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 | 632,723,491 | 2,352 | sundy-li | > What do these a1, c1, a2, c2... parameters represent
It's just a simple demo equation. In your case, if you want to sort the events by sequential comparison by (event_time, pid), you can create an equation based on your data. | 2020-05-22T14:32:46 | 2020-05-22T14:32:46 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 633,013,587 | 11,154 | alexey-milovidov | Not good enough exception message (usability issue). | 2020-05-23T09:24:34 | 2020-05-23T09:24:34 | {} | MEMBER |
ClickHouse | ClickHouse | 633,019,736 | 10,818 | alexey-milovidov | > will lead to sustained memory usage of 26 GiB despite the fact that the query only takes 4..5 GiB.
It was lowered to 13 GiB on current master. Still higher than needed. | 2020-05-23T10:11:15 | 2020-05-23T10:11:15 | {
"eyes": 1
} | MEMBER |
ClickHouse | ClickHouse | 633,062,228 | 10,973 | alexey-milovidov | Does it support generating tables that are suitable for multiple SELECT queries? | 2020-05-23T14:16:03 | 2020-05-23T14:16:03 | {} | MEMBER |
ClickHouse | ClickHouse | 633,296,085 | 10,931 | alexey-milovidov | Submodules were changed accidentially, need to revert them back to what we have in master. | 2020-05-24T20:42:49 | 2020-05-24T20:42:49 | {} | MEMBER |
ClickHouse | ClickHouse | 632,640,469 | 9,545 | And42 | @akuzm @alexey-milovidov Is there any way to suppress a specific style warning? Having double (and even tripple) whitespaces in the comment as an example of a valid value: https://clickhouse-test-reports.s3.yandex.net/9545/e3b756e65f969fa0be5f69b71eb501da2e6fbfe6/style_check/output.txt
P.S. apart from that this requ... | 2020-05-22T11:20:27 | 2020-05-22T11:23:45 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 632,958,485 | 11,154 | akuzm | @qoega FYI | 2020-05-23T00:56:33 | 2020-05-23T00:56:33 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 633,312,435 | 11,163 | azat | >Does it perform a test at runtime that this feature is available in kernel?
No (will take this into account, thanks) | 2020-05-24T22:53:02 | 2020-05-24T22:53:02 | {} | MEMBER |
ClickHouse | ClickHouse | 633,325,382 | 11,153 | azat | > I keeped the server opened for few hours without sending any query and seems that counter didn't reseted. All queries are still failing..
Have you installed the server setting `max_server_memory_usage_to_ram_ratio` from the comment?
And what RSS for the clickhouse-server process? | 2020-05-25T00:40:09 | 2020-05-25T00:40:18 | {} | MEMBER |
ClickHouse | ClickHouse | 633,474,902 | 11,173 | filimonov | https://github.com/ClickHouse/ClickHouse/issues/9637 | 2020-05-25T09:24:13 | 2020-05-25T09:24:13 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 632,694,463 | 11,138 | alexey-milovidov | https://launchpad.net/~ubuntu-toolchain-r/+archive/ubuntu/test
Status: building (3 days ago). | 2020-05-22T13:37:13 | 2020-05-22T13:37:13 | {} | MEMBER |
ClickHouse | ClickHouse | 632,828,766 | 11,106 | alesapin | All affected tests passed. | 2020-05-22T17:52:56 | 2020-05-22T17:52:56 | {} | MEMBER |
