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 | 511,843,664 | 5,997 | Schaefer-Thomas | Good question!
I was running clickhouse in vmware and the database in vmhgfs (host guest file system) because lack of local space.
After copying it back to ext4 the problem disappeared!
Thanks for the hint!
| 2019-07-16T14:35:23 | 2019-07-16T14:35:23 | {} | NONE |
ClickHouse | ClickHouse | 512,201,176 | 3,853 | alexey-milovidov | But that's only for queries from clickhouse-client. | 2019-07-17T10:40:39 | 2019-07-17T10:40:39 | {} | MEMBER |
ClickHouse | ClickHouse | 512,234,869 | 5,959 | filimonov | How many rows do you insert at once? How big each row is? What are the datatypes of columns? | 2019-07-17T12:34:53 | 2019-07-17T12:34:53 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 513,068,103 | 6,062 | andyyzh | nice! | 2019-07-19T02:41:39 | 2019-07-19T02:41:39 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 513,126,457 | 6,064 | amosbird | What's the issue, and what version do you use? | 2019-07-19T07:35:13 | 2019-07-19T07:35:13 | {} | COLLABORATOR |
ClickHouse | ClickHouse | 513,137,837 | 6,064 | vasyaabr | Use
`SELECT count() FROM mop3 WHERE (key = 'category') AND (value = 'lips')` | 2019-07-19T08:13:59 | 2019-07-19T08:13:59 | {} | NONE |
ClickHouse | ClickHouse | 511,756,397 | 5,947 | filimonov | Related:
https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=926371
https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=926234
https://bugzilla.altlinux.org/show_bug.cgi?id=36972
https://bugzilla.redhat.com/show_bug.cgi?format=multiple&id=1726630
| 2019-07-16T10:15:43 | 2019-07-16T10:15:43 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 511,806,617 | 2,847 | vitlibar | It seems it's already done.
https://github.com/yandex/ClickHouse/pull/5567 | 2019-07-16T13:04:07 | 2019-07-16T13:04:07 | {} | MEMBER |
ClickHouse | ClickHouse | 512,000,286 | 5,999 | alexey-milovidov | I also cannot find it in documentation (I thought it was there).
Note that parsing numbers starting with zero can be quite dangerous. Example in JavaScript:
```
var obj = {"x":0123};
alert(obj.x);
var json = '{"x":0123}';
alert(JSON.parse(json).x);
```
In the first example, it will output 83, because Java... | 2019-07-16T21:45:36 | 2019-07-16T21:45:36 | {} | MEMBER |
ClickHouse | ClickHouse | 512,004,487 | 5,999 | alexey-milovidov | High ratio of zeros in datasets is very common case.
In "NYC taxi" dataset, 60% of all integers are zero.
In "Yandex.Metrica" dataset, 76% of all integers are zero. | 2019-07-16T22:00:14 | 2019-07-16T22:00:14 | {} | MEMBER |
ClickHouse | ClickHouse | 512,059,907 | 4,101 | abraithwaite | I would also like this better documented. Struggling to keep our active part count under 200 for our raw data tables. Each insert is 60k rows in a batch at ~800B/row uncompressed (text data). Our volume is ~150k rows/s spread across 9 nodes. | 2019-07-17T01:04:37 | 2019-07-17T01:04:37 | {} | NONE |
ClickHouse | ClickHouse | 512,126,073 | 5,999 | MahmoudGoda0 | Thanks for this workaround, but i have a table with more than 3 billion rows, and by this workaround it will take lot of time, specially we receive a file (~ 700 K rows) every 11 second. | 2019-07-17T06:46:02 | 2019-07-17T06:46:02 | {} | NONE |
ClickHouse | ClickHouse | 512,205,529 | 5,813 | alexey-milovidov | https://github.com/yandex/ClickHouse/issues/2099
This was my attempt but it's abandoned:
https://github.com/yandex/ClickHouse/commit/a763a4a6c94bc0448cd35d174746f82e985338fa
CC @amosbird
It should be very easy to finish. | 2019-07-17T10:55:50 | 2019-07-17T10:57:20 | {} | MEMBER |
ClickHouse | ClickHouse | 512,550,336 | 6,041 | alexey-milovidov | It will allow to remove arbitrary directory inside `detached` by its name. But be careful - name must not contain path components (example: `../../../../../data`). | 2019-07-17T20:12:35 | 2019-07-17T20:12:35 | {} | MEMBER |
ClickHouse | ClickHouse | 512,620,940 | 5,959 | renkun2020 | > How many rows do you insert at once? How big each row is? What are the datatypes of columns?
