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 | 514,450,361 | 6,124 | alexey-milovidov | libunwind is not signal safe:
```
Thread 75 (Thread 0x7f840b57a700 (LWP 872742)):
#0 __lll_lock_wait () at ../sysdeps/unix/sysv/linux/x86_64/lowlevellock.S:135
#1 0x00007f856a766098 in __GI___pthread_mutex_lock (mutex=0x7f856adab990 <_rtld_global+2352>) at ../nptl/pthread_mutex_lock.c:113
#2 0x00007f8569f2ede... | 2019-07-24T02:09:25 | 2019-07-24T02:09:25 | {} | MEMBER |
ClickHouse | ClickHouse | 514,524,863 | 6,126 | alesapin | Seems like `tokenbf_v1` is not able to work with non-UTF-8/ASCII encoded strings. | 2019-07-24T08:01:23 | 2019-07-24T08:01:23 | {} | MEMBER |
ClickHouse | ClickHouse | 514,591,919 | 6,134 | akuzm | What version do you use?
On 19.11, I get maxIf(null, 1) = null, as expected.
```
:) select maxIf(null, 1);
┌─maxIf(NULL, 1)─┐
│ ᴺᵁᴸᴸ │
└────────────────┘
``` | 2019-07-24T11:21:52 | 2019-07-24T11:21:52 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 514,687,892 | 5,976 | 4ertus2 | It would be fixed in #6131 | 2019-07-24T15:45:15 | 2019-07-24T15:45:15 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 514,786,910 | 6,126 | alesapin | Integration test failure is not connected with code. | 2019-07-24T20:26:57 | 2019-07-24T20:26:57 | {} | MEMBER |
ClickHouse | ClickHouse | 514,865,206 | 4,903 | kindred77 | I think arrow:ORC is ok.
https://github.com/apache/arrow/blob/master/cpp/src/arrow/adapters/orc/adapter-test.cc | 2019-07-25T02:04:36 | 2019-07-25T02:04:36 | {} | NONE |
ClickHouse | ClickHouse | 515,542,728 | 6,181 | 4ertus2 | closes #6090 | 2019-07-26T17:49:59 | 2019-07-26T17:49:59 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 514,262,396 | 6,121 | alesapin | fix https://github.com/yandex/ClickHouse/pull/6126/files | 2019-07-23T15:31:43 | 2019-07-23T15:31:43 | {
"+1": 2
} | MEMBER |
ClickHouse | ClickHouse | 514,366,044 | 6,124 | alexey-milovidov | ```
0x000000000b8bc0b1 in libunwind::LocalAddressSpace::get64 (this=0x10471a08 <libunwind::LocalAddressSpace::sThisAddressSpace>, addr=18446744073709551576)
at ../contrib/libunwind/src/AddressSpace.hpp:203
203 memcpy(&val, (void *)addr, sizeof(val));
(gdb) bt
#0 0x000000000b8bc0b1 in libunwind::LocalA... | 2019-07-23T20:13:16 | 2019-07-23T20:13:16 | {} | MEMBER |
ClickHouse | ClickHouse | 514,416,551 | 6,124 | alexey-milovidov | Note that the address that libunwind wants to read (and it think that it's the value of `rbp` register) corresponds to the value of function argument a few stack frames above:
#0 0x000000001402be70 in libunwind::LocalAddressSpace::get64 (this=0x14175fc0 <libunwind::LocalAddressSpace::sThisAddressSpace>, addr=**5242... | 2019-07-23T23:11:32 | 2019-07-23T23:11:32 | {} | MEMBER |
ClickHouse | ClickHouse | 514,434,817 | 6,124 | alexey-milovidov | > Probably it will work if we limit DWARF version to 3 for ClickHouse build.
Not true. | 2019-07-24T00:46:01 | 2019-07-24T00:46:01 | {} | MEMBER |
ClickHouse | ClickHouse | 514,538,076 | 6,115 | masosky | Hi,
1. we don't need it
2. Currently, less than 10 but it seems that it will increase rapidly.
3. Yes, we will implement it for clients who have massive data
| 2019-07-24T08:41:05 | 2019-07-24T08:41:05 | {} | NONE |
ClickHouse | ClickHouse | 514,853,350 | 6,123 | alexey-milovidov | Integration tests: CI failure. | 2019-07-25T01:00:20 | 2019-07-25T01:00:20 | {} | MEMBER |
ClickHouse | ClickHouse | 515,663,565 | 6,139 | kreuzerkrieg | Thats it, `addr2line` AFAIR, we have used it once upon a time, but our project was way smaller than CH. Though I dont remember we had a problem with licensing, but what do I know about licensing.