ClickHouse | ClickHouse | 633,275,600 | 11,153 | silviucpp | @azat I keeped the server opened for few hours without sending any query and seems that counter didn't reseted. All queries are still failing.. | 2020-05-24T18:46:03 | 2020-05-24T18:46:03 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 633,349,008 | 11,153 | silviucpp | I added max_server_memory_usage_to_ram_ratio and this fixed all problems. But without this once the problems appears the app becomes unusable even if the clickhouse-server process doesn't have more than 300 MB of memory on a 16 GB server... Even if you wait 1 hour will not become usable till you restart. | 2020-05-25T02:44:31 | 2020-05-25T02:45:10 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 632,865,231 | 11,140 | akuzm | I'll close this one as a duplicate. | 2020-05-22T19:10:03 | 2020-05-22T19:10:03 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 632,945,702 | 11,147 | azat | Actually it is possible already (#9919):
From the [doc](https://github.com/ClickHouse/ClickHouse/blob/master/docs/en/operations/system-tables.md#systemtables-system-tables) (cannot use link to the official documentation since anchor for `system.tables` broken by the `system tables` - article header):
- `total_r... | 2020-05-22T23:45:52 | 2020-05-22T23:45:52 | {
"+1": 3
} | MEMBER |
ClickHouse | ClickHouse | 633,211,390 | 11,149 | azat | Also seems that now the upstream cppkafka can be used (your PR is merged, and other build fixes already there too) | 2020-05-24T10:39:49 | 2020-05-24T10:39:49 | {} | MEMBER |
ClickHouse | ClickHouse | 632,972,980 | 10,917 | huang--yijian | > I don't really understand what is the problem here. CH able to use all server's memory. What is the difference is it used all the time or only part of time?
>
> > Is it possible for users to configure ClickHouse to use an alternative memory allocator, so that memory is freed earlier?
>
> Old versions of CH used... | 2020-05-23T02:35:28 | 2020-05-23T02:38:19 | {
"+1": 1
} | NONE |
ClickHouse | ClickHouse | 633,168,028 | 10,917 | huang--yijian | > I re-read a thread. So every 6 hours the query eat +30GB, and after 2 days memory usage drops itself? Is it right?
Yes, you are right.
"every 6 hours the query eat +30GB" is caused by every 6 hour big application queries, resolved.
"after 2 days memory usage drops" is not expected, too late because most har... | 2020-05-24T02:34:37 | 2020-05-24T02:35:38 | {} | NONE |
ClickHouse | ClickHouse | 633,207,564 | 11,137 | alexey-milovidov | Squashing is Ok. | 2020-05-24T10:04:08 | 2020-05-24T10:04:08 | {} | MEMBER |
ClickHouse | ClickHouse | 633,309,484 | 11,137 | alexey-milovidov | Looks all right but "Arcadia" build has to be fixed separately. | 2020-05-24T22:24:59 | 2020-05-24T22:24:59 | {} | MEMBER |
ClickHouse | ClickHouse | 632,616,855 | 11,084 | alexey-milovidov | > but my main concern about affecting mixed payload (performance tests are a little bit synthetic I guess), but maybe it will works well
Let's be optimistic. | 2020-05-22T10:20:12 | 2020-05-22T10:20:12 | {} | MEMBER |
ClickHouse | ClickHouse | 632,617,039 | 11,042 | simPod | Hi, what version is this targeted for? I'd love to try it as soon as it's released, thanks! | 2020-05-22T10:20:45 | 2020-05-22T10:20:45 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 632,911,426 | 11,146 | silviucpp | My initial conclusion seems wrong. flush can happened for other reasons. Using `wait_end_of_query` provides accurate results into the stats. Without this setting the stats are wrong. Is this expected ? | 2020-05-22T21:15:30 | 2020-05-22T21:15:30 