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 | 467,452,151 | 4,510 | alexey-milovidov | When you calculate a single value of aggregate function, the merging phase is not parallelized. (It is parallelized for different aggregate function states.)
E.g. if you calculate multiple small-medium uniqExacts in a single query (with GROUP BY), it will be fast, but when you calculate just a single uniqExact, it i... | 2019-02-26T14:10:27 | 2019-02-26T14:10:43 | {} | MEMBER |
ClickHouse | ClickHouse | 468,138,797 | 751 | abraithwaite | I think the docs should be updated to be explicit about requiring the all the key fields in the `GROUP BY` clause. I ran into this when I was first using it too. | 2019-02-28T05:03:37 | 2019-02-28T05:03:37 | {
"+1": 2
} | NONE |
ClickHouse | ClickHouse | 468,236,328 | 4,529 | simPod | And yes, I have found in error log, that I have incorrect column types. Defined. Then I saw
> 2019.02.28 11:17:20.473187 [ 777 ] {578bc0ef-be9f-4cf8-90e2-0fb1e4b95081} <Error> HTTPHandler: Code: 27, e.displayText() = DB::Exception: Cannot parse input: expected { before: \0: (at row 2)
But adding `kafka_row_delimi... | 2019-02-28T11:19:41 | 2019-02-28T11:19:41 | {
"+1": 1
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 468,438,964 | 4,522 | filimonov | I think that described behaviour is acceptable for alters which do really make some significant change in the table structure.
But with Enums... it's quite stupid, it should be really cheap operation, as no data should be changed, only type description. | 2019-02-28T21:08:22 | 2019-02-28T21:08:22 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 468,618,255 | 4,543 | alesapin | doesn't help | 2019-03-01T10:25:43 | 2019-03-01T10:25:43 | {} | MEMBER |
ClickHouse | ClickHouse | 468,915,346 | 3,669 | Enmk | > @Enmk : Out of curiosity, why is IPv6 represented using `FixedString(16)` instead of `UInt128`?
Well, the major reason is compatibility. Existing IPv6-functions work with `FixedString(16)` and hence work with IPv6 domain too, also that allows for seamless comparing of old `FixedString(16)`-IPv6 column values with ... | 2019-03-02T12:16:18 | 2019-03-02T12:16:18 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 468,940,466 | 4,390 | nezed | @alexey-milovidov There is still an issue with CH own meta-files.
Server can't restart and always fails if table with column ``` `index` ``` was created in 19.3.5.
Also you can't upgrade to 19.3 if there is `index` or ``` `index` ``` columns.
```
2019.03.02 19:30:04.445373 [ 1 ] {} <Debug> Application: Destroye... | 2019-03-02T17:20:50 | 2019-03-02T17:20:50 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 469,276,348 | 2,174 | simPod | Encountered the same issue. I intended to chain it
```
kafka -> mv stream A -> table A -> mv stream B -> table B
```
but had to create `MV stream B` from `MV stream A` instrad of from `table A`
```
kafka -> mv stream A -> table A
-> mv stream B -> table B
```
Glad it works tho | 2019-03-04T14:41:51 | 2019-03-04T14:41:51 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 467,445,616 | 4,510 | den-crane | uniqExact is always the worst.
Connected to ClickHouse server version 19.3.4
case1. create table u_perf(a Int64, v String) Engine=MergeTree order by a;
case2. create table u_perf(a Int64, v String) Engine=MergeTree order by v;
case3. create table u_perf(a Int64, v LowCardinality(String)) Engine=MergeTree order ... | 2019-02-26T13:51:51 | 2019-02-26T13:51:51 | {
"+1": 6
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 468,228,513 | 4,532 | filimonov | Recently (v19.1.0) `boundingRatio` function was added. It's quite simple - calculates the rate of growth by 2 border points (so it suits only for some limited number of cases).
Usage:
```
select sensorid, boundingRatio(time, value) from table group by sensorid;
```
See some details/comments in https://github.... | 2019-02-28T10:53:07 | 2019-02-28T10:53:07 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 468,320,042 | 4,156 | simPod | Happening to me as well on 19.3.5
```2019.02.28 15:19:45.425559 [ 1567 ] {} <Error> BaseDaemon: (from thread 1178) Terminate called without an active exception
2019.02.28 15:19:45.425849 [ 1567 ] {} <Error> BaseDaemon: ########################################
2019.02.28 15:19:45.425974 [ 1567 ] {} <Error> BaseDaem... | 2019-02-28T15:40:10 | 2019-02-28T15:40:10 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 468,729,976 | 4,551 | temoon | Thanks for reply, Alexey.
