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 | 701,291,854 | 15,448 | alesapin | I hate TTL-move tests, but this one failure seems to be related... | 2020-09-30T10:00:55 | 2020-09-30T10:00:55 | {} | MEMBER |
ClickHouse | ClickHouse | 701,539,546 | 15,474 | den-crane | @John-belt Do you use k8s kubernetes ? It looks like you run CH in some docker or kubernetes? | 2020-09-30T17:39:54 | 2020-09-30T17:39:54 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 702,099,218 | 15,510 | den-crane | No way to distinguish in TSV and CSV and JSONEachRow what is "1" enum_value or enum_id
https://github.com/ClickHouse/ClickHouse/issues/7971#issuecomment-624222686
https://github.com/ClickHouse/ClickHouse/issues/10682 | 2020-10-01T12:26:14 | 2020-10-01T12:26:14 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 702,236,987 | 15,314 | stcharitak | Unfortunately, I am still unable to make fast SELECT queries after 150.000.000 inserts. I have the following table:
```
CREATE TABLE default.products_sharded
(
`request_time` DateTime DEFAULT now(),
`un_id` UUID,
`organization_id` LowCardinality(String),
`investor` String,
`provider` Strin... | 2020-10-01T16:04:23 | 2020-10-01T16:04:23 | {} | NONE |
ClickHouse | ClickHouse | 702,250,836 | 5,300 | 4ertus2 | Reproduced on master
```
CREATE TABLE currencyRateShard (`date` Date, `base` FixedString(3)) ENGINE = MergeTree ORDER BY (date);
CREATE TABLE currencyRate AS currencyRateShard ENGINE = Distributed(test_cluster_two_shards_localhost, currentDatabase(), currencyRateShard);
INSERT INTO currencyRateShard VALUES (0, 'a... | 2020-10-01T16:26:44 | 2020-10-01T16:26:44 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 701,133,426 | 15,449 | amosbird | https://github.com/ClickHouse/ClickHouse/blob/master/src/Storages/MergeTree/MergeTreeData.cpp#L413 Here we only consider physical columns. It's easy to extend. | 2020-09-30T03:21:01 | 2020-09-30T03:21:01 | {} | COLLABORATOR |
ClickHouse | ClickHouse | 701,222,775 | 15,461 | filimonov | When version need to be changed with keepeing possibility to use tracelog:
```sql
CREATE TABLE system.addresses_dict
(
`revision` UInt32,
`address` UInt64,
`line` String,
`symbol` String,
`demangle` String
)
ENGINE = MergeTree
ORDER BY (revision, address)
SETTINGS index_granularity = 128... | 2020-09-30T07:46:22 | 2020-09-30T07:49:57 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 701,254,821 | 4,282 | filimonov | I believe it was caused by Kafka issue
```
Thread 30 (Thread 0x7f62eccff700 (LWP 851)):
#0 0x00007f690b824f57 in pthread_join () from /lib64/libpthread.so.0
#1 0x000000000651c005 in thrd_join ()
#2 0x000000000649585f in rd_kafka_destroy ()
#3 0x0000000005005182 in DB::StorageKafka::shutdown() ()
#4 0x00000... | 2020-09-30T08:48:50 | 2020-09-30T08:48:50 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 701,413,394 | 15,420 | den-crane | hmm, try to use a new create syntax instead of obsoleted.
>ENGINE = ReplacingMergeTree (DATE,(daemon,server,port),8192,server_time)
ENGINE = ReplacingMergeTree(server_time) partition by toYYYYMM(date) order by (daemon,server,port) settings index_granularity=8192 | 2020-09-30T14:05:44 | 2020-09-30T14:06:35 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 701,443,741 | 15,420 | filimonov | Maybe some weird filesystem in your VM?
