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 | 570,837,228 | 6,811 | alexey-milovidov | > Is it true that sharding key can't be anything other than a constant value or a column of table?
No, you can use arbitrary expressions as a sharding key. Your example is perfectly valid.
> When data distibutes with a shading key, can we predict which shard the data is insert?
Yes. Data will be placed by the ... | 2020-01-05T02:06:16 | 2020-01-05T02:06:16 | {
"+1": 3
} | MEMBER |
ClickHouse | ClickHouse | 570,866,921 | 7,892 | filimonov | > you can pass UUID to the hash function
That is not true (at least it was't at the moment of writing my prev comment :))
> You may have heard that xxHash64 is the "fastest" hash function
I agree and I'm aware of that. I've just adjusted @den-crane sample a bit to make it more reliable. xxHash is the best fo... | 2020-01-05T05:34:08 | 2020-01-05T05:44:01 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 570,955,957 | 8,534 | alexey-milovidov | It should be included in changelog. | 2020-01-05T22:47:41 | 2020-01-05T22:47:41 | {} | MEMBER |
ClickHouse | ClickHouse | 570,973,873 | 8,531 | samz406 | ok. thank you | 2020-01-06T01:46:11 | 2020-01-06T01:46:11 | {} | NONE |
ClickHouse | ClickHouse | 571,616,119 | 7,646 | ivanprostran | > According to your logs the issue is about too heavy message load - CH can't keep up with incoming messages between poll intervals, and consumers get dropped. Also they require time to reconnect again later. You can try to tune the `stream_poll_timeout_ms` setting to a greater value.
Hi could you please be more spe... | 2020-01-07T14:44:18 | 2020-01-07T14:44:18 | {} | NONE |
ClickHouse | ClickHouse | 571,620,375 | 8,562 | alexey-milovidov | Мы хотим сделать, чтобы документация по однородным частям системы была расположена прямо в коде, была доступна из системных таблиц, и из неё уже генерировалась бы обычная документация для сайта.
Это как раз входит в наш roadmap, но так как задача довольно объёмная, скорее всего мы только начнём работу по ней в Q1, ... | 2020-01-07T14:54:17 | 2020-01-07T14:54:17 | {} | MEMBER |
ClickHouse | ClickHouse | 571,703,802 | 6,168 | nikitamikhaylov | Sorry :( I'll merge it after executing tests. | 2020-01-07T18:08:14 | 2020-01-07T18:08:14 | {} | MEMBER |
ClickHouse | ClickHouse | 570,707,187 | 8,515 | den-crane | The issue is not in the 00915_simple_aggregate_function.sql
I suspect (from previous conversation in telegram) that Sergio executed the 00915_simple_aggregate_function.sql (without drop table part) at his production server and left the table. After some time CH started to merge it and caught crash-loop. | 2020-01-03T21:43:51 | 2020-01-03T21:44:10 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 570,708,585 | 8,515 | azat | >And this is not related only to SimpleAggregateFunction but to LowCardinality(Nullable())
Ignore this, this is only within `SimpleAggregateFunction` | 2020-01-03T21:50:20 | 2020-01-03T21:50:20 | {} | MEMBER |
ClickHouse | ClickHouse | 570,731,811 | 8,515 | azat | > alexey-milovidov added the obsolete version label
This is reproduced even using version from upstream
AFAICS the problem is that output stream has `LowCardinality(Nullable())` while the input stream has `Nullable()` due to stripped `LowCardinality` in `AggregatingSortedBlockInputStream` (`RemovingLowCardinalit... | 2020-01-03T23:46:34 | 2020-01-03T23:46:34 | {} | MEMBER |
ClickHouse | ClickHouse | 571,470,889 | 8,473 | beartom | > I got the same error. Could you please share what configuration is wrong?
I found out the wrong configuration my self. Value of Interserver_http_host in one of cluster node is wrong. | 2020-01-07T07:32:15 | 2020-01-07T07:32:15 | {} | NONE |
ClickHouse | ClickHouse | 571,549,065 | 8,545 | zhang2014 | It changes the result of the following statement(Is that expected?)