Would you implement it?
| 2019-07-27T07:57:49 | 2019-07-27T07:57:49 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 514,050,297 | 6,031 | yurriy | I used --with-pdo-mysql=mysqlnd since you've sent a result of phpinfo (and it was notable in gdb that code from ext/mysqlnd is used). The error "MySQL server has gone away" occurs, but not with SSL options. Could you share a result of ``setrace -s 1000 -- php -f test.php`` with SSL options in test.php? | 2019-07-23T04:33:57 | 2019-07-23T04:34:22 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 514,510,499 | 6,130 | alesapin | @den-crane's answer is correct. Also we have plans to add query `ALTER TABLE xxx MODIFY SETTING replicated_deduplication_window=yyyy`. | 2019-07-24T07:14:14 | 2019-07-24T07:14:14 | {} | MEMBER |
ClickHouse | ClickHouse | 514,643,355 | 6,127 | Enmk | Regarding name: I personally don't think that `boundingBox` is a good choice, since that has a well-defined meaning. `boundingBoxes` is kind of confusing in that regard.
Given the rule that the more specific functions is, longer the name, maybe we shall rename that to `geohashCoverAreaWithBoxes()` ? | 2019-07-24T13:57:29 | 2019-07-24T13:57:29 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 514,722,626 | 6,128 | millin | Why you need to build it?
All testing releases are available https://repo.yandex.ru/clickhouse/deb/testing/main/ | 2019-07-24T17:20:21 | 2019-07-24T17:20:21 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 515,429,650 | 6,168 | nvartolomei | Isn’t this similar to what replace partition already does? | 2019-07-26T12:22:05 | 2019-07-26T12:22:05 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 514,199,584 | 5,815 | alesapin | ```
# yum install yum-utils
# yum-config-manager --add-repo https://repo.yandex.ru/clickhouse/rpm/prestable/x86_64
# yum install clickhouse-server clickhouse-client
added from: https://repo.yandex.ru/clickhouse/rpm/prestable/x86_64 ... | 2019-07-23T13:01:24 | 2019-07-23T13:01:24 | {} | MEMBER |
ClickHouse | ClickHouse | 514,506,801 | 6,105 | Imomoi | @alexey-milovidov We cannot reproduce this bug on our test servers (neither 19.9 or 19.11).
We are currently testing 19.11 before install it on our release servers. | 2019-07-24T07:00:30 | 2019-07-24T07:00:30 | {} | NONE |
ClickHouse | ClickHouse | 514,820,735 | 6,132 | alexey-milovidov | > Build failed without build log. It's inner CI problem. Maybe something wrong with docker or host. Just wait for build restart. | 2019-07-24T22:18:37 | 2019-07-24T22:18:37 | {} | MEMBER |
ClickHouse | ClickHouse | 515,016,643 | 6,149 | 4ertus2 | @BayoNet Предлагаю взять этот PR за основу, закрыть его и сделать правильный. Готов проконсультировать в чатике или лично. | 2019-07-25T12:03:21 | 2019-07-25T12:03:21 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 515,435,096 | 6,149 | BayoNet | Да. Давай так и сделаем. | 2019-07-26T12:40:55 | 2019-07-26T12:40:55 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 515,626,194 | 6,166 | alexey-milovidov | Yes, we can enable it for gcc. (You can send a PR or I'll try not forget to do it myself...)
PS. Why it isn't enabled right now? - We use both gcc and clang in our CI and after we enabled maximum possible warnings for clang (inside `dbms` directory with our main code base), there is less motivation to add them in gc... | 2019-07-26T23:16:28 | 2019-07-26T23:18:25 | {} | MEMBER |
ClickHouse | ClickHouse | 514,858,572 | 6,026 | alexey-milovidov | @abyss7 ? | 2019-07-25T01:29:44 | 2019-07-25T01:29:44 | {} | MEMBER |
ClickHouse | ClickHouse | 514,344,746 | 6,124 | alexey-milovidov | In gdb type
```
handle SIGUSR1 noprint
handle SIGUSR2 noprint
```
to not stop on profiling signals. | 2019-07-23T19:13:21 | 2019-07-23T19:14:48 | {} | MEMBER |
ClickHouse | ClickHouse | 514,404,818 | 6,115 | den-crane | 1. Do you need user access isolation? CH does not have table level grants only database level grants.
