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 | 620,821,137 | 10,551 | alexey-milovidov | We can also allow to attach existing tables. | 2020-04-28T19:55:21 | 2020-04-28T19:55:21 | {} | MEMBER |
ClickHouse | ClickHouse | 620,823,359 | 10,516 | azat | Also maybe it is a good idea to add a setting to disable this optimization (even more safe stuff has done under some switch) and even disable it by default, since I know at least one way to fool it, by using as sharding key something like `if(col1>0, col1, col2)` (although this is not common sharding key I would say, b... | 2020-04-28T19:59:52 | 2020-04-28T19:59:52 | {} | MEMBER |
ClickHouse | ClickHouse | 621,508,512 | 10,585 | alexey-milovidov | > I don't know if there exists one that can be used for performance testing and with ENABLE_TESTS=0.
It should be the only clang/relwithdebinfo/no-sanitizer build... We'll see. | 2020-04-29T22:49:12 | 2020-04-29T22:49:12 | {} | MEMBER |
ClickHouse | ClickHouse | 620,101,898 | 10,035 | vitlibar | It does work on my computer without those two lines:
```
from sqlalchemy.dialects import registry
registry.register("clickhouse", "base", "dialect")
```
What version of clickhouse-sqlalchemy are you using? | 2020-04-27T16:44:07 | 2020-04-27T16:44:07 | {} | MEMBER |
ClickHouse | ClickHouse | 620,601,349 | 9,447 | qoega | We have to add good documentation about this behavior. | 2020-04-28T13:18:18 | 2020-04-28T13:18:18 | {} | MEMBER |
ClickHouse | ClickHouse | 620,664,418 | 5,491 | CurtizJ | I meant, that it's possible to remove empty parts, only if they are continious prefix of all parts sorted by block numbers in partition, because we can't remove random part from the middle due to merge algorithm. But it will cover most of cases. BTW, it's not implemented yet. | 2020-04-28T15:05:02 | 2020-04-28T15:05:02 | {} | MEMBER |
ClickHouse | ClickHouse | 620,823,559 | 10,562 | alexey-milovidov | Ok.
BTW, we already decided to migrate all Python scripts to Python 3. | 2020-04-28T20:00:18 | 2020-04-28T20:00:18 | {} | MEMBER |
ClickHouse | ClickHouse | 621,033,703 | 8,158 | filimonov | #7659 | 2020-04-29T07:24:00 | 2020-04-29T07:24:00 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 621,205,069 | 9,334 | filimonov | @youssefqarri can you follow the recommendations above to collect debug info? | 2020-04-29T13:27:03 | 2020-04-29T13:27:03 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 621,409,855 | 10,584 | alesapin | I haven't tested it. | 2020-04-29T19:16:38 | 2020-04-29T19:16:38 | {} | MEMBER |
ClickHouse | ClickHouse | 619,989,846 | 4,393 | alexey-milovidov | @simPod I see that the documentation was edited according to your suggestions. | 2020-04-27T13:35:20 | 2020-04-27T13:35:20 | {} | MEMBER |
ClickHouse | ClickHouse | 620,239,867 | 10,128 | filimonov | #5491 | 2020-04-27T21:17:38 | 2020-04-27T21:17:38 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 620,275,905 | 9,707 | oandrew | @alexey-milovidov sure, I'll make a PR sometime soon. You can assign it to me | 2020-04-27T22:49:42 | 2020-04-27T22:49:57 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 620,409,149 | 9,948 | simPod | @alesapin any idea in which version this will be included? I went with `20.1.11.73` but it's not there yet. Thanks! | 2020-04-28T06:32:46 | 2020-04-28T06:32:46 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 620,475,685 | 10,546 | qoega | If you need some help, fell free to contact us directly. Especially to add tests in CI. | 2020-04-28T08:58:55 | 2020-04-28T08:58:55 | {} | MEMBER |
ClickHouse | ClickHouse | 620,666,339 | 10,553 | qoega | I thought about it also, but is seems less explicit. If you think it is OK, I'll try. | 2020-04-28T15:08:22 | 2020-04-28T15:08:22 | {} | MEMBER |
ClickHouse | ClickHouse | 620,856,813 | 10,516 | azat | >setting to disable this optimization (even more safe stuff has done under some switch) and even disable it by default
Added `optimize_distributed_group_by_sharding_key` (disabled by default, due to trickery). | 2020-04-28T21:10:34 | 2020-04-28T21:10:34 | {} | MEMBER |
ClickHouse | ClickHouse | 621,195,934 | 10,565 | filimonov | 1 - will brake backward compatibity
2 - may work (some extra checks needed)
3 - quite safe option, but schema migration may be tricky from the user perspecive.