```
SELECT * FROM (SELECT a, runningDifference(a) rd FROM (SELECT * FROM (SELECT 1 a UNION ALL SELECT 2 a ) ORDER BY a DESC ));
```
before:
```
┌─a─┬─rd─┐
│ 2 │ 0 │
│ 1 │ -1 │
└───┴────┘
```
after:
```
┌─a─┬─rd─┐
│ 2 │ 0 │
└───┴───... | 2020-01-07T11:30:17 | 2020-01-07T11:45:33 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 571,683,671 | 8,314 | javisantana | so, it'd ok to just rewrite the PR so only OPTIMIZE queries check for active replicas?
| 2020-01-07T17:17:06 | 2020-01-07T17:17:06 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 570,926,640 | 8,531 | den-crane | It's unclear how are you going to `delete a part`.
`alter table ... drop partition` does not affect performance. | 2020-01-05T16:27:51 | 2020-01-05T17:23:58 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 570,816,680 | 8,528 | excitoon | Okay, this is Ninja issue, because it creates output directories by itself:
> Output directories are always implicitly created before running the command that relies on them.
And `make` does not.
> If "crypto" is disabled, why we are going to this directory and interpreting this CMakeLists file?
I meant th... | 2020-01-04T20:22:05 | 2020-01-04T20:22:05 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 570,946,002 | 8,530 | alexey-milovidov | https://github.com/ClickHouse/ClickHouse/commit/31e4dc0e084737fb41e8c8ea1ea48124f7ce78b4 | 2020-01-05T20:45:32 | 2020-01-05T20:45:32 | {} | MEMBER |
ClickHouse | ClickHouse | 570,983,639 | 8,527 | den-crane | No haha, you executed the wrong SQL.
I specifically asked about
```
select * from
where length(toString(RDATE)) < 8 or length(toString(PDATE))<8
limit 100
```
not about
```
select * from
where length(toString(RDATE)) < 8 or length(toString(PDATE))<8
RDATE=20200104
limit 100
```
There is no or... | 2020-01-06T02:49:59 | 2020-01-06T03:06:23 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 570,793,514 | 8,527 | den-crane | Seems some of your RDATE column's values has less than 6 digits.
Use parseDateTimeBestEffortOrNull
```
SELECT toDate(parseDateTimeBestEffortOrNull('20040101'))
┌─toDate(parseDateTimeBestEffortOrNull('20040101'))─┐
│ 2004-01-01 │
└───────────────────────────────────────... | 2020-01-04T15:18:07 | 2020-01-04T15:27:06 | {
"+1": 2
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 570,821,341 | 8,529 | excitoon | If it will not, it shall be fixed. | 2020-01-04T21:26:43 | 2020-01-04T21:26:43 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 570,837,670 | 8,300 | kolen | The problem was fixed in #8297. No longer occurs for me (although similar problem occurs, #8530). | 2020-01-05T02:17:35 | 2020-01-05T02:17:35 | {} | NONE |
ClickHouse | ClickHouse | 571,116,044 | 7,507 | yuzhichang | The dddl hang problem occurred today on a node of a cluster. Before that the node machine rebooted several times for unknown reason.
Following log indicates that this node refused to execute the DDL:
```
2020.01.06 19:33:45.721174 [ 143 ] {19d35336-2ece-4938-a110-1cb1c696c0af} <Debug> executeQuery: (from [::1]:... | 2020-01-06T12:09:48 | 2020-01-06T14:28:53 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 571,330,958 | 8,547 | mrtarsa | Yes, I executed CREATE TABLE statements at all 3 nodes. Sorry, forgot to mention that.
And *distributed_product_mode* was set to *local* on config level at every node. | 2020-01-06T21:52:03 | 2020-01-06T21:53:28 | {} | NONE |
ClickHouse | ClickHouse | 571,394,423 | 8,542 | alexey-milovidov | > One query was slowed down, need further investigation:
Yes, `Ryu` library is slower than `doble-conversion` on small integers.
This is because the order of branches in this function:
```
static inline uint32_t decimalLength17(const uint64_t v) {
// This is slightly faster than a loop.
// The average out... | 2020-01-07T01:42:11 | 2020-01-07T01:45:37 | {} | MEMBER |
ClickHouse | ClickHouse | 571,486,318 | 7,507 | yuzhichang | I reproduced the hang problem with rebooting the machine on which a very light ck instance (loading tables costs only ~3 seconds) run.
