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 | 311,139,559 | 925 | alexey-milovidov | Ok. Great solution! | 2017-06-26T18:16:50 | 2017-06-26T18:16:50 | {} | MEMBER |
ClickHouse | ClickHouse | 312,237,075 | 790 | genestack-solomatin | Reproduced once again after upgrade to `Docker 17.06.0-ce-mac18 (18433)`
> There is this other closed issue that may be relevant here: #721
Yes, it is. This comment helped me to workaround my troubles. https://github.com/yandex/ClickHouse/issues/721#issuecomment-296400878
| 2017-06-30T10:47:54 | 2017-06-30T10:47:54 | {
"+1": 2
} | NONE |
ClickHouse | ClickHouse | 312,772,687 | 903 | alexey-milovidov | Is it worth to add a logic to round block size to a number, that divide or divides granule size?
For example, if we have index_granularity = 8192 and block size is calculated to 8000, we could round up it to 8192 or round down to 4096. | 2017-07-04T03:36:10 | 2017-07-04T03:42:20 | {} | MEMBER |
ClickHouse | ClickHouse | 313,739,660 | 955 | robot-metrika-test | Can one of the admins verify this patch? | 2017-07-07T17:09:03 | 2017-07-07T17:09:03 | {} | NONE |
ClickHouse | ClickHouse | 313,922,485 | 951 | alexey-milovidov | Awesome!
The only missing part is test. You could look for `00318_pk_tuple_order.sql` as an example.
(The test could use `max_rows_to_read` setting to prove that index is used. Also there are `force_primary_key` and `force_index_by_date` settings.) | 2017-07-09T14:16:01 | 2017-07-09T14:16:01 | {} | MEMBER |
ClickHouse | ClickHouse | 313,997,120 | 945 | alexey-milovidov | Now we have two methods `formatUUID` and two methods `parseUUID` in ReadHelpers, WriteHelpers.
One of them is using `strtoull` and `snprintf`, which are very inefficient, and also an excessive copying. | 2017-07-10T04:08:37 | 2017-07-10T04:08:37 | {} | MEMBER |
ClickHouse | ClickHouse | 314,769,693 | 975 | hagen1778 | Hi @m1schka
You could try http://clickhouse.readthedocs.io/en/latest/reference_en.html#arrayFilter function to filter all elements with i=0.
But it would be better to filter these not interesting results at your application.
Btw, there is a telegram-channel where I believe you can find answers much faster - https:/... | 2017-07-12T13:31:48 | 2017-07-12T13:32:15 | {} | NONE |
ClickHouse | ClickHouse | 310,902,863 | 924 | alexey-milovidov | Ok, added a task for our QA team for auto publishing. | 2017-06-25T13:32:33 | 2017-06-25T13:32:33 | {} | MEMBER |
ClickHouse | ClickHouse | 312,649,323 | 944 | robot-metrika-test | Can one of the admins verify this patch? | 2017-07-03T13:44:03 | 2017-07-03T13:44:03 | {} | NONE |
ClickHouse | ClickHouse | 312,657,825 | 937 | alexey-milovidov | Missing test.
Look at `dbms/tests/queries` directory. | 2017-07-03T14:19:06 | 2017-07-03T14:19:06 | {} | MEMBER |
ClickHouse | ClickHouse | 312,932,854 | 935 | ludv1x | Could you provide your queries?
The following examples work correctly for me:
```
vludv@vludv-ub14:~/distcc-ClickHouse$ echo "select 1" | clickhouse-client
1
vludv@vludv-ub14:~/distcc-ClickHouse$ echo "select 1, 2, 3" | clickhouse-client
1 2 3
vludv@vludv-ub14:~/distcc-ClickHouse$ echo -e "select 1, 2, 3;\n se... | 2017-07-04T18:35:24 | 2017-07-04T18:35:24 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 313,217,741 | 23 | cce | This seems to be resolved by https://github.com/yandex/ClickHouse/pull/787 | 2017-07-05T20:29:32 | 2017-07-05T20:29:32 | {} | NONE |
ClickHouse | ClickHouse | 313,243,284 | 835 | ztlpn | fixed in v1.1.54245 | 2017-07-05T22:22:38 | 2017-07-05T22:22:38 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 313,862,901 | 947 | alexey-milovidov | Don't worry, we are always interested in optimization possibilities.