also there is a 4: create a new columns `_timestamp_ms` or similar. | 2020-04-29T13:13:57 | 2020-04-29T13:14:10 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 620,069,514 | 10,480 | azat | >Can I change clickhouse logger level on the fly, or I have to always restart service?
I guess the answer is [no](https://github.com/ClickHouse/ClickHouse/blob/f34a6dc4821accaaa1b4f01b784abd3446c9df60/programs/server/Server.cpp#L471-L474) | 2020-04-27T15:47:45 | 2020-04-27T15:47:45 | {} | MEMBER |
ClickHouse | ClickHouse | 620,273,900 | 10,545 | alexey-milovidov | @sundy-li Maybe we should continue with #4672? | 2020-04-27T22:43:29 | 2020-04-27T22:43:29 | {} | MEMBER |
ClickHouse | ClickHouse | 620,349,843 | 4,283 | Fanduzi | > There is `queue_max_wait_ms` setting.
> It is user level setting, that can be set up per user profile in `users.xml` configuration file.
> When the max wait time is exceeded, ClickHouse will throw an exception nevertheless.
Is this parameter described in the documentation?, I found that many parameters of clickh... | 2020-04-28T03:02:46 | 2020-04-28T03:02:46 | {
"+1": 3
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 620,420,645 | 9,948 | alex-zaitsev | @simPod , this is for 20.4. You can check the tags of merge commits:
https://github.com/ClickHouse/ClickHouse/commit/f1e0f155565ca2601312ade404c8964de231a36d | 2020-04-28T07:02:24 | 2020-04-28T07:02:24 | {
"+1": 2
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 620,468,513 | 9,545 | akuzm | One thing that bothers me is that the perf event descriptors apparently count toward the open file limit, and it is not unusual for us to have thousands of threads, so we may easily run into the limit. It should be possible to check by running something like `select count(*) from system.zeros max_treads = 4096`. The re... | 2020-04-28T08:45:04 | 2020-04-28T08:45:04 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 620,605,634 | 9,450 | qoega | Probably we need to add help or man for clickhouse-client with binds and other features. | 2020-04-28T13:24:28 | 2020-04-28T13:24:28 | {} | MEMBER |
ClickHouse | ClickHouse | 620,781,984 | 10,553 | alexey-milovidov | I think it's better. Don't have to always keep it in mind writing new tests. | 2020-04-28T18:32:55 | 2020-04-28T18:32:55 | {} | MEMBER |
ClickHouse | ClickHouse | 620,818,776 | 10,486 | azat | >ClickHouse build check — 16/17 builds are OK
>Performance — 1 faster, 6 slower, 144 unstable
check failures looks unrelated | 2020-04-28T19:50:31 | 2020-04-28T19:50:31 | {} | MEMBER |
ClickHouse | ClickHouse | 621,153,823 | 10,530 | 4ertus2 | It looks like it use column from right table for `count(*)` as smallest one. But it's not correct for some reason.
```
create table t engine Memory as select * from numbers(2);
select count(number) from t cross join numbers(2) r;
select count(r.number) from t cross join numbers(2) r;
```
```
4
2
```
Resul... | 2020-04-29T11:52:48 | 2020-04-29T12:57:43 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 620,159,291 | 10,483 | alexey-milovidov | @vitlibar Let's suppress these messages. We should not log for every dictionary function invocation.