How fast can you execute the command
`ls -r /var/lib/clickhouse/data`? | 2020-09-30T14:54:18 | 2020-09-30T14:54:18 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 701,509,556 | 15,474 | John-belt | [new_log.txt](https://github.com/ClickHouse/ClickHouse/files/5307001/new_log.txt)
Attached the log | 2020-09-30T16:43:57 | 2020-09-30T16:44:08 | {} | NONE |
ClickHouse | ClickHouse | 701,532,830 | 15,474 | den-crane | > "Failed to get size of directory DB_1: lstat /bp2/data/data/DB_1/tempo_for_ls/tmp_insert_20200921_20200921_92285_92285_0: no such file or directory
It does not look like CH logs. And I see weird errors. Probably because of K8s. | 2020-09-30T17:27:15 | 2020-09-30T17:27:15 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 701,541,016 | 15,474 | John-belt | /bp2 is the naming convention we use for directory in our prod servers. This is where all the clickhouse data, metadata, schema are present. | 2020-09-30T17:42:37 | 2020-09-30T17:44:15 | {} | NONE |
ClickHouse | ClickHouse | 701,810,171 | 15,486 | den-crane | It's a hell of the work to support all cases including all 30 Februaries cases and 5/6 days weeks.
As I see even PG does not support Russia calendar shift.
```
set time zone 'Europe/Moscow';
select timestamp with time zone '1918-02-14 00:00:00.000000+00' - '3 days'::interval;
?column?
-------------... | 2020-10-01T03:05:23 | 2020-10-01T03:06:27 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 702,060,886 | 15,087 | qoega | > In production environment, hundreds of machines hold thousands of ReplicatedMergeTree tables.
> The only zookeeper cluster is seriously overloaded even good performance machine used.
> In some scenarios, we want to change the current zookeeper cluster to another one smoothly.
I guess it will be very useful if so... | 2020-10-01T11:07:26 | 2020-10-01T11:07:26 | {
"+1": 1
} | MEMBER |
ClickHouse | ClickHouse | 702,101,130 | 15,504 | den-crane | It seems you have or had a database named "store".
Can you show a result of SQL **show databases** | 2020-10-01T12:29:57 | 2020-10-01T12:29:57 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 702,240,899 | 15,505 | alesapin | Failure not related to changes | 2020-10-01T16:10:41 | 2020-10-01T16:10:41 | {} | MEMBER |
ClickHouse | ClickHouse | 702,602,362 | 15,504 | Teeeto |
> It's not documented for now, but `store/` directory contains data and metadata of all tables in all databases with `ENGINE=Atomic`. Database `system` is created with Atomic engine by default since 20.8. However, `system` database should not produce so many subfolders and files in `store/`.
>
> Which ClickHou... | 2020-10-02T08:37:48 | 2020-10-02T08:37:48 | {} | NONE |
ClickHouse | ClickHouse | 702,617,689 | 15,428 | amosbird | @haoch missing tests. You can refer to https://github.com/ClickHouse/ClickHouse/blob/master/tests/queries/0_stateless/00732_quorum_insert_select_with_old_data_and_without_quorum_zookeeper.sql and write a simple test to ensure it works. | 2020-10-02T09:12:18 | 2020-10-02T09:12:18 | {} | COLLABORATOR |
ClickHouse | ClickHouse | 701,276,864 | 14,414 | 4ertus2 | Stress (thread)
https://clickhouse-test-reports.s3.yandex.net/15242/0db8b6d647b67ffb480790626015a13b2b929035/stress_test_(thread).html#fail1 | 2020-09-30T09:30:27 | 2020-09-30T09:30:27 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 701,307,677 | 15,354 | zhang2014 | Can you provide more information(for example: clickhouse log file)? it seems should work fine(county column is synchronized in my environment). | 2020-09-30T10:34:23 | 2020-09-30T10:34:38 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 701,358,074 | 15,424 | CurtizJ | Test `01482_move_to_prewhere_and_cast` actually shows a bug, that is not fixed yet. See https://github.com/ClickHouse/ClickHouse/issues/10694#issuecomment-701357258. | 2020-09-30T12:28:16 | 2020-09-30T12:28:16 | {} | MEMBER |
ClickHouse | ClickHouse | 701,548,425 | 15,474 | John-belt | It is a data volume.
