 | alexey-milovidov | Now OpenSSL works with AArch64 build:
```
milovidov@example:~/ClickHouse/build_output_folder$ docker run --network=host --rm -it -v/home/milovidov/ClickHouse/build_output_folder:/build multiarch/ubuntu-core:arm64-bionic /bin/bash
root@example:/# /build/clickhouse local --query "SELECT * FROM url('https://ya.ru/', TS... | 2019-12-15T06:29:51 | 2019-12-15T06:29:51 | {} | MEMBER |
ClickHouse | ClickHouse | 565,849,634 | 7,517 | alexey-milovidov | > Does the Clickhouse re-sort the data to make granule`s info? or uses the data already sorted by primary key?
ClickHouse only sorts data by order key (primary key). And the granules of secondary indices cover existing ranges of primary key. | 2019-12-15T21:40:25 | 2019-12-15T21:41:15 | {} | MEMBER |
ClickHouse | ClickHouse | 566,203,799 | 7,900 | alexey-milovidov | Two suggestions:
1. Let's add underscores after metric group: `ClickHouseAsyncMetrics_`
2. Let's replace dots with underscores: `jemalloc_background_thread_run_interval` | 2019-12-16T19:21:20 | 2019-12-16T19:21:20 | {} | MEMBER |
ClickHouse | ClickHouse | 566,402,558 | 8,151 | amosbird | some progress:
exception unwinding fails at this frame
https://github.com/yandex/ClickHouse/blob/master/dbms/src/Interpreters/executeQuery.cpp#L574
1. surrounding it with a `try {...} catch (...) { throw; }` works
2. unwinding directly without exception works
3. compile executeQuery with `-fno-omit-frame-poi... | 2019-12-17T06:32:42 | 2019-12-17T06:32:42 | {
"+1": 1
} | COLLABORATOR |
ClickHouse | ClickHouse | 566,033,962 | 7,900 | vdimir | > Hm, but this is prometheus server, not the exporter (while this is an exporter)
Thanks for note, I was wrong. Exporters uses ports starting at 9100.
https://prometheus.io/docs/instrumenting/writing_exporters/#port-numbers
> Feel free to grab the next free port number when developing your exporter, preferably bef... | 2019-12-16T12:07:24 | 2019-12-16T12:08:26 | {} | MEMBER |
ClickHouse | ClickHouse | 565,570,568 | 8,206 | akuzm | 00834_kill_mutation_replicated_zookeeper flaps | 2019-12-13T19:12:00 | 2019-12-13T19:12:00 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 566,467,491 | 8,245 | alex-zaitsev | Live View currently does not support distributed tables. The error message is misleading. | 2019-12-17T09:52:29 | 2019-12-17T09:52:29 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 565,565,925 | 8,168 | zhang2014 | Well, I'm just surprised. I mistakenly thought I had missed an important piece of logic :grin:
| 2019-12-13T18:59:27 | 2019-12-13T19:04:52 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 565,569,954 | 5,378 | alexey-milovidov | This issue was fixed in https://github.com/ClickHouse/ClickHouse/pull/7252, version 19.16. | 2019-12-13T19:10:22 | 2019-12-13T19:10:22 | {} | MEMBER |
ClickHouse | ClickHouse | 565,709,661 | 7,792 | vzakaznikov | @KochetovNicolai, your changes caused a regression. The https://github.com/ClickHouse/ClickHouse/issues/7773#issuecomment-562382283 manual test case does not work any more.
Code: 47. DB::Exception: Received from localhost:9000. DB::Exception: Missing columns: 'A.id' while processing query: 'SELECT A.id, name FROM te... | 2019-12-14T11:42:04 | 2019-12-14T11:42:46 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 565,849,058 | 2,908 | alexey-milovidov | True fixed size database setup is not likely to be implemented, because we use copy-on-write strategy for data parts and it will be very difficult to approach data size limit while still control the maximum data size exactly. Instead, it's recommended to estimate your data size depending on storage duration and setup T... | 2019-12-15T21:33:16 | 2019-12-15T21:33:16 | {} | MEMBER |
ClickHouse | ClickHouse | 565,849,424 | 7,488 | alexey-milovidov | Yes, let's use `deserializeTextEscaped` for parameters.
We will only pay for extra escaping (that's required only for tabs, newlines and backslashes)
and we get support for \N for that price.
Don't forget to include as backward incompatible change into changelog.