Or: why we check access everytime? It should be enough on creation of function object. | 2020-04-27T18:35:35 | 2020-04-27T18:35:35 | {} | MEMBER |
ClickHouse | ClickHouse | 620,172,943 | 10,488 | surekhasaharan | another update, i added below to users.xml
```
<max_memory_usage>1000000000</max_memory_usage>
<max_memory_usage_for_all_queries>1000000000</max_memory_usage_for_all_queries>
```
And now i get
```
Code: 210. DB::NetException: Connection reset by peer, while writing to socket (127.0.0.1:900... | 2020-04-27T19:02:04 | 2020-04-27T19:02:04 | {} | NONE |
ClickHouse | ClickHouse | 620,272,877 | 10,545 | den-crane | ```
Connected to ClickHouse server version 20.1.9 revision 54431.
SELECT count()
FROM numbers(10000000000)
WHERE number % 4
┌────count()─┐
│ 7500000000 │
└────────────┘
1 rows in set. Elapsed: 4.690 sec. Processed 10.00 billion rows, 80.00 GB (2.13 billion rows/s., 17.06 GB/s.)
SELECT count()
FR... | 2020-04-27T22:40:22 | 2020-04-27T23:20:36 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 620,368,610 | 10,200 | AlexanderJLiu | @KochetovNicolai
I've disabled `input_format_parallel_parsing`.
There are still three crashes this time. According to logs. One is `MergeTreeReadTask` fatal errors as usual. One is `TCPHandler` fatal errors shown as below:
```shell
2020.04.28 10:49:23.199596 [ 3394 ] {} <Fatal> BaseDaemon: #################... | 2020-04-28T04:07:48 | 2020-04-28T04:07:48 | {} | NONE |
ClickHouse | ClickHouse | 620,469,506 | 10,546 | qoega | Did you find any bugs while running new fuzzing tests?
Can you add README for other developers how to run fuzz tests? Actually we better add correct way to run fuzz tests as docs are outdated in this case https://clickhouse.tech/docs/en/development/tests/#fuzzing
You created dictionaries from system tables? Can y... | 2020-04-28T08:47:04 | 2020-04-28T08:47:04 | {} | MEMBER |
ClickHouse | ClickHouse | 620,503,914 | 7,057 | zhang2014 | @alexey-milovidov
Can we continue this task? I will resolve the conflict as soon. | 2020-04-28T09:55:52 | 2020-04-28T09:55:52 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 621,567,092 | 10,573 | kyle-cx91 | if needed, I will use latest version of CH, some data of CH come from kafka, and I dont know weather "CH kafka engine table" ganrantee the exactly-once semantics | 2020-04-30T01:59:50 | 2020-04-30T01:59:50 | {} | NONE |
ClickHouse | ClickHouse | 620,148,905 | 8,287 | filimonov | @traceon says it not actual anymore | 2020-04-27T18:15:25 | 2020-04-27T18:15:25 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 620,274,468 | 9,707 | alexey-milovidov | @oandrew There is relevant issue #5213. Maybe you could help? | 2020-04-27T22:45:12 | 2020-04-27T22:45:12 | {} | MEMBER |
ClickHouse | ClickHouse | 620,451,611 | 9,545 | akuzm | A couple of interesting files with new metrics from the performance test. You can find them in the performance test output archive (click "test output" at the bottom of the performance test report). These are the metrics for two of the troublesome queries -- the `visitParam` one often changes performance with no appar... | 2020-04-28T08:11:08 | 2020-04-28T08:11:22 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 620,468,808 | 10,200 | KochetovNicolai | Yes, it is compatible is you don't use new features available only after 19.16. | 2020-04-28T08:45:41 | 2020-04-28T08:45:41 | {} | MEMBER |
ClickHouse | ClickHouse | 620,599,362 | 10,477 | alesapin | I expected this to be a simple task, but it turned out that the virtual columns were made very messy, and I had to rewrite a lot :( | 2020-04-28T13:14:49 | 2020-04-28T13:14:49 | {} | MEMBER |
ClickHouse | ClickHouse | 620,674,965 | 10,555 | alesapin | yes, unfortunately, mutations don't work between different versions. Try to update all nodes to 20.3.8 (latest). | 2020-04-28T15:23:04 | 2020-04-28T15:23:04 | {} | MEMBER |
ClickHouse | ClickHouse | 621,238,219 | 7,878 | filimonov | Fixed in #8788 in version 19.16.13 on newer. | 2020-04-29T14:13:53 | 2020-04-29T14:13:53 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 620,256,790 | 10,510 | Quid37 | > Have you checked if it is hard to switch from OpenTracing-> OpenTelemetry or they are compatible?
