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 | 354,507,365 | 1,689 | silviucpp | @ztlpn I see the difference In backgroundprocessingpool it's used wake_event.wait_for(lock,
std::chrono::duration<double>(sleep_seconds
+ std::uniform_real_distribution<double>(0, sleep_seconds_random_part)(rng)));
and in worst case the task is executed a bit later and w... | 2017-12-29T22:23:10 | 2017-12-29T22:23:10 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 355,747,530 | 1,575 | rtsisyk | Do you actually need boost 1.65? Why not to support boost 1.53+ from apt/yum and get rid of bundled boost and submodules?
Ubuntu Xenial: 1.58
Ubuntu Zesty: 1.62
Debian Stretch: 1.62.0.1
Fedora: 1.64.0
RHEL 7: 1.53.0
| 2018-01-06T13:37:43 | 2018-01-06T13:37:43 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 356,078,580 | 1,169 | filimonov | Yep. AFAIR on September's presentation @alexey-milovidov said that they plan to work on fixing that in some close perspective (as they need to do some resharding in their datacenter).
For now - they propose just to increase weigth of new server(s) for inserts. | 2018-01-08T19:58:52 | 2018-01-08T19:58:52 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 355,483,710 | 1,726 | badvir | My log data is inserting to each log table that is a merge tree engine.
**In Global_Master, all query is fine on the distributed table when configured as above picture.**
Does it support distributed tab... | 2018-01-05T06:38:56 | 2018-01-09T00:08:32 | {} | NONE |
ClickHouse | ClickHouse | 356,641,920 | 1,665 | cyfdecyf | I made a dirty hack to import float64 precisely. This works for me as a temporary workaround for now. | 2018-01-10T15:45:45 | 2018-01-10T15:45:45 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 356,936,405 | 1,743 | ztlpn | This is the intended behavior. You can view the container logs with the `docker logs` command.
If you still need the server to log to a file, you should override the container entrypoint. Something like this:
```
$ docker run -d --name some-clickhouse-server --ulimit nofile=262144:262144 --entrypoint="" yandex/cli... | 2018-01-11T13:39:20 | 2018-01-11T15:08:05 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 357,226,159 | 1,753 | ztlpn | Yes, this is a bug. Merge assigned by OPTIMIZE are not properly coordinated with the regular merge scheduling process. This can lead to replication log entries that cannot be executed (as in your case) and even to intersecting parts. | 2018-01-12T12:28:11 | 2018-01-12T12:28:11 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 357,678,790 | 1,754 | ztlpn | In general you can expect performance of queries with FINAL to be significantly worse than ordinary queries, because they are executed in the single thread. By contrast GROUP BY statements are executed in parallel (that doesn't depend on which MergeTree engine is used). Note that you can emulate FINAL to the extent wit... | 2018-01-15T13:11:40 | 2018-01-15T13:11:40 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 354,175,475 | 1,696 | alexey-milovidov | Merged in master. | 2017-12-27T20:47:32 | 2017-12-27T20:47:32 | {} | MEMBER |
ClickHouse | ClickHouse | 356,115,017 | 217 | harlinb | This bug has regressed to the point where the max_execution_time does not work for any queries against a distributed table:
event_rep is a replicated table
event_dist is a distributed table over event_rep
```
select * from system.settings where name='max_execution_time'
┌─name───────────────┬─value─┬─changed─┐
... | 2018-01-08T22:19:45 | 2018-01-08T22:50:00 | {
"+1": 2
} | NONE |
ClickHouse | ClickHouse | 357,596,237 | 1,765 | wormen | @Slach в C++ к сожалению нет опыта | 2018-01-15T06:50:06 | 2018-01-15T06:50:06 | {} | NONE |
ClickHouse | ClickHouse | 354,506,260 | 1,689 | silviucpp | @ztlpn I'll review again all your observations. delayExecutionThreadFunction is more or less inspired from BackgroundProcessingPool::threadFunction where we have more or less same locking logic.
