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 | 474,994,602 | 4,525 | alexey-milovidov | OpenSSL (and LibreSSL) contains two libraries:
- SSL;
- Crypto.
We have used OpenSSL from the beginning.
Hash functions are available inside `Crypto` component of `OpenSSL` library. | 2019-03-20T19:30:05 | 2019-03-20T19:30:05 | {} | MEMBER |
ClickHouse | ClickHouse | 475,195,316 | 4,731 | 4ertus2 | I'll try to make some solution this or next week in master branch. But there's some risks that it will not fix the problem. | 2019-03-21T11:32:28 | 2019-03-21T11:32:28 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 475,242,956 | 4,747 | olegkv | @alexey-milovidov maybe nodeNumber() would be better?
Query can run on some arbitrary set of nodes, these nodes are not always part of any cluster or can be parts of multiple clusters, and using word "shard" could be confusing with real shards, like alex-krash mentioned..
| 2019-03-21T14:08:33 | 2019-03-21T14:08:33 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 473,753,919 | 4,703 | xxxxzr | > > Besides when I login in to xxx.xx.xx.32,I can't connect to xxx.xx.xx.35 using clickhouse-client -host xxx.xx.xx.35 --password xxx
>
> So you have some connectivity issues.
>
> Please check your network/ports/listen configuration (for example ensure that you listen needed network interfaces), also firewalls/ip... | 2019-03-18T03:08:43 | 2019-03-18T03:08:43 | {} | NONE |
ClickHouse | ClickHouse | 475,041,340 | 4,737 | alexey-milovidov | `сlickhouse-server-base` is an obsolete package that has embedded debug info.
Most of its size is debug information. There is a lot of debug info due to large number of C++ template instantiations. You can obtain debug info separately from `clickhouse-common-static-dbg` package.
The size of `clickhouse-common-stati... | 2019-03-20T21:46:48 | 2019-03-20T21:49:51 | {} | MEMBER |
ClickHouse | ClickHouse | 473,934,626 | 3,840 | zombiemonkey | I think I might have this wrong. When I rebooted the previous version it also lost the definition so it wasn't working previously either. I found that if I wrapped the Array in a Tuple then everything worked. So it seems if you use Arrays of Tuples these are converted to nested tables. This behaviour is fine so closing... | 2019-03-18T14:32:57 | 2019-03-18T14:32:57 | {} | NONE |
ClickHouse | ClickHouse | 473,980,421 | 4,616 | wawaka | If it helps somehow: we are having segmentation faults using kafka+mview with kafka tables from 2 different kafka clusters. For some reason, when we removed some tables and used only ones from one cluster - segfaults disappeared. Not sure, how it makes the difference, just an observation. | 2019-03-18T16:08:40 | 2019-03-18T16:08:40 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 474,344,172 | 4,712 | alesapin | Locally it's working good, but I still cannot run it inside CI. Lets merge. | 2019-03-19T12:37:56 | 2019-03-19T12:37:56 | {} | MEMBER |
ClickHouse | ClickHouse | 474,715,158 | 720 | kszucs | Well, I forgot about this one. For long term the best would be to implement a native clickhouse client in https://github.com/apache/arrow with python bindings, then We could efficiently read data from clickhouse to arrow/pandas.
Any help would be appreciated! | 2019-03-20T07:19:37 | 2019-03-20T07:20:01 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 474,887,422 | 4,727 | den-crane | >ok, but why insert in not null datetime is working?
because CH did not support Null(nullable) a year ago, and substitutes them with default values for each type String '', Date 0, Int64 0
CREATE TABLE IF NOT EXISTS test (name String
really means
CREATE TABLE IF NOT EXISTS test (name String Default '' | 2019-03-20T15:33:17 | 2019-03-20T15:33:17 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 474,987,698 | 4,525 | code-of-kpp | AFAIR `md5` and `shaXXX` where there long before you added SSL support. How it was implemented back then? | 2019-03-20T19:09:40 | 2019-03-20T19:09:40 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 475,056,153 | 4,741 | olegkv | > It is possible with `randConstant` function. It will give you a constant expression that is random for each remote query. It's a little bit tricky. There are a few examples in tests...