here is the table column datatype
┌─type─────┐
│ String │
│ String │
│ String │
│ String │
│ String │
│ String │
│ Int64 │
│ Int64 │
│ String │
│ String │
│ UInt8 │
│ UInt8 │
│ UIn... | 2019-07-18T00:46:23 | 2019-07-18T00:46:23 | {} | NONE |
ClickHouse | ClickHouse | 513,190,031 | 6,042 | abyss7 | @florida-hehe can you provide your Kafka server configuration regarding the auth settings? | 2019-07-19T11:09:06 | 2019-07-19T11:09:06 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 512,078,846 | 6,028 | amosbird | Heh, just found a relevant issue https://github.com/yandex/ClickHouse/issues/5813 . | 2019-07-17T02:44:09 | 2019-07-17T02:44:09 | {
"rocket": 1
} | COLLABORATOR |
ClickHouse | ClickHouse | 512,191,940 | 6,028 | alexey-milovidov | Ok.
Now the next step:
We have exact information about the number of rows in MergeTree tables. If the user do simple `SELECT count() ... FROM table` query, we can return the answer instantly. | 2019-07-17T10:09:54 | 2019-07-17T10:09:54 | {
"+1": 1
} | MEMBER |
ClickHouse | ClickHouse | 512,207,315 | 5,981 | 4ertus2 | There's one more interesting issue here. If we block new allocation we block it not only for query but also for some library activities. It looks like in some cases server caches MemoryTracker errors (that are just a DB::Exception) and retries allocate same memory in loop. I think we could adapt fault probability for t... | 2019-07-17T11:01:57 | 2019-07-17T11:40:10 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 512,349,076 | 5,999 | MahmoudGoda0 | Just to make sure I understand these lines correctly, it will copy the data from from a file "data.csv" (in my case it will be the TSV file, then insert it into "target_table" with the correct type, to remove the zero ?
If this is true, this will work if i pass the folder path to insert all files ?
like this cat *.ts... | 2019-07-17T16:11:38 | 2019-07-17T16:11:38 | {} | NONE |
ClickHouse | ClickHouse | 512,822,141 | 6,050 | alexey-milovidov | Dictionaries will load successfully regardless to this warning.
This warning present, because some config elements get substituted into `/etc/clickhouse-server/dnl_dictionary.xml`. You can check it in `/etc/clickhouse-server/preprocessed/...` directory.
Extraneous config elements are substituted from `conf.d` dir... | 2019-07-18T13:45:06 | 2019-07-18T13:45:06 | {} | MEMBER |
ClickHouse | ClickHouse | 512,946,425 | 5,981 | 4ertus2 | NOTE: JOINs perf boost is fake. We're joining less data now to avoid out-of-memory. | 2019-07-18T19:04:53 | 2019-07-18T19:05:08 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 513,046,739 | 469 | anjia0532 | my python script
https://gist.github.com/anjia0532/6db48b0886d91d9a663e5a9fd19f2aaa
src csv
```csv
val1,val2,val3
aa,bb,cc
a
a,bb,cc
aa
,bb,cc
aa,
bb,cc
a\a,bb,cc
```
python cmd `python clean_csv.py --src=src.csv --dest=dest.csv --chunksize=50000 --cols --encoding=utf-8 --delimiter=,`
dest csv
``... | 2019-07-19T00:47:24 | 2019-07-19T00:47:24 | {} | NONE |
ClickHouse | ClickHouse | 511,768,865 | 5,947 | filimonov | I.e. the problem caused by issue inside the compiler itself: https://gcc.gnu.org/bugzilla/show_bug.cgi?id=89906
It was patched in Debian, but it looks like redhat / centos maintainers missed that patch.