On a first look I cant spot why there works :) | 2017-12-29T22:10:18 | 2017-12-29T22:10:18 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 354,942,404 | 1,713 | shangshujie365 | Hi, after add some debug log and do some tests, I found:
1) the 'inner aggregate function with group by' works just like what I want
2) the 'outer aggregate function' is grouped by the distributed keys, so it should be aggregated first in sub-data-node before the result send to the query-node, so, it seems ch lack s... | 2018-01-03T06:01:08 | 2018-01-03T06:01:08 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 356,351,440 | 1,734 | alexey-milovidov | Ok. | 2018-01-09T17:15:36 | 2018-01-09T17:15:36 | {} | MEMBER |
ClickHouse | ClickHouse | 356,354,077 | 1,733 | alexey-milovidov | Looks Ok. Minor issues remain. | 2018-01-09T17:24:53 | 2018-01-09T17:24:53 | {} | MEMBER |
ClickHouse | ClickHouse | 356,364,500 | 1,720 | alexey-milovidov | :+1: | 2018-01-09T18:01:24 | 2018-01-09T18:01:24 | {} | MEMBER |
ClickHouse | ClickHouse | 356,910,517 | 1,469 | ztlpn | @kszucs You are right. `runningIncome()` is a pretty misleading name for this function, so in the next release it will be renamed to `runningDifferenceStartingWithFirstValue()`. For the first row it outputs the input value instead of 0:
```
:) select number + 123 as x, runningDifference(x), runningDifferenceStartingW... | 2018-01-11T11:43:05 | 2018-01-11T11:43:05 | {
"+1": 7
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 357,456,681 | 1,758 | kmatt | Its possible based on counts in table and from source files that all inserts are succeeding, but backlog of .bin files grows with new inserts, and those .bin files are not logged as successful or with error. | 2018-01-13T18:39:20 | 2018-01-13T18:39:20 | {} | NONE |
ClickHouse | ClickHouse | 357,652,004 | 1,765 | wormen | @filimonov спасибо за информативность | 2018-01-15T11:07:12 | 2018-01-15T11:07:12 | {} | NONE |
ClickHouse | ClickHouse | 354,156,421 | 1,458 | alexey-milovidov | > We have problems with 3.4.11 so had to rollback to 3.4.9
I would like to know, what issues do you have? This information will be viable for our ops. | 2017-12-27T18:36:50 | 2017-12-27T18:36:50 | {} | MEMBER |
ClickHouse | ClickHouse | 354,326,716 | 1,710 | amosbird | ## Before this patch
```
QPS: 2.528, RPS: 257865127.011, MiB/s: 983.677, result RPS: 2.528, result MiB/s: 0.000.
0.000% 0.363 sec.
10.000% 0.369 sec.
20.000% 0.370 sec.
30.000% 0.372 sec.
40.000% 0.374 sec.
50.000% 0.377 sec.
60.000% 0.398 sec.
70.000% 0.414 sec.
80.000% 0.424 sec.
90.000% 0.441 sec.
95.0... | 2017-12-28T17:44:17 | 2017-12-28T17:47:55 | {} | COLLABORATOR |
ClickHouse | ClickHouse | 356,355,536 | 1,735 | proller | Please test it: https://github.com/yandex/ClickHouse/commit/694c6ed2430448a45ba6c99b65b234f555630bfb | 2018-01-09T17:29:50 | 2018-01-09T17:29:50 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 356,495,204 | 1,737 | MetikovVadim | The problem was in file permissions. | 2018-01-10T04:16:43 | 2018-01-10T04:16:43 | {
"+1": 3
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 356,669,152 | 1,724 | ztlpn | Vitaliy will investigate. | 2018-01-10T17:08:35 | 2018-01-10T17:08:35 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 356,927,081 | 1,728 | ztlpn | Unfortunately the docs for this table are not written yet :( I've created a task for our documenting team. | 2018-01-11T13:00:01 | 2018-01-11T13:00:01 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 357,223,356 | 1,724 | ludv1x | It is because of that commit: https://github.com/yandex/ClickHouse/commit/ad59a1460eac895f3158237ae190ca1e268c19fb
Result size check is done only in the next `read()` call, so it makes possible to push data in the output buffer even if the data size is greater than `max_result_bytes`. | 2018-01-12T12:12:55 | 2018-01-12T12:13:56 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 357,646,444 | 1,765 | filimonov | Информация "побольше": в настоящее время у разработчиков ClickHouse имеется немало разнородных задач, направленных на улучшение функциональности продукта (о ближайщих планах можно прочитать в ROADMAP: https://clickhouse.yandex/docs/en/roadmap.html ).
