Monitoring_

Every native MySQL database ships with programmatic health checks, and the engine's own instrumentation is fully available to you. There is nothing to install in the database for these checks.

Database health

Poll the database status for live health information after the database is ready: uptime, connection counts, replica state, and storage volumes. Use it from your own monitoring:

For the database lifecycle state (ready, scaling, restoring, and friends), read the status field of the database object itself. Use that field from deploy pipelines to wait for the database; see lifecycle states.

Inspect activity in the engine

MySQL's own instrumentation works unmodified. From the mysql client or any driver:

-- live connections and their current statements
SELECT id, user, host, db, command, time, state, info
FROM performance_schema.processlist;

-- kill a runaway connection
KILL <id>;

-- execution plan for a slow query
EXPLAIN ANALYZE SELECT ...;

A healthy OLTP database keeps its working set in memory. If throughput drops while disk reads climb, move up a specification so the buffer pool fits the working set.