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 | 510,902,546 | 5,988 | abraithwaite | We are doing inserts with the Go client from a kafka topic. We write in batches of 10-20k records which happens on a 1-2s period. The number of parts grew past `parts_to_delay_insert` and was hovering around 200 parts for the most recent partition of data.
After stopping our workers entirely, I noticed that insert... | 2019-07-12T14:14:46 | 2019-07-12T14:14:46 | {} | NONE |
ClickHouse | ClickHouse | 511,045,308 | 5,958 | alexey-milovidov | I'll add by myself. | 2019-07-12T21:52:15 | 2019-07-12T21:52:15 | {} | MEMBER |
ClickHouse | ClickHouse | 511,355,626 | 5,947 | filimonov | it's make/ninja output, not cmake :)
I.e. output of that command: ```cmake3 ..``` executed in empty build folder, or in folder w/o CMakeCache.txt It show all components & libraries which will be used for building
BTW: does it build without tests?
Try
```
cmake3 .. -DENABLE_TESTS=OFF
```
(better on clean... | 2019-07-15T10:51:14 | 2019-07-15T10:51:14 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 511,361,106 | 5,974 | filimonov | Rather should be fixed other way, as we prefer to have compatibility with older kernel & glibc.
Do you have full stacktraces of error? Usually can be found near exception in log. | 2019-07-15T11:12:35 | 2019-07-15T11:12:35 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 511,526,382 | 6,003 | den-crane | and be aware that * can cause a wrong result https://github.com/yandex/ClickHouse/issues/4974 | 2019-07-15T18:55:53 | 2019-07-15T18:55:53 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 510,404,179 | 5,926 | filimonov | Can't reproduce:
```sql
CREATE TABLE default.tca_revision
(
`trade_date` Date,
`revision_id` FixedString(20),
`instrument` LowCardinality(String),
`side` LowCardinality(String),
`limit` UInt32,
`start_time` UInt64,
`end_time` UInt64,
`state` LowCardinality(String),
... | 2019-07-11T09:16:23 | 2019-07-11T09:17:34 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 510,442,828 | 5,500 | igor-sh8 | Thanks for detailed info.
We set min_merge_bytes_to_use_direct_io to 0 and made bigger parts in Buffer tables.
But we were still afraid IO problem in some near future so in our case (we do not need really store such big table) we decided to use Null Engine.
| 2019-07-11T11:12:48 | 2019-07-11T11:12:48 | {} | NONE |
ClickHouse | ClickHouse | 509,128,683 | 5,883 | mobilesfinks | /cc @proller | 2019-07-08T08:23:02 | 2019-07-11T14:27:35 | {} | NONE |
ClickHouse | ClickHouse | 511,636,868 | 6,013 | ngoanpv | Very clear, thank you. @den-crane | 2019-07-16T02:15:18 | 2019-07-16T02:15:18 | {} | NONE |
ClickHouse | ClickHouse | 511,717,519 | 5,999 | alexey-milovidov | @filimonov It is documented behaviour and should not be fixed without proper performance test with real dataset. | 2019-07-16T08:23:44 | 2019-07-16T08:24:15 | {} | MEMBER |
ClickHouse | ClickHouse | 511,749,859 | 5,990 | akuzm | First, please make sure that the type of column is Nullable(Int32), or else it won't be able to store nulls. Then, use '\N' for null values, not 'NULL'. Alternatively, you can set `input_format_defaults_for_omitted_fields = 1`, and then empty values like in your second example will be treated as nulls. | 2019-07-16T09:55:54 | 2019-07-16T09:55:54 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 510,392,339 | 5,969 | filimonov | > 2019.07.10 13:23:57.251952 [ 65 ] {} BaseDaemon: (version **19.8.3.8** (official build)) (from thread 55) Received signal Segmentation fault (11).
> ...