Please use newer (or older) compiler version for now, or disable tests during building. | 2019-07-16T10:58:23 | 2019-07-16T10:58:23 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 512,170,348 | 5,182 | crakjie | The merger of OpenTracing and OpenCensus is called OpenTelemetry https://opentelemetry.io/ | 2019-07-17T09:07:47 | 2019-07-17T09:07:47 | {} | NONE |
ClickHouse | ClickHouse | 512,195,480 | 5,056 | alexey-milovidov | Ok.
We can simply implement a mode to continue after printing an error.
PS. To perform load testing with query_log on testing environment, I just filter some queries and ensure that all other queries execute successfully:
```
grep -Fv "\\'banners\\'" queries.tsv | grep -Fv "\\'phrases\\'" | grep -v _data | gr... | 2019-07-17T10:21:21 | 2019-07-17T10:22:00 | {} | MEMBER |
ClickHouse | ClickHouse | 512,196,282 | 5,056 | alexey-milovidov | "easy task" but better to not implementing reconnections. | 2019-07-17T10:24:11 | 2019-07-17T10:24:11 | {} | MEMBER |
ClickHouse | ClickHouse | 512,203,948 | 6,030 | filimonov | It is not a bug. `formatReadableSize` return strings, so they are ordered as strings.
Use the approach shown in @den-crane 's answer, | 2019-07-17T10:50:04 | 2019-07-17T10:50:04 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 512,239,723 | 5,056 | filimonov | Context: I was using `clickhouse-benchmark` as a stress test tool. You probably saw that: https://www.altinity.com/blog/clickhouse-in-the-storm-part-1
As with each kind of DoS - after reaching some critical load (in terms of concurrency and qps)
proper queries are not able to be served properly anymore (like too ma... | 2019-07-17T12:49:20 | 2019-07-17T12:49:20 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 512,774,400 | 2,847 | vitlibar | I still think the problem is already fixed (in another place). | 2019-07-18T11:16:50 | 2019-07-18T11:16:50 | {} | MEMBER |
ClickHouse | ClickHouse | 513,138,550 | 6,064 | masosky | Oops, I forgot the single quotes.
But I do not understand I tried that previously and I didn't work.
But now it is working!
Sorry for the inconvenience and thanks for the fast answer. | 2019-07-19T08:16:19 | 2019-07-19T08:16:19 | {} | NONE |
ClickHouse | ClickHouse | 513,190,666 | 6,063 | alexey-milovidov | CC @zhang2014 | 2019-07-19T11:11:25 | 2019-07-19T11:11:25 | {} | MEMBER |
ClickHouse | ClickHouse | 512,194,132 | 5,059 | alexey-milovidov | MinMax index of partition key is build on columns participating in expression (`dt`) and not on expression itself (`toDate(toDateTime(dt / 1000))`).
You can write `dt BETWEEN toUInt64(toDateTime(toDate('2019-04-18'))) * 1000 AND toUInt64(toDateTime(toDate('2019-04-19'))) * 1000 - 1` | 2019-07-17T10:16:52 | 2019-07-17T10:16:52 | {} | MEMBER |
ClickHouse | ClickHouse | 512,420,196 | 6,040 | alexey-milovidov | I will add a test. | 2019-07-17T17:19:05 | 2019-07-17T17:19:05 | {} | MEMBER |
ClickHouse | ClickHouse | 512,580,280 | 5,775 | alexey-milovidov | This was not a bugfix but bug-introduce. | 2019-07-17T21:43:26 | 2019-07-17T21:43:26 | {
"laugh": 1
} | MEMBER |
ClickHouse | ClickHouse | 512,828,181 | 6,050 | mikeeremin | ```
clickhouse-test-node1 :) system reload dictionaries;
SYSTEM RELOAD DICTIONARIES
Ok.
0 rows in set. Elapsed: 0.005 sec.