I tried to use rand() but the problem is that I don't know what rand generates in each of two cases,
and, as a result, I would no... | 2019-03-20T22:37:45 | 2019-03-20T22:41:01 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 475,134,954 | 4,727 | n00bik | thank you! | 2019-03-21T07:35:06 | 2019-03-21T07:35:06 | {} | NONE |
ClickHouse | ClickHouse | 473,570,361 | 4,706 | alexey-milovidov | Approved. But missing details for the changelog: what is the issue, in what version the bug has appeared. | 2019-03-16T17:59:27 | 2019-03-16T17:59:27 | {} | MEMBER |
ClickHouse | ClickHouse | 473,911,976 | 4,695 | SaltTan | It seems that any request to any dictionary during the initial (not lazy) loading will cause CH to hang | 2019-03-18T13:35:03 | 2019-03-18T13:35:03 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 474,673,582 | 720 | cloudronin | I am guessing It should allow to reading and writing [dask](https://dask.org/) dataframes from clickhouse using the HTTP API similar to how @kszucs did for pandas here: https://github.com/kszucs/pandahouse
| 2019-03-20T03:46:48 | 2019-03-20T03:48:51 | {} | NONE |
ClickHouse | ClickHouse | 474,767,935 | 4,724 | BHYCHIK | > Looks like a draft: no tests, the function implemented only for constant argument.
Thanks for review. It was done for internal needs. I will update MR with tests and non-constants arguments. | 2019-03-20T10:14:39 | 2019-03-20T10:14:39 | {} | NONE |
ClickHouse | ClickHouse | 474,801,088 | 4,717 | KochetovNicolai | Now it's not possible, but I think not very difficult to implement (the order is fixed now in src/programms/Server.cpp). Probably, we need special setting for this (otherwise, queries which don't need dictionary will also wait). | 2019-03-20T12:04:06 | 2019-03-20T12:04:06 | {} | MEMBER |
ClickHouse | ClickHouse | 474,971,561 | 4,731 | 4ertus2 | Hi,
Performance depend on JOINs order. Could you try with switched j1 and j2. There's no logic that switches them to the best order yet.
The second thing that you may check is would it depend on count of columns that are not in the query or not. I can imagine a situation then it would get more (up to all) columns... | 2019-03-20T18:26:11 | 2019-03-20T18:26:11 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 475,048,418 | 4,741 | olegkv | @alexey-milovidov
I tried to make it work on one node only by using remote('localhost,localhost') but
I am not sure how to make remote('localhost,localhost') merge two different result sets - it merges queries from same table.
So, I created test table with engine=memory and added two different insert queries, one fo... | 2019-03-20T22:11:31 | 2019-03-20T22:21:32 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 475,154,382 | 4,736 | aegorov05 | > `VersionedCollapsingMergeTree` is a very specific table engine. Do you have a reason to use it?
> Does the issue persist if you use `MergeTree`?
Changing engine to standard MergeTree does not make any difference. Instead of 3 million records we only got 2941414. | 2019-03-21T09:03:53 | 2019-03-21T09:04:22 | {} | NONE |
ClickHouse | ClickHouse | 475,226,446 | 4,731 | Slind14 | Thanks It will be great if this can be solved on CH side. | 2019-03-21T13:23:12 | 2019-03-21T13:23:12 | {} | NONE |
ClickHouse | ClickHouse | 473,581,087 | 4,712 | alesapin | WIP | 2019-03-16T19:59:49 | 2019-03-16T19:59:49 | {} | MEMBER |
ClickHouse | ClickHouse | 474,880,439 | 4,737 | den-crane | you need only 3 debs
```
103M Mar 18 12:57 clickhouse-common-static_19.5.1.122_amd64.deb
12K Mar 18 12:57 clickhouse-client_19.5.1.122_all.deb
24K Mar 18 12:57 clickhouse-server_19.5.1.122_all.deb
```
242M /usr/bin/clickhouse | 2019-03-20T15:19:56 | 2019-03-20T15:21:48 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 475,048,281 | 4,727 | alexey-milovidov | It is possible to implement `input_format_null_as_default` and `input_format_empty_as_default` settings in future. | 2019-03-20T22:11:01 | 2019-03-20T22:11:01 | {} | MEMBER |
ClickHouse | ClickHouse | 473,978,320 | 4,716 | den-crane | IMHO skip `INDEX s_key_idx (s_key) TYPE set(0) GRANULARITY 1000` just has no sense.