It looks like 19.8.3 not 19.9 | 2019-07-11T08:44:52 | 2019-07-11T08:45:09 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 510,545,422 | 5,683 | den-crane | I believe the reason is different partitioning. | 2019-07-11T15:55:23 | 2019-07-11T15:55:23 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 511,108,578 | 5,984 | filimonov | Less typing and no subselects:
```sql
WITH
[(1, 'one'), (2, 'two'), (3, 'three')] AS rows_array,
arrayJoin(rows_array) AS row_tuple
SELECT
row_tuple.1 AS number_decimal,
row_tuple.2 AS number_string
┌─number_decimal─┬─number_string─┐
│ 1 │ one │
│ 2 │ ... | 2019-07-13T09:49:05 | 2019-07-13T09:52:28 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 511,422,925 | 5,376 | tarabanton | Of course we know about it. That's why we set `UseServerSidePrepare = 0`
We are using postgres packages from https://wiki.postgresql.org/wiki/Apt | 2019-07-15T14:21:36 | 2019-07-15T14:22:50 | {} | NONE |
ClickHouse | ClickHouse | 511,515,585 | 5,541 | KochetovNicolai | @vzakaznikov I actually went through only couple of files, because diff is big. I think it will take some time.
Anyway, implementation is not intrusive indeed, so it will likely be merged. Thank you for it. | 2019-07-15T18:24:49 | 2019-07-15T18:24:49 | {} | MEMBER |
ClickHouse | ClickHouse | 511,520,909 | 6,015 | alexey-milovidov | Mimalloc has memory leak (aka unbounded fragmentation) + leak of memory mappings in some typical scenario. | 2019-07-15T18:39:46 | 2019-07-15T18:39:46 | {} | MEMBER |
ClickHouse | ClickHouse | 511,283,887 | 5,264 | filimonov | @bzon, can you check your `<remote_servers>` config? You need to ensure that all servers are listed there with proper name. | 2019-07-15T06:09:57 | 2019-07-15T06:09:57 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 511,350,600 | 6,004 | filimonov | https://github.com/yandex/ClickHouse/blob/e1271ae1f2a9bfb635ef83b23e48e9791195cce3/dbms/src/Interpreters/Join.h#L65-L120 | 2019-07-15T10:31:23 | 2019-07-15T10:31:23 | {
"+1": 2
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 511,434,672 | 5,969 | kondratspb | >
>
> Does it depend on data inserted? Could you make a reproducible sample that does not need special data?
No, it doesn't depend. It repeats if no data in table | 2019-07-15T14:50:01 | 2019-07-15T14:50:01 | {} | NONE |
ClickHouse | ClickHouse | 511,509,832 | 5,940 | alesapin | Fix https://github.com/yandex/ClickHouse/pull/6018. | 2019-07-15T18:08:46 | 2019-07-15T18:08:46 | {} | MEMBER |
ClickHouse | ClickHouse | 510,396,378 | 5,959 | filimonov | That sound suspicious. Can you please:
1) recheck `select * from system.settings where changed=1` (especially block sizes)
2) do a `tee` to file in your pipe chain to dump what exactly was passed. It's possible that something is wrong with delimiters and CH interprets some big check of data as one huge field. I.e. so... | 2019-07-11T08:56:04 | 2019-07-11T08:57:05 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 510,769,779 | 5,987 | ddliu | Tried to specify all columns instead of "*", it's much faster(0.009 sec), about twice of `1.1.54380`(0.005 sec).
```
SELECT
database,
name,
engine,
is_temporary,
data_path,
metadata_path
FROM system.tables
WHERE name = 'test_997'
┌─database─┬─name─────┬─engine────┬─is_temporary─┬─... | 2019-07-12T06:52:03 | 2019-07-12T06:52:03 | {} | NONE |
ClickHouse | ClickHouse | 511,233,795 | 5,601 | Maiakov | Kafka nowadays is one of the most popular event streaming platform and the most preferred message format for it is Avro. I really believe that adding avro connector extremely important for my company and whole the community in general. | 2019-07-14T20:41:47 | 2019-07-14T20:41:47 | {} | NONE |
ClickHouse | ClickHouse | 510,940,211 | 4,402 | akuzm | After debugging this for a couple of days, I came to the conclusion that it must be a race and ran it under tsan, and sure, here it is:
```
WARNING: ThreadSanitizer: data race (pid=25547)
Read of size 8 at 0x7b10001080e8 by thread T58:
#0 DB::Arena::alignedAlloc(unsigned long, unsigned long) /home/akuzm/ch3/c... | 2019-07-12T15:58:29 | 2019-07-12T15:59:10 | {
"+1": 1
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 511,056,560 | 5,997 | den-crane | Please provide DDL of tickets_flat. Do you use Date(Int16 / Int8) ... CODEC(Delta ? | 2019-07-12T22:51:33 | 2019-07-12T22:51:33 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 511,362,168 | 5,994 | filimonov | Please provide stacktrace (try repeating query with `clickhouse-client --stacktrace`).
