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 | 615,053,711 | 10,305 | filimonov | > I assumed `new_avg = avg((existing_val + new_val) / 2)` which only requires existing state and new state. ... Am I missing something?
You are.
((10+50)/2 +90)/2=60 (proper avg for thouse 3 numbers is 50).
Avg = sum / count
Sum and count can both be simple aggregate functions. Just divide one by another at s... | 2020-04-17T05:40:32 | 2020-04-17T05:43:45 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 615,152,415 | 10,305 | filimonov | Just change the order to get completely another result.
`((90+50)/2 +10)/2=40`
Or: ((10+10)/2 + (20+20+20+20+20+20+20+20+20)/9)/2
Are you sure that 15 here is a correct answer?
That approach is not usable. | 2020-04-17T09:46:01 | 2020-04-17T09:46:23 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 615,243,688 | 10,332 | im-gc | @den-crane thank you . but i need to change the engine keeping on serving . | 2020-04-17T13:27:32 | 2020-04-17T13:27:32 | {} | NONE |
ClickHouse | ClickHouse | 615,286,434 | 10,200 | KochetovNicolai | @AlexanderJLiu can you please update to 20.3.7? | 2020-04-17T14:47:11 | 2020-04-17T14:47:11 | {} | MEMBER |
ClickHouse | ClickHouse | 615,362,636 | 10,336 | den-crane | could not reproduce
```
cat /etc/clickhouse-server/conf.d/user_default.xml
<?xml version="1.0" ?>
<yandex>
<users>
<default>
<networks>
<ip>::1</ip>
<ip>127.0.0.1</ip>
</networks>
</default>
</users>
</yandex>
```
clickho... | 2020-04-17T17:10:31 | 2020-04-17T17:10:39 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 615,876,110 | 10,349 | alexey-milovidov | Yes. | 2020-04-18T13:57:21 | 2020-04-18T13:57:21 | {} | MEMBER |
ClickHouse | ClickHouse | 614,895,432 | 332 | azat | >If we do GROUP BY injective function of sharding key
Indeed, I already though about that kind of optimization (automatically set `distributed_group_by_no_merge`, this can speedup distributed queries a lot in such cases)
>and the setting skip_unavailable_shards is active
Am I understand this correctly, the pro... | 2020-04-16T21:05:09 | 2020-04-16T21:05:09 | {} | MEMBER |
ClickHouse | ClickHouse | 614,928,384 | 10,312 | tavplubix | Looks like it also fixes some other `00993_system_parts_race_condition_drop_zookeeper`-related issues from CI Fest. I ran this test 100 times with debug build and 100 times with tsan debug build, no logical errors, races or segfaults found | 2020-04-16T22:25:25 | 2020-04-16T22:25:25 | {
"+1": 1
} | MEMBER |
ClickHouse | ClickHouse | 615,080,312 | 10,319 | filimonov | > Our version number: 19.16.4.12
Update to the latest 19.16 release.
> gubinjie closed this 4 hours ago
Please write some note about why did you close the issue
Also please provide the required information in https://github.com/ClickHouse/ClickHouse/issues/10284
| 2020-04-17T07:03:32 | 2020-04-17T07:03:32 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 615,181,492 | 7,748 | Girgitt | This issue could be related to #8939 (?) | 2020-04-17T10:54:51 | 2020-04-17T10:54:51 | {} | NONE |
ClickHouse | ClickHouse | 615,379,152 | 10,334 | XavierQuiniou | Thanks,
One more, it’s great.
Yes, I had already found a solution by using arrays.
Principle of code to compute diff:
<pre>Select …
groupArray(Date) as Date_arr,
arrayEnumerate(Date_arr) as indexes,
groupArray(Qty) as Qty_arr,
arrayMap( i -> Qty _arr[i] - Qty_arr[i-1], indexes) as Qty_arr,
… tw... | 2020-04-17T17:46:56 | 2020-04-17T17:46:56 | {} | NONE |
ClickHouse | ClickHouse | 615,378,770 | 10,336 | den-crane | it could be that you corrupted settings for default user and your xml is not applied at all. So default settings works -- empty password and access from everywhere.