About `Replacing`: the main concern is that `Vertical` merge algorithm is not enabled for this mode.
This is easily doable and with `Vertical` algorithm, merges will run significantly faster (in case of wide tables). | 2017-07-08T15:28:54 | 2017-07-08T15:28:54 | {} | MEMBER |
ClickHouse | ClickHouse | 314,071,748 | 960 | robot-metrika-test | Can one of the admins verify this patch? | 2017-07-10T10:49:02 | 2017-07-10T10:49:02 | {} | NONE |
ClickHouse | ClickHouse | 314,971,314 | 971 | alexey-milovidov | New column in existing rows will have default values.
(Although this default values are not physically written on disk during ALTER.) | 2017-07-13T05:01:34 | 2017-07-13T05:01:34 | {} | MEMBER |
ClickHouse | ClickHouse | 311,381,985 | 687 | inv2004 | I have created duplicate: https://github.com/yandex/ClickHouse/issues/932 | 2017-06-27T14:48:05 | 2017-06-27T14:48:05 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 312,439,200 | 942 | alexey-milovidov | Thanks!
Changes will be available on the website soon. | 2017-07-01T15:44:06 | 2017-07-01T15:44:06 | {
"+1": 1
} | MEMBER |
ClickHouse | ClickHouse | 314,245,458 | 329 | alexey-milovidov | This is still in our longlist.
| 2017-07-10T21:11:53 | 2017-07-10T21:11:53 | {} | MEMBER |
ClickHouse | ClickHouse | 314,534,243 | 968 | alexey-milovidov | :+1: | 2017-07-11T18:37:01 | 2017-07-11T18:37:01 | {} | MEMBER |
ClickHouse | ClickHouse | 314,846,148 | 971 | ludv1x | Confirmed.
It applies `WHERE` to left table (`implus.test_a`) instead of joined table in the following query:
```
SELECT *
FROM implus.test_a
ANY INNER JOIN
(
SELECT
OldColumn,
NewColumn
FROM implus.test_b
) USING (OldColumn)
WHERE NewColumn != ''
``` | 2017-07-12T17:51:02 | 2017-07-12T17:51:02 | {
"+1": 1
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 315,201,579 | 981 | alexey-milovidov | :+1: | 2017-07-13T21:03:03 | 2017-07-13T21:03:03 | {} | MEMBER |
ClickHouse | ClickHouse | 310,937,924 | 925 | robot-metrika-test | Can one of the admins verify this patch? | 2017-06-26T00:14:02 | 2017-06-26T00:14:02 | {} | NONE |
ClickHouse | ClickHouse | 311,296,015 | 928 | alexey-milovidov | Thanks! | 2017-06-27T08:54:04 | 2017-06-27T08:54:04 | {} | MEMBER |
ClickHouse | ClickHouse | 311,715,397 | 931 | valyala | Yes, the only thing is `max_bytes_to_merge_at_max_space_in_pool` that changed from 100Gb to 500Gb.
clickhouse version is 1.1.54236 , installed from the official deb package `deb http://repo.yandex.ru/clickhouse/xenial stable main` | 2017-06-28T16:32:19 | 2017-06-28T16:32:19 | {} | NONE |
ClickHouse | ClickHouse | 313,224,780 | 23 | alexey-milovidov | Now we have 100% translated in `dbms` directory. Only few files in `libs` remains.