clickhouse-test-node1 :) select * from system.dictionaries\G
SELECT *
FROM system.dictionaries
Row 1:
──────
name: programreleasefiles
status: NOT_L... | 2019-07-18T14:00:11 | 2019-07-18T14:00:11 | {} | NONE |
ClickHouse | ClickHouse | 512,905,483 | 5,813 | amosbird | I'll see what I can do :) | 2019-07-18T17:10:41 | 2019-07-18T17:10:41 | {} | COLLABORATOR |
ClickHouse | ClickHouse | 512,063,496 | 1,686 | abraithwaite | Did you guys find a resolution to this? We're experiencing something similar. | 2019-07-17T01:24:57 | 2019-07-17T01:24:57 | {} | NONE |
ClickHouse | ClickHouse | 512,112,706 | 4,101 | nvartolomei | @abraithwaite default server settings? 60k sounds like a small batch size, I had great experience with batch size > 1M. | 2019-07-17T05:51:02 | 2019-07-17T05:51:02 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 512,200,298 | 3,853 | alexey-milovidov | Accidentially solved in https://github.com/yandex/ClickHouse/pull/5630
Now queries are logged after reformatting and all comments are lost:
```
milovidov-Pro-P30 :) SET send_logs_level = 'trace'
SET send_logs_level = 'trace'
Ok.
0 rows in set. Elapsed: 0.007 sec.
milovidov-Pro-P30 :) SELECT /* test ... | 2019-07-17T10:37:43 | 2019-07-17T10:37:43 | {} | MEMBER |
ClickHouse | ClickHouse | 512,290,387 | 5,951 | alexey-milovidov | Continued in #6038 | 2019-07-17T14:52:01 | 2019-07-17T14:52:01 | {} | MEMBER |
ClickHouse | ClickHouse | 512,324,522 | 5,056 | alexey-milovidov | Ok. BTW, clickhouse-benchmark creates fixed number of connections at startup and use them all the time. | 2019-07-17T15:21:01 | 2019-07-17T15:21:01 | {} | MEMBER |
ClickHouse | ClickHouse | 512,539,503 | 6,041 | alexey-milovidov | And a setting to enable this query, disabled by default.
Use case: in cloud providers the user has no access to filesystem and has no way to deal with detached parts. | 2019-07-17T19:40:00 | 2019-07-17T19:40:00 | {} | MEMBER |
ClickHouse | ClickHouse | 511,756,774 | 5,964 | alexey-milovidov | Maybe we can directly use "channel" without patching Poco? | 2019-07-16T10:16:55 | 2019-07-16T10:16:55 | {} | MEMBER |
ClickHouse | ClickHouse | 511,924,132 | 5,951 | alexey-milovidov | I have started code review and planning to merge today.
> Do we really need the brackets
We need brackets, because
`table.COLUMNS('a\\d+')`
will look much better than
`table.COLUMNS 'a\\d+'`
(but it probably doesn't work in any of these variants right now?)
> Also, in the second step I wanted to extend t... | 2019-07-16T18:04:28 | 2019-07-16T18:05:09 | {} | MEMBER |
ClickHouse | ClickHouse | 512,062,613 | 6,002 | den-crane | Check sharding key at distributed table. With rand() every time a row goes to a random shard. You need to set some expression like cityHash64(user_id), | 2019-07-17T01:20:27 | 2019-07-17T01:20:27 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 512,838,330 | 5,545 | k0st1an | I seem to have the same problem. Version 19.11. The consumer is connected (this is seen on the Kafka side), but the lag increases. At the same time on version 19.7.3 this problem is also there, but it is not so common. Data in the JSON topic. Tried to catch something in the logs, but there is silence. | 2019-07-18T14:24:44 | 2019-07-18T14:24:44 | {} | NONE |
ClickHouse | ClickHouse | 512,881,404 | 5,981 | alexey-milovidov | > There's some performance degradation in JSONExtract*
Don't need to fix, it's up to @vitlibar | 2019-07-18T16:06:00 | 2019-07-18T16:06:00 | {} | MEMBER |
ClickHouse | ClickHouse | 511,765,836 | 6,003 | heneyin | I get it, thank. | 2019-07-16T10:47:43 | 2019-07-16T10:47:43 | {} | NONE |
ClickHouse | ClickHouse | 512,235,889 | 5,182 | filimonov | > The merger of OpenTracing and OpenCensus is called OpenTelemetry https://opentelemetry.io/
 | 2019-07-17T12:38:04 | 2019-07-17T12:38:04 | {
"laugh": 1
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 512,067,179 | 1,686 | abraithwaite | Another thing we noticed is that clickhouse is for some reason prioritizing merging materialized views parts over raw data table parts. I'm guessing this is because table selection for merges is somewhat random, but it would be nice if clickhouse considered the active part count for the table in it's decision when sel... | 2019-07-17T01:42:58 | 2019-07-17T01:42:58 | {} | NONE |
ClickHouse | ClickHouse | 512,204,255 | 6,030 | alexey-milovidov | @abraithwaite
Consider more simple example:
```
SELECT toString(number) AS x FROM numbers(100) ORDER BY x
```
It's obvious that this query sort data alphabetically (by String value). And it is the only way it should work.