And if possible - reproducible sample. I.e. create table statement + some data | 2019-07-15T11:17:01 | 2019-07-15T11:17:01 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 511,490,396 | 2,847 | den-crane | @vitlibar Can you fix it, please? It should be pretty easy fix. Need just a check that the partial dictionary update was successful before shifting(updating) last updating time?
And would be good to have a lag (-1 sec) configurable in dictionary config, because 1 second is not enough for **most** of cases (time drif... | 2019-07-15T17:16:10 | 2019-07-15T17:17:30 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 510,616,372 | 5,018 | simPod | @4ertus2 I no longer remember the context and the test case I provided is not sufficient :/ Let's consider it solved, thanks! | 2019-07-11T19:11:20 | 2019-07-11T19:11:20 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 510,697,868 | 5,984 | alexey-milovidov | We can implement it with a little bit different syntax:
```
SELECT * FROM VALUES('a UInt64, s String', (1, 'one'), (2, 'two'), (3, 'three'))
``` | 2019-07-12T00:15:23 | 2019-07-12T00:15:23 | {
"+1": 6
} | MEMBER |
ClickHouse | ClickHouse | 510,701,122 | 5,436 | alexey-milovidov | Probably we can implement it as `{name:Identifier}`.
`Identifier` will be processed as a special case and generate ASTIdentifier instead of ASTLiteral.
(you can implement it for this task) | 2019-07-12T00:34:38 | 2019-07-12T00:35:04 | {} | MEMBER |
ClickHouse | ClickHouse | 510,913,712 | 5,988 | abraithwaite | If we're not inserting into the distributed table, do we have to write to both replica's of the shard ourselves (since internal_replication is a distributed table setting)?
The distributed table docs make it sound like we will have to write to both replicas and the replicatedmergetree won't copy over the data itself... | 2019-07-12T14:45:36 | 2019-07-12T14:47:46 | {} | NONE |
ClickHouse | ClickHouse | 510,984,278 | 5,981 | 4ertus2 | related #5844 | 2019-07-12T18:19:49 | 2019-07-12T18:19:49 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 511,039,360 | 5,988 | abraithwaite | A better question might be: How do we appropriately monitor distributed table insertions? As far as I can tell, there's no settings to adjust or metrics to watch to tune and monitor distributed table insertion lag.
We've setup our own (just does `select max(dt) from global_raw_events` and compare that value to the ... | 2019-07-12T21:29:13 | 2019-07-12T21:29:39 | {} | NONE |
ClickHouse | ClickHouse | 511,046,157 | 5,988 | abraithwaite | Found the "DelayedInserts" metric which seems to be for distributed tables. That's helpful at least :-) | 2019-07-12T21:56:24 | 2019-07-12T21:56:24 | {} | NONE |
ClickHouse | ClickHouse | 511,510,378 | 5,376 | akuzm | Could you run a query against your ODBC source with isql or some other client, and look at the postgres logs to confirm that it doesn't use prepared statements? That way we can make sure that the ODBC source is configured correctly. | 2019-07-15T18:10:23 | 2019-07-15T18:10:23 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 510,334,938 | 5,436 | zhang2014 | OK, I haven't looked at this part of the code on query parameters, will it support identifier type parameters in the future? | 2019-07-11T05:28:49 | 2019-07-11T05:28:49 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 510,428,371 | 4,923 | agaurav | @filimonov sorry got lost in some ongoing work. I will generate the data and schema to reproduce during the weekend. | 2019-07-11T10:26:07 | 2019-07-11T10:26:07 | {} | NONE |
ClickHouse | ClickHouse | 510,435,961 | 5,969 | kondratspb | If I change schema to this (remove some columns from ORDER BY index):
CREATE TABLE ... (localDate Date DEFAULT toDate(localDateTime), localDateTime DateTime, timestamp UInt32, id String, mac String, zone Int32, assoc Int8, radio String, rssi Float64, freq Float64, vendor Int64) ENGINE = ReplicatedMergeTree('/clickhous... | 2019-07-11T10:50:51 | 2019-07-11T10:50:51 | {} | NONE |
ClickHouse | ClickHouse | 510,795,690 | 5,947 | blind-oracle | @filimonov
Here's the failed build log: https://novg.net/build.log.txt
Made on `CentOS` + `gcc-8.3.1` using simple:
```
# cd /usr/src
# git clone --single-branch --branch 19.9 --recursive https://github.com/yandex/ClickHouse ClickHouse-19.9
# cd ClickHouse-19.9
# mkdir build && cd build
# cmake3 ..
# make VER... | 2019-07-12T08:15:52 | 2019-07-12T08:18:20 | {} | NONE |
ClickHouse | ClickHouse | 510,915,964 | 5,984 | sklyar | > We can implement it with a little bit different syntax:
>
> ```
> SELECT * FROM VALUES('a UInt64, s String', (1, 'one'), (2, 'two'), (3, 'three'))
> ```
That would be great! Now I can do so:
```
SELECT *
FROM (
SELECT toUInt64(tupleElement(arr, 1)) as digit_number,
tupleEleme... | 2019-07-12T14:51:45 | 2019-07-12T14:51:45 | {
"+1": 2
} | NONE |
ClickHouse | ClickHouse | 510,999,572 | 5,996 | alexey-milovidov | https://github.com/yandex/ClickHouse/commit/731a47996908dab48e07f6a9bf7ce4bde4ebea2a | 2019-07-12T19:06:45 | 2019-07-12T19:06:45 | {} | MEMBER |
ClickHouse | ClickHouse | 511,438,957 | 6,003 | Akazz | Current ClickHouse behavior conforms with MySQL:
```
mysql> select * from test.test
+--------+--------------+-------+----------------+
| int_id | int_nullable | float | float_nullable |
+--------+--------------+-------+----------------+
| 1 | NULL | 2 | NULL |
| 2 | 1 |... | 2019-07-15T15:00:49 | 2019-07-15T15:00:49 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 510,909,081 | 5,965 | dimarub2000 | I have a test for primary key as well, should I add it after checks are completed? | 2019-07-12T14:32:38 | 2019-07-12T14:32:38 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 511,338,384 | 5,846 | abyss7 | @josepowera thanks for the test, we'll look into the problem further. | 2019-07-15T09:46:49 | 2019-07-15T09:46:49 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 511,353,234 | 5,997 | filimonov | What the filesystem is? Do other / simpler inerts behave normally?
Can you see some other records containing `tmp_insert_5_69_69_0` in `clickhouse-server.log`? | 2019-07-15T10:41:46 | 2019-07-15T11:28:17 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 510,472,585 | 5,977 | alesapin | Need to update docs because rpm requires to import public repo key on client machine:
`$ rpm --import robot-clickhouse-pub-key.gpg`
| 2019-07-11T12:52:46 | 2019-07-11T12:52:46 | {} | MEMBER |
ClickHouse | ClickHouse | 510,692,482 | 5,983 | filimonov | Relates to #5932? | 2019-07-11T23:43:03 | 2019-07-11T23:43:03 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 510,759,712 | 5,988 | blinkov | @abraithwaite there are different options on how to do insertion with various trade-offs. Major choice is whether to write to Distributed table or directly to shards (the latter is harder to implement, but works better), then there are settings like `insert_quorum` that allow to wait for data to be written on more than... | 2019-07-12T06:10:09 | 2019-07-12T06:10:09 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 510,723,727 | 5,436 | zhang2014 | Ok, I will start this task after I finish [json-type](https://github.com/zhang2014/ClickHouse/tree/feature/json_data_type) work(which takes about a week or two). | 2019-07-12T02:46:05 | 2019-07-12T06:16:19 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 510,782,721 | 5,991 | lordk911 | root cause is :
<Error> executeQuery: Code: 252, e.displayText() = DB::Exception: Too many partitions for single INSERT block (more than 5000). The limit is controlled by 'max_partitions_per_insert_block' setting. Large number of partitions is a common misconception. It will lead to severe negative performance impact,... | 2019-07-12T07:36:13 | 2019-07-12T07:36:13 | {} | NONE |
ClickHouse | ClickHouse | 510,801,805 | 5,993 | fasoal | ClickHouse 18.16.1
query sql: | 2019-07-12T08:34:39 | 2019-07-12T08:34:39 | {} | NONE |
ClickHouse | ClickHouse | 510,841,189 | 5,974 | lordk911 | after update linux kernel to 5.2.0 , I can read parquet from HDFS. | 2019-07-12T10:44:22 | 2019-07-12T10:44:22 | {} | NONE |
ClickHouse | ClickHouse | 510,911,330 | 5,988 | abraithwaite | Gotcha. So for distributed writes, since the data can end up on any node, clickhouse buffers the data instead of streaming it to the ReplicatedMergeTree tables.