In total we have translated 97.1% of comments. | 2017-07-05T20:57:52 | 2017-07-05T20:57:52 | {} | MEMBER |
ClickHouse | ClickHouse | 313,792,615 | 947 | innerr | I'm sorry I made a big mistake.
What I actually saying is ReplacingSortedBlockInputStream, some how I mixed it up with MergingSortedBlockInputStream, apologize again.
The ReplacingSortedBlockInputStream have a much simpler implement in the "merge" method, and truely have the problem as I mentioned.
It caused a s... | 2017-07-07T21:04:47 | 2017-07-07T21:05:18 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 313,856,460 | 868 | Slach | after add
```
<background_pool_size>2</background_pool_size>
```
and add total memory limit 2048Gb in my Vagrant environment
i was able to import my data sample successcull | 2017-07-08T13:39:25 | 2017-07-08T14:12:00 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 314,023,945 | 959 | alex-zaitsev | Yandex does not provide packages for CentOS. You can build from sources or use Altinity repo with prebuilt RPMs: https://packagecloud.io/altinity/clickhouse | 2017-07-10T07:11:07 | 2017-07-10T07:11:07 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 314,282,055 | 957 | alexey-milovidov | Buffer tables just doing ordinary INSERTs to destination table.
That's why they could not have columns, that are MATERIALIZED in destination table.
This is understandable, but inconvenient behavior.
We could do one of the following:
- add a flag, that will allow to INSERT into MATERIALIZED column, if INSERT is ... | 2017-07-10T23:52:56 | 2017-07-10T23:52:56 | {} | MEMBER |
ClickHouse | ClickHouse | 314,523,548 | 965 | alexey-milovidov | Ok.
BTW we have not much settings, that can be effectively set only for default profile. | 2017-07-11T17:57:20 | 2017-07-11T17:57:20 | {} | MEMBER |
ClickHouse | ClickHouse | 312,935,392 | 936 | ludv1x | Yes, it is designed behavior.
> I'm not sure if this a bug or not but based on documentation: "INSERT and ALTER are replicated"
It is related to `Replicated*` tables, not to `Distributed`.
From the [doc](https://clickhouse.yandex/docs/en/single/index.html#document-table_engines/distributed):
> The Distributed... | 2017-07-04T18:59:46 | 2017-07-04T18:59:46 | {
"+1": 5
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 314,019,874 | 936 | silviucpp | Ah ok got it. Might be better to add this in documentation as well :) I'm sure it's confusing for other peoples as well. | 2017-07-10T06:47:38 | 2017-07-10T06:47:38 | {
"+1": 1
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 314,771,970 | 976 | vavrusa | 😓 Thanks! | 2017-07-12T13:39:46 | 2017-07-12T13:39:46 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 314,969,769 | 971 | osbertngok | Related question, when we do `ALTER TABLE implus.test_a ADD COLUMN NewColumn String DEFAULT '' AFTER OldColumn;`, what exactly happens to the existing row in `implus.test_a`? Is the expected behavior that it would be padded with NewColumn = '', or the column of that row will just remain undefined? | 2017-07-13T04:48:47 | 2017-07-13T04:48:47 | {} | NONE |
ClickHouse | ClickHouse | 311,584,224 | 924 | silviucpp | Hello I see v1.1.54242-stable was released in github but not yet in the deb repository. It's a known behavior ? | 2017-06-28T07:49:27 | 2017-06-28T07:49:27 | {
"confused": 1
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 312,939,891 | 936 | silviucpp | Hello thanks,
Very useful comment. I'll try the new feature "distributed DDL" but still you are telling that "Distributed table is just proxies INSERT and SELECT queries." if it's like this then why in order to alter a replicated table I need to:
1. alter all replicated tables on all nodes where I have it.