При этом задача замены Zookeepera, не привнося новых функций и не... | 2018-01-15T10:43:58 | 2018-01-15T10:43:58 | {
"+1": 5
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 354,320,462 | 1,710 | alexey-milovidov | What about performance tests?
If the performance decrease (at least 1%), it's important to choose, for what feature we will trade the performance.
For example, I am ready to trade few percent of performance to add support of parsing DateTime in full ISO 8601 format like `2017-12-28T15:01:29+00:00` and `2017-12-28... | 2017-12-28T17:01:14 | 2017-12-28T17:01:14 | {
"+1": 1
} | MEMBER |
ClickHouse | ClickHouse | 354,392,170 | 1,713 | shangshujie365 | I sugest clickhouse can call merge between the execute threads before they send the results to the initial thread, and that is what merge means.
For execution plan now clickhouse has, the merge/ serialize / deserialize is useless. | 2017-12-29T03:16:35 | 2017-12-29T03:16:35 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 356,020,144 | 1,169 | rpfennin | Guys I think this is a key feature for mass adoption. Without this, and I'm in the middle of going through this process, it is a pain in the butt when you grow a cluster and want to evenly distribute historical data. | 2018-01-08T16:38:51 | 2018-01-08T16:38:51 | {} | NONE |
ClickHouse | ClickHouse | 357,113,753 | 1,744 | naah69 | but 2018-02-31...
is there 02-31?
it return 2018-03-01 | 2018-01-12T01:12:44 | 2018-01-12T01:12:44 | {} | NONE |
ClickHouse | ClickHouse | 357,217,592 | 1,729 | filimonov | About Pretty* format - it's not so clear how to deal with that. One of the possible options is just "document & let user choose".
For now that behavior leads to some misunderstanding what's in DB, especially if you're inserting some strings with slashes or special characters, and trying to check how the data look li... | 2018-01-12T11:42:18 | 2018-01-12T11:42:18 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 357,648,231 | 1,761 | filimonov | May be that will work?
```sql
SYSTEM RELOAD DICTIONARIES
```
| 2018-01-15T10:51:15 | 2018-01-15T10:51:15 | {
"+1": 1
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 354,711,840 | 520 | bobrik | It definitely disappears at the end, but it takes hours to take to the end. My main issue is that read queries run a lot slower because there's no throttling. We see writes at 1.8GB/s when reads from 2 replicas saturate disks at ~900MB/s.
Should I file an issue to throttle recovery bandwidth (hopefully with zookeepe... | 2018-01-02T07:04:49 | 2018-01-02T07:04:49 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 354,872,643 | 1,322 | silviucpp | @alexey-milovidov This bug is still present on master | 2018-01-02T20:53:58 | 2018-01-02T20:53:58 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 355,746,986 | 1,733 | rtsisyk | @proller, please review this follow-up for your fixes. | 2018-01-06T13:27:22 | 2018-01-06T13:27:22 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 357,594,821 | 1,755 | windreamer | OK, we should use `reverse` to fix the endianness
```
SELECT
hex(1234) AS str,
reinterpretAsUInt32(reverse(unhex(str))) AS integer,
reinterpretAsUInt32(unhex('D204'))
┌─str──┬─integer─┬─reinterpretAsUInt32(unhex(\'D204\'))─┐
│ 04D2 │ 1234 │ 1234 │
└──────┴────... | 2018-01-15T06:39:29 | 2018-01-15T06:39:29 | {} | NONE |
ClickHouse | ClickHouse | 357,597,473 | 1,765 | wormen | ... ожидал информации по этому поводу побольше ... | 2018-01-15T06:58:26 | 2018-01-15T06:58:26 | {} | NONE |
ClickHouse | ClickHouse | 357,650,051 | 1,756 | filimonov | In ClickHouse date is stored as number of days since '1970-01-01'. So the '1970-01-01' itself is 'zero days since 1970-01-01'.