```
2020.01.07 16:14:15.117528 [ 76 ] {} <Debug> DDLWorker: Started DDLWorker thread
2020.01.07 16:14:15.117682 [ 88 ] {} <Debug> DDLWorker: Started DDLWorker cleanup thread
202... | 2020-01-07T08:27:43 | 2020-01-07T08:29:21 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 571,604,090 | 8,557 | alexey-milovidov | That's not correct, SimpleSortCursor doesn't respect the ordering between streams that is important for background merges. | 2020-01-07T14:16:18 | 2020-01-07T14:16:18 | {} | MEMBER |
ClickHouse | ClickHouse | 570,794,901 | 8,296 | den-crane | @Har01d Are you sure?
```
create table blocks (id Int64, A String) Engine=MergeTree order by id;
insert into blocks select number, 'aaa' from numbers(600000);
SELECT count()
FROM blocks;
┌─count()─┐
│ 600000 │
└─────────┘
SELECT id
FROM blocks
ORDER BY id DESC
LIMIT 10
FORMAT JSON
....
"rows_befor... | 2020-01-04T15:36:07 | 2020-01-04T15:36:36 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 570,817,948 | 7,892 | alexey-milovidov | @filimonov @den-crane Sorry, both answers are not good enough.
1. You should not use `toString` as now you can pass UUID to the hash function directly.
2. @filimonov You may have heard that `xxHash64` is the "fastest" hash function but actually it depends on the distribution of string lengths and on the way how t... | 2020-01-04T20:38:14 | 2020-01-04T20:40:26 | {} | MEMBER |
ClickHouse | ClickHouse | 570,972,915 | 8,531 | den-crane | `ALTER TABLE table DELETE` is very heavy operation in comparison with `drop partition` .
`ALTER TABLE table DELETE` causes a huge I/O and CPU usage. After it finishes it does not affect performance.
Consider to use `alter table ... drop partition` and table with monthly / daily partitioning.
` drop partition` muc... | 2020-01-06T01:38:45 | 2020-01-06T01:55:29 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 571,100,820 | 8,093 | huang--yijian | > Debug package is not necessarily to install and usually it's not installed.
> The only benefit is to see source files and line numbers in stack traces of exception messages.
Thank you for the quick response.
We will upgrade by installing the following 4 RPM packages only:
clickhouse-client-19.17.6.36-2.noarch.r... | 2020-01-06T11:11:59 | 2020-01-06T11:11:59 | {} | NONE |
ClickHouse | ClickHouse | 571,173,879 | 8,544 | den-crane | MV uses columns names for insert not a columns order
```
CREATE MATERIALIZED VIEW test6pba.table_MV_raw2_sum1hour TO test6pba.table_sum1hour AS
SELECT
tenantid,
eventDate,
toStartOfHour(eventTimeStamp) AS eventTimeStamp, <<<<<<--------
dimension1,
SUM(wins) AS wins
FROM test6pba.table_ra... | 2020-01-06T15:09:09 | 2020-01-06T15:09:09 | {
"+1": 1
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 571,337,945 | 8,547 | den-crane | probably you've missed a database qualifier `db0` and created Distributed tables in a `default` database.
check `desc db0.a1` vs `desc default.a1`
```
CREATE TABLE db0.a1 (
day Date,
id UInt32
)
ENGINE = Distributed('ha1', 'db0', a1_replicated, id);
CREATE TABLE db0.b1 (
day Date,
id UIn... | 2020-01-06T22:11:25 | 2020-01-06T23:12:29 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 571,373,257 | 8,545 | alexey-milovidov | ```
SET max_memory_usage = 100000000000, max_bytes_before_external_sort = 50000000000;
SELECT number FROM numbers_mt(1000000) ORDER BY intHash64(number) LIMIT 900000, 10;
old: 0.149 sec.
new: 0.119 sec.
SELECT number FROM numbers_mt(10000000) ORDER BY intHash64(number) LIMIT 9000000, 10;
old: 2.622 sec.
new:... | 2020-01-07T00:16:56 | 2020-01-07T00:18:53 | {} | MEMBER |
ClickHouse | ClickHouse | 571,568,850 | 8,558 | filimonov | Fix delivered with v19.19.1.1892-testing or newer.
Probably we can consider backporting of that fix. | 2020-01-07T12:35:42 | 2020-01-07T12:35:42 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 570,701,603 | 8,515 | azat | >After run "as is" sample code:
> 00915_simple_aggregate_function.sql
I guess you really means at least [this version](https://github.com/ClickHouse/ClickHouse/blob/458f596e187c75c4e77669bfc0831cfe20b3b986/dbms/tests/queries/0_stateless/00915_simple_aggregate_function.sql) (since in #8485 support for `groupBit*` ad... | 2020-01-03T21:20:28 | 2020-01-03T21:20:28 | {} | MEMBER |
ClickHouse | ClickHouse | 570,828,150 | 7,892 | den-crane | I used toString cityHash64 does not support UUID
```
SELECT cityHash64(toUUID('00da1615-acd3-48ec-8dfc-1cc78cae98ce'))
Received exception from server (version 19.17.5):
Code: 43. DB::Exception: Received from localhost:9000. DB::Exception: Unexpected type UUID of argument of function cityHash64.