2. How many clients do you have?
3. Do you plan to use replicated tables (CH replicated engine has some resource overhead for each replicated table).
| 2019-07-23T22:19:35 | 2019-07-23T22:19:35 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 514,748,293 | 6,128 | alexey-milovidov | Better to do full recursive checkout from GitHub, then pack everything in tarball. Then you can build ClickHouse from this tarball.
> I am able to mostly work around this by installing various libs through the system package manager
Don't do it. With libraries from OS packages ClickHouse will work in a slightly d... | 2019-07-24T18:32:34 | 2019-07-24T18:32:34 | {} | MEMBER |
ClickHouse | ClickHouse | 515,269,233 | 6,103 | alexey-milovidov | https://github.com/yandex/ClickHouse/pull/6164/commits/e5df077c21a339338cbf962b3d8e5eb1f67415d7
- documentation for ClickHouse DBA in default server config;
https://github.com/yandex/ClickHouse/pull/6164/commits/2a9d606d4b6ec50a079591ca28d57b020c5e02f3
- enable text log when you run clickhouse-server from sour... | 2019-07-26T01:03:39 | 2019-07-26T01:03:39 | {} | MEMBER |
ClickHouse | ClickHouse | 514,158,858 | 6,120 | TCeason | I also think this is not appropriate. Can you recommend some more elegant ways?
---Original---
From: "Alex Krash"<notifications@github.com>
Date: Tue, Jul 23, 2019 18:33 PM
To: "yandex/ClickHouse"<ClickHouse@noreply.github.com>;
Cc: "Author"<author@noreply.github.com>;"TCeason"<tai_chong@foxmail.com>;
S... | 2019-07-23T10:50:18 | 2019-07-23T10:50:18 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 514,368,501 | 6,124 | alexey-milovidov | ```
#0 0x000000000b8bc0b1 in libunwind::LocalAddressSpace::get64 (this=0x10471a08 <libunwind::LocalAddressSpace::sThisAddressSpace>, addr=18446744073709551576)
at ../contrib/libunwind/src/AddressSpace.hpp:203
val = 19721751088857089
#1 0x000000000b8bc1b1 in libunwind::LocalAddressSpace::getRegister (t... | 2019-07-23T20:20:44 | 2019-07-23T20:20:44 | {} | MEMBER |
ClickHouse | ClickHouse | 514,858,966 | 6,016 | alexey-milovidov | Waiting for Sandbox to become alive. | 2019-07-25T01:31:55 | 2019-07-25T01:31:55 | {} | MEMBER |
ClickHouse | ClickHouse | 515,000,177 | 6,145 | akuzm | Please try:
```
git submodule sync
git submodule update --init --recursive
``` | 2019-07-25T11:02:03 | 2019-07-25T11:02:03 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 515,574,695 | 6,128 | knusbaum | @millin I want to build for a linux other than debian. It's helpful to have a tarball for a release with a full set of sources, to create reproducible builds.
https://repo.yandex.ru/clickhouse/deb/testing/main/ looks like it had something similar to what I want, except there aren't any tarballs beyond 18.6.0. | 2019-07-26T19:37:16 | 2019-07-26T19:37:16 | {} | NONE |
ClickHouse | ClickHouse | 515,631,840 | 6,103 | alexey-milovidov | Also please add `function` column (from `__PRETTY_FUNCTION__` macro in a location of `LOG_*`) | 2019-07-26T23:57:01 | 2019-07-26T23:57:01 | {} | MEMBER |
ClickHouse | ClickHouse | 514,163,781 | 6,104 | 4ertus2 | We can replace currentDatabase() call with actual value in AddDefaultDatabaseVisitor. | 2019-07-23T11:05:54 | 2019-07-23T11:05:54 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 514,296,624 | 6,127 | alexey-milovidov | If it returns multiple boxes, maybe `*Boxes` is a better name? | 2019-07-23T16:59:59 | 2019-07-23T16:59:59 | {
"+1": 1
} | MEMBER |
ClickHouse | ClickHouse | 514,434,389 | 6,124 | alexey-milovidov | Probably it will work if we limit DWARF version to 3 for ClickHouse build. | 2019-07-24T00:43:49 | 2019-07-24T00:43:49 | {} | MEMBER |
ClickHouse | ClickHouse | 514,441,513 | 6,124 | alexey-milovidov | Deadlock is possible if:
- signal handler is invoked while malloc mutex is hold;
- the trace info pipe becomes full because there are many traces. | 2019-07-24T01:22:50 | 2019-07-24T01:22:50 | {} | MEMBER |
ClickHouse | ClickHouse | 514,122,037 | 6,109 | masosky | As far as I know, the answer is yes.
