HOWTO: Execute a PostgreSQL Failover with Patroni¶

Purpose: Execute a controlled or emergency PostgreSQL failover in a Patroni-managed cluster with structured run-record capture.

Difficulty: Advanced

Track: Disaster Recovery Automation

Overview¶

Both paths (planned switchover and emergency failover) produce the same output: before/after Patroni cluster state, replication lag at the point of promotion, and application reconnection confirmation. Capture these at each step; they form the run record.

1. Pre-Failover State Capture¶

Patroni cluster status snapshot.
Replication lag baseline.
Application connection pool state.

2. Controlled Switchover¶

patronictl -c /etc/patroni/config.yml switchover.
Pre-switchover health check.
Execution and timing.
Post-switchover state confirmation.

3. Emergency Failover¶

Detecting primary failure: Patroni leader election log.
Understanding the DCS-based election process.
Manual trigger if automatic election is inhibited.

4. Post-Failover Validation¶

New primary confirmed in Patroni cluster list.
Replication from old primary (now replica) established.
Application reconnection confirmed.
pgBackRest reconfigured for new primary.

5. Run-record capture and Run Record¶

Storing before/after cluster state under <runtime-root>/logs/dr/postgresql-failover/.
Timing data for RTO measurement.
Linking to the DR drill run record.

References¶

License: MIT-0 for code, CC-BY-4.0 for documentation unless otherwise stated.