I use of about 70 million rows sized external dictionary which has frequently updates (~5k entries per second), and full dump tooks few minutes :((
Even is this form of functionality is very helpful for me! | 2019-03-01T16:45:18 | 2019-03-01T16:45:18 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 469,007,000 | 4,442 | vladimir-golovchenko | Adding option `kafka_row_delimiter = '\n'` to Kafka-engine definition fixs this problem.
| 2019-03-03T10:00:03 | 2019-03-03T10:00:03 | {
"+1": 3
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 467,916,073 | 4,423 | dr1nf3rn0 | @0i
Rolling back to 18.6.1 version fixed this problem for us. Stop all clickhouse services and apt-get install specific version of **clickhouse-server clickhouse-client clickhouse-common-static** packages | 2019-02-27T15:51:18 | 2019-02-27T15:51:18 | {} | NONE |
ClickHouse | ClickHouse | 465,839,351 | 4,212 | CurtizJ | TODO
- [x] save min-max TTL for specific columns and table | 2019-02-21T02:41:36 | 2019-02-28T11:37:26 | {} | MEMBER |
ClickHouse | ClickHouse | 468,532,954 | 4,533 | den-crane | "properties": {
"latitude": "28.920899",
"page_id": "010400000000"
},
oh, CH does not support such nested, it's not an array []. | 2019-03-01T03:52:35 | 2019-03-01T03:52:35 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 468,577,786 | 4,538 | javisantana | ok, got it, thanks
About the partitioning I was thinking on creating partitions by geographical area but it'd more something like
```
ZCurve(toInt32(x/K), toInt32(y/K))
```
| 2019-03-01T08:00:03 | 2019-03-01T08:00:03 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 468,752,169 | 3,669 | The-Alchemist | @Enmk : Out of curiosity, why is IPv6 represented using `FixedString(16)` instead of `UInt128`?
@KochetovNicolai : Would a PR changing it to `UInt128` be considered? | 2019-03-01T17:53:42 | 2019-03-01T17:56:19 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 468,929,739 | 4,560 | nezed | @filimonov can you help me to understand?)
1. Is `SELECT datetime FROMAT Pretty` are the same with `SELECT toString(datetime) FROMAT Pretty `?
2. Are TZ correction applied for `INSERT INTO tz_sensetive VALUES ( '2019-01-01', '2019-01-01 00:00:00' )` done on server with non-UTC TZ?
Will it be converted into UTC w... | 2019-03-02T15:21:13 | 2019-03-02T15:21:13 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 468,940,894 | 4,560 | nezed | Finally I have a "deep dive" in https://clickhouse.yandex/docs/en/data_types/datetime/
And now I understand:
1. Yes, at time of query executions
2. Yes, at time of insert. Yes, it is stored in UTC
@filimonov Thanks a lot for you attention! | 2019-03-02T17:25:32 | 2019-03-02T17:25:32 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 469,079,701 | 4,442 | josepowera | We are trying to break problem down... This is what we know at the moment:
How we are using kafka-clickhouse combination in 18.16.1
The message sent to kafka is:
-CSV message delimited with "," - around 20 fields in total
-CSV fields are plain - no quotes for string fields : NOT "stringdata1","stringdata2",8000... | 2019-03-03T23:59:26 | 2019-03-03T23:59:26 | {} | NONE |
ClickHouse | ClickHouse | 467,938,289 | 4,511 | alesapin | Fix https://github.com/yandex/ClickHouse/pull/4523/files. | 2019-02-27T16:45:56 | 2019-02-27T16:45:56 | {} | MEMBER |
ClickHouse | ClickHouse | 468,017,655 | 4,512 | igor-sh8 | Great. Thanks.
We will try it. | 2019-02-27T20:28:20 | 2019-02-27T20:28:20 | {} | NONE |
ClickHouse | ClickHouse | 468,253,358 | 4,534 | hohaftf | Moved [here](https://github.com/alex-krash/clickhouse-jdbc-bridge/issues/4) | 2019-02-28T12:23:10 | 2019-02-28T12:23:10 | {} | NONE |
ClickHouse | ClickHouse | 468,351,333 | 4,522 | ztlpn | Is the table replicated or not?