It makes a query slower and doubles the column stored data.
```
create table BX(
I Int64,
S String,
INDEX Sx S TYPE set(0) GRANULARITY 1000
) Engine=MergeTree order by I;
insert into BX select number, toString(rand()) fro... | 2019-03-18T16:04:40 | 2019-03-18T16:07:04 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 474,437,034 | 4,721 | alesapin | Merge after 19.3 and 19.1 will be released with new system. | 2019-03-19T15:50:39 | 2019-03-19T15:50:39 | {} | MEMBER |
ClickHouse | ClickHouse | 474,595,014 | 4,724 | alexey-milovidov | Looks like a draft: no tests, the function implemented only for constant argument. | 2019-03-19T21:39:07 | 2019-03-19T21:39:07 | {} | MEMBER |
ClickHouse | ClickHouse | 474,841,603 | 4,717 | SaltTan | I don't think it's bad that all queries will wait.
ClickHouse is supposed to shut down anyway if dictionaries_lazy_load is false and there were errors during dictionaries loading.
Moreover, it works exactly like this in all versions earlier than 19.x, all queries have to wait because dictionaries are loaded even befo... | 2019-03-20T14:00:03 | 2019-03-20T14:00:03 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 475,003,390 | 4,741 | alexey-milovidov | Can you provide a test with
```
SELECT ... FROM remote('127.0.0.{1,2}', ...)
```
?
`remote('127.0.0.{1,2}', ...)` is to make a distributed query with single server. | 2019-03-20T19:55:41 | 2019-03-20T20:11:19 | {} | MEMBER |
ClickHouse | ClickHouse | 475,144,679 | 4,737 | nitso | Thanks a lot for explanation. | 2019-03-21T08:25:25 | 2019-03-21T08:25:25 | {} | NONE |
ClickHouse | ClickHouse | 475,186,975 | 4,749 | makrushin-evgenii | > а_гг_регируют is also a typo.
Заметил еще одну опечатку, поправил. | 2019-03-21T11:00:20 | 2019-03-21T11:01:40 | {
"+1": 1
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 473,377,829 | 4,218 | Slind14 | Any news on this? | 2019-03-15T17:35:19 | 2019-03-15T17:35:19 | {} | NONE |
ClickHouse | ClickHouse | 473,821,373 | 4,716 | KochetovNicolai | I think the reason is that `set(0)` is an empty set, so empty index was created. Try to use, for example, `set(100)`. 100 here is the max set size which is created per granule (so, if granule has more than 100 distinct values, set won't be created, and granule will never be dropped). | 2019-03-18T08:52:41 | 2019-03-18T08:52:41 | {
"+1": 1,
"-1": 1
} | MEMBER |
ClickHouse | ClickHouse | 474,466,440 | 4,501 | madm1keru | Update: clickhouse restart without any data modifications or merges - fixes error. | 2019-03-19T16:51:18 | 2019-03-19T16:51:18 | {} | NONE |
ClickHouse | ClickHouse | 474,511,358 | 4,714 | vkosh | I also can't reproduce it yet, even under the same conditions. But another issue occurs: after ~1-2 hours of continuos inserts (1 per 2-3 seconds) every further query leads to "Cannot allocate memory" or "bad_alloc" (see below).
Some more details about the conditions that may help:
* the table columns wasn't altere... | 2019-03-19T18:21:29 | 2019-03-19T18:21:29 | {} | NONE |
ClickHouse | ClickHouse | 474,841,220 | 4,735 | den-crane | > It would be nice if ALTER TABLE could accept system.parts.partition_id as a partition identifier.