The function `deserializeAsWholeText` will be ... | 2019-12-15T21:37:55 | 2019-12-15T21:38:29 | {} | MEMBER |
ClickHouse | ClickHouse | 565,850,464 | 7,327 | stavrolia | I couldn't reproduce improvement from this optimization and stucked | 2019-12-15T21:50:35 | 2019-12-15T21:50:35 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 565,884,260 | 2,455 | alexey-milovidov | @namikhnenko
@evillique
@kekekekule | 2019-12-16T02:52:09 | 2019-12-16T02:52:09 | {} | MEMBER |
ClickHouse | ClickHouse | 565,925,737 | 8,227 | onine007 | there are some dictionaries which start with broken_ ,ignored_ like ignored_20191203_902_907_1_908 and broken_20191022_17_17_0_try1 in detached dictionary | 2019-12-16T06:37:34 | 2019-12-16T06:37:34 | {} | NONE |
ClickHouse | ClickHouse | 565,905,619 | 8,151 | amosbird | > Maybe it's enough to only build some libraries with this option. libcxxabi maybe?
It's hard to tell which libs break libcxx unwinding, but libcxxabi is the least likely one as it's only used to unwinding others, not itself. (I've also tested with `-fno-omit-frame-pointer` in both libcxx and libcxxabi just in case)... | 2019-12-16T05:01:06 | 2019-12-16T06:52:21 | {} | COLLABORATOR |
ClickHouse | ClickHouse | 565,969,669 | 8,201 | alesapin | Cannot reproduce:
```
./clickhouse client --join_use_nulls=0
ClickHouse client version 19.19.1.1.
Connecting to localhost:9000 as user default.
Connected to ClickHouse server version 19.19.1 revision 54430.
click.sas.yp-c.yandex.net :) select 1
SELECT 1
┌─1─┐
│ 1 │
└───┘
1 rows in set. Elapsed: 0.00... | 2019-12-16T09:06:18 | 2019-12-16T09:06:18 | {} | MEMBER |
ClickHouse | ClickHouse | 566,042,129 | 7,900 | azat | Also there are problems with jemalloc statistics, there is `.` in it, i.e.:
```
ClickHouseAsyncMetricsjemalloc.background_thread.run_interval 0
ClickHouseAsyncMetricsjemalloc.background_thread.num_runs 0
ClickHouseAsyncMetricsjemalloc.background_thread.num_threads 0
ClickHouseAsyncMetricsjemalloc.metadata 663630... | 2019-12-16T12:33:13 | 2019-12-16T12:33:26 | {} | MEMBER |
ClickHouse | ClickHouse | 566,207,771 | 8,177 | stavrolia | Failed tests don't influenced by this PR | 2019-12-16T19:30:58 | 2019-12-16T19:30:58 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 566,170,472 | 8,223 | den-crane | I am not sure why `select final` is single threaded. Probably several threads process own set of parts. The same key happens in different parts. Threads don't see keys in other threads.
IMHO: PostgreSQL MVCC, and other ideas they just don't fit into the paradigm of MergeTree / immutable parts. | 2019-12-16T17:54:45 | 2019-12-16T20:13:15 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 565,782,975 | 7,970 | liukun4515 | ```
cmake .. -DCMAKE_CXX_COMPILER=`which clang++` -DCMAKE_C_COMPILER=`which clang`
```
is ok to compile project.
llvm --version
```
Apple LLVM version 10.0.1 (clang-1001.0.46.4)
Target: x86_64-apple-darwin18.6.0
Thread model: posix
InstalledDir: /Library/Developer/CommandLineTools/usr/bin
``` | 2019-12-15T06:44:06 | 2019-12-15T06:44:59 | {} | NONE |
ClickHouse | ClickHouse | 565,783,042 | 8,141 | liukun4515 | I meet the same problem for [#7970 ](https://github.com/ClickHouse/ClickHouse/issues/7970), and test
```
cmake .. -DCMAKE_CXX_COMPILER=`which clang++` -DCMAKE_C_COMPILER=`which clang`
```
in my mac.
I can compile the project successfully using the clang compiler in mac.
| 2019-12-15T06:46:14 | 2019-12-15T06:47:16 | {} | NONE |
ClickHouse | ClickHouse | 565,877,204 | 8,218 | alexey-milovidov | The issue is caused by wrong "CFA expression".
Something around line 809 in crypto/sha/asm/sha512-x86_64.pl.
I have disabled ASM version of SHA256.
Alternative solution is to remove all `.cfi_*` directives from assembly source code (btw, libressl doesn't have them at all). The cost is that stack traces will end on... | 2019-12-16T02:11:37 | 2019-12-16T02:11:37 | {} | MEMBER |
ClickHouse | ClickHouse | 566,065,432 | 8,228 | KochetovNicolai | It probably means that you have table with `MySQL` engine which can't connect to MySQL.
It also strange that we have `nullptr` in error message. May be a misconfiguration.
Can you please check that all you MySQL configurations are correct?
And also find full stacktrace after this error in logs? | 2019-12-16T13:39:58 | 2019-12-16T13:39:58 | {} | MEMBER |
ClickHouse | ClickHouse | 566,207,094 | 8,151 | alexey-milovidov | > clang build with libcxx is not performance centric
Right now it is, but we keep the following in mind:
1. We want to enable libcxx also for gcc.
2. clang looks to keep up with gcc soon and we will switch to clang whenever it happens.