(at t... | 2017-07-04T19:43:45 | 2017-07-04T19:43:45 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 311,321,812 | 930 | robot-metrika-test | Can one of the admins verify this patch? | 2017-06-27T10:44:02 | 2017-06-27T10:44:02 | {} | NONE |
ClickHouse | ClickHouse | 311,645,528 | 925 | alexey-milovidov | https://github.com/yandex/ClickHouse/commit/c6b83a1c60b611af2ec47ca0e1251fbb73cc51ee
https://github.com/yandex/ClickHouse/commit/39ddcf1c746cadbb8295215964c5821220a3fad8
https://github.com/yandex/ClickHouse/commit/2b67782845395db55c0b6d2194362a5558969505 | 2017-06-28T12:30:56 | 2017-06-28T12:30:56 | {} | MEMBER |
ClickHouse | ClickHouse | 311,877,007 | 931 | hagen1778 | Any workaround how to avoid RAM exhaustion and increase part size? | 2017-06-29T06:43:39 | 2017-06-29T06:43:39 | {} | NONE |
ClickHouse | ClickHouse | 312,438,401 | 942 | robot-metrika-test | Can one of the admins verify this patch? | 2017-07-01T15:29:03 | 2017-07-01T15:29:03 | {} | NONE |
ClickHouse | ClickHouse | 313,243,669 | 839 | ztlpn | fix was released in v1.1.54245 | 2017-07-05T22:24:39 | 2017-07-05T22:24:39 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 313,529,357 | 947 | ludv1x | Thank you for your detail proposal :+1: .
There are two usecases of MergingSortedBlockInputStream: implementation of ORDER BY and background merges.
1. Prefetching is implemented for ORDER BY, see [here](https://github.com/yandex/ClickHouse/blob/master/dbms/src/Interpreters/InterpreterSelectQuery.cpp#L1131) and A... | 2017-07-06T21:51:16 | 2017-07-06T21:51:16 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 314,775,361 | 975 | m1schka | thanks for the valuable insights!
I ended up with the following statement, which gave me the results I was looking for (see the `i != 1` as enumerate starts with 1):
```SQL
SELECT
case_id,
arrayFilter((x, i) -> (i != 1), groupArray(runningDifference(activity_datetime)) AS arr, arrayEnumerate(arr)) AS delta... | 2017-07-12T13:50:50 | 2017-07-12T13:50:50 | {
"+1": 1
} | NONE |
ClickHouse | ClickHouse | 314,846,711 | 979 | alexey-milovidov | Time type will have almost no advantages over DateTime.
In memory, we need more than two bytes for representation of Time up to seconds resolution, and due to alignment and for simpler arithmetic, it will be four bytes, just as for DateTime.
On disk, full DateTime will compress to similar size as Time, because ma... | 2017-07-12T17:52:54 | 2017-07-12T17:52:54 | {} | MEMBER |
ClickHouse | ClickHouse | 315,212,226 | 980 | alexey-milovidov | > It would be nice if you share eclipse settings or at least formatting style to all contributors. Ideally, a whole development environment in a VM.
You could try to use `clang-format` for new or rewritten files.
(For this you should have latest clang (4.0). Use it as following: `clang-format -style=file -i filenam... | 2017-07-13T21:51:03 | 2017-07-13T21:52:23 | {} | MEMBER |
ClickHouse | ClickHouse | 315,764,776 | 987 | khrisanfov | Thank you. Now it works fine. | 2017-07-17T14:02:37 | 2017-07-17T14:02:37 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 310,886,463 | 924 | silviucpp | Hello Alexey,
I think community can provide also fast feedback in case you will have another channel where releases can happened much faster.. (maybe at each commit in master or daily)
Silviu | 2017-06-25T06:59:46 | 2017-06-25T06:59:46 | {
"+1": 2
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 311,752,141 | 931 | ztlpn | Yes, problem acknowledged. When you raised the part size limit, the server suddenly became aware that it can do a lot of new merges. As with the default settings ClickHouse can execute 8 merges without any size limit, 8 concurrent heavy merges were launched. Together they ate up all the RAM.