> We would not need any complicated logic then to generate ALTER statements...
I concur
and PARTITION expects an expression (201802 === 201802/2*2+0)
so I prefer PARTITION_ID or ( PARTITION ID )
`ALTER TABLE ... | 2019-03-20T13:59:12 | 2019-03-20T13:59:25 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 474,842,457 | 4,528 | igor-sh8 | Hi.
Thanks for the explanation.
> MV is an on insert trigger. It works over inserting buffer.
So I also can not create MV behind another MV, or MV behind MV To Table ? | 2019-03-20T14:02:11 | 2019-03-20T14:02:11 | {} | NONE |
ClickHouse | ClickHouse | 474,942,683 | 4,723 | 4ertus2 | All the fix is in the last patch. | 2019-03-20T17:23:34 | 2019-03-20T17:23:34 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 475,176,029 | 4,749 | alexey-milovidov | а*гг*регируют is also a typo. | 2019-03-21T10:20:48 | 2019-03-21T10:20:48 | {} | MEMBER |
ClickHouse | ClickHouse | 475,216,029 | 4,388 | fandyushin | I added define BROTLI_BUILD_PORTABLE, and its works. But I have no idea why unbandled tests fails. According to logs: connection breaks by timeout, but I cant reproduce it locally and other types of same tests runs ok. Can you explain what difference compared to other stateless tests? | 2019-03-21T12:50:30 | 2019-03-21T12:50:30 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 473,753,143 | 4,703 | xxxxzr | > > Besides when I login in to xxx.xx.xx.32,I can't connect to xxx.xx.xx.35 using clickhouse-client -host xxx.xx.xx.35 --password xxx
>
> So you have some connectivity issues.
>
> Please check your network/ports/listen configuration (for example ensure that you listen needed network interfaces), also firewalls/ip... | 2019-03-18T03:04:08 | 2019-03-18T03:04:08 | {} | NONE |
ClickHouse | ClickHouse | 474,191,013 | 4,670 | hushengyue | Thanks den, what about to select distinct values of multiple columns? I guess there is no optimization for such case. | 2019-03-19T04:09:15 | 2019-03-19T04:09:15 | {} | NONE |
ClickHouse | ClickHouse | 474,812,108 | 4,733 | Slind14 | ahh, duplicate: https://github.com/yandex/ClickHouse/issues/4658 | 2019-03-20T12:41:53 | 2019-03-20T12:41:53 | {} | NONE |
ClickHouse | ClickHouse | 474,838,088 | 4,735 | SaltTan | It would be nice if ALTER TABLE could accept system.parts.partition_id as a partition identifier.
We would not need any complicated logic then to generate ALTER statements... | 2019-03-20T13:51:51 | 2019-03-20T13:51:51 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 475,030,962 | 4,736 | filimonov | I thought @den-crane helped yu to find the root of the problem in telegram chat?
As I understand the reason of "missing" rows was wrong `order by` columns in VersionedCollapsingMergeTree? Or it was not the only problem? | 2019-03-20T21:14:16 | 2019-03-20T21:14:16 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 475,073,898 | 4,736 | den-crane | > I thought @den-crane helped yu to find the root of the problem in telegram chat?
> As I understand the reason of "missing" rows was wrong `order by` columns in VersionedCollapsingMergeTree? Or it was not the only problem?
probably you've confused two different cases. It's another issue.
And probably it's not re... | 2019-03-21T00:04:21 | 2019-03-21T00:50:44 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 475,135,106 | 4,730 | BayoNet | Поправил ссылки в языковых версиях. Локально документация собирается без ошибок. | 2019-03-21T07:36:01 | 2019-03-21T07:36:01 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 475,191,325 | 4,731 | 4ertus2 | There's no workaround for now. We need to make some improvement in code. | 2019-03-21T11:16:55 | 2019-03-21T11:16:55 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 475,232,988 | 4,736 | den-crane | I think kafka engine waits for {kafka_max_block_size} rows with **blocking** read and after that flushes by reaching max_insert_block_size or 7500ms.