check err.log (/var/log/clickhouse-server/clickhouse-server.err.log) for
```
2020.04.17 17:46:55.394418 [ 413480 ] {} <Error> Access(users.xml): Could... | 2020-04-17T17:46:05 | 2020-04-17T17:47:40 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 615,380,059 | 10,336 | TimonKK | I mean what If I have empty password tag into users.xml:
```xml
<?xml version="1.0" ?>
<yandex>
<users>
<test>
<password></password>
<networks replace="replace">
<ip>127.0.0.1</ip>
<ip>::1</ip>
</networks>
</test>
</users>
</yandex>
``... | 2020-04-17T17:48:56 | 2020-04-17T17:48:56 | {} | NONE |
ClickHouse | ClickHouse | 615,461,734 | 10,336 | den-crane | I could not reproduce the issue using your users.xml
Do you have nginx at the same host?
Do you have any files at /etc/clickhouse-server/conf.d/ or /etc/clickhouse-server/users.d ? | 2020-04-17T21:01:29 | 2020-04-17T21:01:29 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 615,532,022 | 10,341 | alexey-milovidov | What I'm wondering before looking into code... is that `distributed_group_by_no_merge` just sets query processing stage to Complete instead of WithMergeableState and it works not only for GROUP BY queries, but will also prevent merging of streams for queries with ORDER BY and without GROUP BY.
Let's check that. | 2020-04-18T01:20:53 | 2020-04-18T01:21:38 | {} | MEMBER |
ClickHouse | ClickHouse | 615,685,550 | 10,263 | azat | >PVS check — Found 1 new errors, total 20 errors
Does not looks related
>test_insert_into_distributed/test.py::test_inserts_batching
Uses `sleep` over `system flush distributed`, and I decided do just increase the sleep time (maybe it will handle some corner cases, even though I tried to cover all of them) | 2020-04-18T07:36:58 | 2020-04-18T07:36:58 | {} | MEMBER |
ClickHouse | ClickHouse | 615,008,162 | 10,300 | land012 | thank you! | 2020-04-17T02:47:51 | 2020-04-17T02:47:51 | {} | NONE |
ClickHouse | ClickHouse | 615,382,766 | 10,336 | den-crane | Still nothing, please show all configs.
Two different users: `test` VS `default`
```
<users> <test>
curl https://domain.com/?
```
| 2020-04-17T17:54:30 | 2020-04-17T17:54:45 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 615,491,679 | 10,293 | alexey-milovidov | Yes, it was the idea but there are three considerations:
- surprisingly enough, there is no counter from jemalloc stats that corresponds to process' resident set size;
- we also allocate some memory without jemalloc using mmap/munmap (but not enough to explain the previous consideration);
- we should not depend on s... | 2020-04-17T22:28:36 | 2020-04-17T22:28:48 | {} | MEMBER |
ClickHouse | ClickHouse | 615,872,986 | 10,349 | hellvesper | Like this?
```
if (${CMAKE_SYSTEM_PROCESSOR} STREQUAL "amd64")
execute_process (COMMAND ${CMAKE_CXX_COMPILER} --print-file-name=libclang_rt.builtins-x86_64.a OUTPUT_VARIABLE BUILTINS_LIBRARY OUTPUT_STRIP_TRAILING_WHITESPACE)
else ()
execute_process (COMMAND ${CMAKE_CXX_COMPILER} --print-file-name=libclang_... | 2020-04-18T13:35:40 | 2020-04-18T13:35:40 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 615,074,998 | 10,305 | filimonov | That has no sense at all.
It will be non-determenistic (depends on insert / merge order) and as accurate as random from the range min(value) ... max(value).
ClickHouse needs a way to merge 2 states of aggregate functions (it doesn't do it row-by-row in insert order).
So if we have a block of data with value = ... | 2020-04-17T06:49:06 | 2020-04-17T06:52:24 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 615,260,888 | 8,045 | den-crane |
```
SELECT
multiIf(isNull(A) AND isNull(B), 'Total', isNull(B) AND isNotNull(A), 'Subtotal', '') AS desc,
A,
B,
s,
c
FROM
(
SELECT
CAST(A, 'Nullable(String)') AS A,
CAST(B, 'Nullable(Int32)') AS B,
sum(C) AS s,
count() AS c
FROM rollup
G... | 2020-04-17T14:00:38 | 2020-04-17T14:00:38 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 615,283,777 | 10,334 | den-crane | using Arrays?