`formatReadableSize` is similar. This function returns `String` data type, not some... | 2019-07-17T10:51:14 | 2019-07-17T10:51:14 | {} | MEMBER |
ClickHouse | ClickHouse | 512,557,578 | 4,959 | alexey-milovidov | If you experience this issue, consider updating to 19.11. | 2019-07-17T20:33:40 | 2019-07-17T20:33:40 | {} | MEMBER |
ClickHouse | ClickHouse | 512,963,070 | 6,050 | mikeeremin | Worked, thanks.
| 2019-07-18T19:54:16 | 2019-07-18T19:54:16 | {} | NONE |
ClickHouse | ClickHouse | 513,033,321 | 6,061 | weiqxu | I also create a PR as a fix for this issue. | 2019-07-18T23:38:50 | 2019-07-18T23:38:50 | {
"+1": 1
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 513,186,545 | 5,959 | filimonov | > [root@master tmp]# du -sh passed_data.csv
> 2.1G passed_data.csv
> [root@master tmp]# wc -l passed_data.csv
> 36953 passed_data.csv
That explains everything. You have avg size of the row about 57Kb. It's very big rows.
ClickHouse tries to form blocks of max_insert_block_size (by default 1048576) to sort ... | 2019-07-19T10:55:22 | 2019-07-19T10:55:22 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 513,198,831 | 5,545 | filimonov | @k0st1an are you able to reproduce the problem with docker compose? | 2019-07-19T11:44:05 | 2019-07-19T11:44:05 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 512,005,136 | 5,999 | alexey-milovidov | @MahmoudGoda0
Workaround:
- create a temporary table with String data type instead of integer;
- load your data as usual;
- insert into destination table with INSERT SELECT query and `toInt32` function. | 2019-07-16T22:02:20 | 2019-07-16T22:02:20 | {} | MEMBER |
ClickHouse | ClickHouse | 512,167,305 | 5,959 | renkun2020 | I took some testing on DEV,
here is the sql,
cat passed_data.csv | clickhouse-client -m --user=**** --password=**** --host=***** --port=9000 --database=test '--format_csv_delimiter=|' '--query=INSERT INTO temp FORMAT CSV'
temp is a local ReplicatedMergeTree table.
1、
<max_memory_usage>2147483648</max_memory_u... | 2019-07-17T08:59:05 | 2019-07-17T08:59:05 | {} | NONE |
ClickHouse | ClickHouse | 512,325,753 | 5,813 | amosbird | OK. Why is it abandoned? | 2019-07-17T15:21:33 | 2019-07-17T15:21:33 | {} | COLLABORATOR |
ClickHouse | ClickHouse | 512,538,522 | 5,813 | alexey-milovidov | I was thought that I can implement it in one hour. After a few hours there were one or two failing tests. Then I switched to more important tasks and never finished. You are the only hope :) | 2019-07-17T19:37:21 | 2019-07-17T19:37:21 | {} | MEMBER |
ClickHouse | ClickHouse | 512,557,405 | 4,959 | alexey-milovidov | @free6k said that the issue no longer reproduces. | 2019-07-17T20:33:06 | 2019-07-17T20:33:06 | {} | MEMBER |
ClickHouse | ClickHouse | 512,834,413 | 5,981 | 4ertus2 | There's some performance degradation in `JSONExtract*`. toString() and materialize() for strings are also not good. | 2019-07-18T14:15:33 | 2019-07-18T14:21:36 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 512,967,707 | 5,981 | alexey-milovidov | Note: this should not be backported due to severity of this change. | 2019-07-18T20:06:55 | 2019-07-18T20:06:55 | {} | MEMBER |
ClickHouse | ClickHouse | 512,969,028 | 5,981 | alexey-milovidov | 71eed6507e | 2019-07-18T20:10:44 | 2019-07-18T20:10:44 | {} | MEMBER |
ClickHouse | ClickHouse | 513,093,320 | 5,999 | filimonov | 1. You need to run several inserts in parallel, i.e. run that pipe on each file separately in parallel.