| 2020-09-30T17:57:15 | 2020-09-30T17:57:15 | {} | NONE |
ClickHouse | ClickHouse | 701,577,649 | 13,597 | Evgeniy-KH | You need to know the binutils version
changed: find_program (OBJCOPY_PATH NAMES "llvm-objcopy" "llvm-objcopy-10" "llvm-objcopy-9" "llvm-objcopy-8" "objcopy" PATHS "/usr/local/Cellar/binutils/**2.35.1**/bin/")
This command became working for me
| 2020-09-30T18:51:24 | 2020-09-30T18:52:30 | {} | NONE |
ClickHouse | ClickHouse | 701,580,774 | 15,420 | OmarBazaraa | > Maybe some weird filesystem in your VM?
> How fast can you execute the command
> `ls -r /var/lib/clickhouse/data`?
It returns instantaneously. | 2020-09-30T18:57:11 | 2020-09-30T18:57:11 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 701,645,128 | 5,293 | alexey-milovidov | @spanningwings
Yes, it will be very appreciated.
This task has low priority but users need it. | 2020-09-30T21:04:49 | 2020-09-30T21:05:17 | {} | MEMBER |
ClickHouse | ClickHouse | 701,973,818 | 15,490 | 4ertus2 | No backport. Need TreeRewriter rename. It would be ported to 20.3 manually. | 2020-10-01T08:20:38 | 2020-10-01T08:20:38 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 690,155,875 | 12,501 | NiyazNz | Also ran into this issue. After submitting the update/delete query, I found it to be in `system.mutations`. But mutation cannot be applied succesfully and the server crushes and shuts downs with error code:
```
clickhouse_1 exited with code 139
```
Has been tested on versions 20.6.3.28, 20.7. *, 20.8.2.3-latest wi... | 2020-09-10T10:58:46 | 2020-10-01T08:50:26 | {} | NONE |
ClickHouse | ClickHouse | 702,607,513 | 15,504 | Teeeto | > It's not documented for now, but `store/` directory contains data and metadata of all tables in all databases with `ENGINE=Atomic`.
Obviously data is not there, since there s is another directory called "data" which contains the DB data (60 GB compressed)
Also to quantify "a lot" from my previous comment - there ... | 2020-10-02T08:49:37 | 2020-10-02T08:49:53 | {} | NONE |
ClickHouse | ClickHouse | 701,377,375 | 15,464 | den-crane | Parquet is excluded from MacOS build, and mysql.
Check make file.
You can use dockerized `clickhouse-client` as well.
docker run -it --rm --link some-clickhouse-server:clickhouse-server yandex/clickhouse-client --host clickhouse-server | 2020-09-30T13:04:54 | 2020-09-30T13:04:54 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 701,378,535 | 15,306 | den-crane | Please don't close it. It's still a bug in arrayEnumerate and it should be addressed. | 2020-09-30T13:06:59 | 2020-09-30T13:06:59 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 701,429,266 | 15,408 | qoega | Now we have to fix them :) | 2020-09-30T14:31:09 | 2020-09-30T14:31:09 | {} | MEMBER |
ClickHouse | ClickHouse | 701,429,594 | 14,414 | qoega | #15475
Stress undefined | 2020-09-30T14:31:40 | 2020-09-30T14:32:00 | {} | MEMBER |
ClickHouse | ClickHouse | 701,447,935 | 14,693 | alexey-milovidov | Not necessary... I think it will be too much burden. | 2020-09-30T15:01:01 | 2020-09-30T15:01:01 | {
"+1": 1
} | MEMBER |
ClickHouse | ClickHouse | 701,624,951 | 5,293 | spanningwings | Is this issue still need a fix? It's been a long time. Sounds like something that I can help to get me familiar with the ClickHouse dev process. Please drop me a message. | 2020-09-30T20:24:42 | 2020-09-30T20:24:42 | {} | NONE |
ClickHouse | ClickHouse | 701,950,044 | 15,474 | John-belt | @den-crane
I ran 'select * from system.metrics LIMIT 1;'
Number of executing queries was 8, this probably means 8 concurrent queries are running which might be bottlenecking??