```
SELECT
n,
nn
FROM
(
SELECT
groupArray(n) AS na,
arrayMap(y -> y, arrayPushFront(arraySlice(na, 1, -1), na[1])) AS na1
FROM
(
SELECT number AS n
FROM numbers(10)
ORDER BY n ASC
)
)
ARRAY JOIN
na AS n,
na1 AS nn... | 2020-04-17T14:42:12 | 2020-04-17T14:42:12 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 615,377,674 | 10,336 | den-crane | >But If I set empty password
What do you mean?
I could not reproduce
```
cat /etc/clickhouse-server/conf.d/user_test.xml
<?xml version="1.0" ?>
<yandex>
<users>
<test>
<password>test</password>
<profile>default</profile>
<quota>default</quota>
<netw... | 2020-04-17T17:43:26 | 2020-04-17T17:43:26 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 615,838,366 | 10,320 | maison2710 | @alexey-milovidov As I made some modification to source code, I want to build Debian Package by running `./release`.
When I followed instruction for building on any Linux, it worked. But when I run `./release`, above error happened. | 2020-04-18T10:30:36 | 2020-04-18T10:30:36 | {} | NONE |
ClickHouse | ClickHouse | 615,887,139 | 10,301 | thhave | > May be your input file have a BOM mark at the very beginning?
Yes. | 2020-04-18T15:11:02 | 2020-04-18T15:11:02 | {} | NONE |
ClickHouse | ClickHouse | 614,899,632 | 332 | alexey-milovidov | > Am I understand this correctly, the problem is that if this setting will be ignored, in case of server is unavailable the server will send all rows from all shards except unavailable while it should just give an error? Nice catch!
No, I messed it with skip_unused_shards :) | 2020-04-16T21:13:48 | 2020-04-16T21:13:48 | {} | MEMBER |
ClickHouse | ClickHouse | 615,170,211 | 8,045 | XavierQuiniou | No answer to my question?
I found a solution here
https://dzone.com/articles/implementing-subtotals-aggregations-in-clickhouse
But it's somewhat complex to do such a simple thing.
| 2020-04-17T10:26:03 | 2020-04-17T10:26:03 | {} | NONE |
ClickHouse | ClickHouse | 615,237,328 | 10,326 | den-crane | Materialized view is an insert trigger it catches data from inserts.
Did you create your MV at all CH nodes?
It would be nice to see DDL of MV. Very often users create MV table with wrong `partition by` / `order by` -- inconsistent with MV select. | 2020-04-17T13:14:22 | 2020-04-17T13:14:48 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 615,369,596 | 10,336 | den-crane | it means replace values inherited from default profile | 2020-04-17T17:25:41 | 2020-04-17T17:25:41 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 615,470,234 | 5,618 | blinkov | @I-akshay, do you have any further questions? | 2020-04-17T21:24:02 | 2020-04-17T21:24:02 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 615,048,133 | 10,305 | vmihailenco | I assumed `new_avg = avg((existing_val + new_val) / 2)` which only requires existing state and new state. I guess min works the same `new_min = min(existing_val, new_val)`. Am I missing something? | 2020-04-17T05:19:09 | 2020-04-17T05:19:09 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 615,126,494 | 8,639 | alesapin | No new failures https://github.com/ClickHouse/ClickHouse/pull/8639#issuecomment-613917326 | 2020-04-17T08:49:51 | 2020-04-17T08:49:51 | {} | MEMBER |
ClickHouse | ClickHouse | 615,247,769 | 10,301 | filimonov | May be your input file have a BOM mark at the very beginning? | 2020-04-17T13:35:36 | 2020-04-17T13:35:36 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 615,395,603 | 10,336 | den-crane | So it's your responsibility to check that config is valid.
If config (or .xml) is invalid it's not used. Then a default settings applied. User `default` without a password and without network restrictions. | 2020-04-17T18:24:24 | 2020-04-17T18:24:51 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 615,533,156 | 5,591 | alexey-milovidov | @BayoNet I cannot find LowCardinality in documentation.