Your proposed solution s... | 2017-06-28T18:47:57 | 2017-06-28T18:47:57 | {
"+1": 3
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 311,863,333 | 938 | vavrusa | I think this is fixed in 28e9e069923e67b8cc9d865117d0fd712e2fa00e it should happen when the cleanup thread was trying to clean a block that was not fully written and didn't have either /checksum or /number. | 2017-06-29T05:04:42 | 2017-06-29T05:04:42 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 313,868,614 | 947 | innerr | Thanks a lot!
Does it means a new method in IColumn for vertical merge like IColumn.filterAndPermute? | 2017-07-08T17:10:31 | 2017-07-08T17:22:13 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 314,621,367 | 951 | vavrusa | Ah yes, I was wondering how to test it. Good idea to use `max_rows_to_read`, I'll make a PR soonish! | 2017-07-12T01:50:43 | 2017-07-12T01:50:43 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 312,108,966 | 936 | silviucpp | Behavior it's pretty odd...
Basically I create a table with a column which is `Nullable(String)` then I alter the table to change the type of that column in `Nullable(Int64)`. (look at `validation_date` in the following example)
```sql
CREATE TABLE project_1_1.ivr_src(`ck.partition` Date,`ck.created_at` DateTime... | 2017-06-29T21:07:47 | 2017-06-29T21:07:47 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 312,115,048 | 924 | silviucpp | By the way for all those that needs and want to test the testing releases they need to change only in `/etc/apt/sources.list`:
`deb http://repo.yandex.ru/clickhouse/trusty stable main`
with
`deb http://repo.yandex.ru/clickhouse/trusty testing main`
I think this should be documented :) | 2017-06-29T21:34:05 | 2017-06-29T21:34:05 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 313,174,219 | 950 | robot-metrika-test | Can one of the admins verify this patch? | 2017-07-05T17:39:03 | 2017-07-05T17:39:03 | {} | NONE |
ClickHouse | ClickHouse | 313,923,802 | 936 | alexey-milovidov | Distributed table is just a view, but it has it's own table structure (list of columns),
that must match table structure of remote tables, or at least have a subset of columns.
This looks redundant, but it is much more simple in implementation: every table has it's own list of columns.
That's why you have to alt... | 2017-07-09T14:38:29 | 2017-07-09T14:38:29 | {} | MEMBER |
ClickHouse | ClickHouse | 314,285,657 | 964 | alexey-milovidov | Also please replace all occurences of `isdigit` to `isNumericASCII` (minor modification). | 2017-07-11T00:14:21 | 2017-07-11T00:14:21 | {} | MEMBER |
ClickHouse | ClickHouse | 313,951,504 | 936 | alexey-milovidov | This is correct.
It's typical to have Distributed table on every node on cluster.
In that case, you just run second `ALTER TABLE y ON CLUSTER`. | 2017-07-09T20:12:25 | 2017-07-09T20:12:35 | {} | MEMBER |
ClickHouse | ClickHouse | 314,288,758 | 426 | alexey-milovidov | I just have run grep of error log across all the servers in our cluster.
We had segfaults in `DB::Block::bytes`, but all of them was in September 2016 (just two days: 27th and 28th September).
I think, this issue is relevant to very old version of ClickHouse.
Nevertheless, if the issue is still present in newer ve... | 2017-07-11T00:35:23 | 2017-07-11T00:35:23 | {} | MEMBER |
ClickHouse | ClickHouse | 314,746,460 | 975 | valyala | See how `$rateColumns` is implemented at [clickhouse-grafana](https://github.com/Vertamedia/clickhouse-grafana/blob/master/README.md) | 2017-07-12T11:53:27 | 2017-07-12T11:53:27 | {} | NONE |
ClickHouse | ClickHouse | 314,853,670 | 979 | alexey-milovidov | Agree. Special time type with custom resolution will be way more convenient. | 2017-07-12T18:17:47 | 2017-07-12T18:17:47 | {} | MEMBER |
ClickHouse | ClickHouse | 315,419,479 | 985 | alexey-milovidov | autotests | 2017-07-14T17:35:00 | 2017-07-14T17:35:00 | {} | MEMBER |
ClickHouse | ClickHouse | 315,741,261 | 987 | proller | You need to have clang binary.