```
And I did ... | 2020-01-04T23:16:29 | 2020-01-04T23:16:29 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 570,975,579 | 8,527 | den-crane | oops. 20200104 - 8 digits not 6
```
select * from
where length(toString(RDATE)) < 8 or length(toString(PDATE))<8
limit 100
``` | 2020-01-06T01:58:57 | 2020-01-06T01:58:57 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 571,264,703 | 8,543 | den-crane | Q1: CH insert single-threaded
https://clickhouse.yandex/docs/en/introduction/performance/
_To improve performance, you can make multiple INSERT queries in **parallel**, and performance will increase **linearly**._
Q2: by default idle_connection_timeout = 3600. You can change it in user's profile.
Maybe JDBC p... | 2020-01-06T18:56:40 | 2020-01-06T18:56:50 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 571,630,365 | 8,551 | den-crane | CH Arrays?
```
CREATE TABLE some_table
(
`col1` Array(Int64)
)
ENGINE = Memory;
cat test.parquet |clickhouse-client -q 'insert into some_table format Parquet'
SELECT *
FROM some_table
┌─col1────┐
│ [1,2,3] │
│ [2,8,9] │
└─────────┘
``` | 2020-01-07T15:17:16 | 2020-01-07T15:17:16 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 570,743,076 | 8,523 | den-crane | I've replaced `ANY INNER JOIN` to `JOIN` because `ANY` became excessive (now) and leads to incorrect result in CH 19.19 (without any_join_distinct_right_table_keys=1) and an error in 19.17 | 2020-01-04T01:08:22 | 2020-01-04T01:08:22 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 570,797,896 | 4,607 | Felixoid | BTW, there's a project which does the job. https://github.com/innogames/graphite-ch-optimizer
The packaging is planned for the following week | 2020-01-04T16:16:08 | 2020-01-04T16:16:08 | {} | MEMBER |
ClickHouse | ClickHouse | 570,812,427 | 8,522 | alexey-milovidov | Don't understand why it's correct. Need review from @KochetovNicolai | 2020-01-04T19:24:49 | 2020-01-04T19:24:49 | {} | MEMBER |
ClickHouse | ClickHouse | 570,970,683 | 8,527 | den-crane | perhaps additional column changes a "query plan", and moves to prewhere different columns.
Check why the columns have less than six digits
```
select * from
where length(toString(RDATE)) < 6 or length(toString(PDATE))<6
limit 100
``` | 2020-01-06T01:22:12 | 2020-01-06T01:22:12 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 570,974,112 | 8,527 | 865357384 | 
the result is null , I'm still very confused. | 2020-01-06T01:48:07 | 2020-01-06T01:48:07 | {} | NONE |
ClickHouse | ClickHouse | 570,994,933 | 8,085 | tisonkun | the specific issue fixed. new build issue arise. Close this one and open the other. | 2020-01-06T04:02:07 | 2020-01-06T04:02:07 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 570,998,909 | 8,503 | zhang2014 | @alexey-milovidov
I think this pr is ready for review. I've made some changes, and now the predicate optimizer does two things:
- Try move predicates from having to where
- Try push the predicate down from the current query to the having of the subquery | 2020-01-06T04:28:02 | 2020-01-06T04:28:02 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 571,392,448 | 8,548 | den-crane | toInt64(60000)
toInt64(30000)
```
SELECT
toInt64(60000) AS x,
toInt64(30000) AS y,
ts,
ts / x,
intDiv(ts, x),
floor(ts / x),
ts / y,
intDiv(ts, y),
intDiv(1578346218965, x),
intDiv(1578346218965, y),
intDiv(60002, x),
intDiv(60002, y)
FROM testI... | 2020-01-07T01:32:57 | 2020-01-07T01:33:12 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 571,636,100 | 8,556 | den-crane | a. **You already have an index** on topic `ORDER BY(topic, domain)` -- as a prefix of a primary index.
b. You can add only skip indexes -- this a special type of indexes and they work by another way, minmax index stores only min max values of an index granula, so for example with 6020, this minmax index will have o... | 2020-01-07T15:30:03 | 2020-01-07T16:40:54 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 571,671,007 | 8,564 | luqasz | > Changes of remote_servers does not require reload clickhouse
So I only need to edit config file and CH will pick those changes ?