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 632,970,007 | 11,144 | akuzm | x300 speedup is really something 😂 | 2020-05-23T02:09:55 | 2020-05-23T02:09:55 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 632,977,080 | 11,139 | filimonov | Maybe we should have 2 variants of config in distribution? Config.xml (as now) and config.xml.optimized (mostly for prod setup)? | 2020-05-23T03:16:48 | 2020-05-23T03:16:48 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 633,011,599 | 11,139 | alexey-milovidov | +200 MB of resident memory looks acceptable even for development laptop. | 2020-05-23T09:09:43 | 2020-05-23T09:09:43 | {} | MEMBER |
ClickHouse | ClickHouse | 633,212,020 | 11,036 | alexey-milovidov | It's strange that no test was added (how we are going to support it?) | 2020-05-24T10:45:59 | 2020-05-24T10:46:09 | {} | MEMBER |
ClickHouse | ClickHouse | 632,652,016 | 10,623 | alexey-milovidov | Performance test is incorrect, it doesn't work at all. | 2020-05-22T11:50:07 | 2020-05-22T11:50:07 | {} | MEMBER |
ClickHouse | ClickHouse | 632,688,693 | 10,623 | alexey-milovidov | The code is absolutely wrong - it builds grid instead to work directly. | 2020-05-22T13:24:12 | 2020-05-22T13:24:12 | {} | MEMBER |
ClickHouse | ClickHouse | 632,722,163 | 11,121 | alexey-milovidov | The SYSTEM query to flush allocation cache can be generally useful. | 2020-05-22T14:29:46 | 2020-05-22T14:29:46 | {} | MEMBER |
ClickHouse | ClickHouse | 632,930,420 | 11,037 | nikitamikhaylov | It has rather few conflicts... But I know nothing in this code | 2020-05-22T22:19:49 | 2020-05-22T22:19:49 | {} | MEMBER |
ClickHouse | ClickHouse | 633,009,903 | 11,139 | alexey-milovidov | For development machines it won't work nevertheless, because it requires CAP_IPC_LOCK (that is set up only on package installation). | 2020-05-23T08:56:35 | 2020-05-23T08:56:35 | {} | MEMBER |
ClickHouse | ClickHouse | 632,762,191 | 11,127 | 4ertus2 | Also you may try `multiple_joins_rewriter_version=2`, but it disabled for now cause of issues in push down predicate. You may try the query with `settings multiple_joins_rewriter_version=2, enable_optimize_predicate_expression=0` and push needed predicates manually. | 2020-05-22T15:45:36 | 2020-05-22T15:45:36 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 632,828,527 | 11,142 | alesapin | Test passed! | 2020-05-22T17:52:20 | 2020-05-22T17:52:20 | {} | MEMBER |
ClickHouse | ClickHouse | 633,326,981 | 11,094 | den-crane | But there is no REAL table in my example https://gist.github.com/den-crane/d03524eadbbce0bafa528101afa8f794
```
CREATE TABLE states_raw(
...
) ENGINE = Null;
```
Engine=Null has no stored data. https://clickhouse.tech/docs/en/engines/table-engines/special/null/
Or you can use input function https://clickhouse... | 2020-05-25T00:48:55 | 2020-05-25T00:50:11 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 633,331,747 | 11,158 | akuzm | > Many queries from `agg_functions_min_max_any` are "unstable".
> Don't see the reason why - they are just usual simple SELECT queries from a table with data.
Looking at the metrics, it's the usual thing with page faults. | 2020-05-25T01:16:24 | 2020-05-25T01:16:24 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 633,452,014 | 10,884 | brefre | This didn't have any effect. I'm setting retries to 10 to see if that will help.