LowCardinality feature is used in production since long time ago. | 2020-04-18T01:25:21 | 2020-04-18T01:25:21 | {} | MEMBER |
ClickHouse | ClickHouse | 615,539,147 | 10,340 | akuzm | https://github.com/ClickHouse/ClickHouse/pull/10340/files#diff-4571d47993d9a8ed40fd8c6c27de8ceaL15
The main offending line was this -- it just copied a block and then mutated all its columns, for no good reason. | 2020-04-18T02:09:37 | 2020-04-18T02:09:37 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 615,887,915 | 10,288 | sfriquet | @alesapin I still see the same issue with version `20.3.7.46` | 2020-04-18T15:16:23 | 2020-04-18T15:16:23 | {} | NONE |
ClickHouse | ClickHouse | 614,900,298 | 332 | alexey-milovidov | > Although brief look at isInjective shows that it is not set correctly for some useful stuff (module, consistent hashing), so this also should not be forgotten.
But modulo and consistent hashing are not injective.
Example of injective function is `toString`.
Suppose we write:
```
SELECT ... GROUP BY toString(... | 2020-04-16T21:15:14 | 2020-04-16T21:15:14 | {} | MEMBER |
ClickHouse | ClickHouse | 614,902,248 | 3,933 | alexey-milovidov | That's awful, the query breaks client-server protocol:
```
milovidov-desktop :) SELECT arrayMap(x -> arrayJoin([x, 1]), [1, 2])
SELECT arrayMap(x -> arrayJoin([x, 1]), [1, 2])
┌─arrayMap(lambda(tuple(x), arrayJoin(array(x, 1))), [1, 2])─┐
│ [1,1] │
└─────... | 2020-04-16T21:19:49 | 2020-04-16T21:19:49 | {} | MEMBER |
ClickHouse | ClickHouse | 615,013,639 | 10,221 | alexey-milovidov | Now another different integration test has failed. | 2020-04-17T03:11:06 | 2020-04-17T03:11:06 | {} | MEMBER |
ClickHouse | ClickHouse | 615,242,603 | 10,331 | filimonov | Most probably a duplicate of:
https://github.com/ClickHouse/ClickHouse/issues/6371
https://github.com/ClickHouse/ClickHouse/issues/5825
Fixed in
https://github.com/ClickHouse/ClickHouse/pull/6378
Please update your ClickHouse. | 2020-04-17T13:25:22 | 2020-04-17T13:26:12 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 614,905,045 | 332 | azat | >No, I messed it with skip_unused_shards :)
Now I got it!
>server is unavailable ... it should just give an error?
Too smart :)
>But modulo and consistent hashing are not injective.
Indeed, my mistake.
Optimization for GROUP BY and for WHERE (optimize_skip_unused_shards) mixed in my head | 2020-04-16T21:26:20 | 2020-04-16T21:26:20 | {} | MEMBER |
ClickHouse | ClickHouse | 615,327,911 | 5,183 | filimonov | Steps to reproduce:
```bash
# open a http connection
exec 3<>/dev/tcp/127.0.0.1/8123
# start sending request
echo "GET / HTTP/1.0" >&3
# now kill the connection (to mimic connection reset)
sudo ss -K dst 127.0.0.1 dport = 8123
``` | 2020-04-17T16:01:18 | 2020-04-17T16:01:18 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 615,365,514 | 10,336 | TimonKK | What does attribute **replace="replace"** mean in networks tags? | 2020-04-17T17:17:08 | 2020-04-17T17:17:08 | {} | NONE |
ClickHouse | ClickHouse | 615,838,375 | 10,336 | TimonKK | Yes, nginx and CH placed at the same server | 2020-04-18T10:30:40 | 2020-04-18T10:32:11 | {} | NONE |
ClickHouse | ClickHouse | 615,843,639 | 8,212 | Slach | duplicated with https://github.com/ClickHouse/ClickHouse/issues/4342 | 2020-04-18T11:15:12 | 2020-04-18T11:15:12 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 614,923,426 | 10,000 | akuzm | ERROR: type should be string, got "https://clickhouse-test-reports.s3.yandex.net/10000/602baf62f00c653975432bd9a9f32e9ad70774fa/performance_comparison/all-queries.html#2.1004\r\n\r\nLooks like a statistically significant 5% regression." | 2020-04-16T22:11:37 | 2020-04-16T22:11:37 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 615,138,884 | 346 | 4ertus2 | You may try and make an issue if there's something unexpected. | 2020-04-17T09:16:07 | 2020-04-17T09:16:34 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 615,536,331 | 10,320 | alexey-milovidov | By the way, it isn't necessarily to use `pbuilder`.