For non-replicated tables the locking proceeds as follows:
1. `data_lock` is locked exclusively. BTW the lock names are super confusing, but this `data_lock` is locked in the shared mode by inserts and merges, so if you lock it exclusively, you can be sure that no new parts appear in... | 2019-02-28T16:59:31 | 2019-02-28T17:01:19 | {
"confused": 2
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 468,506,262 | 4,536 | alexey-milovidov | Ok, but I don't understand what is the source of performance improvement. In this PR we still store keys in lookup table and the FixedHashSet/Map doesn't look much different to the HashSet/HashMap. | 2019-03-01T01:27:43 | 2019-03-01T01:27:43 | {} | MEMBER |
ClickHouse | ClickHouse | 468,702,909 | 4,350 | arctica | @ztlpn Thanks for that information. That's a situation that albeit sub-optimal, I can make it work in our use-case. | 2019-03-01T15:28:17 | 2019-03-01T15:28:17 | {} | NONE |
ClickHouse | ClickHouse | 468,846,563 | 4,542 | filimonov | I think that that it would be also cool to have general "value alignment" function.
Something like `arrayAlignByInterpolation( array of tuples (x, y), array of needed x )` => return interpolated values for needed x positions.
In that case you can just do something like:
```
with (select groupUniqArray(timestam... | 2019-03-01T23:26:12 | 2019-03-01T23:41:01 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 467,453,777 | 4,513 | 4ertus2 | #3783 | 2019-02-26T14:14:23 | 2019-02-26T14:14:23 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 467,688,575 | 3,822 | lyl2008dsg | I had a similar problem.
ClickHouse client version 19.1.6.
Connected to ClickHouse server version 19.1.6 revision 54413.
```
2019.02.27 09:37:44.812881 [ 21 ] {} <Warning> default.app_log_tradenum (Data): Removing temporary directory /data/lib/clickhouse/data/default/app_log_tradenum/tmp_merge_20190226_8_... | 2019-02-27T01:41:17 | 2019-02-27T01:41:17 | {} | NONE |
ClickHouse | ClickHouse | 467,822,920 | 4,511 | alesapin | Similar problem with `COMMENT`:
```
CREATE TABLE test.check_query_comment_column
(
first_column UInt8 DEFAULT 1 COMMENT 'comment 1',
second_column UInt8 MATERIALIZED first_column COMMENT 'comment 2',
third_column UInt8 ALIAS second_column COMMENT 'comment 3',
fourth_column UInt8 COMMENT 'comme... | 2019-02-27T11:13:42 | 2019-02-27T11:14:00 | {} | MEMBER |
ClickHouse | ClickHouse | 468,012,133 | 4,525 | alexey-milovidov | SSL can also be disabled.
As it is used through Poco library, to disable it, you should specify `-D ENABLE_POCO_NETSSL=0`, if I remember correctly. | 2019-02-27T20:11:22 | 2019-02-27T20:11:22 | {} | MEMBER |
ClickHouse | ClickHouse | 468,129,647 | 4,467 | sogngenWang | SYSTEM DROP MARK CACHE
It can really solve the problem . My ClickHouse server version 19.3.3 revision 54415 | 2019-02-28T04:07:14 | 2019-02-28T04:07:14 | {
"+1": 7
} | NONE |
ClickHouse | ClickHouse | 468,272,771 | 4,538 | javisantana | is it like:
```
create table ( x Float32, y Float32, ....) ENGINE=MergeTree partition by (ZCurve(x, y)) order by ZCurve(x, y)
```
how would you query a range (for example, a bounding box in 2D)
| 2019-02-28T13:30:26 | 2019-02-28T13:30:26 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 468,750,472 | 4,522 | nvartolomei | @ztlpn it does make sense now. This happened for replicated table, didn't pay attention that there are different alter implementations.
@filimonov I fully agree with you, let's create a feature request for it if one does not exist yet.
later edit: it is already metadata only change, however table metadata is modi... | 2019-03-01T17:48:06 | 2019-03-01T17:59:17 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 468,932,492 | 4,544 | den-crane | Does not matter.
MV does not read real tableB (except at a populate stage).