I'm guessing `insert_quorum` doesn't affect distributed table inserts as well? FWIW, the docs say the default for insert_quorum is 0 not 1. 1 makes sense... | 2019-07-12T14:39:00 | 2019-07-12T14:39:00 | {} | NONE |
ClickHouse | ClickHouse | 511,366,750 | 5,999 | MahmoudGoda0 | So we dont have any solution right now for this issue ? | 2019-07-15T11:34:02 | 2019-07-15T11:34:02 | {} | NONE |
ClickHouse | ClickHouse | 511,530,848 | 6,013 | den-crane | system.parts has information about parts of MergeTree tables.
If XX is empty it has no parts, and the size of the X is 0.
For engines *Log you can use OS utilities only [du -sh /var/lib/clickhouse/data/db/XX]
For engine Memory no way at all. | 2019-07-15T19:08:25 | 2019-07-15T19:08:47 | {
"heart": 1
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 511,619,381 | 6,009 | alexey-milovidov | Ok. | 2019-07-16T00:38:25 | 2019-07-16T00:38:25 | {} | MEMBER |
ClickHouse | ClickHouse | 510,467,868 | 5,683 | blind-oracle | @den-crane
Nope, nothing changed. If I remove `Delta` codec then this `UInt32` column explodes 10 times in size. The same with ZSTD1 compression. | 2019-07-11T12:37:43 | 2019-07-11T12:37:43 | {} | NONE |
ClickHouse | ClickHouse | 510,697,030 | 5,983 | vincenzon | Thanks @filimonov I arrived at the change of '::' to '0.0.0.0' after reading https://www.altinity.com/blog/2019/3/15/clickhouse-networking-part-1 and that fixed it. I'm not sure why ipv6 is turned off on the new machine, but thanks for your help. | 2019-07-12T00:10:06 | 2019-07-12T00:10:06 | {} | NONE |
ClickHouse | ClickHouse | 510,749,846 | 5,974 | lordk911 | logs from error.log :
2019-07-11 19:08:43.884803, p26368, th140517123806976, ERROR Unexpected exception N5boost10wrapexceptINS_5uuids13entropy_errorEEE: getrandom
2019-07-11 19:25:18.846773, p26368, th140517123806976, ERROR Unexpected exception N5boost10wrapexceptINS_5uuids13entropy_errorEEE: getrandom
| 2019-07-12T05:24:21 | 2019-07-12T05:24:21 | {} | NONE |
ClickHouse | ClickHouse | 510,787,863 | 5,961 | vitlibar | Fixed in master. | 2019-07-12T07:52:18 | 2019-07-12T07:52:18 | {} | MEMBER |
ClickHouse | ClickHouse | 510,788,200 | 5,935 | vitlibar | Fixed in master. | 2019-07-12T07:53:25 | 2019-07-12T07:53:25 | {} | MEMBER |
ClickHouse | ClickHouse | 511,528,628 | 6,015 | alexey-milovidov | https://github.com/microsoft/mimalloc/issues/100 | 2019-07-15T19:02:09 | 2019-07-15T19:02:09 | {} | MEMBER |
ClickHouse | ClickHouse | 511,692,985 | 5,999 | MahmoudGoda0 | Could you please elaborate more what should i do to apply this fix in my case ?