Do you have other suggestions of where to look for errors? | 2020-05-25T08:34:09 | 2020-05-25T08:34:09 | {} | NONE |
ClickHouse | ClickHouse | 632,637,096 | 11,119 | alexey-milovidov | I'm Ok if we will just dismiss this test (already done in master). | 2020-05-22T11:12:16 | 2020-05-22T11:12:16 | {} | MEMBER |
ClickHouse | ClickHouse | 632,864,103 | 11,140 | akuzm | https://github.com/ClickHouse/ClickHouse/issues/9473 | 2020-05-22T19:07:00 | 2020-05-22T19:07:00 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 633,088,854 | 10,973 | alexey-milovidov | It will be very useful. The user gives to us a bunch of SELECT queries and we have to guess from what tables these queries will successfully read data. | 2020-05-23T16:30:12 | 2020-05-23T16:30:12 | {} | MEMBER |
ClickHouse | ClickHouse | 633,279,065 | 11,163 | alexey-milovidov | > JEMALLOC_PURGE_MADVISE_FREE
Does it perform a test at runtime that this feature is available in kernel? | 2020-05-24T19:06:14 | 2020-05-24T19:06:24 | {} | MEMBER |
ClickHouse | ClickHouse | 632,630,726 | 11,084 | alexey-milovidov | And server RSS is 35 GiB, so the issue #10818 is not solved. | 2020-05-22T10:56:12 | 2020-05-22T10:56:12 | {} | MEMBER |
ClickHouse | ClickHouse | 632,861,154 | 11,149 | oandrew | Thank you! I was just about to make a PR fixing this same issue. In our case 500ms reschedule was limiting rate to about 300k rows per second. With a similar fix - 1-2M+/s | 2020-05-22T19:01:27 | 2020-05-22T19:01:27 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 632,883,773 | 11,147 | den-crane | No. | 2020-05-22T19:39:57 | 2020-05-22T19:39:57 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 633,009,655 | 11,139 | alexey-milovidov | On my machine, fresh startup:
```
$ echo 3 | sudo tee /proc/sys/vm/drop_caches
$ ./clickhouse-server
...
$ vmtouch -f clickhouse-server
Files: 1
Directories: 0
Resident Pages: 17820/407775 69M/1G 4.37%
Elapsed: 0.011254 seconds
```
So, 69M of the binary is needed for startup. | 2020-05-23T08:54:27 | 2020-05-23T08:54:57 | {} | MEMBER |
ClickHouse | ClickHouse | 633,011,442 | 11,139 | alexey-milovidov | I ran a few another experiments on a slow server:
```
milovidov@example:~/ClickHouse/build_gcc9$ sudo setcap cap_ipc_lock=+ep /opt/milovidov/ClickHouse/build_gcc9/programs/clickhouse
milovidov@example:~/ClickHouse/build_gcc9$ echo 3 | sudo tee /proc/sys/vm/drop_caches
milovidov@example:~/ClickHouse/build_gcc9$ prog... | 2020-05-23T09:08:33 | 2020-05-23T09:08:33 | {} | MEMBER |
ClickHouse | ClickHouse | 633,012,698 | 11,042 | CurtizJ | It will be available in 20.5. | 2020-05-23T09:18:10 | 2020-05-23T09:18:10 | {
"heart": 3
} | MEMBER |
ClickHouse | ClickHouse | 633,060,596 | 11,036 | alexey-milovidov | > Caught exception while loading metadata: Code: 499, e.displayText() = DB::Exception: The specified bucket does not exist
Looks like already fixed in #11142 | 2020-05-23T14:06:53 | 2020-05-23T14:06:53 | {} | MEMBER |
ClickHouse | ClickHouse | 633,071,131 | 11,158 | alexey-milovidov | TODO: enable all `long` tests.
All tests exist for reason and we cannot simply skip them. | 2020-05-23T15:01:12 | 2020-05-23T15:01:12 | {} | MEMBER |
ClickHouse | ClickHouse | 633,103,621 | 11,149 | azat | Actually there are tests failures that looks related (well maybe not directly, but they should not fail), @filimonov can you take a look? | 2020-05-23T17:43:06 | 2020-05-23T17:43:06 | {} | MEMBER |
ClickHouse | ClickHouse | 633,211,940 | 11,153 | silviucpp | @azat I fund out that also If I do 10K small inserts one after the other without any `wait_end_of_query` receive same issue. There is any way I can disable this memory limit check ? | 2020-05-24T10:45:16 | 2020-05-24T10:45:16 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 633,314,589 | 11,114 | azat | @KochetovNicolai can you take a look on PR?
>Integration tests (release) — failed:1, passed:584, error:0
The test `test_insert_distributed_load_balancing` fails even without this changes, will take a look | 2020-05-24T23:15:00 | 2020-05-24T23:15:00 | {} | MEMBER |
ClickHouse | ClickHouse | 633,392,565 | 11,094 | shubham721 | Okay, we can achieve it with Engine = Null.
