Last week I've noticed a sudden spike in the ReadIOPS activity across all our RDS instances. It happened in two batches, actually, but it affected all of them:
Digging into this with the Performance Insights monitoring panel I can tell these spikes correlate with activity on a database called rdsadmin that seems to be present on every RDS instance:
And if I look at the SQL section I see this query being executed at the moment:
SELECT DATNAME,
LARGEST_DB_BYTES,
ALL_DB_BYTES,
OLDEST_XACT_SEC,
OLDEST_XACT_XID
FROM (
SELECT DATNAME,
DATALLOWCONN,
CONNECTIONS,
BYTES,
MAX(BYTES) OVER () LARGEST_DB_BYTES,
SUM(BYTES) OVER () ALL_DB_BYTES,
MAX(SEC) OVER () OLDEST_XACT_SEC,
MAX(XID) OVER () OLDEST_XACT_XID
FROM (
SELECT D.DATNAME,
D.DATALLOWCONN,
PG_CATALOG.PG_DATABASE_SIZE(D.OID) BYTES,
SUBQ_DB_BY_ACTIVITY.CONNECTIONS,
SUBQ_DB_BY_ACTIVITY.SEC,
SUBQ_DB_BY_ACTIVITY.XID
FROM (
SELECT A.DATID,
COUNT(*) CONNECTIONS,
MAX(EXTRACT(EPO...
Does someone know what this query is about and why it triggers across every RDS instance? Does it depend on activity on other databases or is it a RDS scheduled maintenance task? If so, is there a way we can anticipate/delay these spikes so they don't potentially affect our databases' performnace?