Just follow the instructions:
https://clickhouse.tech/docs/en/development/build/#how-to-build-clickhouse-on-any-linux | 2020-04-18T01:47:26 | 2020-04-18T01:47:26 | {} | MEMBER |
ClickHouse | ClickHouse | 615,865,460 | 10,349 | alexey-milovidov | But we support build not only for x86_64.
Maybe better to add if that will replace amd64 to x86_64 for this particular case? | 2020-04-18T12:58:09 | 2020-04-18T12:58:09 | {} | MEMBER |
ClickHouse | ClickHouse | 615,304,553 | 10,327 | den-crane | ```
SELECT
rand() = rand(),
rand(),
rand()
FROM numbers(10)
┌─equals(rand(), rand())─┬─────rand()─┬─────rand()─┐
│ 1 │ 391924423 │ 391924423 │
│ 1 │ 1383619981 │ 1383619981 │
│ 1 │ 1060033606 │ 1060033606 │
│ 1... | 2020-04-17T15:20:14 | 2020-04-17T15:20:14 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 615,845,616 | 10,325 | alexey-milovidov | clang-tidy can be suppressed with `// NOLINT` | 2020-04-18T11:25:35 | 2020-04-18T11:25:35 | {} | MEMBER |
ClickHouse | ClickHouse | 615,867,099 | 10,133 | alexey-milovidov | @blinkov is a robot (and very primitive one) and he has assigned this task by mistake. | 2020-04-18T13:05:47 | 2020-04-18T13:05:47 | {} | MEMBER |
ClickHouse | ClickHouse | 615,055,349 | 10,305 | vmihailenco | I use `sum / count` trick when I can and that is usually called `mean` value.
But sometimes I have a single value and want to merge them by calculating value in between `(x+y)/2`. It seems to be better than using `min/max/any`. | 2020-04-17T05:46:46 | 2020-04-17T05:47:29 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 615,231,401 | 10,331 | astudnev | [err.txt](https://github.com/ClickHouse/ClickHouse/files/4493030/err.txt)
| 2020-04-17T13:01:55 | 2020-04-17T13:01:55 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 615,388,987 | 10,334 | den-crane | Obviously for millions rows pivoted into Array it will require a lot of memory.
I was trying to avoid arrayEnumerate and do not create an additional array but I am not sure that it helped because of arraySlice. | 2020-04-17T18:08:50 | 2020-04-17T18:09:04 | {
"+1": 1
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 615,449,337 | 10,336 | TimonKK | [It](https://pastebin.com/eSX3BzMz) is my correct valid users.xml and the clickhouse-server starts without errors.
This config contains:
- empty password for default user
- section with limitation of non localhost query
and sstill allows execute query
```bash
curl 'https://somedomain.com?query=SELECT%20*%20F... | 2020-04-17T20:28:23 | 2020-04-17T20:48:37 | {} | NONE |
ClickHouse | ClickHouse | 615,496,965 | 9,595 | tavplubix | Похоже, сегфолт исправлен [здесь](https://github.com/ClickHouse/ClickHouse/commit/2d75b3d3c8789720acd4267e881ad70fbf240cd2#diff-9f4b16b65f37a3b4498f7af66270b27cR139), но [сломалось](https://github.com/ClickHouse/ClickHouse/commit/918e145ea57a7ee03083d5b9a922911d98e6fe15) пробрасывание настроек запроса из `SETTINGS` в т... | 2020-04-17T22:50:21 | 2020-04-17T22:51:08 | {} | MEMBER |
ClickHouse | ClickHouse | 615,867,961 | 10,126 | alexey-milovidov | @blinkov is actually a robot script and it has assigned this task to a few random people by mistake. | 2020-04-18T13:09:55 | 2020-04-18T13:13:02 | {} | MEMBER |
ClickHouse | ClickHouse | 615,087,004 | 322 | ltylty | can clickhouse implement this sql? rollup subgroup.
```sql
SELECT
warehouse, product, SUM(quantity)
FROM
inventory
GROUP BY warehouse, ROLLUP (product);
``` | 2020-04-17T07:20:54 | 2020-04-17T07:20:54 | {
"+1": 4
} | NONE |
ClickHouse | ClickHouse | 615,136,029 | 346 | crakjie | is join_use_nulls affect a function like `emptyArrayToSingle` ?