Thanks @den-crane
https://clickhouse.tech/docs/en/sql-reference/data-types/aggregatefunction/
Here, why does it says uniq, anyif and quantiles are only aggregate functions supported in click house. I suppose Its not updated, right? | 2020-05-25T06:00:33 | 2020-05-25T06:00:33 | {} | NONE |
ClickHouse | ClickHouse | 632,683,927 | 10,128 | alex-zaitsev | #10591 | 2020-05-22T13:13:11 | 2020-05-22T13:13:11 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 632,964,273 | 11,139 | akuzm | No difference in perf test. The mlock might be a problem on systems with low memory such as laptops, so new contributors have to be aware of this and change the default config. | 2020-05-23T01:36:02 | 2020-05-23T01:36:28 | {
"confused": 1
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 632,969,107 | 11,081 | akuzm | Looking at the code, it seems there is no explicit record in xml for NO PASSWORD auth, right? Maybe it would be better to require it. Imagine if someone edits the config and makes a typo in password tag, and it just silently switches to accepting any password. The correct password is also accepted, so the user would th... | 2020-05-23T02:03:47 | 2020-05-23T02:03:47 | {
"+1": 3
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 633,066,676 | 10,981 | alexey-milovidov | Yes, it is profitable.
```
clickhouse-benchmark <<< "SELECT rand64() AS x FROM numbers(1000000) ORDER BY x"
```
New:
```
Queries executed: 710.
localhost:9000, queries 710, QPS: 11.472, RPS: 12029649.997, MiB/s: 91.779, result RPS: 11472368.238, result MiB/s: 87.527.
0.000% 0.079 sec.
10.000% ... | 2020-05-23T14:38:00 | 2020-05-23T14:38:00 | {} | MEMBER |
ClickHouse | ClickHouse | 633,130,993 | 11,158 | akuzm | > I supported this behavior, so the tests that don't have a required table are ignored with a message (yes, I know it's a very bad message).
```
2019.10.19 07:40:08.738542 <Information> PerformanceTestSuite: Preconditions for test 'nyc_taxi' are not fullfilled, skip run
```
Looks like this behavior was deliber... | 2020-05-23T20:13:06 | 2020-05-23T20:13:06 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 633,229,051 | 11,162 | azat | Rebased, previous HEAD - 7d196d08e4af015664675beac6d2f6714db673a5 (contains tests) | 2020-05-24T13:12:37 | 2020-05-24T13:12:37 | {} | MEMBER |
ClickHouse | ClickHouse | 633,310,715 | 11,170 | alexey-milovidov | See #11137 | 2020-05-24T22:36:56 | 2020-05-24T22:36:56 | {} | MEMBER |
ClickHouse | ClickHouse | 633,408,244 | 11,036 | excitoon | > It's strange that no test was added (how we are going to support it?)
Will be fixed in #11174 | 2020-05-25T06:47:39 | 2020-05-25T06:47:39 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 633,437,120 | 11,153 | azat | >clickhouse-server process doesn't have more than 300 MB of memory
of RSS?
Can you post output of the following?