Ran query on metric_log and the result is attached below :
 -- I am not joking.
cat abc.tsv
1 4
What is "4" in this case? | 2020-10-01T12:42:03 | 2020-10-01T12:42:03 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 702,247,304 | 15,444 | akuzm | Stress tests broken in 20.7 | 2020-10-01T16:20:54 | 2020-10-01T16:20:54 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 701,111,788 | 15,127 | alexey-milovidov | We have discussed it and the variant with copying the list of blocks to readers looks Ok. | 2020-09-30T01:54:05 | 2020-09-30T01:54:05 | {} | MEMBER |
ClickHouse | ClickHouse | 701,217,574 | 15,461 | filimonov | OK, i see `revision` there, is it enough to idenify the version?
| 2020-09-30T07:36:02 | 2020-09-30T07:36:02 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 701,384,499 | 15,377 | cagev | documentation | 2020-09-30T13:17:49 | 2020-09-30T13:17:49 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 701,449,004 | 15,477 | alesapin | Previously it worked because we can have multiple leaders and all replicas finally get here https://github.com/ClickHouse/ClickHouse/pull/15477/files#diff-2f1161f797903e43711d21fd16137b00R930. It's extremely hard to add a test for this, but we already have one for long-running `ON CLUSTER` queries https://github.com/Cl... | 2020-09-30T15:02:44 | 2020-09-30T15:02:44 | {} | MEMBER |
ClickHouse | ClickHouse | 701,972,218 | 15,487 | 4ertus2 | Build error in GatherUtils. It's probably OOM.
```
2020-09-30 23:48:49 {standard input}: Assembler messages:
2020-09-30 23:48:49 {standard input}:1445734: Warning: end of file not at end of a line; newline inserted
2020-09-30 23:48:49 {standard input}:1446162: Error: unknown pseudo-op: `.cfi_def_cfa_'
2020-09-30 2... | 2020-10-01T08:17:35 | 2020-10-01T08:17:35 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 702,102,996 | 15,510 | Grian | Why? You can create hash table of enum values + enum numbers:
```
{
'1': 1,
'a': 1,
'2': 2,
'b': 2,
...
}
```
And parse as string values? | 2020-10-01T12:33:35 | 2020-10-01T12:33:35 | {} | NONE |
ClickHouse | ClickHouse | 702,214,458 | 15,479 | akuzm | Turns out shellcheck doesn't work in CI (never did, a PR that tried to add it didn't complete the job). It does work if you run it locally, so I'll merge. | 2020-10-01T15:27:50 | 2020-10-01T15:27:50 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 702,231,672 | 15,516 | nvartolomei | The error message before that was still hard to understand and required digging through the code.
```sql
CREATE TABLE mt (v UInt8) ENGINE = ReplicatedMergeTree('/clickhouse/tables/test_01497/mt') order by tuple()
```
```
TCPHandler: Code: 62, e.displayText() = DB::Exception: Macro 'uuid' and empty arguments of... | 2020-10-01T15:55:40 | 2020-10-01T15:58:25 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 701,119,236 | 15,458 | alexey-milovidov | But we already have this option. Just specify all servers as replicas in your cluster configuration and don't specify `<internal_replication>true</internal_replication>`. | 2020-09-30T02:23:53 | 2020-09-30T02:23:53 | {} | MEMBER |
ClickHouse | ClickHouse | 701,397,169 | 15,474 | den-crane | it could be merges or `parts_to_delay_insert`
## parts\_to\_delay\_insert {#parts-to-delay-insert}
If the number of active parts in single partition exceeds the value of `parts_to_delay_insert` ` INSERT` artificially slows down.
Possible values:
- Any positive integer.