Try apt install clang
also you can pass custom path to clang:
env DEB_CLANG=/usr/bin/clang-5.0 ./release | 2017-07-17T12:26:16 | 2017-07-17T12:26:16 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 311,643,309 | 930 | alexey-milovidov | Build should work without LZMA. This was fixed in https://github.com/yandex/ClickHouse/pull/928 | 2017-06-28T12:20:55 | 2017-06-28T12:20:55 | {} | MEMBER |
ClickHouse | ClickHouse | 312,701,985 | 945 | alexey-milovidov | autotests | 2017-07-03T17:42:24 | 2017-07-03T17:42:24 | {} | MEMBER |
ClickHouse | ClickHouse | 312,937,629 | 329 | innerr | Is this feature deliver yet?
A further question: is "grid" indexes in the plan?
("grid" indexes: keeping small aggregates (min, max, all unique values if not many) for each block of data, to allow skipping of data blocks - by alexey-milovidov)
@ludv1x | 2017-07-04T19:21:29 | 2017-07-04T19:43:28 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 313,428,920 | 945 | YiuRULE | So, little erratum, just waited to fix all the things for telling this, my machine is in little endian (and would be actually easier to this case if it was in little indian) and not big endian, sorry for the misunderstood. :)
Ok, so everything was fix, `UUID` is now a type different of `UInt128` using the `STRONG_TY... | 2017-07-06T15:21:59 | 2017-07-06T15:22:10 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 313,598,079 | 945 | alexey-milovidov | Ok. I need about two days for further review. | 2017-07-07T06:25:50 | 2017-07-07T06:25:50 | {} | MEMBER |
ClickHouse | ClickHouse | 314,374,035 | 426 | ztlpn | Looks similar to #844 which was fixed in v1.1.54236 | 2017-07-11T08:36:01 | 2017-07-11T08:36:01 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 315,636,147 | 935 | yctn | @ludv1x you are piping and not STDIN. piping works fine. only STDIN doesnt cause clickhouse-client does not understand \n
to test stdin to clickhouse-client i did echo "query\n" > /proc/pid/fd/0 the line appeared in the clickhouse-client as "query\n" so my guess is that clickhouse-client does not accept STDIN correc... | 2017-07-16T20:50:36 | 2017-07-16T20:51:02 | {} | NONE |
ClickHouse | ClickHouse | 311,382,107 | 932 | inv2004 | looks like duplicate: https://github.com/yandex/ClickHouse/issues/687 | 2017-06-27T14:48:24 | 2017-06-27T14:48:24 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 311,809,273 | 937 | robot-metrika-test | Can one of the admins verify this patch? | 2017-06-28T22:24:03 | 2017-06-28T22:24:03 | {} | NONE |
ClickHouse | ClickHouse | 312,440,356 | 942 | keijiyoshida | @alexey-milovidov Thanks for your checking and the great DBMS. I watched your presentation at Data@Scale 2017, got interested in ClickHouse and I'm trying to use it now. | 2017-07-01T16:05:09 | 2017-07-01T16:05:09 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 312,442,028 | 941 | arduanov | thanks for the answer, i use mac only for tests | 2017-07-01T16:36:36 | 2017-07-01T16:36:36 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 312,702,380 | 945 | alexey-milovidov | > my machine is in big endian
I wonder, does ClickHouse work on this machine? There are quite a lot of code, that rely on little endianess. | 2017-07-03T17:44:32 | 2017-07-03T17:44:32 | {} | MEMBER |
ClickHouse | ClickHouse | 314,284,531 | 957 | alexey-milovidov | > add a flag, that will allow to INSERT into MATERIALIZED column, if INSERT is generated internally (it will also solve the same issue in Distributed tables);
That is the preferred solution.