In order to synchronize replication, you must have a Zookeeper instance. Clickhouse recommends having at least 3 instances of Zookeeper in production.
Here, you can find a link that explain hard concepts in an easy way. Furthermore, you maybe find some answers about replication:
... | 2019-07-23T09:04:32 | 2019-07-23T09:04:32 | {
"+1": 1,
"heart": 1
} | NONE |
ClickHouse | ClickHouse | 514,592,745 | 6,134 | CAspeling | ClickHouse 19.5.3.8 | 2019-07-24T11:25:00 | 2019-07-24T11:25:00 | {} | NONE |
ClickHouse | ClickHouse | 515,192,008 | 2,232 | scottlepp | brew tap arduanov/clickhouse
brew install gcc@8
brew install --cc=gcc-8 clickhouse
You many encounter dependency errors during the above install. If a dependency fails to install, identify that dependency in the error and:
1. brew install [dependency] (without the --cc flag).
2. brew install --cc=gcc-8 clickho... | 2019-07-25T19:58:29 | 2019-07-25T19:59:00 | {} | NONE |
ClickHouse | ClickHouse | 514,346,086 | 6,124 | alexey-milovidov | It still can cause segfault:
```
Thread 65 "ParalInputsProc" received signal SIGSEGV, Segmentation fault.
[Switching to Thread 0x7fff232fa700 (LWP 565980)]
0x000000000b900c1d in libunwind::DwarfInstructions<libunwind::LocalAddressSpace, libunwind::Registers_x86_64>::stepWithDwarf (addressSpace=..., pc=<optimized ... | 2019-07-23T19:17:00 | 2019-07-23T19:17:00 | {} | MEMBER |
ClickHouse | ClickHouse | 514,415,258 | 6,124 | alexey-milovidov | It happens not only inside malloc, also in normal C++ code:
```
#0 0x000000001402be70 in libunwind::LocalAddressSpace::get64 (this=0x14175fc0 <libunwind::LocalAddressSpace::sThisAddressSpace>, addr=524288) at ../contrib/libunwind/src/AddressSpace.hpp:203
#1 0x000000001402cecd in libunwind::LocalAddressSpace::getRe... | 2019-07-23T23:05:24 | 2019-07-23T23:05:24 | {} | MEMBER |
ClickHouse | ClickHouse | 514,568,318 | 6,080 | abyss7 | > All ZooKeeper tests have timed out: https://nda.ya.ru/3VmJmN
That's because of `StorageReplicatedMergeTree::startup()`, which waits for some event from task dispatched to `BackgroundSchedulePool`. Since nobody knows what will happen if I remove this conditional wait, I dropped the idea to postpone `BackgroundSched... | 2019-07-24T10:03:51 | 2019-07-24T10:03:51 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 514,853,304 | 6,123 | alexey-milovidov | drop_truncate_memory_table
replace_partition_from_table_zookeeper
- flappy tests. | 2019-07-25T01:00:05 | 2019-07-25T01:00:05 | {} | MEMBER |
ClickHouse | ClickHouse | 514,856,268 | 6,080 | alexey-milovidov | Ok. (Debug output in integration test - does not relate to this pr - up to you). | 2019-07-25T01:17:19 | 2019-07-25T01:17:19 | {} | MEMBER |
ClickHouse | ClickHouse | 515,199,462 | 6,160 | olegkv | I tried single quotes like '1test' and it did not work, did not try double quotes.