MV gets inserted buffer (from insert query) and evaluates select over this buffer. | 2019-03-02T15:54:24 | 2019-03-02T15:54:40 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 468,959,852 | 4,442 | pdeva | this issue has rendered clickhouse unusable for anyone using kafka. cannot upgrade to 19.x till this is fixed | 2019-03-02T21:02:19 | 2019-03-02T21:02:19 | {} | NONE |
ClickHouse | ClickHouse | 467,535,727 | 3,997 | mcarbonneaux | i think the problem is because you use your cmakelist... and hardcode librdkafka flag...
in https://github.com/yandex/ClickHouse/blob/master/contrib/librdkafka-cmake/config.h you commented sasl flag...
```
// #define ENABLE_SASL 1
// #define WITH_SASL_SCRAM 1
```
the good version working with sasl/scram wit... | 2019-02-26T17:33:56 | 2019-02-26T18:00:57 | {} | NONE |
ClickHouse | ClickHouse | 468,042,255 | 4,522 | nvartolomei | So far couldn't reproduce, maybe the cause lies somewhere else. | 2019-02-27T21:42:11 | 2019-02-27T21:42:11 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 468,287,593 | 4,533 | den-crane | 18.14.9 Nested data structures can be read as nested objects in JSONEachRow format. Added the input_format_import_nested_json setting.
```
echo "create table N(a String, KV Nested(K String, V String)) engine=MergeTree order by a;"|clickhouse-client
echo '{"a":"xxx", "KV" : { "K" : ["a", "b"], "V":["1","2"]} }'... | 2019-02-28T14:15:18 | 2019-02-28T14:19:15 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 468,506,723 | 4,536 | alexey-milovidov | It's Ok as a first step, if the motivation will be mentioned in comments. | 2019-03-01T01:30:04 | 2019-03-01T01:30:04 | {} | MEMBER |
ClickHouse | ClickHouse | 468,653,150 | 4,538 | alexey-milovidov | > About the partitioning I was thinking on creating partitions by geographical area but it'd more something like ZCurve(toInt32(x/K), toInt32(y/K))
Partitions are independent. If you partition by `ZCurve(x, y)` it is the same as if you partition by tuple `(x, y)`. Moreover, partitions are not a replacement of primar... | 2019-03-01T12:44:31 | 2019-03-01T12:44:59 | {} | MEMBER |
ClickHouse | ClickHouse | 468,659,094 | 4,538 | javisantana | > If you partition by `ZCurve(x, y)` it is the same as if you partition by tuple `(x, y)`
I don't exactly know how CH sorts the data when the primary key is a tuple (I read it first order by x and then by y for each x). In the geospatial work the access pattern is usually different so if the curve function is able t... | 2019-03-01T13:07:56 | 2019-03-01T13:07:56 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 467,867,084 | 2,416 | simPod | @alexey-milovidov is it true that
> if you have identical subqueries, ClickHouse only actually runs it once
are there any preconditions for it to work?
I have created this query
```sql
SELECT *
FROM (
SELECT (intDiv(toUInt32(`time`), 300) * 300) AS `time`,
avg(volume) AS volume
FROM (
... | 2019-02-27T13:45:56 | 2019-02-27T13:45:56 | {
"+1": 1
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 465,646,350 | 3,852 | bgranvea | I have a first working version:
```
create table data_test (ts DATETIME,id VARCHAR, v0 SimpleAggregateFunction(anyLast,Nullable(Float64)), v1 SimpleAggregateFunction(sum,Nullable(Float64)), v2 SimpleAggregateFunction(min,Nullable(Float64)), v3 SimpleAggregateFunction(max,Nullable(Float64)), p0 SimpleAggregateFuncti... | 2019-02-20T16:16:25 | 2019-02-27T14:11:13 | {
"+1": 7
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 468,527,959 | 4,533 | kindred77 | no error in log:
2019.03.01 11:19:49.121865 [ 130 ] {} <Trace> StorageKafka (event_topic): Committing message with offset 752877126
2019.03.01 11:19:49.128301 [ 21 ] {} <Trace> StorageKafka (event_topic): Committing message with offset 724775060
2019.03.01 11:19:49.151314 [ 92 ] {} <Trace> StorageKafka (event_topi... | 2019-03-01T03:21:14 | 2019-03-01T03:21:14 | {} | NONE |
ClickHouse | ClickHouse | 468,660,101 | 4,538 | alexey-milovidov | > I don't exactly know how CH sorts the data when the primary key is a tuple (I read it first order by x and then by y for each x). In the geospatial work the access pattern is usually different so if the curve function is able to keep close things (in the real world) closer in the disk is much better.
