Track connection count against max_connections and capture slow queries (e.g. via pg_stat_statements) with alerting thresholds.
Without visibility into connection saturation and slow queries, incidents surface only as user-facing outages.
BeforeMerge scans your pull requests against this rule and dozens more. Get actionable feedback before code ships.
Connection exhaustion and slow queries are the most common managed-database incidents, and both are detectable in advance. Alert before you hit the connection ceiling and surface the slowest statements so they can be optimized.
-- Bad: no monitoring; learn about saturation from a 500
-- Good: watch usage and slow statements
SELECT count(*) FROM pg_stat_activity; -- vs max_connections
SELECT query, mean_exec_time FROM pg_stat_statements
ORDER BY mean_exec_time DESC LIMIT 10;