Using double quotes solves it, but it is still sort of inconvenient because tables always need to be referenced like "1test".table1 instead of 1test.table1, but that's minor.
Thanks!
| 2019-07-25T20:21:29 | 2019-07-25T20:21:29 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 514,134,532 | 5,894 | blinkov | @alex-krash ping? | 2019-07-23T09:38:50 | 2019-07-23T09:38:50 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 514,226,424 | 6,109 | filimonov | Yes. To have reliable replication you need to install zookeeper & set internal_replication flag to true.
Please read official documentation
https://clickhouse.yandex/docs/en/operations/table_engines/replication/
You can also check:
https://blog.uiza.io/replicated-vs-distributed-on-clickhouse-part-1/
https://... | 2019-07-23T14:09:23 | 2019-07-23T14:09:23 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 514,235,547 | 6,105 | filimonov | > ```
> 2019.07.22 12:46:00.869158 [ 1560 ] {} <Error> BaseDaemon: ########################################
> 2019.07.22 12:46:00.869244 [ 1560 ] {} <Error> BaseDaemon: (version 19.9.3.31 (official build)) (from thread 1519) Received signal Segmentation fault (11).
> ```
Can you please try to find what was happen... | 2019-07-23T14:30:48 | 2019-07-23T14:32:01 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 514,405,425 | 6,130 | den-crane | https://github.com/yandex/ClickHouse/issues/2607#issuecomment-403454286 | 2019-07-23T22:21:57 | 2019-07-23T22:21:57 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 514,496,449 | 4,785 | denisgolius | > * RBAC, LDAP: moved to spring
Spring 2020 ? | 2019-07-24T06:19:58 | 2019-07-24T06:19:58 | {} | NONE |
ClickHouse | ClickHouse | 514,668,105 | 6,127 | alexey-milovidov | > Regarding name: I personally don't think that `boundingBox` is a good choice, since that has a well-defined meaning. `boundingBoxes` is kind of confusing in that regard.
>
> Given the rule that the more specific functions is, longer the name, maybe we shall rename that to `geohashCoverAreaWithBoxes()` ?
Ok. | 2019-07-24T14:57:38 | 2019-07-24T14:57:38 | {} | MEMBER |
ClickHouse | ClickHouse | 515,020,540 | 6,080 | millin | I hope this change fixes #6071 and it will be released soon, because currently it is impossible to use kafka + mv in 19.11. | 2019-07-25T12:17:29 | 2019-07-25T12:17:29 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 515,441,618 | 6,127 | alex-zaitsev | geohashCoverAreaWithBoxes() seems confusing to me in 'WithBoxes' part. It assumes there are multiple boxes and we do something with them. But there is single 'box argument'. Bounding Box is a commonly used term. Probably we could name it geohashArrayFromBoundingBox() if you want a long descriptive name? | 2019-07-26T13:01:14 | 2019-07-26T13:01:14 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 515,570,350 | 5,984 | ramazanpolat | > We can use least common supertype (the same way as it works for arrays).
Determine column type from first row.
If it is not explicit, use supertype:
```sql
VALUES (1, 2.3, 'one', '2019-01-02', '2019-01-02 01:02:03)),
(2, 3.4, 'two', '2019-01-03', '2019-01-02 01:02:04'),
(3, 4.5, 'three', '201... | 2019-07-26T19:21:35 | 2019-07-26T19:21:35 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 515,660,197 | 6,115 | den-crane | CH replicated tables have some CPU/Zookeeper overhead and recommended to keep number of replicated tables less than 1000. | 2019-07-27T07:17:41 | 2019-07-27T07:17:41 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 514,132,119 | 6,096 | abyss7 | It looks very similar to #5846 which is in progress. | 2019-07-23T09:32:11 | 2019-07-23T09:32:11 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 514,158,773 | 365 | jbwl | Same error happens on Mac OS 10.14.6 trying to build from master branch with these instructions: https://clickhouse.yandex/docs/en/development/build_osx/#caveats | 2019-07-23T10:50:02 | 2019-07-23T10:50:02 | {} | NONE |
ClickHouse | ClickHouse | 514,296,304 | 6,127 | alexey-milovidov | Is it officially named "cover box"?