Default value: 150.
ClickH... | 2020-09-30T13:39:17 | 2020-09-30T13:39:17 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 701,543,295 | 15,474 | den-crane | No, that's not what I am asking.
Can you show compose or docker command.
What is bp2 ?
* data volume
* mounted host directory
* network attached | 2020-09-30T17:47:05 | 2020-09-30T17:47:05 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 701,702,400 | 15,319 | bharatnc | Also I don't think Stress test failure is related ? I can rebase against master again if needed. | 2020-09-30T23:39:41 | 2020-09-30T23:39:41 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 702,199,954 | 14,448 | genzgd | We are also seeing weird issues with uniqCombined -- merges on a uniqCombinedState seem to break and lead to unmerged parts. We're running 20.8.2.3 | 2020-10-01T15:06:20 | 2020-10-01T15:55:40 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 702,314,222 | 15,504 | tavplubix | It's not documented for now, but `store/` directory contains data and metadata of all tables in all databases with `ENGINE=Atomic`. Database `system` is created with Atomic engine by default since 20.8. However, `system` database should not produce so many subfolders and files in `store/`.
Which ClickHouse version d... | 2020-10-01T18:20:40 | 2020-10-01T18:28:52 | {} | MEMBER |
ClickHouse | ClickHouse | 702,658,002 | 15,454 | KochetovNicolai | I suppose we don't need a new parameter for dictionary. It will only make things even more complex ...
Now pre-allocation is useful almost always.
With an exception that we could insert into dictionary a lot of equal keys (every new equal key will be skipped, so hash table will be pre-allocated too much). But, I sup... | 2020-10-02T10:37:48 | 2020-10-02T10:37:48 | {} | MEMBER |
ClickHouse | ClickHouse | 701,137,009 | 15,456 | alexey-milovidov | > Stress test (thread) — No errors found
Maybe just a coincidence. | 2020-09-30T03:34:56 | 2020-09-30T03:34:56 | {} | MEMBER |
ClickHouse | ClickHouse | 701,240,775 | 15,435 | tavplubix | Integration tests (release) - test_always_fetch_merged/test.py::test_replica_always_download - flacky | 2020-09-30T08:21:13 | 2020-09-30T08:21:13 | {
"+1": 1
} | MEMBER |
ClickHouse | ClickHouse | 701,357,258 | 10,694 | CurtizJ | Unfortunately, it was not fixed. For some period after enabling compact parts by default, stateless tests ran with this type of parts only, but prewhere doesn't work with them and it could seem, that issue is fixed. With setting `min_bytes_for_wide_part = 0` added test fails. [Report](https://clickhouse-test-reports.s3... | 2020-09-30T12:26:43 | 2020-09-30T12:29:04 | {} | MEMBER |
ClickHouse | ClickHouse | 701,380,733 | 15,464 | pan3793 | > Parquet is excluded from MacOS build, and mysql.
> Check make file.
>
> You can use dockerized `clickhouse-client` as well.
> docker run -it --rm --link some-clickhouse-server:clickhouse-server yandex/clickhouse-client --host clickhouse-server
Thanks for explanation, it works in docker. | 2020-09-30T13:10:58 | 2020-09-30T13:11:44 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 701,435,380 | 15,242 | 4ertus2 | There's the same stress test error in not related PR based on the same master: #15472 | 2020-09-30T14:40:46 | 2020-09-30T14:40:46 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 701,499,368 | 15,461 | alexey-milovidov | Revision is enough to distinguish two versions. | 2020-09-30T16:25:19 | 2020-09-30T16:25:19 | {} | MEMBER |
ClickHouse | ClickHouse | 701,311,876 | 15,306 | cybertjofi | Den, this solved our problem:
select .... partner_id, client_id, case_id
from table array join partner_ids as partner_id, client_ids as client_id, case_ids as case_id
Thank you very much! | 2020-09-30T10:44:19 | 2020-09-30T10:44:19 | {} | NONE |
ClickHouse | ClickHouse | 701,403,589 | 15,420 | OmarBazaraa | I also created the following table but didn't include its logs:
```
CREATE TABLE default.internal_logs (
date Date DEFAULT toDate(time),
time DateTime,
server String,
port Int32,
type String,
table String,
volume Int64,
message String,
content String,
source String,
server_time Date... | 2020-09-30T13:50:14 | 2020-09-30T13:50:14 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 702,022,936 | 14,741 | KochetovNicolai | `Performance — 27 faster, 254 unstable`
I have no idea why, but I like it. | 2020-10-01T09:50:24 | 2020-10-01T09:50:24 | {} | MEMBER |
ClickHouse | ClickHouse | 702,576,177 | 15,463 | wutchzone | Thank you for your answer. If you do not mind I have one more question.