> Each CH node can be member of any number of clusters in any configuration. What are you going to discover automatically?
So one CH instance can participate in multiple clusters. ... | 2020-01-07T16:46:09 | 2020-01-07T16:48:34 | {} | NONE |
ClickHouse | ClickHouse | 571,681,377 | 8,563 | excitoon | Cherry-picked to #8422 | 2020-01-07T17:11:09 | 2020-01-07T17:11:09 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 570,796,611 | 8,466 | pachico | Have a look at chproxy, you won't need to expose real ch users and you can set rate limits too. | 2020-01-04T15:58:53 | 2020-01-04T15:58:53 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 570,909,602 | 8,530 | kolen | BTW, it's still possible to build it, with disabled ICU. I ran `ccmake`, set `NO` for `ENABLE_ICU` and finally got Mac OS binary, but without ICU feature. (I don't know how to do the same with cmake command line params). | 2020-01-05T12:49:51 | 2020-01-05T12:49:51 | {} | NONE |
ClickHouse | ClickHouse | 570,997,100 | 8,527 | 865357384 | Thank you very much. | 2020-01-06T04:17:08 | 2020-01-06T04:17:08 | {} | NONE |
ClickHouse | ClickHouse | 571,684,230 | 8,564 | den-crane | >So I only need to edit config file and CH will pick those changes ?
yes. Immediately. You can verify it with `select * from system.clusters`
>Scaling here is important since this changes number of nodes.
The big problem is that CH does not have automatic re-sharding. There is no easy way to add a new shard ... | 2020-01-07T17:18:33 | 2020-01-07T17:18:33 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 570,836,998 | 6,049 | alexey-milovidov | In recent ClickHouse version it's possible to configure.
You can enable `text_log` in configuration file and then create a MATERIALIZED VIEW on top of it with `File` engine:
```
CREATE MATERIALIZED VIEW system.text_log_json ENGINE = File(JSONEachRow) AS SELECT * FROM system.text_log
```
Data will be availabl... | 2020-01-05T02:01:25 | 2020-01-05T02:01:25 | {
"+1": 2
} | MEMBER |
ClickHouse | ClickHouse | 571,329,397 | 8,547 | den-crane | Your expectations are wrong.
CH does not create tables at other nodes automatically. All CH nodes are independent and are not aware about cluster. You need to execute all 4 `create table` at all 3 nodes manually. Or use `on cluster` helper https://clickhouse.yandex/docs/en/query_language/create/#create-table-query bu... | 2020-01-06T21:47:52 | 2020-01-06T21:49:32 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 571,642,665 | 8,548 | cw9 | even with `toInt64()` the result of `intDiv(60002, toInt64(30000))` should be 2 instead of 0?
```
SELECT
intDiv(60002, 30000),
intDiv(60002, toInt64(30000)),
intDiv(toInt64(60002), toInt64(30000))
┌─intDiv(60002, 30000)─┬─intDiv(60002, toInt64(30000))─┬─intDiv(toInt64(60002), toInt64(30000))─┐
│ ... | 2020-01-07T15:44:53 | 2020-01-07T15:46:18 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 570,813,184 | 8,528 | excitoon | Perl code generation scripts require destination directory to exist,
otherwise they output files to `STDOUT`. CMake automatically creates binary
directories only if corresponding source directory was added with
`add_subdirectory`. Everything works fine on other OSes probably because
`crypto` from system is being used (... | 2020-01-04T19:34:00 | 2020-01-04T19:34:00 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 570,813,606 | 8,528 | alexey-milovidov | > CMake automatically creates binary
directories only if corresponding source directory was added with
`add_subdirectory`.
If "crypto" is disabled, why we are going to this directory and interpreting this CMakeLists file?
> Everything works fine on other OSes probably because
`crypto` from system is being used... | 2020-01-04T19:39:47 | 2020-01-04T19:39:47 | {} | MEMBER |
ClickHouse | ClickHouse | 570,816,187 | 6,128 | alexey-milovidov | @den-crane We have tarball releases but they don't contain installation script.
(You still can run ClickHouse without installation.)
Will reopen.