@aegorov05 try to write (3000 * 65536 + 1) = 196608001 rows | 2019-03-21T13:42:05 | 2019-03-21T13:48:18 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 473,570,908 | 4,687 | alexey-milovidov | Ok. I have checked that it works fine on old system (Ubuntu 12.04). | 2019-03-16T18:05:55 | 2019-03-16T18:05:55 | {} | MEMBER |
ClickHouse | ClickHouse | 473,911,970 | 4,708 | nvartolomei | @4ertus2 thanks! | 2019-03-18T13:35:01 | 2019-03-18T13:35:01 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 474,893,214 | 720 | kszucs | If anyone interested to develop what @martindurant suggested, use [clickhouse-driver](https://github.com/mymarilyn/clickhouse-driver) or [ibis's clickhouse backend (uses clickhouse-driver under the hood)](https://github.com/ibis-project/ibis) instead of pandahouse. | 2019-03-20T15:44:40 | 2019-03-20T15:45:10 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 474,933,977 | 4,718 | alexey-milovidov | Bad description. Missing test. | 2019-03-20T17:05:26 | 2019-03-20T17:05:26 | {} | MEMBER |
ClickHouse | ClickHouse | 475,051,810 | 4,741 | alexey-milovidov | > I am not sure how to make remote('127.0.0.{1,2}') merge two different result sets - it merges queries from same table.
It is possible with `randConstant` function. It will give you a constant expression that is random for each remote query. It's a little bit tricky. There are a few examples in tests...
> So, I ... | 2019-03-20T22:22:31 | 2019-03-20T22:22:31 | {} | MEMBER |
ClickHouse | ClickHouse | 475,052,815 | 4,736 | alexey-milovidov | `VersionedCollapsingMergeTree` is a very specific table engine. Do you have a reason to use it?
Does the issue persist if you use `MergeTree`? | 2019-03-20T22:25:54 | 2019-03-20T22:25:54 | {} | MEMBER |
ClickHouse | ClickHouse | 475,154,322 | 4,453 | AlexAkulov | Кажется после обновления до 19.4.0 проблема исчезла. | 2019-03-21T09:03:38 | 2019-03-21T09:03:38 | {} | NONE |
ClickHouse | ClickHouse | 475,241,836 | 4,528 | den-crane | >So I also can not create MV behind another MV, or MV behind MV To Table ?
No. No MV cascading. But you can attach several MVs to a table.
| 2019-03-21T14:05:53 | 2019-03-21T14:05:53 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 473,571,255 | 4,672 | alexey-milovidov | > I have considered other names like allow_constant_folding_logic_expressions, allow_logic_expressions_constant_folding
Ok. | 2019-03-16T18:10:10 | 2019-03-16T18:10:10 | {} | MEMBER |
ClickHouse | ClickHouse | 474,972,488 | 4,731 | Slind14 | I tested it the other way around as well which is slightly slower. 2-4% | 2019-03-20T18:28:41 | 2019-03-20T18:28:41 | {} | NONE |
ClickHouse | ClickHouse | 475,060,544 | 4,388 | alexey-milovidov | UBSan test has failed due to
```
static BROTLI_INLINE uint64_t BrotliUnalignedRead64(const void* p) {
return *(const uint64_t*)p;
}
```
We should define `BROTLI_BUILD_PORTABLE`.
```
BROTLI_BUILD_PORTABLE disables dangerous optimizations, like unaligned
read and overlapping memcpy; this reduces decomp... | 2019-03-20T22:56:49 | 2019-03-20T22:56:49 | {} | MEMBER |
ClickHouse | ClickHouse | 475,061,940 | 4,747 | alexey-milovidov | For #4741 | 2019-03-20T23:03:02 | 2019-03-20T23:03:02 | {} | MEMBER |
ClickHouse | ClickHouse | 475,163,222 | 4,736 | aegorov05 | > > I thought @den-crane helped yu to find the root of the problem in telegram chat?
> > As I understand the reason of "missing" rows was wrong `order by` columns in VersionedCollapsingMergeTree? Or it was not the only problem?