2. Check `max_threads` setting. Also for small enough queries ClickHouse can decrease the number of threads automatically. Also if you can have concurrent queries - using all cores by single query can be suboptimal.... | 2019-07-19T05:14:52 | 2019-07-19T05:14:52 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 511,775,043 | 5,964 | alexey-milovidov | ```
LOG_INFO(log, (detach ? "Detached " : "Removed ") << parts_to_remove.size() << " parts inside partition ID " << partition_id << ".");
```
This is string formatting. Expression `(detach ? "Detached " : "Removed ") << parts_to_remove.size() << " parts inside partition ID " << partition_id << "."` is lazy evaluat... | 2019-07-16T11:20:16 | 2019-07-16T11:20:16 | {} | MEMBER |
ClickHouse | ClickHouse | 512,541,962 | 5,981 | alexey-milovidov | Consider the following scenario:
- we want to allocate 16 bytes, but throw an exception because memory limit exceeded;
- to print exception message, we allocate another 32 bytes.
Maybe we can avoid it by flushing delta before throwing exception. | 2019-07-17T19:46:56 | 2019-07-17T19:47:18 | {} | MEMBER |
ClickHouse | ClickHouse | 512,627,982 | 4,914 | yingfeng | When could this feature be production ready? | 2019-07-18T01:24:26 | 2019-07-18T01:24:26 | {} | NONE |
ClickHouse | ClickHouse | 513,130,280 | 4,903 | akonyaev90 | I would like to help with this task. Can you give any recommendations for implementation?
Which format implementation can I take as a model for study and a pattern? May be Parquet?
Arrow::parquet used for parquet-format. Does it good idea use arrow::ORC for orc-format? | 2019-07-19T07:48:09 | 2019-07-19T07:48:09 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 513,136,385 | 6,064 | masosky | I am using last version released
`ClickHouse client version 19.9.2.4.
Connecting to localhost:9000 as user default.
Connected to ClickHouse server version 19.9.2 revision 54421.`
Error:
`Received exception from server (version 19.9.2):
Code: 47. DB::Exception: Received from localhost:9000, ::1. DB::Exception: M... | 2019-07-19T08:09:11 | 2019-07-19T08:14:07 | {} | NONE |
ClickHouse | ClickHouse | 511,757,570 | 5,964 | alexey-milovidov | Logger looks like a very thin wrapper over Channel and we don't use most of its features. | 2019-07-16T10:19:44 | 2019-07-16T10:19:57 | {} | MEMBER |
ClickHouse | ClickHouse | 511,759,067 | 5,964 | nikitamikhaylov | Logger saves internal log level and has some pretty functions to compare level of message with level of logger. I think we will rewrite with macroses if we don't use that class. | 2019-07-16T10:24:30 | 2019-07-16T10:24:30 | {} | MEMBER |
ClickHouse | ClickHouse | 511,765,361 | 5,964 | nikitamikhaylov | I don't understand you clearly. Is it true that string formatting is to add date and some other useful information to our message? And, the previous version of *logger_useful.h* contains lines such as *if ((logger)->trace)* which compare logger's level with the level of message. | 2019-07-16T10:46:05 | 2019-07-16T10:46:05 | {} | MEMBER |
ClickHouse | ClickHouse | 512,070,979 | 6,030 | den-crane | this is contradicts SQL.
you can use
select formatReadableSize(size)
...
order by size | 2019-07-17T02:04:29 | 2019-07-17T02:04:29 | {
"+1": 3
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 512,083,913 | 5,722 | amosbird | case 1:
> around midterm the allocation drops close to 0 and the CLI progress bar freezes for a few seconds (probably the join)
Most likely it's doing https://github.com/yandex/ClickHouse/blob/c1/dbms/src/Interpreters/Join.cpp#L514 which is single threaded.