```sql
select metric, formatReadableSize(value) from system.metrics where metric = 'MemoryTracking'
```
*this query should work since it uses less then `max_untracked_memory`, by default 4MB*
... | 2020-05-25T08:00:14 | 2020-05-25T08:00:14 | {} | MEMBER |
ClickHouse | ClickHouse | 633,458,194 | 11,081 | vitlibar | > Looking at the code, it seems there is no explicit record in xml for NO PASSWORD auth, right? Maybe it would be better to require it. Imagine if someone edits the config and makes a typo in password tag, and it just silently switches to accepting any password. The correct password is also accepted, so the user would ... | 2020-05-25T08:48:04 | 2020-05-25T08:48:04 | {} | MEMBER |
ClickHouse | ClickHouse | 632,643,497 | 11,111 | alexey-milovidov | #11138 | 2020-05-22T11:27:37 | 2020-05-22T11:27:37 | {} | MEMBER |
ClickHouse | ClickHouse | 632,653,182 | 9,545 | alexey-milovidov | @And42 Two ways:
- use at least 4 whitespaces;
- rephrase this comment; | 2020-05-22T11:53:02 | 2020-05-22T11:53:02 | {} | MEMBER |
ClickHouse | ClickHouse | 632,953,539 | 2,352 | hongyuzhou | > > What do these a1, c1, a2, c2... parameters represent
>
> It's just a simple demo equation. In your case, if you want to sort the events by sequential comparison by (event_time, pid), you can create an equation based on your data.
Thanks a lot... I try to make the description of my confusion clearer again. Act... | 2020-05-23T00:22:48 | 2020-05-23T00:23:32 | {} | NONE |
ClickHouse | ClickHouse | 633,008,732 | 11,139 | alexey-milovidov | > The mlock might be a problem on systems with low memory such as laptops, so new contributors have to be aware of this and change the default config.
I also had the same thoughts and run some experiments. For whatever reason, I did not observe any slowdown.
Notes:
- only small part of the binary is mapped (e.g.... | 2020-05-23T08:46:07 | 2020-05-23T08:46:07 | {} | MEMBER |
ClickHouse | ClickHouse | 633,137,278 | 11,153 | silviucpp | Hello @azat,
I can replicate the problem with your query. Please look into the logs (those are spammed with `DynamicQueryHandler: Code: 241, e.displayText() = DB::Exception: Memory limit (total) exceeded:`) . If you look at the memory this will not increasing. There is no memory leak, the problem is in the MemoryTra... | 2020-05-23T20:47:49 | 2020-05-23T20:48:33 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 633,209,593 | 11,036 | alexey-milovidov | Sorry, I did not notice that the test was broken. Will fix. | 2020-05-24T10:22:36 | 2020-05-24T10:22:36 | {} | MEMBER |
ClickHouse | ClickHouse | 633,212,510 | 11,153 | azat | >I fund out that also If I do 10K small inserts one after the other without any wait_end_of_query receive same issue.
Yes, this is kind of expecting behaviour
> There is any way I can disable this memory limit check ?
You can add the following into `config.xml`:
```xml
<max_server_memory_usage_to_ram_ratio... | 2020-05-24T10:50:43 | 2020-05-24T11:00:46 | {
"+1": 3
} | MEMBER |
ClickHouse | ClickHouse | 633,409,285 | 11,174 | excitoon | FYI @alexey-milovidov | 2020-05-25T06:50:29 | 2020-05-25T06:50:29 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 633,010,126 | 11,139 | alexey-milovidov | Another experiment:
```
$ sudo setcap cap_ipc_lock=+ep /home/milovidov/work/ClickHouse/build/programs/clickhouse
$ ./clickhouse-server
...
$ vmtouch -f clickhouse-server
Files: 1
Directories: 0
Resident Pages: 65256/407775 254M/1G 16%
Elapsed: 0.016184 seconds
```
Yes, it works... | 2020-05-23T08:58:39 | 2020-05-23T08:58:39 | {} | MEMBER |
ClickHouse | ClickHouse | 633,010,640 | 11,139 | alexey-milovidov | An experiment on a server with HDDs:
```
milovidov@example:~$ time /usr/bin/clickhouse server 2>&1 | head -n1
Processing configuration file 'config.xml'.
real 0m0,104s
user 0m0,080s
sys 0m0,020s
milovidov@example:~$ echo 3 | sudo tee /proc/sys/vm/drop_caches
3
milovidov@example:~$ time /usr/bin/cl... | 2020-05-23T09:02:22 | 2020-05-23T09:02:56 | {} | MEMBER |
ClickHouse | ClickHouse | 633,070,436 | 10,917 | den-crane | I re-read a thread. So every 6 hours the query eat +30GB, and after 2 days memory usage drops itself? Is it right? | 2020-05-23T14:57:21 | 2020-05-23T14:57:21 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 633,100,922 | 10,621 | azat | Looks like some tests stuck?
