HOWTO: Run a DR Failback After PostgreSQL Recovery¶
Failback completes the DR loop: after a failover, the original primary must be safely restored as a replica, re-synchronised, and optionally promoted back to primary. This HOWTO covers the full failback sequence in a Patroni-managed cluster: rebuilding the original node via pgBackRest, reattaching it as a streaming replica, and executing a controlled switchback if required. The goal is a clean cluster topology with a complete run-record trail that closes the DR drill record.
What this covers:
- Rebuilding the original primary as a replica using
pgbackrest restorewith Patroni integration - Replica re-synchronisation validation: replication lag, WAL catch-up, and Patroni member state
- Optional controlled switchback: patronictl switchover to restore original primary role with matching run records