> For example... | 2019-03-01T13:11:54 | 2019-03-01T13:11:54 | {} | MEMBER |
ClickHouse | ClickHouse | 469,348,873 | 4,486 | reflection | Any update on this? We'd love to test out the new data skipping indices feature, but this is a blocker since all of our tables utilize the LowCardinality(Nullable(String)) type extensively. Thanks! | 2019-03-04T17:50:18 | 2019-03-04T17:50:18 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 469,377,515 | 4,426 | Felixoid | If this PR will be merged before #4584, then the second PR should adjust tests and fix conflicts. | 2019-03-04T18:57:16 | 2019-03-04T18:57:16 | {} | MEMBER |
ClickHouse | ClickHouse | 469,377,517 | 4,486 | KochetovNicolai | I'll fix it tomorrow. | 2019-03-04T18:57:16 | 2019-03-04T18:57:16 | {} | MEMBER |
ClickHouse | ClickHouse | 467,671,890 | 4,507 | kindred77 | it works, thanks | 2019-02-27T00:27:12 | 2019-02-27T00:27:12 | {} | NONE |
ClickHouse | ClickHouse | 468,278,718 | 4,538 | achulkov2 | > is it like:
>
> ```
> create table ( x Float32, y Float32, ....) ENGINE=MergeTree partition by (ZCurve(x, y)) order by ZCurve(x, y)
> ```
> how would you query a range (for example, a bounding box in 2D)
I am not sure that partitioning by ZCurve(x, y) would be a very good idea because the function is inject... | 2019-02-28T13:48:55 | 2019-02-28T13:48:55 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 469,267,756 | 3,520 | vglocus | I run a 20 node replicated cluster (10 shards * 2 replicas). Cluster is configured with internal_replication and using user "distributed" for inter-cluster communication.
I have encountered this same issue both using distributed queries (ON CLUSTER) and local (single Replica Pair).
Either `ALTER TABLE "foo"."b... | 2019-03-04T14:16:35 | 2019-03-04T14:16:35 | {} | NONE |
ClickHouse | ClickHouse | 469,335,808 | 4,580 | SaltTan | This behaviour was introduced in 19.1.7 and 19.3.3 | 2019-03-04T17:14:13 | 2019-03-04T17:14:13 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 469,377,192 | 4,584 | Felixoid | If this PR will be merged before #4426, then the second PR should adjust tests and fix conflicts. | 2019-03-04T18:56:23 | 2019-03-04T18:56:23 | {} | MEMBER |
ClickHouse | ClickHouse | 467,941,950 | 2,369 | vitlibar | Protobuf IO format has been implemented in master. All types of fields (repeated/optional/required) and nested messages are supported now. See [documentation](https://github.com/yandex/ClickHouse/blob/master/docs/en/interfaces/formats.md) for details. | 2019-02-27T16:55:17 | 2019-02-27T16:56:19 | {
"rocket": 6
} | MEMBER |
ClickHouse | ClickHouse | 468,138,124 | 3,866 | abraithwaite | I'm actually curious about the differences too. Did you ever find out? | 2019-02-28T05:00:04 | 2019-02-28T05:00:04 | {} | NONE |
ClickHouse | ClickHouse | 468,167,124 | 2,833 | 452806278 | @kshvakov I have the same problem, we are connecting to clickhouse through python library. | 2019-02-28T07:27:54 | 2019-02-28T07:27:54 | {
"+1": 2
} | NONE |
ClickHouse | ClickHouse | 468,293,516 | 4,528 | den-crane | >So it turned out the discrepancy of the same data in the two Materialized Views.
it's expected. No atomicity with multiple mviews (even with one MV insert can be succeeded with MV and failed with table)
>Is it possible to reload for example only one day in Materialized View ?
No. Only manual fix with / alter... | 2019-02-28T14:31:36 | 2019-02-28T14:40:28 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 468,470,873 | 4,530 | alexey-milovidov | Ok. | 2019-02-28T22:50:05 | 2019-02-28T22:50:05 | {} | MEMBER |
ClickHouse | ClickHouse | 468,579,979 | 2,178 | NanoBjorn | I had an issue with docker — you have yo use `--network=host` to make it work and it is not save. This happens because you place actual server ip in `config.yml` file but by default docker iptables config container could not connect to the host itself so replicating does not work properly.
