mirror of
https://github.com/ClickHouse/ClickHouse.git
synced 2024-09-21 17:20:50 +00:00
e2960e1a52
In case of Buffer table has columns of AggregateFunction type, aggregate states for such columns will be allocated from the query context but those states can be destroyed from the server context (in case of background flush), and thus memory will be leaked from the query since aggregate states can be shared, and eventually this will lead to MEMORY_LIMIT_EXCEEDED error. To avoid this, prohibit sharing the aggregate states. But note, that this problem only about memory accounting, not memory usage itself. Signed-off-by: Azat Khuzhin <a.khuzhin@semrush.com>
37 lines
1001 B
SQL
37 lines
1001 B
SQL
-- Tags: long
|
|
|
|
drop table if exists buffer_02231;
|
|
drop table if exists out_02231;
|
|
drop table if exists in_02231;
|
|
drop table if exists mv_02231;
|
|
|
|
-- To reproduce leak of memory tracking of aggregate states,
|
|
-- background flush is required.
|
|
create table buffer_02231
|
|
(
|
|
key Int,
|
|
v1 AggregateFunction(groupArray, String)
|
|
) engine=Buffer(currentDatabase(), 'out_02231',
|
|
/* layers= */1,
|
|
/* min/max time */ 86400, 86400,
|
|
/* min/max rows */ 1e9, 1e9,
|
|
/* min/max bytes */ 1e12, 1e12,
|
|
/* flush time */ 1
|
|
);
|
|
create table out_02231 as buffer_02231 engine=Null();
|
|
create table in_02231 (number Int) engine=Null();
|
|
|
|
-- Create lots of INSERT blocks with MV
|
|
create materialized view mv_02231 to buffer_02231 as select
|
|
number as key,
|
|
groupArrayState(toString(number)) as v1
|
|
from in_02231
|
|
group by key;
|
|
|
|
insert into in_02231 select * from numbers(10e6) settings max_memory_usage='300Mi';
|
|
|
|
drop table buffer_02231;
|
|
drop table out_02231;
|
|
drop table in_02231;
|
|
drop table mv_02231;
|