> I believe that OpenTracing will be supported further, but probably it is better to use their newer project.
According to the roadmap https://medium.com/opentracing/a-roadmap-to-convergence-b074e5815289, they will s... | 2020-04-27T21:55:59 | 2020-04-27T21:55:59 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 620,336,360 | 10,506 | BernieJiangDL | > The root cause of everything is "merges are processing significantly slower than inserts".
>
> [#6720 (comment)](https://github.com/ClickHouse/ClickHouse/issues/6720#issuecomment-526045768)
Thanks~ | 2020-04-28T02:17:30 | 2020-04-28T02:17:30 | {} | NONE |
ClickHouse | ClickHouse | 620,819,712 | 8,760 | alexey-milovidov | @qoega That's obvious, the code won't build otherwise.
It works both for gcc and clang. | 2020-04-28T19:52:24 | 2020-04-28T19:52:24 | {} | MEMBER |
ClickHouse | ClickHouse | 620,821,402 | 7,057 | alexey-milovidov | Yes, we should continue and merge. | 2020-04-28T19:55:50 | 2020-04-28T19:55:50 | {
"+1": 1
} | MEMBER |
ClickHouse | ClickHouse | 620,018,297 | 6,050 | Naveen071110 | when i type system reload dictionary dictionary_name
Then it shows that connection to all replicas failed!
Can you explain why?
| 2020-04-27T14:20:22 | 2020-04-27T14:20:22 | {} | NONE |
ClickHouse | ClickHouse | 620,218,635 | 9,707 | maqroll | @oandrew
Default values doesn't seem to be working in Kafka tables neither for avro nor avroconfluent format.
If I push a null value defined as union(string,null) through clickhouse-local and read it using String DEFAULT 'default' I get 'default' (as in some your tests) but I try the same using Kafka tables I ... | 2020-04-27T20:33:18 | 2020-04-27T20:33:18 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 620,545,743 | 9,545 | And42 | > PERF_CUSTOM_INSTRUCTIONS_PER_CPU_CYCLE_SCALED looks kinda strange, sometimes it's 1 and sometimes 4636. What might be the reason? The values are from system.query_log.ProfileEvents.
Seems like a bug, as `PERF_CUSTOM_INSTRUCTIONS_PER_CPU_CYCLE_SCALED` = `PERF_COUNT_HW_INSTRUCTIONS` / `PERF_COUNT_HW_CPU_CYCLES`. And... | 2020-04-28T11:23:33 | 2020-04-28T11:23:33 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 620,625,061 | 8,643 | Kozloff | I can reproduce with another case:
```
CREATE DATABASE IF NOT EXISTS test;
CREATE TABLE IF NOT EXISTS test.test
(
d Date,
k String,
v UInt64
)
ENGINE MergeTree PARTITION BY d ORDER BY k;
-- failed with code 246
INSERT INTO test.test
SELECT
reinterpretAsDate(k) d,
reinterpretAsString(ra... | 2020-04-28T13:58:16 | 2020-04-28T13:58:16 | {} | NONE |
ClickHouse | ClickHouse | 620,806,886 | 10,477 | alesapin | Sanitizer assert and build failure are expected. | 2020-04-28T19:24:26 | 2020-04-28T19:24:26 | {} | MEMBER |
ClickHouse | ClickHouse | 621,210,482 | 9,334 | youssefqarri | @filimonov I already tried without any success. I stopped using Kafka Engine because it is not reliable.
As I am running the functionality in production, I preferred to switch to importing CSV with a custom script using CRON TAB.
By the way, I also tried JSON in first place but import stops working if my logs co... | 2020-04-29T13:34:28 | 2020-04-29T13:34:28 | {} | NONE |
ClickHouse | ClickHouse | 621,637,368 | 10,577 | filimonov | I think the only safe option to reconnect is to reuse the previosly created session with unique id (which should support that, like for http).