It seems that there is no ... | 2019-03-01T08:09:05 | 2019-03-01T08:09:05 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 469,313,751 | 3,852 | bgranvea | I have difficulties when data types are recursively nested like for example `SimpleAggregateFunction(Nullable(Float64))`. You could also imagine define a column of type `SimpleAggregateFunction(LowCardinality(Nullable(Float64)))` (not tested yet).
There are multiple locations in the code where the data type is direc... | 2019-03-04T16:18:18 | 2019-03-04T16:19:42 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 469,335,270 | 4,579 | alesapin | Tried to build with `gcc-7` + `asan`+ `compiled_expression_cache_size=5` and reproduce cache overflow. All works fine, expressions are removed, no asserts or segfaults in destructor.
Additionaly we found, that llvm symbol resolution can get into deadlock. | 2019-03-04T17:12:49 | 2019-03-04T17:17:45 | {} | MEMBER |
ClickHouse | ClickHouse | 468,629,578 | 467 | PangKuo | What if many replicas in one shard? Can I change the ENGINE of materialized view to ReplicatedMergeTree?
| 2019-03-01T11:05:28 | 2019-03-01T11:05:28 | {} | NONE |
ClickHouse | ClickHouse | 468,716,794 | 4,551 | alexey-milovidov | The code (not yours) is terrible and should be rewritten.
`std::string getUpdateFieldAndDate();`
There is no comment about this method.
The name of the method is misleading.
It is not qualified as const despite its name and it changes update time by surprise.
```
char buffer[80];
struct tm * timeinfo;
timei... | 2019-03-01T16:08:22 | 2019-03-01T16:10:09 | {} | MEMBER |
ClickHouse | ClickHouse | 468,726,518 | 4,551 | alexey-milovidov | The logic in code is also flawed. We cannot use client side time to compare it on server side. | 2019-03-01T16:35:21 | 2019-03-01T16:35:21 | {} | MEMBER |
ClickHouse | ClickHouse | 467,897,752 | 4,508 | 4ertus2 | In this place we just try to call this instruction on any registers with any data. Just to detect that CPU supports AVX. The problem is vaddpd expects 3 registers, but __asm__ construction in this line means something else (probably 2 source registers and one of them is resulting). If you can change this line to someth... | 2019-02-27T15:07:50 | 2019-02-27T15:09:11 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 468,134,627 | 4,467 | PhantomPhreak | > We need to check one hypothesis. Does query `SYSTEM DROP MARK CACHE` solves the problem?
I can't reproduce this issue now and can't check if it works or not.
Last time i solved the problem by deleting broken part and server restarting, ClickHouse server found missing part and fetched it from replica. | 2019-02-28T04:38:18 | 2019-02-28T04:40:34 | {} | NONE |
ClickHouse | ClickHouse | 468,276,321 | 4,537 | den-crane | To download data try to use non-human formats TSV / CSV / JSONEachRow
https://clickhouse.yandex/docs/en/interfaces/formats/
select * from numbers(10) format TSV
10000 -- controlled by output_format_pretty_max_rows
```
set output_format_pretty_max_rows = 3;
select * from numbers(10)
┌─number─┐
│ 0 │
│ ... | 2019-02-28T13:41:29 | 2019-02-28T13:41:29 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 468,795,817 | 4,409 | alexey-milovidov | I have completed the implementation, you can read code along with comments. | 2019-03-01T20:09:55 | 2019-03-01T20:09:55 | {} | MEMBER |
ClickHouse | ClickHouse | 468,894,820 | 4,542 | filimonov | BTW: nice and short presentation of how similar feature works in wavefront https://youtu.be/9LnDszVrJs4 | 2019-03-02T07:14:52 | 2019-03-02T07:14:52 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 468,922,355 | 4,559 | B1anker | @filimonov thanks | 2019-03-02T13:49:58 | 2019-03-02T13:49:58 | {} | NONE |
ClickHouse | ClickHouse | 468,923,976 | 4,449 | valyala | There is another option - do not use `O_DIRECT` and use periodic calls to `fsync() + posix_fadvise(POSIX_FADV_DONTNEED)` on the output file during big merge. This will instruct the OS to free pages occupied by output file data. I successfully use this technique in VictoriaMetrics. Contrary to `O_DIRECT`, it works witho... | 2019-03-02T14:11:04 | 2019-03-02T14:11:04 | {} | NONE |
ClickHouse | ClickHouse | 468,927,810 | 4,560 | filimonov | Actually it currently works exactly as expected.