case 2:
> only one core allocated for 15 seconds, no... | 2019-07-17T03:09:57 | 2019-07-17T03:09:57 | {} | COLLABORATOR |
ClickHouse | ClickHouse | 512,339,394 | 6,026 | abyss7 | Thank you for the fix - will review it tomorrow in more detail | 2019-07-17T15:48:52 | 2019-07-17T15:48:52 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 512,516,051 | 6,043 | 4ertus2 | It needs fix for
```
10: 00933_ttl_with_default: [ FAIL ] - return code 62
10: Error on processing query: insert into ttl_00933_2 values (toDateTime('2000-10-10 00:00:00'), 2);
10: Code: 62, e.displayText() = DB::Exception: Syntax error (column parser): failed at posi... | 2019-07-17T18:34:20 | 2019-07-17T18:34:20 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 512,536,785 | 5,999 | alexey-milovidov | Yes, it will work as expected.
`clickhouse-local` works in streaming fashion. | 2019-07-17T19:32:35 | 2019-07-17T19:32:35 | {} | MEMBER |
ClickHouse | ClickHouse | 512,669,554 | 4,631 | florida-hehe | > I got same error `void DB::StorageKafka::streamThread(): std::exception. Code: 1001, type: cppkafka::HandleException, e.what() = Local: Invalid argument or configuration`. Drop Kakfa engine table and recreate help. ClickHouse server version 19.3.7
<Error> void DB::StorageKafka::streamThread(): std::exception. Cod... | 2019-07-18T05:24:11 | 2019-07-18T05:24:11 | {} | NONE |
ClickHouse | ClickHouse | 513,138,608 | 6,065 | javisantana | oops, I realized that I inserted wrong data, this is working as expected. I'm leaving the right insert just in case someone asks the same question
```
eb9e0bb5b003 :) insert into test_mt format CSV ,,1234
INSERT INTO test_mt FORMAT CSV
Ok.
1 rows in set. Elapsed: 0.005 sec.
eb9e0bb5b003 :) select * from t... | 2019-07-19T08:16:30 | 2019-07-19T08:16:30 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 513,191,450 | 6,042 | filimonov | @florida-hehe SASL support was added recently, and it looks it first introduced in 19.8 version
https://github.com/yandex/ClickHouse/pull/5253/commits/67cf1bc48c5a4ac147a6abd2159d0e4f8ed6d814
It looks like you're using 19.7.54419. Please try 19.11, it also contains some more Kafka related fixes.
About tests: c... | 2019-07-19T11:14:33 | 2019-07-19T11:14:33 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 511,964,989 | 6,026 | bopohaa | Perhaps this error is described in these tasks.
https://github.com/yandex/ClickHouse/issues/5957
https://github.com/yandex/ClickHouse/issues/4616
https://github.com/yandex/ClickHouse/issues/2480 | 2019-07-16T20:04:58 | 2019-07-16T20:04:58 | {} | NONE |
ClickHouse | ClickHouse | 512,200,596 | 3,853 | alexey-milovidov | It's also written in query_log without comments.
That is an issue, because users rely on these comments. | 2019-07-17T10:38:43 | 2019-07-17T10:39:11 | {
"+1": 1,
"heart": 1
} | MEMBER |
ClickHouse | ClickHouse | 512,555,280 | 5,999 | MahmoudGoda0 | Thanks guys for your support.
I have 2 question please :)
1- Do we have parallel import ? So I can speed up the import time ?
2 - After imported all files, i run query that take some time (23 seconds) and i realized the clcikhouse use only 10 CPUs during the query run time, how i tell clickhouse to use more CPUs( i ... | 2019-07-17T20:27:06 | 2019-07-17T20:27:06 | {} | NONE |
ClickHouse | ClickHouse | 512,577,400 | 5,929 | alexey-milovidov | > please remove those packages with prefix 2 from testing repo (OR add 2: to all of them).