Maybe "bounding box" is a better name? | 2019-07-23T16:59:09 | 2019-07-23T16:59:09 | {
"+1": 1
} | MEMBER |
ClickHouse | ClickHouse | 514,415,578 | 6,124 | alexey-milovidov | ```
#0 0x000000001402be70 in libunwind::LocalAddressSpace::get64 (this=0x14175fc0 <libunwind::LocalAddressSpace::sThisAddressSpace>, addr=524288) at ../contrib/libunwind/src/AddressSpace.hpp:203
val = 0
#1 0x000000001402cecd in libunwind::LocalAddressSpace::getRegister (this=0x14175fc0 <libunwind::LocalAdd... | 2019-07-23T23:06:57 | 2019-07-23T23:06:57 | {} | MEMBER |
ClickHouse | ClickHouse | 514,503,302 | 6,052 | 4ertus2 | There’s 50/50 probability that it works with commited #6043 without any modifications needed. It takes lots of time to deploy real distributed configuration. It would be great if you could test the issue with newer version with #6043 merged. Could you? | 2019-07-24T06:47:02 | 2019-07-24T06:47:02 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 514,855,770 | 6,108 | alexey-milovidov | TSan report: fixed in master. | 2019-07-25T01:14:28 | 2019-07-25T01:14:28 | {} | MEMBER |
ClickHouse | ClickHouse | 515,004,931 | 6,148 | blinkov | cc @zhang2014 @sundy-li | 2019-07-25T11:19:18 | 2019-07-25T11:19:18 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 515,172,328 | 6,160 | alexey-milovidov | You have to quote database name like this:
```
CREATE DATABASE `1test`
```
or this:
```
CREATE DATABASE "1test"
```
| 2019-07-25T18:59:40 | 2019-07-25T18:59:40 | {} | MEMBER |
ClickHouse | ClickHouse | 515,199,124 | 6,157 | alexey-milovidov | It can be used for every kind of MergeTree (Aggregating, Graphite, ...) except plain MergeTree. | 2019-07-25T20:20:30 | 2019-07-25T20:20:30 | {} | MEMBER |
ClickHouse | ClickHouse | 515,252,925 | 6,103 | alexey-milovidov | Please pull changes from my branch: #6164
TODO:
- source file and line don't work at all; you have to add this information via LOG_ macros;
- it logs only events related to queries and merges (where ThreadGroup exists). But we need all logs from some point after startup to some point before shutdown. Maybe better ... | 2019-07-25T23:36:08 | 2019-07-25T23:36:36 | {} | MEMBER |
ClickHouse | ClickHouse | 515,261,384 | 6,124 | alexey-milovidov | TODO: timer offset randomization. For example, we may want to set up timer interval of one second but be able to sample some queries with duration less than a second. | 2019-07-26T00:20:10 | 2019-07-26T00:20:10 | {} | MEMBER |
ClickHouse | ClickHouse | 515,482,208 | 5,984 | akuzm | The syntax is unusual in that it uses table definition inside a string literal. This also makes it somewhat troublesome to parse. Can we determine column types from the first row? It would also be consistent with the above array example. | 2019-07-26T14:45:04 | 2019-07-26T14:45:04 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 514,057,481 | 6,031 | yurriy | From capability flags in strace result it seems that you've started ClickHouse without SSL support. In this case client behaves the same way as without SSL options. In order to use SSL you should generate certificateFile, privateKeyFile and dhParamsFile using commands from [config.xml](https://github.com/yandex/ClickHo... | 2019-07-23T05:16:34 | 2019-07-23T05:16:34 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 514,070,276 | 6,031 | foolfish3 | OK, thank you. yes i do start a server not support ssl. i will generate ssl file next time.