It should not reconnect silenly in any other scenarios, as you can be in between of some sequence of DDLs, like
```
create new_table as select * from old
-- Do some chec... | 2020-04-30T06:22:17 | 2020-04-30T06:23:34 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 619,998,561 | 10,498 | vzakaznikov | @jason-godden, for the case of something like
```sql
CREATE MATERIALIZED VIEW x TO replicated_merge_tree AS
WATCH some_live_view
```
it would not work. But what will work is `INSERT INTO replicated_merge_tree WATCH some_live_view` where you "insert right into the inner table". You would not need MATERIALIZED VIEW... | 2020-04-27T13:49:26 | 2020-04-27T13:49:26 | {
"+1": 1
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 620,125,705 | 10,483 | azat | >Once dictGet or dictGetOrDefault is used, for each row it creates a log. /var/log/clickhouse/clickhouse-server.log has many log related to it and it makes log unreadable.
I guess it is not for each row, but for each block
> A workaround would be creating a named pipe at /var/log/clickhouse/clickhouse-server.log.... | 2020-04-27T17:29:15 | 2020-04-27T17:29:15 | {} | MEMBER |
ClickHouse | ClickHouse | 620,160,987 | 10,483 | azat | >Let's suppress these messages
Added #10542 just in case (anyway I don't care about RBAC, so I will suppress them all). | 2020-04-27T18:39:02 | 2020-04-27T18:39:02 | {} | MEMBER |
ClickHouse | ClickHouse | 620,307,806 | 5,733 | den-crane | the same behavior 19.13.7.57 | 2020-04-28T00:34:36 | 2020-04-28T00:34:36 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 620,306,190 | 10,545 | azat | FWIW the difference between versions for for the query without `AND 1` looks related to processors (this numbers for the 20.1.11.73):
```sql
SET experimental_use_processors = 0;
SELECT count()
FROM numbers(10000000000)
WHERE number % 4;
┌────count()─┐
│ 7500000000 │
└────────────┘
1 rows in set. Elapse... | 2020-04-28T00:29:15 | 2020-04-28T00:43:33 | {} | MEMBER |
ClickHouse | ClickHouse | 620,519,784 | 10,150 | qoega | ```
[2563/4894] Linking CXX executable base/common/tests/date_lut2
FAILED: base/common/tests/date_lut2
: && /usr/bin/ccache /usr/bin/clang++-9 --target=x86_64-pc-freebsd12.1 --sysroot=/build/cmake/freebsd/../toolchain/freebsd-x86_64 -std=c++17 -fdiagnostics-color=always -std=c++2a -fsized-deallocation -gdwarf-ara... | 2020-04-28T10:26:49 | 2020-04-28T10:26:49 | {} | MEMBER |
ClickHouse | ClickHouse | 620,088,122 | 10,526 | alesapin | test failures are OK.
| 2020-04-27T16:18:58 | 2020-04-27T16:19:11 | {} | MEMBER |
ClickHouse | ClickHouse | 620,207,026 | 10,436 | filimonov | > And my version 20.3.5.21, is the latest release.
It is NOT. There were 3 bugfix releases since that (look on their third number in version)
https://github.com/ClickHouse/ClickHouse/releases/tag/v20.3.8.53-lts
The log fragments you send are not useful.
Please update. | 2020-04-27T20:10:33 | 2020-04-27T20:10:33 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 620,212,874 | 10,488 | surekhasaharan | i have the query working with `<max_memory_usage>64000000000</max_memory_usage>`, i'll close this issue, thanks . | 2020-04-27T20:22:10 | 2020-04-27T20:22:10 | {} | NONE |
ClickHouse | ClickHouse | 620,651,278 | 10,496 | qoega | We need to document current approach of memory tracking. Both in parts about limits and metrics. | 2020-04-28T14:43:02 | 2020-04-28T14:43:02 | {} | MEMBER |
ClickHouse | ClickHouse | 621,176,275 | 3,990 | junneyang | https://github.com/Altinity/clickhouse-operator | 2020-04-29T12:38:33 | 2020-04-29T12:38:33 | {
"+1": 1
} | NONE |
ClickHouse | ClickHouse | 621,248,991 | 8,282 | filimonov | Actually it looks like that message hides the original error message.