Another question, i have around 7800 files, and i need parallel import, how i can do it ? and how i can tell clickhouse to use more cpus ? | 2019-07-16T07:02:59 | 2019-07-16T07:02:59 | {} | NONE |
ClickHouse | ClickHouse | 511,752,686 | 5,815 | alesapin | Packages already signed. Fixes for deps problems https://github.com/yandex/ClickHouse/pull/6023. | 2019-07-16T10:03:43 | 2019-07-16T10:03:43 | {} | MEMBER |
ClickHouse | ClickHouse | 510,400,655 | 5,945 | filimonov | > ```
> std::string userInfo = uri.getUserInfo();
> if (userInfo.front() != ':') {
> char *user = strtok(const_cast<char *> (userInfo.c_str()), ":");
> if (user != NULL) {
> hdfsBuilderSetUserName(builder.get(), user);
> }
> }
> ```
>
> I think this code is goo... | 2019-07-11T09:07:06 | 2019-07-11T09:07:06 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 510,416,808 | 5,954 | filimonov | That is problably one of the most popular issue when start using ClickHouse. When you write something like `expression( col ) as col` it hides original `col` value.
Try to avoid redefining STR_VALUE.
Neverless that low-cardinality incompatibility in that case is rather an issue. | 2019-07-11T09:51:10 | 2019-07-11T09:54:07 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 510,504,323 | 5,975 | blind-oracle | https://github.com/yandex/ClickHouse/commit/4e1286e1e562d7ec3e7ddf51fb982a5b75359bd1
This commit didn't make it to 19.9, needs to be backported. | 2019-07-11T14:15:59 | 2019-07-11T14:15:59 | {
"+1": 1
} | NONE |
ClickHouse | ClickHouse | 510,288,418 | 1,510 | den-crane | @shivacr you can change 300 to 3000 temporarily for maintenance time. 300 it's just a limiter / alerter.
```
<yandex>
<merge_tree>
<parts_to_throw_insert>3000</parts_to_throw_insert>
</merge_tree>
</yandex>
``` | 2019-07-11T01:19:48 | 2019-07-11T01:19:48 | {
"+1": 1
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 510,397,543 | 3,141 | Axelfoley85 | Still observing this issue on 19.7.3.9
```
2019.07.11 10:27:49.763877 [ 290 ] {} <Error> BaseDaemon: (version 19.7.3.9 (official build)) (from thread 8) Received signal Segmentation fault (11).
2019.07.11 10:27:49.763919 [ 290 ] {} <Error> BaseDaemon: Address: 0x7effe17ecdd8
2019.07.11 10:27:49.763937 [ 290 ] {} ... | 2019-07-11T08:59:01 | 2019-07-11T09:25:09 | {} | NONE |
ClickHouse | ClickHouse | 510,417,529 | 5,945 | akonyaev90 | https://github.com/yandex/ClickHouse/pull/5946
Already merged | 2019-07-11T09:53:10 | 2019-07-11T09:53:10 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 510,424,105 | 4,923 | filimonov | @agaurav any news here? | 2019-07-11T10:12:31 | 2019-07-11T10:12:31 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 510,452,320 | 5,975 | citrin | Also it looks like at least in this case time zone could be determined just reading `/etc/localtime` content. No need to look where symlink is pointing to, but probably it will not work for non-UTC timezones. | 2019-07-11T11:46:47 | 2019-07-11T11:48:10 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 510,692,145 | 5,983 | filimonov | > DNS error: EAI: -9
Most problably you have IPv6 disabled. Try to replace listen address from `::` to `0.0.0.0` | 2019-07-11T23:41:08 | 2019-07-11T23:41:21 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 510,752,738 | 5,974 | lordk911 | uname -r
3.10.0-514.el7.x86_64
should be 3.17+ ? | 2019-07-12T05:38:51 | 2019-07-12T05:38:51 | {} | NONE |
ClickHouse | ClickHouse | 511,052,004 | 5,988 | den-crane | >The distributed table docs make it sound like we will have to write to both replicas and the >replicatedmergetree won't copy over the data itself, if you write to one of the replicas but not both.
There are two different approaches
1. no zookeeper. no replicated. internal_replication=false. Distributed do "replica... | 2019-07-12T22:25:13 | 2019-07-12T22:28:34 | {
"+1": 2
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 511,057,663 | 5,997 | Schaefer-Thomas | I only use DateTime and no special compressions:
CREATE TABLE test.tickets_flat (`isid` Int32, `sid` String, `dim_ticket_sid` String, `duration` Float32, `quantity` Int32, `additions` Int32, `inquiries` Int32, `forwards` Int32, `working_duration` Float32, `q_contact` Int32, `registration_duration` Float32, `parking... | 2019-07-12T22:58:12 | 2019-07-12T22:58:12 | {} | NONE |
ClickHouse | ClickHouse | 511,295,751 | 6,002 | nvartolomei | Not sure I understood your question.