3. We can debug unwinding and figure out the exact reason why it cannot work. | 2019-12-16T19:29:21 | 2019-12-16T19:29:21 | {} | MEMBER |
ClickHouse | ClickHouse | 566,215,571 | 8,141 | kolen | With this configuration I go further in compilation process, but then I get the following error:
```
[1194/4432] Building ASM object contrib/openssl-cmake/CMakeFiles/crypto.dir/__/openssl/crypto/aes/aesni-mb-x86_64.s.o
FAILED: contrib/openssl-cmake/CMakeFiles/crypto.dir/__/openssl/crypto/aes/aesni-mb-x86_64.s.o
/... | 2019-12-16T19:51:12 | 2019-12-16T19:51:12 | {} | NONE |
ClickHouse | ClickHouse | 566,414,544 | 8,228 | ashkank66 | 2019.12.17 10:46:30.000314 [ 44 ] {} <Information> Application: MYSQL: Connecting to database@host:0 as user user
2019.12.17 10:46:30.001630 [ 44 ] {} <Error> Application: mysqlxx::ConnectionFailed
2019.12.17 10:46:30.001943 [ 44 ] {} <Error> void DB::AsynchronousMetrics::run(): Poco::Exception. Code: 1000, e.code() ... | 2019-12-17T07:17:18 | 2019-12-17T07:17:18 | {} | NONE |
ClickHouse | ClickHouse | 566,414,655 | 8,027 | jiugem | @amosbird Our server is Huawei Taishan server. I am not a developer. I do n’t use GitHub. I try to test it. | 2019-12-17T07:17:40 | 2019-12-17T07:17:40 | {} | NONE |
ClickHouse | ClickHouse | 565,753,690 | 7,477 | alexey-milovidov | > Hi! I want to work on this as my first issue.
That's great!
> Am I correct that I should replace thread_number assignment logic with os_thread_id one (for example here)?
Yes.
> Firstly I thought I can change getThreadNumber, but have found that getThreadNumber used not only in logging as it said in getThr... | 2019-12-14T21:12:47 | 2019-12-14T21:13:26 | {} | MEMBER |
ClickHouse | ClickHouse | 565,760,503 | 8,218 | alexey-milovidov | Still have some troubles both with old and new version:
libressl:
```
~/ClickHouse/build_gcc9$ /usr/bin/clickhouse local --query="SELECT * FROM url('https://danluu.com/', TSV, 's String')"
SSL Exception: error:14004410:SSL routines:CONNECT_CR_SRVR_HELLO:sslv3 alert handshake failure
```
openssl:
```
~/Click... | 2019-12-14T22:58:47 | 2019-12-14T22:58:47 | {} | MEMBER |
ClickHouse | ClickHouse | 565,783,553 | 8,141 | liukun4515 | The Chinese docs should also be updated synchronously.
If Chinese is not easy for you, I would like to help you. | 2019-12-15T06:57:58 | 2019-12-15T06:57:58 | {} | NONE |
ClickHouse | ClickHouse | 565,849,188 | 7,435 | alexey-milovidov | https://github.com/ClickHouse/ClickHouse/issues/8167 | 2019-12-15T21:34:43 | 2019-12-15T21:34:43 | {} | MEMBER |
ClickHouse | ClickHouse | 565,903,920 | 8,223 | den-crane | >I was thinking that second proposed solution might be straightforward approach.
Isn't it the same as FINAL? `select ... from table FINAL` ? | 2019-12-16T04:51:04 | 2019-12-16T04:51:04 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 566,440,447 | 8,223 | krnaveen14 | I understand this is not something which is possible with MergeTree. That's why we are having this thread right now.
Just like how ReplacingMergeTree and *CollapsingMergeTree have come up based on the use cases, we need to look it from a different perspective rather than in a fixed perspective of it doesn't suit the... | 2019-12-17T08:38:52 | 2019-12-17T08:38:52 | {} | NONE |
ClickHouse | ClickHouse | 565,563,852 | 4,116 | alexey-milovidov | The same as #8056 | 2019-12-13T18:54:04 | 2019-12-13T18:54:04 | {} | MEMBER |
ClickHouse | ClickHouse | 565,855,328 | 7,900 | alexey-milovidov | But works perfectly. | 2019-12-15T22:49:26 | 2019-12-15T22:49:26 | {} | MEMBER |
ClickHouse | ClickHouse | 566,219,946 | 8,034 | stavrolia | Why don't we want to prefix C function from h3? | 2019-12-16T20:03:03 | 2019-12-16T20:03:03 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 576,582,907 | 8,738 | hhell | > Possible alternative in 'general' case (for all formats): #6294
Yes, that one is most relevant; it is even preferable to the example above; but it is not exactly 'for all formats': serialization for all the events can vary (schema / progress / totals / statistics / summary / error / ...). | 2020-01-21T08:59:40 | 2020-01-21T09:00:25 | {} | CONTRIBUTOR |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.