>Functional stateless tests (thread) Pending — Started
>Stress test (address) Pending — Started
>Stress test (memory) Pending — Started
>Stress test (thread) Pending — Started
Everything else is green (including performance) | 2020-05-23T17:27:22 | 2020-05-23T17:27:22 | {} | MEMBER |
ClickHouse | ClickHouse | 633,142,043 | 11,153 | silviucpp | @azat I have no idea :) Now I wanted to use `wait_end_of_query` . But everything seems broken once the problem occurs. Even config auto reload is not longer working and throw same problem.
| 2020-05-23T21:20:46 | 2020-05-23T21:20:46 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 633,153,879 | 11,159 | alexey-milovidov | No visible difference in perf test. | 2020-05-23T23:31:42 | 2020-05-23T23:31:42 | {} | MEMBER |
ClickHouse | ClickHouse | 633,281,465 | 10,998 | alexey-milovidov | Reopened.
1. Need to verify that it is fixed in master.
2. The issue with bad error message is still relevant. | 2020-05-24T19:20:37 | 2020-05-24T19:21:01 | {} | MEMBER |
ClickHouse | ClickHouse | 632,730,497 | 10,925 | nicbaz | thanks! :+1: | 2020-05-22T14:47:18 | 2020-05-22T14:47:18 | {} | NONE |
ClickHouse | ClickHouse | 633,028,046 | 11,106 | nikitamikhaylov | Decided not to cherry pick it because of "volumes" and etc | 2020-05-23T11:06:08 | 2020-05-23T11:06:08 | {} | MEMBER |
ClickHouse | ClickHouse | 633,058,433 | 10,484 | alexey-milovidov | Let's merge with master. | 2020-05-23T13:54:48 | 2020-05-23T13:54:48 | {} | MEMBER |
ClickHouse | ClickHouse | 633,082,676 | 10,973 | ZeDRoman | > Does it support generating tables that are suitable for multiple SELECT queries?
Now it supports JOIN operation.
It's easy to make it support multiple SELECT | 2020-05-23T16:00:38 | 2020-05-23T16:00:38 | {} | NONE |
ClickHouse | ClickHouse | 633,239,167 | 11,167 | shivakumarss | - Another point to highlight is, day by day the partition size is being increased. ( yes the partition key has YYYY,DD,HH )
- Total 5 keys in partition | 2020-05-24T14:27:02 | 2020-05-24T14:36:11 | {} | NONE |
ClickHouse | ClickHouse | 632,962,602 | 1,118 | Fanduzi |
> @BlackEric001 nope, not scheduled yet. Though you can already use something like this to get some idea on how query is executed:
>
> ```
> % clickhouse-client --send_logs_level=debug <<< 'select max(number) from numbers(100) where number % 2 union all select 123 as number;' > /dev/null
> [....yandex.net] 2... | 2020-05-23T01:25:44 | 2020-05-23T01:25:44 | {
"+1": 1
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 632,965,741 | 11,053 | akuzm | Please add a functional test for this to `0_stateless` (see the description here: https://clickhouse.tech/docs/en/development/tests/#functional-tests) | 2020-05-23T01:48:49 | 2020-05-23T01:48:49 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 633,327,262 | 11,158 | alexey-milovidov | Many queries from `agg_functions_min_max_any` are "unstable".