Removed. | 2019-07-17T21:33:46 | 2019-07-17T21:33:46 | {} | MEMBER |
ClickHouse | ClickHouse | 513,194,401 | 6,046 | filimonov | That is very uncommon: usually CH is using whole cpu power even on single query :)
Most probably you're testing something that can't be efficiently parallelized, or you reached some I/O limitation, or something similar. You can try to monitor system during queries using standard tools like `dstat`, or try some debu... | 2019-07-19T11:26:20 | 2019-07-19T11:26:20 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 511,757,182 | 5,964 | alexey-milovidov | At least there is `Logger::getChannel` method. | 2019-07-16T10:18:23 | 2019-07-16T10:18:23 | {} | MEMBER |
ClickHouse | ClickHouse | 511,761,282 | 5,964 | alexey-milovidov | Yes, and we don't use these functions... | 2019-07-16T10:31:51 | 2019-07-16T10:31:51 | {} | MEMBER |
ClickHouse | ClickHouse | 511,761,513 | 5,964 | alexey-milovidov | (The purpose of macros is to avoid string formatting if we don't need to output log message). | 2019-07-16T10:32:34 | 2019-07-16T10:32:34 | {} | MEMBER |
ClickHouse | ClickHouse | 512,350,678 | 6,030 | abraithwaite | Thanks! I was aware that it was being converted to a string but didn't realize I could order by the original field. Works great! | 2019-07-17T16:15:54 | 2019-07-17T16:15:54 | {} | NONE |
ClickHouse | ClickHouse | 512,351,734 | 5,894 | alex-krash | I've finished here. | 2019-07-17T16:18:44 | 2019-07-17T16:18:44 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 512,537,736 | 6,039 | alexey-milovidov | I've changed TotalNumberOfTables to NumberOfTables, because it will be more handy to see these metrics near each other in the list. | 2019-07-17T19:35:12 | 2019-07-17T19:35:12 | {} | MEMBER |
ClickHouse | ClickHouse | 512,871,901 | 6,050 | vitlibar | `SYSTEM RELOAD DICTIONARIES` reloads only those dictionaries which have been loaded before.
Clickhouse has not loaded `programreleasefiles` before so the command `SYSTEM RELOAD DICTIONARIES` doesn't reload it.
If you want `programreleasefiles` to be loaded use
```
SYSTEM RELOAD DICTIONARY programreleasefiles
`... | 2019-07-18T15:43:06 | 2019-07-18T15:43:48 | {
"+1": 1
} | MEMBER |
ClickHouse | ClickHouse | 511,846,116 | 2,847 | den-crane | @vitlibar I am not sure that it's fixed.
Check this https://github.com/yandex/ClickHouse/blob/master/dbms/src/Dictionaries/HTTPDictionarySource.cpp#L57
```
auto tmp_time = update_time; // time for current partial update (http_get & updated=tmp_time -1
update_time = std::chrono::system_clock::no... | 2019-07-16T14:41:05 | 2019-07-16T14:45:20 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 512,232,438 | 5,999 | filimonov | > Thanks for this workaround, but i have a table with more than 3 billion rows, and by this workaround it will take lot of time, specially we receive a file (~ 700 K rows) every 11 second.
I've gave you a [link](https://www.altinity.com/blog/2019/6/11/clickhouse-local-the-power-of-clickhouse-sql-in-a-single-command)... | 2019-07-17T12:27:19 | 2019-07-17T12:27:19 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 512,577,611 | 5,981 | 4ertus2 | I don’t understand the case with 16/32 bytes. If we cannot allocate 16 bytes it means we already out of ~Gb limit for 4 Mb. +-100 bytes make no sense. The block in MemoryTracker just breaks the loop new -> MemoryTracker.alloc -> new -> MemoryTracker.alloc | 2019-07-17T21:34:31 | 2019-07-17T21:34:31 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 513,175,251 | 5,710 | filimonov | More potential sources of leaks: system.mutations (query), system.zookeeper, system.tables (create table statements with passwords). Also (theoretically) last exception filed in dictionaries / merges / replication_queue?
Something else? | 2019-07-19T10:12:07 | 2019-07-19T10:12:07 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 512,970,470 | 2,847 | den-crane | https://github.com/yandex/ClickHouse/blob/master/dbms/src/Dictionaries/HTTPDictionarySource.cpp#L57
update_time = std::chrono::system_clock::now();
Речь про частичное обновление.
Мы запоминаем новое время, до запроса изменений из бекэнда.
Если обновление неуспешно, в следующем цикле partial update мы
прочитае... | 2019-07-18T20:15:12 | 2019-07-18T20:15:12 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 523,710,455 | 6,600 | alexey-milovidov | No test required. | 2019-08-22T01:34:23 | 2019-08-22T01:34:23 | {} | MEMBER |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.