[MDEV-16520] Out-Of-Memory running big aggregate query on Spider Engine - Jira

XML

Word

Printable

When running 6 concurrent queries (2 queries per 3 tables) the mysqld process gets killed by the OOM killer.

The queries was:

select min(id), max(id), avg(id), sum(id), min(id2), max(id2), avg(id2), sum(id2), count(*) FROM t2; -- Finished

select min(id), max(id), avg(id), sum(id), min(id2), max(id2), avg(id2), sum(id2), count(*) FROM t3;

select min(id), max(id), avg(id), sum(id), count(*) FROM t1;

select min(id), max(id), avg(id), bit_xor(id), min(id2), max(id2), avg(id2), bit_xor(id2), count(*) FROM t2; -- Finished

select min(id), max(id), avg(id), bit_xor(id), min(id2), max(id2), avg(id2), bit_xor(id2), count(*) FROM t3;

select min(id), max(id), avg(id), bit_xor(id), count(*) FROM t1;

Where the tables has more than 100M rows.

relates to

MDEV-16249 CHECKSUM TABLE for a spider table is not parallel and saves all data in memory in the spider head by default

MDEV-16279 Spider crashes on CHECKSUM TABLE with spider_quick_mode=3

MDEV-16880 Provide checksum aggregate functions, and partition-level checksums

Error rendering 'com.xiplink.jira.git.jira_git_plugin:git-issue-webpanel'. Please contact your Jira administrators.