And we still don't have source tarballs. | 2020-01-04T20:16:46 | 2020-01-04T20:17:56 | {} | MEMBER |
ClickHouse | ClickHouse | 570,816,814 | 7,934 | alexey-milovidov | @den-crane
All these ideas are viable:
- support for monotonicity analysis for functions with multiple arguments if all but one argument are constant - so `substring` will start to use index;
- add a function `startsWithAnyOf` and make it use index;
- index usage for regexp when it implies one or multiple prefix... | 2020-01-04T20:23:29 | 2020-01-04T20:23:29 | {} | MEMBER |
ClickHouse | ClickHouse | 571,070,157 | 8,515 | SergioSafonov | > The issue is not in the 00915_simple_aggregate_function.sql
> I suspect (from previous conversation in telegram) that Sergio executed the 00915_simple_aggregate_function.sql (without drop table part) at his production server and left the table. After some time CH started to merge it and caught crash-loop.
Exactly | 2020-01-06T09:36:15 | 2020-01-06T09:36:15 | {} | NONE |
ClickHouse | ClickHouse | 571,374,762 | 8,545 | alexey-milovidov | @KochetovNicolai This has to be ported to "Processors" branch. | 2020-01-07T00:22:16 | 2020-01-07T00:22:16 | {} | MEMBER |
ClickHouse | ClickHouse | 571,453,935 | 8,511 | YiuRULE | I rebase the branch from master.
Small note: The CI tests who fail are not related to this PR. (reproducible to master) | 2020-01-07T06:24:37 | 2020-01-07T06:24:37 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 571,579,523 | 8,471 | filimonov | Duplicate https://github.com/ClickHouse/ClickHouse/issues/7690 | 2020-01-07T13:11:04 | 2020-01-07T13:11:04 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 570,742,439 | 8,515 | azat | At first glance - #8522 | 2020-01-04T01:02:35 | 2020-01-04T01:02:35 | {} | MEMBER |
ClickHouse | ClickHouse | 570,792,772 | 8,296 | Har01d | > optimize_read_in_order = 0 ?
That doesn’t help unfortunately.
The latest version which supports the old behaviour is 19.13.6.51.
@alexey-milovidov @den-crane | 2020-01-04T15:08:45 | 2020-01-04T15:08:45 | {} | NONE |
ClickHouse | ClickHouse | 570,837,326 | 6,808 | alexey-milovidov | We should also implement bitwise ops on FixedString and IPv6. | 2020-01-05T02:08:39 | 2020-01-05T02:08:39 | {} | MEMBER |
ClickHouse | ClickHouse | 570,839,256 | 8,530 | tisonkun | same problem | 2020-01-05T02:46:05 | 2020-01-05T02:46:05 | {
"-1": 1
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 571,390,879 | 8,542 | alexey-milovidov | Performance test is Ok and shows 2% improvement across all the test cases:
https://clickhouse-test-reports.s3.yandex.net/8542/828e43cf9432e3624e6cdfb5513bc5d851f91209/performance_test/comparison_max_rps_gcc_9.html
One query was slowed down, need further investigation:
```
SELECT count() FROM system.numbers WHERE ... | 2020-01-07T01:26:29 | 2020-01-07T01:26:29 | {} | MEMBER |
ClickHouse | ClickHouse | 571,485,995 | 8,542 | alexey-milovidov | Now we have 4% overall performance increase:
https://clickhouse-test-reports.s3.yandex.net/8542/c1ccb427d5b1de0a13ee456458b9a8fbbade92c8/performance_test/comparison_max_rps_gcc_9.html
And there is no slower queries. | 2020-01-07T08:26:41 | 2020-01-07T08:26:41 | {} | MEMBER |
ClickHouse | ClickHouse | 571,600,300 | 8,545 | alexey-milovidov | It's not guaranteed how the returned data will be split in blocks.
Yes, this result is valid.
| 2020-01-07T14:05:54 | 2020-01-07T14:05:54 | {} | MEMBER |
ClickHouse | ClickHouse | 571,601,193 | 8,545 | alexey-milovidov | @zhang2014 Let's just adapt the remaining test and we will merge #8503. | 2020-01-07T14:08:26 | 2020-01-07T14:08:26 | {} | MEMBER |
ClickHouse | ClickHouse | 571,689,030 | 8,564 | filimonov | Also check `from_zk`
https://github.com/ClickHouse/ClickHouse/search?utf8=%E2%9C%93&q=from_zk&type= | 2020-01-07T17:30:20 | 2020-01-07T17:31:41 | {
"+1": 1
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 570,837,458 | 7,840 | alexey-milovidov | @nikitamikhaylov This code doesn't work at all.