Don't see the reason why - they are just usual simple SELECT queries from a table with data. | 2020-05-25T00:50:45 | 2020-05-25T00:50:45 | {} | MEMBER |
ClickHouse | ClickHouse | 633,408,419 | 11,036 | excitoon | @alexey-milovidov Sorry for that! | 2020-05-25T06:48:10 | 2020-05-25T06:48:10 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 633,427,447 | 11,177 | excitoon | Progress on #10650 | 2020-05-25T07:37:45 | 2020-05-25T07:37:45 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 632,754,874 | 11,106 | alesapin | Error in integration test expected | 2020-05-22T15:29:58 | 2020-05-22T15:34:57 | {} | MEMBER |
ClickHouse | ClickHouse | 632,864,910 | 9,473 | akuzm | As a rule, REPL-like things don't exit if you spam Ctrl+C (python, psql, etc). Personally, I found the different old behavior of clickhouse-client inconvenient, and prefer the standard way -- exit on Ctrl+D. | 2020-05-22T19:09:12 | 2020-05-22T19:09:12 | {
"+1": 2
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 632,922,595 | 11,146 | silviucpp | Looking to the code, this is expected. | 2020-05-22T21:52:24 | 2020-05-22T21:52:24 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 632,951,205 | 11,147 | Tri0L | Great, thanks a lot! | 2020-05-23T00:08:28 | 2020-05-23T00:08:28 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 633,045,994 | 8,441 | alexey-milovidov | I compare master vs 19.13.4 and clickhouse-client vs curl.
I don't see significant difference in performance between versions (only a slight difference):
master:
```
milovidov@milovidov-desktop:~/work/ClickHouse$ time curl "http://localhost:8123/?query=INSERT%20INTO%20log%20format%20CSV" --data-binary @test.csv... | 2020-05-23T12:43:38 | 2020-05-23T12:44:02 | {} | MEMBER |
ClickHouse | ClickHouse | 633,112,183 | 11,153 | azat | Cannot reproduce (even on that particular commit), 10k requests, ~1k in parallel everything is fine (RSS/VIRT does not grows):
```
$ time yes '127.1:8123/?wait_end_of_query=1' | head -n10000 | xargs -P100
00 curl -s -d 'select 1' | grep -x -c 1 ... | 2020-05-23T18:28:16 | 2020-05-23T18:28:16 | {} | MEMBER |
ClickHouse | ClickHouse | 633,128,959 | 11,158 | akuzm | > It's a shame that some of our tests were not working and no one cares.
If no one cared nothing would work at all, so I think you're exaggerating.
> It has happened at #6478
It happened long before that -- hits_1000m, nyc_taxi and other tables were never downloaded, and the old performance test silently ignor... | 2020-05-23T20:02:04 | 2020-05-23T20:05:50 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 633,139,220 | 11,153 | azat | Reproduced (forgot to set memory limit for server)
>Using the version a2ed826
Hm, and this not happens before? | 2020-05-23T20:58:27 | 2020-05-23T20:58:27 | {} | MEMBER |
ClickHouse | ClickHouse | 633,189,648 | 11,094 | shubham721 | Hey @den-crane
Thanks. But in this case, we would need to create a source raw table, Which will contain all of the duplicate rows with the same primary keys.
And whenever we want to insert, we have to insert only in the source table and materialized view will trigger updates on the table.
Is there a way we can achi... | 2020-05-24T07:03:14 | 2020-05-24T07:03:14 | {} | NONE |
ClickHouse | ClickHouse | 641,562,501 | 11,401 | azat | >Experiment #3: muzzy decay 10 ms, MADV_FREE, no MADV_DONTNEED
@akuzm AFAIK flamegraphs are build automatically, and they are in archive, which one you posted in the https://github.com/ClickHouse/ClickHouse/pull/11401#issuecomment-641314974 (tried play with SHA in the link - did not work)? | 2020-06-09T20:51:07 | 2020-06-09T20:51:07 | {} | MEMBER |
ClickHouse | ClickHouse | 641,958,081 | 11,572 | alexey-milovidov | Looks like to have the same reason as #10389 | 2020-06-10T12:03:35 | 2020-06-10T12:03:35 | {} | MEMBER |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.