But zero for Date type in ClickHouse has a special meaning 'unknown' / 'undefined' date, and is shown as '0000-00-00'.
And that works the same without any timezones:
```
SELECT toDat... | 2018-01-15T10:58:39 | 2018-01-15T11:01:14 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 354,154,757 | 1,458 | andr83 | We have problems with 3.4.11 so had to rollback to 3.4.9 | 2017-12-27T18:24:29 | 2017-12-27T18:24:29 | {} | NONE |
ClickHouse | ClickHouse | 354,163,502 | 520 | alexey-milovidov | It depends... Does the queue disappear after successful recovery? Does the recovery goes slower than expected? Do you see errors in log about failed fetches?
This issue is about using excessive number of network connections during replica recovery,
that can lead to slower recovery and periodic failures in fetching ... | 2017-12-27T19:13:18 | 2017-12-27T19:13:18 | {} | MEMBER |
ClickHouse | ClickHouse | 354,926,996 | 1,665 | cyfdecyf | I'm also affected by the float parsing precision problem. I am trying to hack `readFloatTextImpl` in `ReadHelpers.h` to work around the problem for now. | 2018-01-03T02:41:55 | 2018-01-03T02:41:55 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 355,533,141 | 1,394 | hagen1778 | [gcsfuse](https://github.com/GoogleCloudPlatform/gcsfuse) is a user-space file system for interacting with Google Cloud Storage. Means you can store your data at any type of Google Cloud Storage. Gcsfuse is written in python and will impact performance (see https://github.com/GoogleCloudPlatform/gcsfuse#performance) b... | 2018-01-05T11:25:10 | 2018-01-05T11:25:35 | {} | NONE |
ClickHouse | ClickHouse | 355,961,635 | 1,169 | komex | Any other fuctions for `RESHARD` does not works too. | 2018-01-08T13:07:21 | 2018-01-08T13:07:21 | {} | NONE |
ClickHouse | ClickHouse | 356,605,165 | 1,738 | ztlpn | This looks like the issue #912 which was fixed in version 1.1.54289. Which ClickHouse version are you using? Can you try the latest version and see if the problem goes away? | 2018-01-10T13:41:18 | 2018-01-10T13:41:18 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 356,930,708 | 1,469 | kszucs | Hey @ztlpn! Thanks for the clarification!
Currently I'm implementing ClickHouse backend for https://github.com/ibis-project/ibis which is a match made in heaven except for the missing window functionality in CH which I'd really like to use.
Any chance for a preliminary implementation for that in the near future?
... | 2018-01-11T13:15:47 | 2018-01-11T13:15:47 | {
"+1": 4
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 357,020,489 | 1,744 | alexey-milovidov | The documentation should state, that the behaviour of parsing incorrect dates is implementation specific.
Returning zero date, throwing an exception or doing "natural" overflow - all are conforming.
Maybe we should change this behaviour to be more strict. | 2018-01-11T18:36:45 | 2018-01-11T18:36:45 | {} | MEMBER |
ClickHouse | ClickHouse | 357,117,810 | 1,665 | alexey-milovidov | I've implemented correct parsing of floats: https://github.com/yandex/ClickHouse/compare/correct-parse-floats?expand=1
Added peformance test: `dbms/tests/performance/float_parsing/float_parsing.xml` for various cases.
In extreme case, it has 1.6 times performance decrease (when parsing integer numbers as floats).