I'm using this to make left `arrayJoin()` that could be nice to have both `LEFT ARRAY JOIN` and `arrayJoin` linked by the same setting. | 2020-04-17T09:10:19 | 2020-04-17T09:10:38 | {} | NONE |
ClickHouse | ClickHouse | 615,246,158 | 10,332 | den-crane | You can do it with a 2 seconds downtime, using Materialized View. | 2020-04-17T13:32:28 | 2020-04-17T13:32:28 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 615,270,114 | 9,884 | filimonov | Why 20.3 conflicts? | 2020-04-17T14:17:43 | 2020-04-17T14:17:43 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 615,370,132 | 10,336 | TimonKK | I have empty password tag
```xml
<password></password>
```
And if I delete this tag than I got message like your. But If I set empty password I can do query like this
```sql
SELECT * FROM system.functions
```
I think what it is works because previosly CllickHouse had empty password by default. And query to ht... | 2020-04-17T17:26:52 | 2020-04-17T17:35:56 | {} | NONE |
ClickHouse | ClickHouse | 615,485,601 | 5,618 | jangorecki | @blinkov are those informations from this issue available somewhere in the documentation? | 2020-04-17T22:08:22 | 2020-04-17T22:08:22 | {} | NONE |
ClickHouse | ClickHouse | 615,527,832 | 10,200 | AlexanderJLiu | @KochetovNicolai Ok, I'll update it later, and pay attention to whether this problem will appear again. Thanks~ | 2020-04-18T01:04:45 | 2020-04-18T01:04:45 | {} | NONE |
ClickHouse | ClickHouse | 615,661,006 | 10,341 | azat | >prevent merging of streams for queries with ORDER BY
Indeed, but there is [`if (select.orderBy()) return false`](https://github.com/ClickHouse/ClickHouse/pull/10341/files#diff-837268132765e5df53e31749c9e68dcdR396) and [test](https://github.com/ClickHouse/ClickHouse/pull/10341/files#diff-b994f578b3921603a36e48ab5d15... | 2020-04-18T07:18:49 | 2020-04-18T07:19:27 | {} | MEMBER |
ClickHouse | ClickHouse | 615,852,232 | 10,263 | alexey-milovidov | test_inserts_batching did not fix. | 2020-04-18T11:57:39 | 2020-04-18T11:57:39 | {} | MEMBER |
ClickHouse | ClickHouse | 615,857,553 | 10,320 | alexey-milovidov | I just realize that we are not using `pbuilder` for release packages.
Here is the log:
https://clickhouse-builds.s3.yandex.net/0/a551e1f33beeaaa9850d88baefbbfeffae8e3181/build_log_655588832_1587179319.txt
(you can find this log by clicking on any check mark for any commit and opening "ClickHouse build check")
... | 2020-04-18T12:20:49 | 2020-04-18T12:20:49 | {} | MEMBER |
ClickHouse | ClickHouse | 614,892,885 | 3,266 | alexey-milovidov | We have a task for students for implementation of various statistical tests of choice.
This issue is perfectly aligned with their work.
PS. As first implementation, we have https://github.com/ClickHouse/ClickHouse/pull/9139 - but it's completely unrelated. | 2020-04-16T21:00:09 | 2020-04-16T21:00:09 | {} | MEMBER |
ClickHouse | ClickHouse | 615,212,428 | 10,158 | wutchzone | I am unable to reproduce it on my local machine. This happens only in production where is a much bigger traffic. But when we were using `qunatileState` instead of `quantileTDigestState` it did not happen. | 2020-04-17T12:17:37 | 2020-04-17T12:17:37 | {} | NONE |
ClickHouse | ClickHouse | 615,234,244 | 10,251 | akuzm | LGTM | 2020-04-17T13:07:58 | 2020-04-17T13:07:58 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 615,245,499 | 10,327 | filimonov | > if(condition(heavy_function_call), heavy_function_call_value, anotherValue)
> in this case heavy_function_call is called twice, consuming twice as cpu needed
Why do you think so?
```
SELECT count(if(sleepEachRow(0.1), sleepEachRow(0.1), 0))
FROM numbers(100)
┌─count(if(sleepEachRow(0.1), sleepEachRow(0.1)... | 2020-04-17T13:31:03 | 2020-04-17T13:31:03 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 615,245,770 | 10,321 | nickevin | I encountered a similar problem, are there configurable parameters that can be provided when using the hdfs engine? | 2020-04-17T13:31:36 | 2020-04-17T13:31:36 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 615,251,451 | 10,310 | filimonov | Why do you think it's a bug?