This is my table which calculates percentile from the raw data
```
CREATE MATERIALIZED VIEW rtt_99p_graph_second
ENGINE = ReplicatedAggregatingMergeTree(...)
PARTITION BY metric_date ORDER BY (metric_date, metric_time, machine_name)
TTL met... | 2020-10-02T07:35:01 | 2020-10-02T07:35:01 | {} | NONE |
ClickHouse | ClickHouse | 701,540,462 | 15,474 | den-crane | @John-belt What is /bp2 ? ` /bp2/data/data/DB_1/tempo_for_ls/ ) `
How /bp2 attached into docker ? | 2020-09-30T17:41:34 | 2020-09-30T17:41:34 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 701,887,632 | 15,127 | ucasfl | If implement multiversion storage, do we still need table-level locks for storagememory? It doesn't seem to be needed anymore?@alexey-milovidov
| 2020-10-01T05:07:54 | 2020-10-01T05:07:54 | {} | COLLABORATOR |
ClickHouse | ClickHouse | 701,927,688 | 15,477 | alesapin | Build check ok | 2020-10-01T06:51:37 | 2020-10-01T06:51:37 | {} | MEMBER |
ClickHouse | ClickHouse | 702,402,695 | 15,420 | filimonov | > > `ls -r /var/lib/clickhouse/data`?
>
> It returns instantaneously.
After server started? | 2020-10-01T21:16:34 | 2020-10-01T21:16:34 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 701,312,155 | 14,605 | zhang2014 | Maybe we can close this issue?
CC: @filimonov @tavplubix | 2020-09-30T10:44:57 | 2020-09-30T10:45:18 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 701,312,695 | 14,795 | zhang2014 | Maybe we can close this issue? | 2020-09-30T10:46:13 | 2020-09-30T10:46:13 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 702,085,329 | 15,442 | alesapin | Tests OK, stress failure is known. | 2020-10-01T11:59:12 | 2020-10-01T11:59:12 | {} | MEMBER |
ClickHouse | ClickHouse | 701,177,664 | 15,373 | dependabot-preview[bot] | Superseded by #15460. | 2020-09-30T06:09:02 | 2020-09-30T06:09:02 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 701,478,781 | 15,474 | John-belt | Changed 'parts_to_delay_insert' from 150 to 200 in config.xml(under <mergetree> tag), restarted the server
Now 'select count(*) from system.parts;' => return 356
and then 'select count(*) from system.parts where active = 1;' => return 54
'select * from system.merges' => return 0
No performance boost | 2020-09-30T15:50:15 | 2020-09-30T15:51:30 | {} | NONE |
ClickHouse | ClickHouse | 702,121,922 | 15,424 | akuzm | (copying what I said on TG)
This seems to be incompatible with parallel testing. Maybe we can just randomly change default settings once a second, while running the tests normally? This would be more stress-test-like, so maybe it belongs there.
Another option is to add this to `CREATE` queries while fuzzing. This... | 2020-10-01T13:07:25 | 2020-10-01T13:07:25 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 702,212,544 | 15,510 | Grian | Obviously, you should prefer string values before numbers, because
You may want do
```
$ house --query "select * from foo FORMAT TSV | house "insert into bar FORMAT TSV".