If you will try to run the same query without ON CLUSTER - most probably you will see the real error message.
```
ALTER TABLE repl.`.inner.bar` ON CLUSTER testcluster
ADD COLUMN
`tokens` AggregateFunction('uniq', UInt64)
┌─host──... | 2020-04-29T14:30:30 | 2020-04-29T14:33:05 | {
"+1": 2
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 620,512,322 | 10,553 | alexey-milovidov | It can be embedded in `${CLICKHOUSE_CURL}` | 2020-04-28T10:12:15 | 2020-04-28T10:12:15 | {} | MEMBER |
ClickHouse | ClickHouse | 620,543,684 | 7,690 | PHaroZ | @Slach The issue still exist with v20.3.7.46 (the latest current version) | 2020-04-28T11:19:19 | 2020-04-28T11:19:19 | {} | NONE |
ClickHouse | ClickHouse | 620,671,414 | 10,558 | qoega | What are you going to use as source of timestamps? | 2020-04-28T15:17:10 | 2020-04-28T15:17:10 | {} | MEMBER |
ClickHouse | ClickHouse | 621,061,082 | 10,572 | akuzm | On a side note, if an exception has a cryptic text that mentions code artifacts (ColumnTuple etc.), its code must be `LOGICAL_ERROR`. | 2020-04-29T08:24:49 | 2020-04-29T08:24:49 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 621,214,059 | 10,573 | den-crane | 1. This statement is valid for only for some engines.
2. CH does not use ` Kafka auto commit `.
What ClickHouse version do you use?
| 2020-04-29T13:39:29 | 2020-04-29T13:39:29 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 620,020,572 | 9,707 | oandrew | @maqroll
Could you test again?
Build from here: https://clickhouse-builds.s3.yandex.net/0/436e4e3f5eabbf03e9493c9183cc0c6f8f628882/report.html | 2020-04-27T14:24:31 | 2020-04-27T14:24:31 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 620,513,257 | 10,551 | qoega | Thank you. I found it already. Actually added this issue, as your fix can be backward incompatible and wanted to leave information is someone also will encounter this problem. | 2020-04-28T10:14:05 | 2020-04-28T10:14:05 | {} | MEMBER |
ClickHouse | ClickHouse | 620,525,537 | 10,546 | kyprizel | Ok, added minimal documentation. Docker configuration for fuzzing will be in separate pull. @qoega | 2020-04-28T10:37:10 | 2020-04-28T11:14:50 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 620,782,315 | 10,553 | alexey-milovidov | And I don't see any reason why any test wants to avoid the usage of this option. | 2020-04-28T18:33:32 | 2020-04-28T18:33:32 | {} | MEMBER |
ClickHouse | ClickHouse | 620,783,405 | 10,411 | alexey-milovidov | We can add implicit constraint on INSERT. It will lower performance of INSERTs to CollapsingMergeTree in some neglible amount.
Or we can simply allow any Sign and only distinct between positive and negative values. | 2020-04-28T18:35:39 | 2020-04-28T18:35:39 | {} | MEMBER |
ClickHouse | ClickHouse | 620,819,439 | 10,516 | azat | >ClickHouse build check — 16/17 builds are OK
>Functional stateless tests (ubsan) — fail: 1, passed: 1788, skipped: 5
>Integration tests (asan) — failed:1, passed:540, error:0
>Performance — 1 faster, 5 slower, 139 unstable
check failures looks unrelated | 2020-04-28T19:51:52 | 2020-04-28T19:51:52 | {} | MEMBER |
ClickHouse | ClickHouse | 620,825,226 | 10,562 | azat | >BTW, we already decided to migrate all Python scripts to Python 3.
Yes I though that it will be better to migrate them to python3, but decided not to look into this since this will (at least) require updating some docker images. | 2020-04-28T20:03:34 | 2020-04-28T20:03:34 | {} | MEMBER |
ClickHouse | ClickHouse | 620,945,035 | 10,436 | gglinux | Thank you. I'll try v20.3.8.53-lts.