By local you mean Replicated*MergeTree table? If yes: https://clickhouse.yandex/docs/en/operations/table_engines/distributed/ see `internal_replication` setting, it should be true.
| 2019-07-15T07:12:30 | 2019-07-15T07:12:30 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 511,374,399 | 5,999 | filimonov | Only workarounds. Like use another format (CSV for example), reformat numbers with some simple regexp, read to temporary table as String, or use clickhouse-local as preprocessor: https://www.altinity.com/blog/2019/6/11/clickhouse-local-the-power-of-clickhouse-sql-in-a-single-command | 2019-07-15T12:03:01 | 2019-07-15T12:03:01 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 511,382,227 | 5,885 | LukeCzmoch | I've modified quantileExact function to be consistent with R-6 from here https://en.wikipedia.org/wiki/Quantile. Could you review and add to next release?
```
#pragma once
#include <Common/PODArray.h>
#include <Common/NaNUtils.h>
#include <Core/Types.h>
#include <IO/WriteBuffer.h>
#include <IO/ReadBuffer.h>
... | 2019-07-15T12:30:06 | 2019-07-15T13:14:44 | {
"+1": 2
} | NONE |
ClickHouse | ClickHouse | 511,415,778 | 5,947 | blind-oracle | @filimonov
My fault. Here are both output and a diff: https://gist.github.com/blind-oracle/7a918e47bc5779f00fdc792c3ace3397
> BTW: does it build without tests?
Yep, it builds fine. | 2019-07-15T14:03:50 | 2019-07-15T14:03:50 | {} | NONE |
ClickHouse | ClickHouse | 510,974,080 | 5,987 | den-crane | because NOW CH reads columns create_table_query engine_full from metadata folder from .sql but other columns from in-memory table. | 2019-07-12T17:47:00 | 2019-07-12T17:48:05 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 511,200,240 | 5,846 | josepowera | We tried to use newest version after fix with #5852 to see if the problem is solved.
Unfortunately the PROBLEM REMAINS OPEN (19.8 works but newer 19.9+ DON'T) .
We tried with both:
-19.10.1.5 (official build) - from official docker build
-19.11.0.830 (official build) from prestable debian repo and after tha... | 2019-07-14T12:52:41 | 2019-07-14T12:52:41 | {} | NONE |
ClickHouse | ClickHouse | 511,378,218 | 5,999 | filimonov | The fix is easy - check here:
https://github.com/yandex/ClickHouse/blob/3c7178a8d1c642b92782e2ae27483e88e14307f0/dbms/src/IO/ReadHelpers.h#L349-L354
But it also require test case. | 2019-07-15T12:16:28 | 2019-07-15T12:16:28 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 511,415,582 | 5,988 | abraithwaite | Questions were answered. Was just hard to find in the docs. Thanks @den-crane and @blinkov for the help! | 2019-07-15T14:03:24 | 2019-07-15T14:03:39 | {} | NONE |
ClickHouse | ClickHouse | 510,341,648 | 5,898 | ucasfl | 已修改 @sundy-li | 2019-07-11T06:00:35 | 2019-07-11T06:00:35 | {} | COLLABORATOR |
ClickHouse | ClickHouse | 510,432,674 | 5,959 | renkun2020 | 
> That sound suspicious. Can you please:
>
> 1. recheck `select * from system.settings where changed=1` (especially block sizes)
> 2. do a `tee` to file in your pipe chain to dump what exactly was pas... | 2019-07-11T10:39:48 | 2019-07-11T10:39:48 | {
"+1": 1
} | NONE |
ClickHouse | ClickHouse | 510,485,459 | 5,947 | blind-oracle | It's kinda strange.