... | 2018-01-12T01:38:26 | 2018-01-12T01:41:37 | {} | MEMBER |
ClickHouse | ClickHouse | 357,240,707 | 1,665 | cyfdecyf | @alexey-milovidov I get following error message when running my own performance test mentioned above:
```
Code: 72. DB::Exception: Cannot read floating point value: (at row 1)
Row 1:
Column 0, name: dt, type: Date, parsed text: "2018-01-11"
Column 1, name: v, type: Float64, ERROR: text "171.721937" is ... | 2018-01-12T13:39:46 | 2018-01-12T13:39:46 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 357,658,722 | 1,751 | silviucpp | Ok thx,
I thought there might be something smarter. not sure how this works for huge tables
Silviu | 2018-01-15T11:38:02 | 2018-01-15T11:38:02 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 355,293,148 | 1,052 | theseusyang | this is a really good point, we will prepare to develop these plugins, if other guys want to join this plan, welcome!
| 2018-01-04T14:18:38 | 2018-01-04T14:18:38 | {} | NONE |
ClickHouse | ClickHouse | 356,353,634 | 1,735 | proller | it will add lot of small tests with "wrong" static link of huge libdbms, it will increase build time and use additional 10g of disk | 2018-01-09T17:23:22 | 2018-01-09T17:23:22 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 357,239,182 | 1,724 | ludv1x | @vavrusa If I correctly understand your motivation then this fix should fix the test and satisfy your needs for Kafka engine?
https://github.com/yandex/ClickHouse/pull/1757/commits/808c32408dc715de32a6202e30d070fb9fb6605a#diff-9e258e8c15508184c7fac6f6ded99afcL38 | 2018-01-12T13:33:32 | 2018-01-14T19:29:18 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 355,985,326 | 1,169 | filimonov | https://github.com/yandex/ClickHouse/issues/198#issuecomment-330916156 | 2018-01-08T14:45:13 | 2018-01-08T14:45:13 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 356,787,117 | 1,469 | kszucs | With the following versions:
> ClickHouse client version 1.1.54310
> Connected to ClickHouse server version 1.1.54310.
It seems like both runningIncome and runningDifference are implementing difference operation.
The [implementation](https://github.com/yandex/ClickHouse/blob/master/dbms/src/Functions/FunctionsM... | 2018-01-11T00:45:19 | 2018-01-11T00:45:59 | {
"+1": 1
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 356,969,871 | 1,469 | ztlpn | Cool! Of course window functions support is a must-have feature but it is a pretty big task and currently there is no definite timeline. You can get pretty far with arrays and ARRAY JOINs though. Your query becomes
```sql
select arrayJoin(values) - avg_value from (
select avg(value) as avg_value, groupArray(value)... | 2018-01-11T15:52:55 | 2018-01-11T15:52:55 | {
"+1": 3
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 354,493,145 | 1,633 | silviucpp | @alexey-milovidov what's the second variant :) You told : Then one of the following variants: but listed only one :) | 2017-12-29T19:54:14 | 2017-12-29T19:54:14 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 354,746,334 | 1,633 | silviucpp | Ohh I forgot . Also using the current way clickhouse it's reading the buffers:
```c++
parse_buf.resize(max_query_size);
parse_buf.resize(istr.read(&parse_buf[0], max_query_size));
```
In case you set the `max_query_size` to 10MB each query will allocate 10 MB temporary until read complete. This kind of approac... | 2018-01-02T11:17:07 | 2018-01-02T11:17:07 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 355,526,170 | 1,394 | theseusyang | Hi @valyala , this is an open-source project? | 2018-01-05T10:48:08 | 2018-01-05T10:48:08 | {} | NONE |
ClickHouse | ClickHouse | 354,159,529 | 1,633 | alexey-milovidov | Reading no more than `max_query_size + 1` into the buffer for parsing.
Then one of the following variants:
- pass `end` at `max_query_size + 1` to lexer; throw proper exception message if query parser required to go after max_query_size (both in cases of syntax error or correctly parsed query fragment);
| 2017-12-27T18:55:40 | 2017-12-27T18:55:40 | {} | MEMBER |
ClickHouse | ClickHouse | 354,987,938 | 1,560 | silviucpp | Not fixed from what I’m aware
On Wed, 3 Jan 2018 at 11:45, Ivan Jobs <notifications@github.com> wrote:
> Is this bug done? I plan to fix this.
>
> —
> You are receiving this because you authored the thread.
