thevar1able/ClickHouse
1
0
Fork 0
mirror of https://github.com/ClickHouse/ClickHouse.git synced 2024-11-21 23:21:59 +00:00
Code Issues Packages Projects Releases Wiki Activity

Improve 01730_distributed_group_by_no_merge_order_by_long

Browse Source 
Add a negativ case to show the difference w/ and w/o
optimize_aggregation_in_order=1
This commit is contained in:
Azat Khuzhin 2021-08-25 09:05:34 +03:00
parent b6d9fe6f88
commit af0a4d93ba
1 changed files with 4 additions and 3 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

7
tests/queries/0_stateless/01730_distributed_group_by_no_merge_order_by_long.sql
View File

@ -2,7 +2,7 @@ drop table if exists data_01730;
-- does not use 127.1 due to prefer_localhost_replica
select * from remote('127.{2..11}', view(select * from numbers(1e6))) group by number order by number limit 20 settings distributed_group_by_no_merge=0, max_memory_usage='100Mi'; -- { serverError 241 }
select * from remote('127.{2..11}', view(select * from numbers(1e6))) group by number order by number limit 20 settings distributed_group_by_no_merge=0, max_memory_usage='100Mi'; -- { serverError MEMORY_LIMIT_EXCEEDED }
-- no memory limit error, because with distributed_group_by_no_merge=2 remote servers will do ORDER BY and will cut to the LIMIT
select * from remote('127.{2..11}', view(select * from numbers(1e6))) group by number order by number limit 20 settings distributed_group_by_no_merge=2, max_memory_usage='100Mi';
@ -10,11 +10,12 @@ select * from remote('127.{2..11}', view(select * from numbers(1e6))) group by n
-- and the query with GROUP BY on remote servers will first do GROUP BY and then send the block,
-- so the initiator will first receive all blocks from remotes and only after start merging,
-- and will hit the memory limit.
select * from remote('127.{2..11}', view(select * from numbers(1e6))) group by number order by number limit 1e6 settings distributed_group_by_no_merge=2, max_memory_usage='100Mi'; -- { serverError 241 }
select * from remote('127.{2..11}', view(select * from numbers(1e6))) group by number order by number limit 1e6 settings distributed_group_by_no_merge=2, max_memory_usage='100Mi'; -- { serverError MEMORY_LIMIT_EXCEEDED }
-- with optimize_aggregation_in_order=1 remote servers will produce blocks more frequently,
-- since they don't need to wait until the aggregation will be finished,
-- and so the query will not hit the memory limit error.
create table data_01730 engine=MergeTree() order by key as select number key from numbers(1e6);
select * from remote('127.{2..11}', currentDatabase(), data_01730) group by key order by key limit 1e6 settings distributed_group_by_no_merge=2, max_memory_usage='100Mi', optimize_aggregation_in_order=1 format Null;
select * from remote('127.{2..11}', currentDatabase(), data_01730) group by key order by key limit 1e6 settings distributed_group_by_no_merge=2, max_memory_usage='40Mi', optimize_aggregation_in_order=1 format Null; -- { serverError MEMORY_LIMIT_EXCEEDED }
select * from remote('127.{2..11}', currentDatabase(), data_01730) group by key order by key limit 1e6 settings distributed_group_by_no_merge=2, max_memory_usage='50Mi', optimize_aggregation_in_order=1 format Null;
drop table data_01730;