This is very easy task (half an hour for those who are familiar with the codebase).
1. Add a setting `insert_allow_materia... | 2017-07-11T00:07:25 | 2017-07-11T00:07:25 | {
"+1": 1
} | MEMBER |
ClickHouse | ClickHouse | 311,673,661 | 925 | vavrusa | Oops, I forgot about this file, sorry 😓 | 2017-06-28T14:17:55 | 2017-06-28T14:17:55 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 313,301,501 | 951 | robot-metrika-test | Can one of the admins verify this patch? | 2017-07-06T05:44:03 | 2017-07-06T05:44:03 | {} | NONE |
ClickHouse | ClickHouse | 314,888,007 | 971 | alexey-milovidov | We have quite a stupid behaviour: asterisk is expanded only to the list of columns from "left" table.
We are going to fix it. | 2017-07-12T20:32:15 | 2017-07-12T20:32:15 | {
"+1": 1
} | MEMBER |
ClickHouse | ClickHouse | 315,745,973 | 987 | proller | get_revision_author bug was caused by
5898c11e9f103b3f1d187286be54383fc7845a52
fixed:
https://github.com/yandex/ClickHouse/commit/f068a3edca934927b6e353161d9b0d9708967db3 | 2017-07-17T12:49:20 | 2017-07-17T12:49:20 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 310,921,327 | 922 | alexey-milovidov | Thanks! | 2017-06-25T19:02:29 | 2017-06-25T19:02:29 | {} | MEMBER |
ClickHouse | ClickHouse | 312,430,046 | 941 | alexey-milovidov | Looks like it is using default malloc library from system libc version for some allocations, although ClickHouse should be linked to tcmalloc. Probably, this is build issue on Mac OS (there are some difficulties with allocators).
The issue should not present if you will run clickhouse from Ubuntu in Docker image. | 2017-07-01T12:38:38 | 2017-07-01T12:38:38 | {} | MEMBER |
ClickHouse | ClickHouse | 312,692,402 | 945 | robot-metrika-test | Can one of the admins verify this patch? | 2017-07-03T16:49:03 | 2017-07-03T16:49:03 | {} | NONE |
ClickHouse | ClickHouse | 313,732,566 | 947 | alexey-milovidov | Also we don't repeatedly push and pop from priority_queue if we have a sequence of values all of them are lower than other: https://github.com/yandex/ClickHouse/blob/55bfaca662f68123cd14d916d43923aadaba3b95/dbms/src/DataStreams/MergingSortedBlockInputStream.cpp#L282
For background merges, priority_queue is not a bot... | 2017-07-07T16:40:32 | 2017-07-07T16:40:32 | {} | MEMBER |
ClickHouse | ClickHouse | 314,255,963 | 937 | alexey-milovidov | I still would like to have a test in our repository.
Because otherwise we could eventually break this function. | 2017-07-10T21:35:05 | 2017-07-10T21:35:05 | {} | MEMBER |
ClickHouse | ClickHouse | 315,401,898 | 985 | robot-metrika-test | Can one of the admins verify this patch? | 2017-07-14T16:19:03 | 2017-07-14T16:19:03 | {} | NONE |
ClickHouse | ClickHouse | 311,218,616 | 928 | robot-metrika-test | Can one of the admins verify this patch? | 2017-06-27T00:29:03 | 2017-06-27T00:29:03 | {} | NONE |
ClickHouse | ClickHouse | 312,121,275 | 939 | robot-metrika-test | Can one of the admins verify this patch? | 2017-06-29T22:04:03 | 2017-06-29T22:04:03 | {} | NONE |
ClickHouse | ClickHouse | 312,978,022 | 948 | inv2004 | Found ~ the same discussion in google-groups: https://groups.google.com/d/msg/clickhouse/EuKcaYw_NaM/iKy1cDpECgAJ | 2017-07-05T01:53:16 | 2017-07-05T01:53:16 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 313,243,778 | 546 | ztlpn | fix was released in v1.1.54245 | 2017-07-05T22:25:20 | 2017-07-05T22:25:20 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 313,932,847 | 952 | alexey-milovidov | --Perfect.--
The only missing part is test. This is your debt for the task. | 2017-07-09T17:17:07 | 2017-07-09T17:20:19 | {} | MEMBER |
ClickHouse | ClickHouse | 313,949,555 | 936 | silviucpp | @alexey-milovidov so means that in order to alter a table you need to :