>
> probably you've confused two different cases. It's another issue.
> And probably it... | 2019-03-21T09:36:28 | 2019-03-21T09:36:28 | {} | NONE |
ClickHouse | ClickHouse | 475,179,330 | 4,747 | alex-krash | @alexey-milovidov , this function will not take into consideration topology of clusters, described in configuration?
Why I am asking - we have host being assigned to multiple clusters/shards.
```
SELECT
groupArray((cluster, shard_num)) AS cluster_and_shard,
host_name
FROM system.clusters
WHERE host_na... | 2019-03-21T10:32:19 | 2019-03-21T10:32:19 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 475,174,994 | 4,731 | 4ertus2 | Trying to reproduce. There's 4 times difference between 1 join vs 2 joins on my data (with same result). And ~5% difference between multiple joins vs joins via subselects.
```
SELECT count(dt.d_year)
FROM date_dim AS dt
INNER JOIN store_sales ON dt.d_date_sk = store_sales.ss_sold_date_sk
WHERE dt.d_year = 1998
... | 2019-03-21T10:17:10 | 2019-03-21T11:07:07 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 473,518,832 | 4,207 | vnnw | Is it possible to support 64bit integer? | 2019-03-16T10:31:20 | 2019-03-16T10:31:20 | {} | NONE |
ClickHouse | ClickHouse | 473,839,960 | 4,716 | zombiemonkey | Ah ok - thanks Nicolai. Have it working now. The docs say that set(0) means no limit but that must be incorrect. Cheers! | 2019-03-18T09:48:30 | 2019-03-18T09:48:30 | {} | NONE |
ClickHouse | ClickHouse | 473,844,874 | 4,714 | KochetovNicolai | I cannot reproduce this bug, but it looks like heap corruption. It happens while checking that columns in part are the same as in table (which can be violated by ALTER MODIFY).
`addr2line -e clickhouse 0x6609a32` says
`/build/obj-x86_64-linux-gnu/../dbms/src/Storages/ITableDeclaration.cpp:199`
Can you reproduce th... | 2019-03-18T10:02:44 | 2019-03-18T10:02:44 | {} | MEMBER |
ClickHouse | ClickHouse | 473,948,129 | 4,708 | alexey-milovidov | Did you mean @KochetovNicolai? https://github.com/yandex/ClickHouse/pull/4556 | 2019-03-18T15:03:48 | 2019-03-18T15:03:48 | {} | MEMBER |
ClickHouse | ClickHouse | 474,163,470 | 4,708 | nvartolomei | Yes. <3 | 2019-03-19T01:33:47 | 2019-03-19T01:33:47 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 474,719,171 | 4,726 | blinkov | Basically it's recommended to upgrade to latest release marked "stable" as soon as you can (with max overall version number, reading from left to right). In non-production environments you can update to releases marked "testing" to notice issues earlier and report/fix them in advance.
You should take extra care when... | 2019-03-20T07:35:54 | 2019-03-20T07:35:54 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 475,044,835 | 4,743 | alexey-milovidov | Thank you! The fix looks fine.
1. It's easy to provide a test. Look at `dbms/tests/queries/0_stateless/*.sh`
2. I'd like to see better error message. BTW, there is a function `throwFromErrno` that better fits for this purpose. | 2019-03-20T21:58:46 | 2019-03-20T21:58:46 | {} | MEMBER |
ClickHouse | ClickHouse | 475,181,483 | 4,747 | alexey-milovidov | It will assign numbers to shards dynamically depending on the set of shards participated in a query. | 2019-03-21T10:40:02 | 2019-03-21T10:40:02 | {} | MEMBER |
ClickHouse | ClickHouse | 473,570,979 | 4,687 | alexey-milovidov | Do we also need to modify `website/index.html`? | 2019-03-16T18:06:40 | 2019-03-16T18:06:52 | {} | MEMBER |
ClickHouse | ClickHouse | 473,651,509 | 2,740 | filimonov | check the replica names / netword routing / NAT.