Example:
`INSERT INTO TABLE FUNCTION file('data.tsv.gz', TSV, 'number UInt64') SELECT number FROM numbers(1000000)`
```
zcat data.tsv.gz | tail
gzip: data.tsv.gz: invalid compressed data--format violated
998903
998904
998905
998906
998907
998908
998909
... | 2020-01-05T02:11:43 | 2020-01-05T02:12:25 | {} | MEMBER |
ClickHouse | ClickHouse | 570,918,505 | 8,534 | excitoon | By this fix package now allows to run ClickHouse server, and not only client. | 2020-01-05T14:42:14 | 2020-01-05T14:42:14 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 570,976,303 | 8,527 | 865357384 | 
Ah ha ha ， this is 8 result. | 2020-01-06T02:03:49 | 2020-01-06T02:03:49 | {} | NONE |
ClickHouse | ClickHouse | 571,024,874 | 4,717 | vitlibar | The current behaviour was implemented to fix hanging of the server when it's starting with some dictionaries which take a lot of time to be loaded. Earlier the server used to load all dictionaries synchronously on its start (if `dictionaries_lazy_load=false`) and back then it was enough to have one long-loading diction... | 2020-01-06T06:44:50 | 2020-01-06T06:50:43 | {} | MEMBER |
ClickHouse | ClickHouse | 571,090,335 | 8,093 | huang--yijian | By the way is clickhouse-common-static-dbg-19.17.6.36-2.x86_64.rpm required to install ClickHouse 19.17.6 ?
It is 896MB in size, very slow to download from Shanghai, China. Other RPMs are much smaller:
clickhouse-client-19.17.6.36-2.noarch.rpm
clickhouse-common-static-19.17.6.36-2.x86_64.rpm
clickhouse-server-19... | 2020-01-06T10:38:36 | 2020-01-06T10:41:08 | {} | NONE |
ClickHouse | ClickHouse | 571,091,652 | 8,093 | alexey-milovidov | Debug package is not necessarily to install and usually it's not installed.
The only benefit is to see source files and line numbers in stack traces of exception messages. | 2020-01-06T10:42:26 | 2020-01-06T10:42:26 | {} | MEMBER |
ClickHouse | ClickHouse | 571,392,864 | 8,548 | den-crane | duplicate https://github.com/ClickHouse/ClickHouse/issues/7764 | 2020-01-07T01:34:46 | 2020-01-07T01:34:46 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 571,574,108 | 8,288 | filimonov | https://github.com/ClickHouse/ClickHouse/issues/6418 | 2020-01-07T12:53:52 | 2020-01-07T12:53:52 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 571,640,310 | 8,564 | den-crane | Each CH node can be member of any number of clusters in any configuration. What are you going to discover automatically?
For example I have 3 CH nodes N1, N2, N3 with 4 clusters simultaneously
Cluster1 (each node as a shard): shard1=N1, shard2=N2, shard3=N3
Cluster2 (one shard 3 replica): shard1=N1,N2,N3
Cluste... | 2020-01-07T15:39:47 | 2020-01-07T15:46:09 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 571,655,977 | 7,646 | filimonov | if you have a message about `max.poll.interval.ms exceeded` that means that there was a too big delay between 2 consecutive consume calls.
I.e. either it takes too much time to flush the previously consumed data to the disk, either all the background threads from the pool were busy in the meanwhile (for example that... | 2020-01-07T16:13:51 | 2020-01-07T16:16:24 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 570,806,939 | 8,525 | alexey-milovidov | It means that we cannot read data from HDFS data nodes. You need to check network connectivity - are data nodes accessible from ClickHouse server or they are closed by firewall. | 2020-01-04T18:13:58 | 2020-01-04T18:13:58 | {
"+1": 1
} | MEMBER |
ClickHouse | ClickHouse | 570,807,078 | 8,524 | alexey-milovidov | @excitoon | 2020-01-04T18:15:59 | 2020-01-04T18:15:59 | {
"eyes": 2
} | MEMBER |
ClickHouse | ClickHouse | 571,374,635 | 8,545 | alexey-milovidov | Performance test shows significant improvement:
https://clickhouse-test-reports.s3.yandex.net/8545/2314bb5a3430c421ddc371daeb9d20134bdaf181/performance_test/comparison_min_time_gcc_9.html | 2020-01-07T00:21:43 | 2020-01-07T00:21:43 | {} | MEMBER |
ClickHouse | ClickHouse | 571,646,446 | 8,548 | den-crane | Obviously both arguments of intDiv have to have the same type and should be casted to overpower the bug #7764 .