Current behavior is compatible with Decimal:
```sql
select toUInt64(toDecimal64(1586275212.123,3));
```
with usual DateTime (returns seconds)
and makes roundtrip possible:
```
select toDateTime64(toUInt64(1586275212),3);
```
May be we should add support for reinterpretAsInt... | 2020-04-17T13:42:48 | 2020-04-17T13:47:05 | {
"+1": 2
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 615,278,097 | 9,884 | akuzm | > Why 20.3 conflicts?
Don't remember now, probably something simple, but I broke the build and decided to revert it. If you make a PR to backport it, we'll merge. | 2020-04-17T14:32:18 | 2020-04-17T14:32:18 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 615,391,100 | 10,336 | TimonKK | Yes, I have error "Could not parse user" If I delete <password></password>
It is my config.xml https://pastebin.com/gF5P9BqT with two users. And I do query
https://somedomain.com/?query=SELECT%20*%20FROM%20system.functions
For test user the query
https://somedomain.com/?query=SELECT%20*%20FROM%20system.function... | 2020-04-17T18:13:51 | 2020-04-17T18:13:51 | {} | NONE |
ClickHouse | ClickHouse | 615,501,897 | 10,336 | den-crane | I meant does Nginx shares a host with CH or not ? | 2020-04-17T23:12:08 | 2020-04-17T23:12:08 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 615,538,360 | 10,340 | akuzm | > Ok but what is the main difference?
One less Block::mutateColumns() call -- you can see it in the perf test flamegraphs for, say, `codec_seq_UInt64_Delta`. I still have to wrap my head around this myself :)
A call to Block::getColumns() that I added can also be removed, so I'll continue.
The codecs_int tests s... | 2020-04-18T02:02:40 | 2020-04-18T02:02:40 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 615,778,040 | 10,325 | Avogar | I can't fix one build error because class MsgPackVisitor must have methods with such names. (Because basic class msgpack::visitor has such names). | 2020-04-18T08:45:54 | 2020-04-18T08:45:54 | {} | MEMBER |
ClickHouse | ClickHouse | 615,866,220 | 10,161 | alexey-milovidov | > Мы их потом оттуда выпилим и оставим только ссылку на новый раздел.
Пожалуйста. | 2020-04-18T13:01:57 | 2020-04-18T13:01:57 | {} | MEMBER |
ClickHouse | ClickHouse | 614,894,198 | 3,575 | alexey-milovidov | It's easy to add new settings
`max_concurrent_insert_queries`
`max_concurrent_select_queries`
Up for grabs. | 2020-04-16T21:02:36 | 2020-04-16T21:02:36 | {} | MEMBER |
ClickHouse | ClickHouse | 614,896,082 | 3,685 | alexey-milovidov | Materialized views with scheduled refresh.
It's a good feature to implement. And it is especially useful when SELECT is performed from external data source.
We can allow to store it with arbitrary table engine in ClickHouse (with exception to Replicated tables) and refresh with atomic tables exchange. | 2020-04-16T21:06:26 | 2020-04-16T21:06:26 | {} | MEMBER |
ClickHouse | ClickHouse | 614,896,553 | 1,967 | alexey-milovidov | Fixed in https://github.com/ClickHouse/ClickHouse/pull/9142 | 2020-04-16T21:07:20 | 2020-04-16T21:07:20 | {} | MEMBER |
ClickHouse | ClickHouse | 615,847,359 | 10,341 | alexey-milovidov | What if there are other operations that require merging, e.g.
LIMIT - calculates preliminary limit on shards, apply final LIMIT on initiator;
DISTINCT - the same...
? | 2020-04-18T11:33:15 | 2020-04-18T11:33:15 | {} | MEMBER |
ClickHouse | ClickHouse | 614,911,734 | 3,889 | alexey-milovidov | Need to add separate method for IColumn. | 2020-04-16T21:41:22 | 2020-04-16T21:41:22 | {} | MEMBER |
ClickHouse | ClickHouse | 615,126,358 | 9,624 | alesapin | No new failures https://github.com/ClickHouse/ClickHouse/pull/9624#issuecomment-614173940. | 2020-04-17T08:49:33 | 2020-04-17T08:49:33 | {} | MEMBER |
ClickHouse | ClickHouse | 615,310,801 | 5,183 | filimonov | But it looks like you tested some other scenario (it was already working properly before).