In addition, which version is the most stable between tls and stable.
| 2020-04-29T01:48:44 | 2020-04-29T01:48:44 | {} | NONE |
ClickHouse | ClickHouse | 620,955,292 | 10,566 | den-crane | ```
20.4.1.3163
SELECT * FROM x ORDER BY t LIMIT 4;
┌───────────────────t─┬─v─┐
│ 2020-04-29 02:12:47 │ 1 │
└─────────────────────┴───┘
1 rows in set. Elapsed: 0.001 sec.
SET experimental_use_processors = 0
SELECT * FROM x ORDER BY t LIMIT 4;
┌───────────────────t─┬─v─┐
│ 2020-04-29 02:12:47 │ 1... | 2020-04-29T02:16:30 | 2020-04-29T02:16:30 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 621,632,573 | 9,334 | filimonov | > @filimonov I already tried without any success.
No success in collecting logs?
> I stopped using Kafka Engine because it is not reliable.
That says nothing about the real problems you face.
I don't know which version did you use, no logs, that exactly went wrong. No clues giving ideas how to help you or ... | 2020-04-30T06:07:24 | 2020-04-30T06:08:16 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 620,064,960 | 10,498 | vzakaznikov | @jason-godden, for an example of how to use live views for computing continuous aggregates I've added a new test 00979_live_view_watch_continuous_aggregates.py (https://github.com/ClickHouse/ClickHouse/pull/10498/commits/9d280665391a3f171ba40bb0833ab6afe0457cbe). | 2020-04-27T15:39:51 | 2020-04-27T15:44:31 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 620,087,495 | 10,480 | ermin24 | Thanks) | 2020-04-27T16:17:43 | 2020-04-27T16:17:43 | {} | NONE |
ClickHouse | ClickHouse | 620,301,316 | 8,936 | vzakaznikov | Hi @byx313, no, live views are still experimental. We still working on fully defining the feature and its implementation. | 2020-04-28T00:12:02 | 2020-04-28T00:12:02 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 620,370,899 | 10,200 | AlexanderJLiu | > You can also try to install 19.16 before 20.3 to exclude possible recent bugs.
Are there some other problems `downgrade` to `19.16`? I mean data is now processed by `20.3`, is it compatible for lower version?
Thanks | 2020-04-28T04:16:55 | 2020-04-28T04:16:55 | {} | NONE |
ClickHouse | ClickHouse | 620,471,950 | 10,546 | kyprizel | Dictionaries are generated from the source code and the scripts are not good enough to share.
Yes, I've found some vulnerabilities and they were fixed already.
Those fuzz tests are for "libfuzzer" engine and will let me add ClickHouse as a part of Google Oss-Fuzz project. I'll add the fuzzing documentation later.
| 2020-04-28T08:51:47 | 2020-04-28T08:51:47 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 620,551,954 | 8,760 | qoega | We should add note for developers that we require to explicitly define overrides and have checks for them. | 2020-04-28T11:38:02 | 2020-04-28T11:38:02 | {} | MEMBER |
ClickHouse | ClickHouse | 620,306,285 | 10,506 | den-crane | The root cause of everything is "merges are processing significantly slower than inserts".
https://github.com/ClickHouse/ClickHouse/issues/6720#issuecomment-526045768
| 2020-04-28T00:29:39 | 2020-04-28T00:30:53 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 620,400,585 | 10,035 | kursatsahin74 | I have solved the issue. Please install clickhouse-sqlalchemy package , not sqlalchemy-clickhouse it has some bugs, and gives the above errors. | 2020-04-28T06:07:38 | 2020-04-28T06:07:38 | {
"+1": 2
} | NONE |
ClickHouse | ClickHouse | 620,453,807 | 2,041 | Naveen071110 | What to do if SYSTEM RELOAD DICTIONARIES query throws
Received from localhost:9000. DB::Exception: Poco::Exception. Code: 1000, e.code() = 0, e.displayText() = Exception: Connections to all replicas failed:
This error. | 2020-04-28T08:15:40 | 2020-04-28T08:15:40 | {} | NONE |
ClickHouse | ClickHouse | 621,143,069 | 10,383 | akuzm | A general question: what is finalization? Presumably, an aggregation function has some complex state in run time, and then we finalize it before we can calculate the result.