```
So 4 ='3'.
And hashtable for parsing should be like
```
{
'1': 2,
'33': 2,
'4': 3,
'11': 4,
'2': 2,
'3': 3, // But ... | 2020-10-01T15:24:52 | 2020-10-01T15:33:08 | {} | NONE |
ClickHouse | ClickHouse | 702,518,981 | 15,428 | haoch | Thanks @amosbird for approval. I made one more small change to fix the compiling. | 2020-10-02T04:29:02 | 2020-10-02T04:29:02 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 701,360,971 | 15,457 | nikitamikhaylov | Hello, if I understood you correctly, you need to copy your historical data from single server to a cluster of several shards? It could be done with [copier](https://altinity.com/blog/2018/8/22/clickhouse-copier-in-practice). You have to define your cluster and choose the sharding key. | 2020-09-30T12:34:08 | 2020-09-30T12:34:08 | {} | MEMBER |
ClickHouse | ClickHouse | 701,380,097 | 14,693 | hustnn | > https://zookeeper.apache.org/doc/r3.6.1/zookeeperProgrammers.html#ch_zkWatches
>
> > With regard to watches, ZooKeeper maintains these guarantees:
> >
> > * Watches are ordered with respect to other events, other watches, and asynchronous replies.
>
> Ok.
@alexey-milovidov Should we add some validate logi... | 2020-09-30T13:09:50 | 2020-09-30T13:10:14 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 701,493,986 | 15,474 | den-crane | Can you provide a clipping from clickhouse-server.log around that long insert ? | 2020-09-30T16:15:43 | 2020-09-30T16:15:43 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 701,702,135 | 15,319 | bharatnc | @filimonov who can help merge this?
cc: @alexey-milovidov a few fixes to my previous PR #14509 | 2020-09-30T23:38:44 | 2020-09-30T23:38:44 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 701,706,171 | 15,428 | haoch | @amosbird I have added query context in `totalRows` and `totalBytes` which may cause the above problem and supported correct`select_sequential_consistency` behavior when used in query like `count(*)` and `system.tables`. Please help review it again. | 2020-09-30T23:54:43 | 2020-09-30T23:55:12 | {
"+1": 1
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 702,217,282 | 15,510 | den-crane | My opinion is irrelevant.
Check this comment https://github.com/ClickHouse/ClickHouse/issues/7971#issuecomment-624222686 | 2020-10-01T15:32:26 | 2020-10-01T15:32:26 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 701,538,592 | 15,474 | John-belt | @den-crane 'tempo_for_ls' is my table name. These are the snippet from the logs in /var/log/clickhouse-server/clickhouse-server.log. '/bp2/data' is my docker volume and DB_1 is my database. | 2020-09-30T17:38:04 | 2020-09-30T17:40:02 | {} | NONE |
ClickHouse | ClickHouse | 701,974,651 | 15,488 | 4ertus2 | Build error:
```
{standard input}: Assembler messages:
{standard input}:146228: Warning: end of file not at end of a line; newline inserted
{standard input}:147028: Error: symbol `.LVU2' is already defined
{standard input}: Error: open CFI at the end of file; missing .cfi_endproc directive
g++-9: fatal error: Kil... | 2020-10-01T08:22:12 | 2020-10-01T08:22:12 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 702,569,774 | 15,518 | alesapin | Closed in favor of https://github.com/ClickHouse/ClickHouse/pull/15239 | 2020-10-02T07:17:42 | 2020-10-02T07:17:42 | {} | MEMBER |
ClickHouse | ClickHouse | 702,583,503 | 15,007 | azat | >Integration tests (release) — fail: 1, passed: 901, error: 0
`test_distributed_over_live_view/test.py::TestLiveViewOverDistributedSuite::test_distributed_over_live_view_order_by_node[lv_over_distributed_table-node0]` is flacky (fixed by increasing buffer size for now)
Apart from this, other tests are ok (except fo... | 2020-10-02T07:53:28 | 2020-10-02T07:53:28 | {} | MEMBER |
ClickHouse | ClickHouse | 701,127,348 | 15,458 | hczhcz | The context is, sometimes people wants to use sharded tables and copied tables together with simple setup. | 2020-09-30T02:57:13 | 2020-09-30T02:57:13 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 701,311,097 | 15,146 | zhang2014 | Maybe you are using `SELECT COUNT() FROM table_name` to determine if the data was deleted?