```
desc (select 60002)
┌─name─┬─type───┬
│ x │ UInt16 │
└──────┴────────┴
```
**U**int16 != Int64 | 2020-01-07T15:53:14 | 2020-01-07T15:55:26 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 570,701,673 | 8,515 | azat | I don't know why this tests makes the problem for you without any changes, but if you add `OPTIMIZE TABLE simple FINAL` before final `DROP TABLE simple` it will trigger the issue. | 2020-01-03T21:20:42 | 2020-01-03T21:20:42 | {} | MEMBER |
ClickHouse | ClickHouse | 570,707,883 | 8,515 | azat | >(without drop table part) at his production server and left the table.
Indeed, and now this makes sense | 2020-01-03T21:46:58 | 2020-01-03T21:46:58 | {} | MEMBER |
ClickHouse | ClickHouse | 570,970,357 | 8,531 | samz406 | no partition, first use ALTER TABLE table DELETE WHERE BA_MONTH='xxx', and insert data,Will this affect the performance query?
| 2020-01-06T01:19:42 | 2020-01-06T01:19:42 | {} | NONE |
ClickHouse | ClickHouse | 571,067,562 | 8,525 | aa8808021 | > It means that we cannot read data from HDFS data nodes. You need to check network connectivity - are data nodes accessible from ClickHouse server or they are closed by firewall.
Ok, I'll check the Internet problem. Thanks
| 2020-01-06T09:28:08 | 2020-01-06T09:28:08 | {} | NONE |
ClickHouse | ClickHouse | 571,153,390 | 8,479 | 4ertus2 | #8431 | 2020-01-06T14:16:01 | 2020-01-06T14:16:01 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 571,340,806 | 7,799 | genzgd | We are also rolling back as a result of this bug. Fixing our various queries has proven very cumbersome. | 2020-01-06T22:19:20 | 2020-01-06T22:19:20 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 571,462,498 | 8,547 | mrtarsa | In my case tables `a1` and `b1` are Distributed in `default` database. That was intended and I cannot move the tables into another database.
I simplified it and just for example explained the problem with single replica for every shard. In such case topology is like
```
node0 node1 node2
s0r0 s1r0 s2r... | 2020-01-07T07:00:26 | 2020-01-07T07:20:21 | {} | NONE |
ClickHouse | ClickHouse | 571,469,301 | 6,168 | YiuRULE | ping @nikitamikhaylov @Akazz, does we got any news concerning this PR ? a bit afraid than the PR is a bit abandoned as it was created in july :p | 2020-01-07T07:26:37 | 2020-01-07T07:26:37 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 571,648,721 | 8,550 | den-crane | Workaround: toDateTime('2019-12-03 00:00:00')
```
ALTER TABLE default.cdp_orders DELETE
WHERE order_time >= toDateTime('2019-12-03 00:00:00')
``` | 2020-01-07T15:58:20 | 2020-01-07T15:58:20 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 570,654,046 | 8,519 | vzakaznikov | @KochetovNicolai, this is a draft of first changes to start refactoring live view storage engine. | 2020-01-03T18:17:54 | 2020-01-04T15:41:01 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 570,969,416 | 8,527 | 865357384 | thanks，My needs are solved. But why not add columns? In MySQL ,the statements is OK change same function.
| 2020-01-06T01:12:25 | 2020-01-06T01:12:25 | {} | NONE |
ClickHouse | ClickHouse | 571,448,192 | 8,473 | beartom | I got the same error. Could you please share what configuration is wrong? | 2020-01-07T05:59:25 | 2020-01-07T05:59:25 | {} | NONE |
ClickHouse | ClickHouse | 577,428,972 | 8,750 | alexey-milovidov | > adding tmp/ in all callers looks error-prone
So, the issue is that after we selected disk, we have to append tmp to its path but it's too error prone because we have to add it in multiple places: external aggregation, external sorting, joins, distributed tables... Do I understand correctly?
> IOW maybe this sho... | 2020-01-22T23:08:07 | 2020-01-22T23:08:07 | {} | MEMBER |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.