Statistics lag causes monitoring to miss active transactions and queries

Statistics views do not update instantaneously. Each server process flushes statistics to shared memory only before going idle, but not more than once per PGSTAT_MIN_INTERVAL (1 second default). Active queries/transactions are not reflected in statistics views, causing displayed totals to lag behind actual activity.

When monitoring active workloads, query pg_stat_activity for real-time current query information (always up-to-date). For cumulative statistics, account for up to 1 second lag in displayed values. Avoid making rapid successive queries to statistics views expecting updated values.