(Pay attention on `Code: 1000, e.code() = 104`)
```
2020.03.24 00:00:00.067459 [ 176 ] {} <Error> ServerErrorHandler: Poco::Exception. Code: 1000, e.code() = 104, e.displayText() = Connection reset by peer
0. 0x100b49eb P... | 2020-04-17T15:30:16 | 2020-04-17T15:31:46 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 615,354,620 | 8,045 | XavierQuiniou | Thanks.
Solution is elegant :CAST (A, 'Nullable (String)') AS A
I will never have made the link between nullable () and my problem, although this instruction described in the documentation.
https://clickhouse.tech/docs/en/sql_reference/data_types/nullable/
| 2020-04-17T16:54:10 | 2020-04-17T16:54:10 | {} | NONE |
ClickHouse | ClickHouse | 615,668,257 | 10,342 | azat | >Let's look why tests failed...
Dumb error :)
```c
- else if (!injective_function_names.count(function->name))
+ else if (function_factory.get(function->name, context)->isInjective(Block{}))
```
Fixed | 2020-04-18T07:24:08 | 2020-04-18T07:24:08 | {} | MEMBER |
ClickHouse | ClickHouse | 614,901,523 | 3,889 | alexey-milovidov | A way for implementation:
Convert arguments to the least common type and call `IColumn::compareAt`. | 2020-04-16T21:18:01 | 2020-04-16T21:18:01 | {} | MEMBER |
ClickHouse | ClickHouse | 615,057,799 | 10,293 | filimonov | BTW: maybe we can just use jemalloc stats? Or you prefer not to rely on particular allocator? | 2020-04-17T05:55:49 | 2020-04-17T05:55:49 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 615,084,438 | 10,305 | vmihailenco | >and as accurate as random from the range min(value) ... max(value).
For your example `10,50,90` it provides better result than random. And somehow world is using `x+y/2` despite all its flaws and it is definitely better than random. I would use better tools like tdigests or ddsketch - but first is broken and 2nd is... | 2020-04-17T07:14:09 | 2020-04-17T07:14:09 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 615,181,105 | 10,324 | alesapin | Unfortunately, you have to update all servers of your cluster to 20.3 to execute `ALTER` queries because this functionality significantly changed under the hood. As I can see from your log you have a problem with 19.15 when alter started from 20.3 server. | 2020-04-17T10:53:49 | 2020-04-17T10:53:49 | {} | MEMBER |
ClickHouse | ClickHouse | 615,240,253 | 10,332 | den-crane | The official way is to `create a new table`, `insert into new select from old`, `rename old to old_old , new to old` | 2020-04-17T13:20:26 | 2020-04-17T13:20:26 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 615,490,067 | 10,336 | TimonKK | Thanks for your help. I have only one nginx. Directories /etc/clickhouse-server/conf.d/ and /etc/clickhouse-server/users.d are empty.
I looked preprocessed /clickhouse/preprocessed_configs/users.xml - it is same like /etc/clickhouse-server/users.xml | 2020-04-17T22:23:09 | 2020-04-17T22:23:09 | {} | NONE |
ClickHouse | ClickHouse | 615,866,499 | 10,058 | alexey-milovidov | @blinkov is a robot and he was assigned this task by mistake. | 2020-04-18T13:03:10 | 2020-04-18T13:03:10 | {
"laugh": 1
} | MEMBER |
ClickHouse | ClickHouse | 631,711,873 | 11,022 | alexey-milovidov | @azat Let's test all three variants.
You can send three draft pull requests, we will wait for performance tests and look.
If memory usage will be lower and there will be no regressions in performance - I strongly appreciate these changes. | 2020-05-20T20:35:44 | 2020-05-20T20:35:44 | {} | MEMBER |
ClickHouse | ClickHouse | 631,909,722 | 11,088 | alexey-milovidov | Wow, that's great. By the way, can you please briefly explain, what is that, where to use it? | 2020-05-21T06:28:58 | 2020-05-21T06:28:58 | {} | MEMBER |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.