> Reply to this email directly, view it on GitHub
> <https://github.com/yandex/ClickHouse/issues/1560#issuecomm... | 2018-01-03T11:11:56 | 2018-01-03T11:11:56 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 356,352,997 | 1,735 | rtsisyk | > We need to add tests always to make debian package clickhouse-test with script clickhouse-test
Why not to add -DENABLE_TESTS=ON to debian/rules?
| 2018-01-09T17:20:58 | 2018-01-09T17:20:58 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 356,673,733 | 1,724 | jbfavre | I think it comes from the fact I don't use libpoco from contrib.
But then, I'd like to update the test so that I can check the compression behaviour. | 2018-01-10T17:23:23 | 2018-01-10T17:23:23 | {} | NONE |
ClickHouse | ClickHouse | 356,825,875 | 1,665 | cyfdecyf | @filimonov Just done a simple performance test, import time is a little more than doubled.
Test data is a two columns (date and double) CSV with 10,000,000 rows, 279MB on disk.
```python
import numpy as np
import pandas as pd
df = pd.DataFrame()
df['v'] = np.random.uniform(size=10000000) * 1000
df['dt'] = ... | 2018-01-11T05:01:42 | 2018-01-11T05:01:42 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 356,922,977 | 1,703 | ztlpn | There is a bug in 54236 that causes bad connections to remain in the connection pool for Distributed inserts. You should upgrade. For CentOS please use rpms provided by Altinity: https://packagecloud.io/altinity/clickhouse | 2018-01-11T12:41:48 | 2018-01-11T12:41:48 | {
"+1": 2
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 354,473,395 | 1,285 | alexey-milovidov | Fixed in master. | 2017-12-29T17:13:19 | 2017-12-29T17:13:19 | {} | MEMBER |
ClickHouse | ClickHouse | 355,597,623 | 1,731 | amosbird | To reproduce the issue, run
```
select count() from (select toUInt64(1) s limit 1) any right join (select number s from numbers(65537)) using (s) ;
```
It should return `65537` but `65538` | 2018-01-05T16:25:03 | 2018-01-05T16:25:03 | {} | COLLABORATOR |
ClickHouse | ClickHouse | 355,604,163 | 1,394 | hagen1778 | You don't need a plugin.
We are using gcsfuse only for RO purposes and I don't know how it will handle writes but you can try it out. What we do:
* install gcsfuse
* uncomment `user_allow_other` option at gcsfuse config `/etc/fuse.conf`
* [mount](https://github.com/GoogleCloudPlatform/gcsfuse#invoking-gcsfuse) Goo... | 2018-01-05T16:49:04 | 2018-01-05T16:49:04 | {} | NONE |
ClickHouse | ClickHouse | 356,733,970 | 1,724 | alexey-milovidov | > I think it comes from the fact I don't use libpoco from contrib.
It's not related. We actually have this test failing right now. | 2018-01-10T20:56:10 | 2018-01-10T20:56:10 | {} | MEMBER |
ClickHouse | ClickHouse | 354,331,913 | 1,713 | alexey-milovidov | > So, why not combine data in local node, and just pass the result of 'MyUdaf' to the query node? It's a small number of data compare to the mass pre-aggregate data
This is exactly how ClickHouse works. On each server, it first combines all aggregation states from each thread of execution and then send combined stat... | 2017-12-28T18:21:18 | 2017-12-28T18:21:18 | {} | MEMBER |
ClickHouse | ClickHouse | 354,420,331 | 1,632 | kshvakov | @yurchenkosv Помогло? | 2017-12-29T09:19:29 | 2017-12-29T09:19:29 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 354,971,569 | 1,560 | IvanJobs | Is this bug done? I plan to fix it.
| 2018-01-03T09:45:38 | 2018-01-03T09:45:49 | {} | NONE |
ClickHouse | ClickHouse | 355,738,404 | 1,732 | alex-zaitsev | Probably the main difference is that In Rainbird you are stick with pre-defined counters and hierarchies. In ClickHouse you have full SQL flexibility. We tried Rainbird approach at some point for an application where it seem to fit, but it proved to be much worse than dedicated RDBMS (Vertica or ClickHouse).
| 2018-01-06T10:36:17 | 2018-01-06T10:36:57 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 356,359,618 | 1,727 | alexey-milovidov | Ok.