* can we add more values to finalized state?
* can we merge finalized states? finalized to unfinalized?
* what state we serialize in aggregate ... | 2020-04-29T11:29:04 | 2020-04-29T11:29:16 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 620,010,293 | 10,520 | qoega | 1) Probably https://github.com/PNixx/clickhouse-activerecord should be added in this list as well.
2) When I opened your link, I could find ClickHouse mention only in this tests section
https://github.com/bitquery/activecube#rspec-tests
3) And nothing about ClickHouse in second one
https://github.com/bitquery/act... | 2020-04-27T14:08:02 | 2020-04-27T14:08:16 | {} | MEMBER |
ClickHouse | ClickHouse | 620,236,605 | 9,707 | oandrew | @maqroll
Looks like Kafka storage does not use `AddingDefaultsBlockInputStream` (it's responsible for the defaults). I'll probably add it at some point.
For now you can just use https://clickhouse.tech/docs/en/sql_reference/functions/functions_for_nulls/#ifnull before inserting into target table | 2020-04-27T21:10:23 | 2020-04-27T21:10:23 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 620,493,587 | 10,551 | zhang2014 | Related: #10431
This is as expected. Change 1.01 to 1 in metadata sql file
This is consistent with document
```
The optional false_positive parameter is the probability of receiving a false positive response from the filter. Possible values: (0, 1). Default value: 0.025.
``` | 2020-04-28T09:35:10 | 2020-04-28T09:41:06 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 620,757,530 | 10,558 | maxkuzn | > What are you going to use as source of timestamps?
It is a part of big feature of [adding transactions to MergeTree engines](https://clickhouse.tech/docs/ru/whats_new/extended_roadmap/#minimalnaia-podderzhka-tranzaktsii-dlia-mnozhestva-vstavokchtenii)
Timestamp - just some monotonically increasing value. | 2020-04-28T17:46:47 | 2020-04-28T17:46:47 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 621,567,534 | 10,573 | kyle-cx91 | Of course, we do not consider the duplication of Kafka ’s own data | 2020-04-30T02:01:10 | 2020-04-30T02:01:10 | {} | NONE |
ClickHouse | ClickHouse | 619,982,098 | 4,469 | alexey-milovidov | Implemented in #10382 and #8361 #9606 #8439 #8456 #8701. | 2020-04-27T13:21:36 | 2020-04-27T13:21:36 | {} | MEMBER |
ClickHouse | ClickHouse | 620,827,082 | 10,562 | alexey-milovidov | No problem, just modify Dockerfiles and we will rebuild images. | 2020-04-28T20:07:32 | 2020-04-28T20:07:32 | {} | MEMBER |
ClickHouse | ClickHouse | 621,010,321 | 10,559 | akuzm | @ildus So I continued the thing I mentioned about statically dispatching on argument type, and also cleaned up the code a bit. Surprisingly, the last perf test shows that the performance improved even for the non-tuple version (this is for the penultimate commit from this PR, see https://clickhouse-test-reports.s3.yand... | 2020-04-29T06:12:55 | 2020-04-29T06:12:55 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 621,146,833 | 10,383 | akuzm | In general, if we can't put finalization into the function that calculates the result -- it means that finalization is some meaningful stage of computation that is distinct from the stage where we compute the final result. We must describe what data we have on this stage, when does it happen as opposed to computation o... | 2020-04-29T11:38:35 | 2020-04-29T11:38:35 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 629,722,397 | 10,745 | zhbdesign | I have the same problem. Have you solved it now?
| 2020-05-17T00:08:15 | 2020-05-17T00:08:15 | {} | NONE |
ClickHouse | ClickHouse | 629,726,838 | 5,746 | zhbdesign | Is this a normal problem? How did you solve it? | 2020-05-17T01:00:11 | 2020-05-17T01:00:11 | {} | NONE |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.