can you look into https://github.com/yandex/ClickHouse/issues/6047
is it your bug? | 2019-07-23T06:19:53 | 2019-07-23T06:19:53 | {} | NONE |
ClickHouse | ClickHouse | 514,152,669 | 5,894 | alex-krash | @blinkov , sorry for delay. | 2019-07-23T10:29:22 | 2019-07-23T10:29:22 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 514,243,316 | 6,124 | alexey-milovidov | Usage example:
```
SELECT count(), arrayStringConcat(arrayMap(x -> symbolizeAddress(x), trace), '\n') AS trace FROM system.trace_log GROUP BY trace ORDER BY count() DESC LIMIT 10 FORMAT Vertical
``` | 2019-07-23T14:48:18 | 2019-07-23T14:48:18 | {} | MEMBER |
ClickHouse | ClickHouse | 514,339,619 | 6,124 | alexey-milovidov | Now we can get usable results:
```
SELECT
count(),
arrayStringConcat(arrayMap(x -> symbolizeAddress(x), trace), '\n') AS trace
FROM system.trace_log
GROUP BY trace
ORDER BY count() DESC
LIMIT 10
FORMAT Vertical
Row 1:
──────
count(): 2767
trace: StackTrace::StackTrace(ucontext_t const&)
... | 2019-07-23T18:58:58 | 2019-07-23T18:58:58 | {} | MEMBER |
ClickHouse | ClickHouse | 514,358,763 | 6,131 | 4ertus2 | There's no fix yet. Just a refactoring. | 2019-07-23T19:52:43 | 2019-07-23T19:52:43 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 514,593,053 | 6,134 | CAspeling | Here is a tester query you can use:
```
select
id, maxIf(value, id == 1) as max
from (
select
number as id,
number * 2 as value
from
numbers(5)
) group by id
``` | 2019-07-24T11:26:03 | 2019-07-24T11:26:03 | {} | NONE |
ClickHouse | ClickHouse | 514,759,095 | 6,124 | alexey-milovidov | TODO:
- thread_id in trace_log;
- revision in trace_log; | 2019-07-24T19:02:04 | 2019-07-24T19:02:04 | {} | MEMBER |
ClickHouse | ClickHouse | 514,774,905 | 6,139 | alexey-milovidov | Also it's rather easy to add file and line of exact location of `throw` but it isn't easy for every call stack frame. | 2019-07-24T19:50:19 | 2019-07-24T19:50:19 | {} | MEMBER |
ClickHouse | ClickHouse | 514,819,961 | 6,126 | alexey-milovidov | > Seems like tokenbf_v1 is not able to work with non-UTF-8/ASCII encoded strings.
True. | 2019-07-24T22:15:58 | 2019-07-24T22:15:58 | {} | MEMBER |
ClickHouse | ClickHouse | 515,193,121 | 6,146 | alexey-milovidov | Fixed in #6156 | 2019-07-25T20:01:39 | 2019-07-25T20:01:39 | {} | MEMBER |
ClickHouse | ClickHouse | 515,259,005 | 6,103 | nikitamikhaylov | What are the changes from your branch aimed at? I’ve already solved the problem with text_log setting in server’s config. | 2019-07-26T00:10:35 | 2019-07-26T00:10:35 | {} | MEMBER |
ClickHouse | ClickHouse | 515,625,753 | 6,139 | alexey-milovidov | We probably can create some index on build time to quickly find file and line from address.
(It won't work for dlopened libraries, but it's fine)
PS. Relevant links:
1. https://github.com/gimli-rs/addr2line/issues/2 https://lib.rs/crates/addr2line
2. https://lore.kernel.org/patchwork/patch/413455/ (we can't use ... | 2019-07-26T23:13:51 | 2019-07-26T23:13:51 | {
"+1": 1
} | MEMBER |
ClickHouse | ClickHouse | 514,775,307 | 6,139 | kreuzerkrieg | Hm... slow search is a good reason to drop that feature... how about google's breakpad? maybe they know to do it in a fast fashion, somehow? | 2019-07-24T19:51:35 | 2019-07-24T19:56:45 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 515,429,580 | 5,932 | projectsbird123 | @filimonov Nothing is printed on the logs. | 2019-07-26T12:21:47 | 2019-07-26T12:21:47 | {} | NONE |
ClickHouse | ClickHouse | 515,533,737 | 5,417 | akuzm | Now that sso is enabled by default, it's strange that we don't see improvement in performance test.
There are also some regressions on queries like `SELECT URL, count() AS c FROM hits_100m_single GROUP BY URL ORDER BY c DESC LIMIT 10`: https://clickhouse-test-reports.s3.yandex.net/5417/11f4f571f9c6e04bbbb3b005ac2f51... | 2019-07-26T17:21:17 | 2019-07-26T17:21:17 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 514,173,912 | 5,976 | 4ertus2 | ```
create table t1 (a UInt32, b String) engine = Memory;
create table t2 (c UInt32, d String) engine = Memory;
select * from t1 join t2 on c == a;
DB::Exception: Missing columns: 'a' while processing query: 'a', required columns: 'a', source columns: 'd' 'c'.
