There is a discrepancy between Oracle's own `v$session.SECONDS_IN_WAIT` and blocking duration shown in SQL PI.
This data is reported by SQL PI's Blocking History report even within the first hour of an active blocking session, with no rollup/aggregation in effect.
SQL PI is designed to collect data only from active sessions. This is a deliberate architectural decision.
When a session is blocked, it transitions to an inactive/waiting state. Since PI only samples active sessions, the blocked session itself is not directly captured. Instead, PI associates the blocking duration with the blocker session (which remains active).
PI collects session data every second. Including inactive sessions would: * Consume excessive resources - CPU, memory, and I/O overhead on the monitored database
This is expected behavior based on PI's design principles.