It:
* does not build on `CentOS` using `gcc-8.3.1` from `devtoolkit-8`
* does not build on `CentOS` using `gcc-9.1.0` vanilla built by `build-gcc-from-sources.sh` script
* **builds** on `Ubuntu` using `gcc-8.3.0` from PPA
* **builds** on `Ubuntu` using `gcc-9.1.0` vanilla built by `build-gcc-... | 2019-07-11T13:28:55 | 2019-07-11T13:28:55 | {} | NONE |
ClickHouse | ClickHouse | 510,980,195 | 5,969 | 4ertus2 | Does it depend on data inserted? Could you make a reproducible sample that does not need special data? | 2019-07-12T18:06:25 | 2019-07-12T18:06:25 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 511,290,085 | 5,987 | ddliu | Got it, thank you for your explain. | 2019-07-15T06:46:25 | 2019-07-15T06:46:25 | {} | NONE |
ClickHouse | ClickHouse | 511,475,015 | 6,001 | alexey-milovidov | TODO: Untangle `attachQueryContext` and `attachQuery`. These methods are non-obvious. | 2019-07-15T16:33:29 | 2019-07-15T16:33:29 | {} | MEMBER |
ClickHouse | ClickHouse | 511,537,908 | 6,018 | alesapin | `Merge with master — Cannot fetch mergecommit` is CI error. | 2019-07-15T19:30:36 | 2019-07-15T19:30:36 | {} | MEMBER |
ClickHouse | ClickHouse | 511,583,496 | 5,998 | alexey-milovidov | 00908_long_http_insert - flappy test
Performance test, JSON functions - flap
Performance test, formats - unknown | 2019-07-15T21:49:48 | 2019-07-15T21:49:48 | {} | MEMBER |
ClickHouse | ClickHouse | 510,401,605 | 5,932 | filimonov | > Again started the clickhouse-server as a service and clickhouse not started, also there are no logs generated.
Nothing at all? /var/log/clickhouse-server/clickhouse-server.log should speak a lot unless you disable logging in config.xml.
Does the clickhouse appears in processlist? | 2019-07-11T09:09:39 | 2019-07-11T09:09:39 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 510,491,428 | 5,926 | caueteixeira | Hi @filimonov .
I am using ClickHouse client version 19.9.2.4.
No, I didn't miss a quote. | 2019-07-11T13:44:04 | 2019-07-11T13:44:04 | {} | NONE |
ClickHouse | ClickHouse | 510,765,955 | 5,990 | lordk911 | if I data is like this :
12501,2018-02-21,1062069,hnhys_PZ1A2500351654,801159913,801159913,OTC,儿科用药,儿童胃肠道用药,儿童消化不良/厌食,健儿清解液,2,,,0,,A10U,,A10U,2017-11-27,19.800000,8.300000
Column 0, name: business_id, type: String, parsed text: "12501"
Column 1, name: dates, type: String, ... | 2019-07-12T06:36:48 | 2019-07-12T06:36:48 | {} | NONE |
ClickHouse | ClickHouse | 510,905,352 | 5,988 | blinkov | @abraithwaite it sounds like you are inserting to table with Distributed engine, you can avoid the behaviour you describe in second paragraph by using it only for SELECTs and doing INSERTs directly to shards (tables with ReplicatedMergeTree engine).
> Reading the insert_quorum docs, what happens when the value is 1?... | 2019-07-12T14:22:35 | 2019-07-12T14:22:43 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 511,352,231 | 5,999 | filimonov | Confirmed:
```
clickhouse-client --query='create table x( c1 Nullable(Int32), c2 Nullable(Int32)) Engine=TinyLog'
printf "0123\t01234\n" | clickhouse-client --query='insert into x format TSV' --stacktrace
Code: 27. DB::Exception: Cannot parse input: expected \t before: 123\t01234\n: (at row 1)
Row 1:
Column 0... | 2019-07-15T10:37:43 | 2019-07-15T10:39:03 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 511,752,055 | 6,021 | Akazz | Thank you for reporting this bug. We will get to it as well as to documenting this INSERT capability for MySQL engine | 2019-07-16T10:01:48 | 2019-07-16T10:01:48 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 517,934,942 | 6,166 | alexey-milovidov | > ‘type’ shadows a member
It's not very convenient, because
```
struct C
{
int x;
C(int x) : x(x) {}
};
```
is a common practice.
We also write
```
struct C
{
int x;
C(int x_) : x(x_) {}
};
```
in many places, but it apparently may be more error-prone. | 2019-08-03T15:57:41 | 2019-08-03T15:58:05 | {} | MEMBER |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.