You have timestamp, i.e. some position in time (which is independent of replication, timezones etc).
Timezones settings define a rules how to show that timestamp in a human readable form.
If two servers use different default timezones they will format same times... | 2019-03-02T14:58:31 | 2019-03-02T14:59:51 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 468,955,906 | 4,561 | filimonov | Maybe related to compile_expression. Try to disable it. | 2019-03-02T20:10:39 | 2019-03-02T20:10:39 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 469,080,083 | 4,442 | josepowera | @vladimir77 @pdeva @sobolevsv could you help to narrow it down - your kafka message type (csv/json/json.., single line/batche, using EndOfLine), DML definition,... | 2019-03-04T00:02:33 | 2019-03-04T00:03:14 | {} | NONE |
ClickHouse | ClickHouse | 469,376,072 | 4,486 | alexey-milovidov | It's trivial issue, easy to fix. | 2019-03-04T18:53:19 | 2019-03-04T18:53:19 | {} | MEMBER |
ClickHouse | ClickHouse | 467,573,120 | 4,493 | vitlibar | I've measured the performance:
```
SELECT * FROM datasets.hits_v1 FORMAT TSV 33.16 real 28.67 user 4.05 sys
SELECT * FROM datasets.hits_v1 FORMAT Protobuf 38.18 real 34.59 user 3.32 sys
SELECT * FROM datasets.hits_v1 FORMAT JSONEachRow 58.02 real 47.56 user 8.97 sys
``` | 2019-02-26T19:14:55 | 2019-02-26T19:14:55 | {} | MEMBER |
ClickHouse | ClickHouse | 467,891,380 | 4,508 | code-of-kpp | My asm-fu is not good enough unfortunately | 2019-02-27T14:52:16 | 2019-02-27T14:52:16 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 467,992,491 | 4,508 | patricklauer | ```
sed -i -e 's/vaddpd %%ymm0, %%ymm0/vaddpd %%ymm0, %%ymm0, %%ymm0/' libs/libdaemon/src/BaseDaemon.cpp
```
This makes the compiler happy enough to proceed, I have no idea about the semantics of it. | 2019-02-27T19:14:09 | 2019-02-27T19:14:09 | {} | NONE |
ClickHouse | ClickHouse | 468,012,686 | 4,525 | alexey-milovidov | > Should failure to build without ssl be considered as a bug to be investigated for unconditioned includes and parts of code?
Yes.
We can also add one more build variant to our CI - "minimal build" with the minimum possible number of components. | 2019-02-27T20:13:01 | 2019-02-27T20:13:01 | {} | MEMBER |
ClickHouse | ClickHouse | 468,157,272 | 4,529 | josepowera | Check if it is related to https://github.com/yandex/ClickHouse/issues/4442 | 2019-02-28T06:43:34 | 2019-02-28T06:44:04 | {} | NONE |
ClickHouse | ClickHouse | 468,470,499 | 4,541 | alexey-milovidov | Thank you! | 2019-02-28T22:48:41 | 2019-02-28T22:48:41 | {} | MEMBER |
ClickHouse | ClickHouse | 469,082,368 | 4,442 | pdeva | we use json. not sure about line endings | 2019-03-04T00:26:37 | 2019-03-04T00:26:37 | {} | NONE |
ClickHouse | ClickHouse | 469,321,720 | 4,579 | alesapin | Relevant issues:
https://github.com/yandex/ClickHouse/issues/4180
https://github.com/yandex/ClickHouse/issues/4561 | 2019-03-04T16:38:20 | 2019-03-04T16:38:40 | {} | MEMBER |
ClickHouse | ClickHouse | 469,349,585 | 4,581 | den-crane | The same issue reported about odbc bridge + PG
t.me/clickhouse_ru/90244
t.me/clickhouse_ru/90553 | 2019-03-04T17:52:20 | 2019-03-04T17:52:20 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 468,527,477 | 4,533 | kindred77 | It seams does not work.