> clickhouse only merges one partition for ReplicatedMergeTree
It's expected as one merge pass can merge only one partition. | 2018-01-09T17:44:05 | 2018-01-09T17:44:05 | {} | MEMBER |
ClickHouse | ClickHouse | 357,016,297 | 1,747 | alexey-milovidov | ? | 2018-01-11T18:22:24 | 2018-01-11T18:22:24 | {} | MEMBER |
ClickHouse | ClickHouse | 357,038,192 | 1,665 | alexey-milovidov | There is `StringToDoubleConverter` in `double-conversion` library.
I think it's the only possible choice. | 2018-01-11T19:39:30 | 2018-01-11T19:39:30 | {} | MEMBER |
ClickHouse | ClickHouse | 357,588,340 | 1,765 | Slach | https://github.com/yandex/ClickHouse/issues/777
https://github.com/Slach/clickhouse-zetcd
Если у вас есть опыт в C++
я был бы благораден за помощь чтобы можно было разобраться почему падает Clickhouse
в текущей версии стенд собирается и работает,
но падает если повторно вставить теже данные
| 2018-01-15T05:47:57 | 2018-01-15T05:47:57 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 357,596,865 | 1,765 | Slach | @wormen тогда имеет смысл закрыть этот тикет, в целях неумножения энтропии | 2018-01-15T06:54:23 | 2018-01-15T06:54:23 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 354,158,221 | 1,710 | alexey-milovidov | I wonder, what is the main motivation for this change? | 2017-12-27T18:49:31 | 2017-12-27T18:49:31 | {} | MEMBER |
ClickHouse | ClickHouse | 354,162,921 | 1,710 | amosbird | The main purpose is to accept flexible datetime formats of data which I just received :-) | 2017-12-27T19:09:12 | 2017-12-27T19:09:12 | {
"+1": 1
} | COLLABORATOR |
ClickHouse | ClickHouse | 354,163,659 | 1,458 | andr83 | I don't have logs more but error was about "Node in Zookeeper already exists" on data insert. | 2017-12-27T19:14:35 | 2017-12-27T19:14:35 | {} | NONE |
ClickHouse | ClickHouse | 354,175,306 | 1,712 | alexey-milovidov | Currently mark cache is shared and we don't have information about a table in mark cache entries.
Nevertheless, it is possible to add.
To do this, we need to add callbacks to LRUCache for storing and eviction; add reference to per-table counter in entries of MarkCache, and use these references in callbacks. | 2017-12-27T20:46:15 | 2017-12-27T20:46:15 | {} | MEMBER |
ClickHouse | ClickHouse | 354,871,738 | 1,645 | silviucpp | Hello @alexey-milovidov Yes it's true I can tune that settings in order to make sure I will never rich this issue. But it's not ok to add in the settings `skip_sanity_checks` as well ? | 2018-01-02T20:49:38 | 2018-01-02T20:49:38 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 355,750,644 | 1,575 | filimonov | @rtsisyk Please, read: https://github.com/yandex/ClickHouse/blob/master/contrib/libboost/README
(probably update instuctions should be actualized after switching to submodule)
And by the way 1.66 already released: http://www.boost.org/users/history/version_1_66_0.html
| 2018-01-06T14:34:08 | 2018-01-06T14:36:12 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 356,017,709 | 1,169 | silviucpp | The resharding code was completely removed from the code while ago | 2018-01-08T16:30:51 | 2018-01-08T16:30:51 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 356,192,885 | 1,606 | zero-master | Depends on what NVMe adds to your setup.
When your data is not already in the pagecache and CPU is not bottleneck:
a) If it's offering you a higher sequential read speed, you'll benefit
b) If it offers you more random read performance, you'll benefit when selecting 80% columns+ of table in your queries.
The... | 2018-01-09T06:30:48 | 2018-01-09T06:30:48 | {} | NONE |
ClickHouse | ClickHouse | 356,912,301 | 1,723 | ztlpn | It is the count of rows that have been read from the undrelying tables to execute the query. | 2018-01-11T11:51:31 | 2018-01-11T11:51:31 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 357,676,422 | 1,768 | filimonov | Quite wierd sample actually. Length function is designed for arrays and strings, not for Date / DateTimes.