1. Run "ALTER TABLE x ON CLUSTER" cluster on one of the nodes of the replicated table
2. Run the alter on all nodes individual where you have a distributed table ?
I'm correct ? seems a bit redundant.. | 2017-07-09T19:36:39 | 2017-07-09T19:36:39 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 314,924,076 | 980 | robot-metrika-test | Can one of the admins verify this patch? | 2017-07-12T23:14:02 | 2017-07-12T23:14:02 | {} | NONE |
ClickHouse | ClickHouse | 315,161,560 | 980 | mfridental | executeImplTyped is intended to be virtual; added "override" in the child class.
Reformatted using a code style without tabs.
It would be nice if you share eclipse settings or at least formatting style to all contributors. Ideally, a whole development environment in a VM.
By the way, in my environment the test 00... | 2017-07-13T18:23:12 | 2017-07-13T18:23:12 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 311,445,723 | 931 | ztlpn | Is increasing `max_bytes_to_merge_at_max_space_in_pool` the only thing that changed? In particular I'm interested if you have left ClickHouse version intact (which version is it BTW?) or did you reinstall ClickHouse from source? | 2017-06-27T18:32:02 | 2017-06-27T18:32:02 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 311,925,098 | 931 | ztlpn | For now you can try lowering the number of concurrent heavy merges. It is controlled by `background_pool_size` (configurable in the default user profile, 16 by default) and `number_of_free_entries_in_pool_to_lower_max_size_of_merge` (configurable in the `merge_tree` config section, but unfortunately only after https://... | 2017-06-29T10:19:52 | 2017-06-29T10:19:52 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 312,701,938 | 945 | alexey-milovidov | Great!
What about if data type will be called UUID (not Uuid) for style consistency (and in all other mentions)? | 2017-07-03T17:42:09 | 2017-07-03T17:42:09 | {} | MEMBER |
ClickHouse | ClickHouse | 314,253,407 | 924 | alexey-milovidov | Ok. Now we have auto-publishing of testing packages; and possibility to use testing repositories was documented. | 2017-07-10T21:29:16 | 2017-07-10T21:29:16 | {} | MEMBER |
ClickHouse | ClickHouse | 314,326,620 | 937 | mfridental | Yes, will add it asap, currently I'm very busy with other things. Hope to have time this week.
On 10 Jul 2017, 23:35, at 23:35, alexey-milovidov <notifications@github.com> wrote:
>I still would like to have a test in our repository.
>Because otherwise we could eventually break this function.
>
>--
>You are receiving ... | 2017-07-11T05:26:07 | 2017-07-11T05:26:07 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 315,523,276 | 980 | mfridental | Here is my environment:
immowelt@ubuntu:~/ClickHouse$ uname -r
4.8.0-36-generic
immowelt@ubuntu:~/ClickHouse$ lsb_release -a
No LSB modules are available.
Distributor ID: Ubuntu
Description: Ubuntu 16.04.2 LTS
Release: 16.04
Codename: xenial
immowelt@ubuntu:~/ClickHouse$ cat /proc/cpuinfo
processor : ... | 2017-07-15T09:48:24 | 2017-07-15T09:48:24 | {} | CONTRIBUTOR |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.