It looks like it was trying to download part which was registed in '9.1.1.80' replica from '9.1.1.142' replica. | 2019-03-17T10:10:11 | 2019-03-17T10:10:11 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 473,877,839 | 4,708 | nvartolomei | @alexey-milovidov can we please stage this to 19.1.* as well? | 2019-03-18T11:47:19 | 2019-03-18T11:47:19 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 474,517,082 | 4,615 | alesapin | Failed flappy test fixed in master. | 2019-03-19T18:36:40 | 2019-03-19T18:36:40 | {} | MEMBER |
ClickHouse | ClickHouse | 474,841,164 | 4,727 | n00bik | ok, but why insert in not null datetime is working?
`
drop table if EXISTS test;
CREATE TABLE IF NOT EXISTS test (
name String,
installdatetime datetime
) ENGINE = MergeTree()
PARTITION BY name ORDER BY installdatetime;
insert into test values('x',)
select * from test;
`
| 2019-03-20T13:59:03 | 2019-03-20T13:59:03 | {} | NONE |
ClickHouse | ClickHouse | 474,853,993 | 4,658 | Slind14 | Thank you @proller
Any chance for a hotfix release? | 2019-03-20T14:26:41 | 2019-03-20T14:26:41 | {
"+1": 3
} | NONE |
ClickHouse | ClickHouse | 475,033,290 | 4,537 | filimonov | ```bash
# insert from csv file
clickhouse-client --query='INSERT INTO table_name FORMAT CSV' < data.csv
# select and put results to csv file
clickhouse-client --query='SELECT * FROM table_name FORMAT CSV' > data.csv
```
| 2019-03-20T21:21:10 | 2019-03-20T21:21:10 | {
"+1": 2
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 475,213,087 | 4,736 | filimonov | > > `VersionedCollapsingMergeTree` is a very specific table engine. Do you have a reason to use it?
> > Does the issue persist if you use `MergeTree`?
>
> Changing engine to standard MergeTree does not make any difference. Instead of 3 million records we only got 2941414.
So the difference is 2941414 - 3000000 =... | 2019-03-21T12:40:12 | 2019-03-21T12:40:12 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 473,421,985 | 4,709 | 4ertus2 | There's an error when a new version of client have been sending a fresh option 'send_logs_level' to old version of server that could not understand it. You need update (client or server or both). | 2019-03-15T19:58:20 | 2019-03-15T19:58:54 | {
"+1": 1
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 473,858,777 | 4,716 | KochetovNicolai | Well, it seems that set(0) should mean no limit, and it is fixed in #4640. Docs was correct. | 2019-03-18T10:42:52 | 2019-03-18T10:42:52 | {} | MEMBER |
ClickHouse | ClickHouse | 474,799,353 | 4,537 | Crazylearner30 | Thanks, I got it.
Now I have a new query wherein I want to download the large file directly from the console.
Is there any command or format to download the CSV file into the server directly from the console?
Thanks,
| 2019-03-20T11:58:32 | 2019-03-20T12:03:20 | {} | NONE |
ClickHouse | ClickHouse | 474,828,307 | 4,658 | silviucpp | Building with -DENABLE_PARQUET=0 can be used as workaround | 2019-03-20T13:28:59 | 2019-03-20T13:28:59 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 473,467,135 | 4,666 | alexey-milovidov | RWLock metrics are related to table locks. As far as I remember, write lock is used for ALTER queries.
So, the question is: do you have significantly more ALTER queries on ClickHouse2? | 2019-03-15T23:01:07 | 2019-03-15T23:01:07 | {} | MEMBER |
ClickHouse | ClickHouse | 473,699,824 | 4,713 | alexey-milovidov | Need to fix up "00646_url_engine" test. | 2019-03-17T18:46:34 | 2019-03-17T18:46:34 | {} | MEMBER |
ClickHouse | ClickHouse | 473,808,747 | 4,687 | SevaCode | > Do we also need to modify `website/index.html`?