I test it on 19.3.5.
select name,value,changed from system.settings where name='input_format_import_nested_json';
-----------------------------------name-------value------changed--------
input_format_import_nested_json 1 1
CREATE TABLE event_topic(
... | 2019-03-01T03:18:28 | 2019-03-01T03:18:28 | {} | NONE |
ClickHouse | ClickHouse | 468,709,607 | 4,552 | alexey-milovidov | Missing description for changelog. | 2019-03-01T15:47:26 | 2019-03-01T15:47:26 | {} | MEMBER |
ClickHouse | ClickHouse | 468,904,626 | 4,542 | LiuYangkuan | > BTW: nice and short presentation of how similar feature works in wavefront https://youtu.be/9LnDszVrJs4
This requirement from that we use ClickHouse to store TimeSeries whit tagk and tagv, provide OpenTSDB query API. Query API can close downsample, offer more original granularity for user to aggregate TimeSeries s... | 2019-03-02T09:39:47 | 2019-03-02T09:39:47 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 469,136,259 | 4,442 | vladimir-golovchenko | On CH 18.16.1 the next config works fine: `ENGINE = Kafka SETTINGS kafka_broker_list = '', kafka_topic_list = 'read-events', kafka_group_name = '', kafka_format = 'JSONEachRow'`.
It stops working on CH 19.3.5, got error 'Cannot parse input: expected { before: \0'.
Addding `kafka_row_delimiter = '\n'` fixs the pr... | 2019-03-04T06:39:44 | 2019-03-04T06:39:44 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 467,945,777 | 4,467 | alesapin | We need to check one hypothesis. Does query `SYSTEM DROP MARK CACHE` solves the problem? | 2019-02-27T17:05:02 | 2019-02-27T17:07:17 | {} | MEMBER |
ClickHouse | ClickHouse | 468,231,743 | 4,529 | simPod | I have not encountered such error yet. Only that "Polled batch of 0 messages" | 2019-02-28T11:03:35 | 2019-02-28T11:03:35 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 468,237,101 | 4,538 | javisantana | could you provide a basic example? | 2019-02-28T11:22:20 | 2019-02-28T11:22:20 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 468,241,359 | 4,534 | alex-krash | @hohaftf , please move this issue to https://github.com/alex-krash/clickhouse-jdbc-bridge
It is problem of jdbc-bridge implementation. | 2019-02-28T11:37:44 | 2019-02-28T11:37:44 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 468,264,717 | 4,538 | achulkov2 | > could you provide a basic example?
For example, if you have two columns X and Y with roughly the same number of SELECT queries with conditions on X and Y, then perhaps it is more efficient to order them by some special function f(X, Y), rather than using (X, Y) or (Y, X) in the key. In this case space-filling curv... | 2019-02-28T13:03:22 | 2019-02-28T13:03:36 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 468,535,334 | 4,533 | kindred77 | Is there another solution? | 2019-03-01T04:08:02 | 2019-03-01T04:08:02 | {} | NONE |
ClickHouse | ClickHouse | 468,733,130 | 4,551 | alexey-milovidov | Ok. I have rewritten it a little, but I doubt that it is covered by tests, so good luck. | 2019-03-01T16:54:39 | 2019-03-01T16:54:39 | {} | MEMBER |
ClickHouse | ClickHouse | 468,911,708 | 4,559 | filimonov | HTTP by default is stateless, to remeber states between different requests you need to use session.
> Similarly, you can use ClickHouse sessions in the HTTP protocol. To do this, you need to add the session_id GET parameter to the request. You can use any string as the session ID. By default, the session is terminat... | 2019-03-02T11:22:39 | 2019-03-02T11:22:39 | {
"+1": 2
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 468,932,815 | 4,544 | simPod | Thanks for explanation! | 2019-03-02T15:57:49 | 2019-03-02T15:57:49 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 469,133,090 | 4,566 | rayastar | i found solution:
sudo cat /home/admin/csv/all.csv | docker run -i --rm --link some-clickhouse-server:clickhouse-server yandex/clickhouse-client -m --host clickhouse-server --query="INSERT INTO default.allexamples FORMAT CSV"
It's perfect work! | 2019-03-04T06:23:20 | 2019-03-04T06:23:20 | {} | NONE |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.