``` | 2019-07-23T11:40:08 | 2019-07-23T11:40:08 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 514,408,774 | 6,124 | alexey-milovidov | It's also reproduced: in debug build; in build with clang; without jemalloc. | 2019-07-23T22:36:14 | 2019-07-23T22:36:14 | {} | MEMBER |
ClickHouse | ClickHouse | 514,592,401 | 6,134 | akuzm | ```
:) select maxIf(null, 0);
┌─maxIf(NULL, 0)─┐
│ ᴺᵁᴸᴸ │
└────────────────┘
```
On the other hand, this looks incorrect, because the condition is False, so maxIf shouldn't see the Null and should return a default value. | 2019-07-24T11:23:36 | 2019-07-24T11:23:36 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 514,774,501 | 6,139 | alexey-milovidov | It is possible to implement, but it requires either:
1. Calling a separate program: `addr2line` or `llvm-symbolizer` or `gdb` or `lldb`. We would either check that the program is installed in system and use it if available or add a dependency for our packages (worse).
2. Directly using LLVM code as `llvm-symboliz... | 2019-07-24T19:49:08 | 2019-07-24T19:51:00 | {} | MEMBER |
ClickHouse | ClickHouse | 515,016,165 | 6,146 | dimarub2000 | Many thanks for your fix. Tests will be added.
UPD: Bug Fixed but in another way | 2019-07-25T12:01:24 | 2019-07-25T16:06:01 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 515,197,041 | 6,163 | alexey-milovidov | This issue was introduced in https://github.com/yandex/ClickHouse/pull/6132 that was merged without CI checks, because there was a window of time when CI was not working (issues with internal "Sandbox" system). | 2019-07-25T20:14:05 | 2019-07-25T20:14:05 | {} | MEMBER |
ClickHouse | ClickHouse | 515,461,712 | 6,168 | YiuRULE | @nvartolomei `REPLACE` replace the partition of the destination table from a source table. But the source table partition is not changed.
For example, on `system.parts` on the test I wrote, you would have 3 partions, one partition in destination table, two partitions in source table.
However the potential `MOVE` ... | 2019-07-26T13:52:32 | 2019-07-26T14:12:11 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 515,486,570 | 5,984 | alexey-milovidov | I've already thinking about it.
> The syntax is unusual in that it uses table definition inside a string literal.
It is consistent with file, url, hdfs table functions.
> Can we determine column types from the first row?
No. For example: `(1, 'one'), (256, 'two')` - will derive UInt8 from first row.
> It... | 2019-07-26T14:57:04 | 2019-07-26T14:57:04 | {} | MEMBER |
ClickHouse | ClickHouse | 514,128,121 | 6,052 | millin | @4ertus2 Did you reproduce this issue? Or you need some additional info? | 2019-07-23T09:21:23 | 2019-07-23T09:21:23 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 514,506,137 | 6,105 | Imomoi | It looks like it was insert command - because insert commands (we are using Clickhouse.Ado for dotNet client) often produce this kind of log records.
Last output by thread 1519 was:
2019.07.14 23:40:58.711455 [ 1519 ] {f86efa41-f1b7-4546-8d20-25ad9e6a4877} <Information> TCPHandler: Processed in 0.039 sec.
... | 2019-07-24T06:57:59 | 2019-07-24T07:03:19 | {} | NONE |
ClickHouse | ClickHouse | 514,510,979 | 6,052 | millin | Looks like #6043 included in [v19.12.2.2-prestable](https://github.com/yandex/ClickHouse/releases/tag/v19.12.2.2-prestable).
I will try to check by installing it. | 2019-07-24T07:16:01 | 2019-07-24T07:16:01 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 515,183,160 | 6,124 | alexey-milovidov | `00568_compile_catch_throw` failed as expected. | 2019-07-25T19:32:01 | 2019-07-25T19:32:01 | {} | MEMBER |
ClickHouse | ClickHouse | 504,300,146 | 5,697 | zhang2014 | fixed in #5704 . BTW, this is because MySQL table contains a column and it's named float | 2019-06-21T05:55:41 | 2019-06-21T05:55:41 | {} | CONTRIBUTOR |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.