if so, it is a bug about using `count` queries. it was fixed in #15471.
if not, could you describe more detail for this problem? | 2020-09-30T10:42:28 | 2020-09-30T10:42:28 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 701,342,645 | 15,376 | excitoon | The bug turned out to be even bigger thing than just wrong reading of argument - it was not used afterwards. | 2020-09-30T11:55:20 | 2020-09-30T11:55:20 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 701,375,652 | 15,463 | den-crane | Yes. It's possible but it depends on CH version.
In the latest version you need to use `from .inner` if the first MV created with implicit storage and `from _{to}_table` if the first MV created with explicite storage.
https://github.com/ClickHouse/ClickHouse/issues/7195#issuecomment-538397320 | 2020-09-30T13:01:43 | 2020-09-30T13:01:43 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 701,395,941 | 15,420 | OmarBazaraa | The strange thing is that the issue happens only with this `daemon_heartbeat` table.
You can see in the logs that I was able to create other tables instantaneously before `daemon_heartbeat`. | 2020-09-30T13:37:11 | 2020-09-30T13:37:11 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 702,215,438 | 15,479 | akuzm | Fuzzer failure is known https://github.com/ClickHouse/ClickHouse/issues/15358 | 2020-10-01T15:29:23 | 2020-10-01T15:29:23 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 701,128,224 | 15,458 | alexey-milovidov | Yes, this is quite popular usecase. For example, people may want some dictionary table to be copied across all servers of the cluster.
Usually they just copy-paste cluster definition under different name with all servers as replicas.
Your feature will make this usecase more convenient... it looks maybe a little b... | 2020-09-30T03:00:44 | 2020-09-30T03:00:44 | {} | MEMBER |
ClickHouse | ClickHouse | 701,302,979 | 15,471 | zhang2014 | CC: @tavplubix | 2020-09-30T10:24:00 | 2020-09-30T10:24:00 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 701,400,716 | 15,420 | den-crane | I see only
`CREATE TABLE IF NOT EXISTS default.internal_logs_buffer AS default.internal_logs ENGINE = Buffer`
It's an `ENGINE = Buffer` it does not have disk storage.
And it's happen right after a CH server start? | 2020-09-30T13:45:25 | 2020-09-30T13:45:25 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 701,539,876 | 15,474 | John-belt | yup running inside docker container | 2020-09-30T17:40:28 | 2020-09-30T17:40:44 | {} | NONE |
ClickHouse | ClickHouse | 701,729,652 | 5,293 | spanningwings | Thanks. I will tackle this issue once I set up my environment. | 2020-10-01T01:13:43 | 2020-10-01T01:13:43 | {} | NONE |
ClickHouse | ClickHouse | 699,607,458 | 15,173 | keliss | Do I understand correctly that ALTER queries used to lock the table in versions prior to 20.6, and it could cause the deadlock that we observed? The issue I'm interested in is not the alert, but the cause itself. | 2020-09-27T09:00:00 | 2020-09-27T09:23:40 | {} | NONE |
ClickHouse | ClickHouse | 699,747,115 | 10,954 | andyyzh | > select * from **clusters**(<you_cluster>, view(
> WITH dictGet('dict', 'version', toUInt64(0)) AS _ver SELECT count(*) AS cnt FROM mytable PREWHERE version = _ver
> ))
The **clusters** function needs to be modified to **cluster** | 2020-09-28T03:19:27 | 2020-09-28T03:19:27 | {
"+1": 1
} | CONTRIBUTOR |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.