Error in percona logs - pg_stat_monitor: pg: out of memory

[ANDiv1976]

The problem of running out of memory is due to the fact that pg_stat_monitor does not collapse the same type of queries.

Pg_stat_monitor 2.2 - Collapsing similar queries

Why doesn’t pg_stat_monitor 2.2 collapse such similar queries? Under high TPS, all buckets get filled with identical queries. As a result, it starts consuming more shared memory. It can collapse similar queries. In pg_stat_monitor, during just a few seconds of testing, there are 47,000 rows, with a limit of 20 buckets and pgsm_max of 192MB. [изображение] Is there a way to collapse similar queries? Which parameter in pg_stat_monitor is needed for this? I couldn’t find it in the documentation. …