So actually you should get that error.
```
SELECT length(now());
Received exception from server:
Code: 43. DB::Exception: Received from localhost:9000, ::1. DB::Exception: Illegal type DateTime of argumen... | 2018-01-15T13:00:50 | 2018-01-15T13:00:50 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 354,572,958 | 1,689 | silviucpp | @ztlpn The bug you discovered seems a regression from my fix with deadlock and zookeper. Anyway I started working on a fix and also I'll keep track of all observations you did | 2017-12-30T22:49:13 | 2017-12-30T22:49:13 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 355,294,583 | 1,053 | theseusyang | I've interested in this. @alexey-milovidov can you lead this development or make a roadmap plan in 2018 Q1 ?
we will join the development on these plugins extending. | 2018-01-04T14:24:30 | 2018-01-04T14:25:57 | {} | NONE |
ClickHouse | ClickHouse | 355,584,328 | 1,394 | theseusyang | @hagen1778 Yes, I mean the plugin that hooked clickhouse on gcsfuse storage, not mean gcsfuse itself.
| 2018-01-05T15:37:10 | 2018-01-05T15:37:50 | {} | NONE |
ClickHouse | ClickHouse | 356,642,701 | 1,665 | filimonov | @cyfdecyf can you check what a performance downgrade is? | 2018-01-10T15:48:04 | 2018-01-10T15:48:04 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 357,159,496 | 1,394 | theseusyang | Thanks @hagen1778 I will try it. | 2018-01-12T07:05:38 | 2018-01-12T07:05:38 | {} | NONE |
ClickHouse | ClickHouse | 357,355,669 | 1,758 | kmatt | It looks like some of the .bin files left in directory for the Distributed table are not getting touched at all.
Picking some at random with a file time of 0000 GMT today (19 hours ago),
-rw-r----- 1 root root 205332 Jan 12 00:00 6350.bin
they do not appear as processed in the log:
grep '/6350.bin`... | 2018-01-12T21:14:05 | 2018-01-12T21:14:05 | {} | NONE |
ClickHouse | ClickHouse | 356,351,595 | 1,735 | proller | We need to add tests always to make debian package clickhouse-test with script clickhouse-test
| 2018-01-09T17:16:07 | 2018-01-09T17:16:07 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 356,355,063 | 1,731 | alexey-milovidov | :+1: | 2018-01-09T17:28:19 | 2018-01-09T17:28:19 | {} | MEMBER |
ClickHouse | ClickHouse | 356,360,748 | 1,733 | alexey-milovidov | It doesn't work at all. | 2018-01-09T17:48:07 | 2018-01-09T17:48:07 | {} | MEMBER |
ClickHouse | ClickHouse | 357,019,675 | 1,729 | alexey-milovidov | The behaviour is motivated by following:
- don't break table formatting in most cases (I see that PostgreSQL do it better);
- use the same method as in TSV or for contents of quoted strings.
> -- it looks like TSVRaw is the only printable format which doesn't do that
Also VerticalRaw.
Currently I'm not sure ... | 2018-01-11T18:33:57 | 2018-01-11T18:34:14 | {} | MEMBER |
ClickHouse | ClickHouse | 357,605,599 | 1,766 | wusihang | IO usage ratio reaches 100% occasionally. But average IO usage ratio is 15%. Will this be the reason of "Bad Adress" error? | 2018-01-15T07:50:25 | 2018-01-15T07:55:11 | {} | NONE |
ClickHouse | ClickHouse | 357,657,064 | 1,751 | sanchezzzhak | run ssh
export data
```
clickhouse-client --host="localhost" --port="9000" --user="default" --password="****" --max_threads="1" --query="select * from default.table_name fromat csv" > /home/csv-dump2017-12-01.csv
```
export schema
```
clickhouse-client --host="localhost" --port="9000" --user="default" --passw... | 2018-01-15T11:30:05 | 2018-01-15T11:33:38 | {
"+1": 16
} | NONE |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.