Done. | 2019-03-18T08:07:50 | 2019-03-18T08:07:50 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 473,841,339 | 4,616 | PhantomPhreak | We have the same problems with ClickHouse v19.3.7 and materialized views consuming from a tables with Kafka engine: CH server crashing with segmentation fault | 2019-03-18T09:52:39 | 2019-03-18T09:52:39 | {} | NONE |
ClickHouse | ClickHouse | 474,812,316 | 4,658 | Slind14 | We ran into the same issue which broke nearly everything for use because it all depends on the timezone. It would be great if this could be resolved and rolled out very soon. | 2019-03-20T12:42:31 | 2019-03-20T13:31:54 | {
"+1": 1
} | NONE |
ClickHouse | ClickHouse | 475,057,526 | 4,726 | alexey-milovidov | > After some time, we tag a revision with two tags, for example: v19.1.6-testing, v19.1.6-stable. Do you have any special consideration to tag the same version with different tags?
Testing version goes to stable without any changes to code - as soon as it was tested well enough. You can consider it as "release candi... | 2019-03-20T22:43:30 | 2019-03-20T22:43:30 | {} | MEMBER |
ClickHouse | ClickHouse | 474,463,798 | 4,501 | madm1keru | Got same error.
Simple query. Works on one replica, doesn't work on second replica (exactly the SAME data, checked md5 for all checksum.txt files). Works for smaller intervals - Oct 1-14 and Oct 15-31, but doesn't work for 1-31.
SELECT
sum(count_responses) AS resps,
sum(count_requests) AS reqs,
... | 2019-03-19T16:45:09 | 2019-03-19T16:45:09 | {} | NONE |
ClickHouse | ClickHouse | 474,533,596 | 4,725 | proller |
| 2019-03-19T19:13:45 | 2019-03-19T19:13:45 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 474,781,765 | 4,727 | KochetovNicolai | I think the reason why ClickHouse can't parse data is that column type should be Nullable to be able to store Null. So, \N for String type was parsed as string "N", but \N for DateTime caused a error. Try to change all columns which can store Null to Nullable type. (String -> Nullable(String), DateTime -> Nullable(Date... | 2019-03-20T10:57:08 | 2019-03-20T10:57:08 | {} | MEMBER |
ClickHouse | ClickHouse | 474,802,619 | 4,676 | BayoNet | Сейчас специально попробовал собрать документацию локально полностью, сборщик проблем не выдал. | 2019-03-20T12:09:46 | 2019-03-20T12:09:46 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 474,884,195 | 720 | martindurant | Worth saying: if you can read to pandas dataframes based on a query, would be simple enough to partition up that query and have dataframes read fro each part, to form a Dask dataframe. You could do this with `dask.delayed`, `dd.from_delayed`; a similar process is used for most of the dataframe input functions in Dask. | 2019-03-20T15:27:20 | 2019-03-20T15:27:20 | {} | NONE |
ClickHouse | ClickHouse | 474,945,683 | 4,525 | proller | Currently build without ssl and with functions impossible:
dbms/src/Functions/FunctionsHashing.h :
#include <openssl/md5.h>
#include <openssl/sha.h>
But its easy to fix by disabling some functions | 2019-03-20T17:29:56 | 2019-03-20T17:29:56 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 475,042,040 | 4,737 | alexey-milovidov | PS. JFYI, number of commits since version 18.16 was increased by 14% and number of LOC in `dbms` directory - by 9%. | 2019-03-20T21:49:20 | 2019-03-20T21:51:00 | {} | MEMBER |
ClickHouse | ClickHouse | 475,183,674 | 4,731 | Slind14 | @4ertus2
> The second thing that you may check is would it depend on count of columns that are not in the query or not. I can imagine a situation then it would get more (up to all) columns from some table. Not only required ones. So would it be the same time if there's only affected columns in tables?
You are ri... | 2019-03-21T10:47:58 | 2019-03-21T10:48:52 | {} | NONE |
ClickHouse | ClickHouse | 475,193,961 | 4,731 | Slind14 | @4ertus2 do you think this is something that will be looked into/resolved soon? | 2019-03-21T11:27:24 | 2019-03-21T11:27:24 | {} | NONE |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.