Failover Slots in PostgreSQL-17:
summary
PostgreSQL 17 adds real‑time replication of selected logical slots to the standby so subscribers can resume without full resync; configure slots with failover=true (or pass failover to pg_create_logical_replication_slot), enable standby settings like primary_slot_name and hot_standby_feedback, and run the manual sync function PG_sync_replication_slots or enable the background worker via sync_replication_slots. Sync cycles run at least every 200 ms (backing off to 30 s), synced slots are read‑only on standby and temporary slots are dropped; table copy restarts if it was mid‑copy.
AI-written summary · about
chapters (11)
- 0:00Introduction
- 0:25Logical replication overview
- 9:25Problem during failover
- 12:53Failover slots in PG17
- 14:21Configuring failover slots
- 17:49Manual slot synchronization
- 21:01Automatic slot synchronization
- 24:13Reducing downtime in PG17
- 27:25Verifying failover readiness
- 30:57Summary and limitations
- 31:40Q&A
related talks
43mpg_failover_slots: The missing piece
43mPostgreSQL 17 and beyond
02h 26mTutorial
53mThe journey towards active replication in PostgreSQL
48mReplication Refresh
48mPostgreSQL Replication: 20 Pitfalls and Solutions
51mPostgreSQL Replication Conflicts: Avoiding Pitfalls
42mPostgreSQL Replication Conflicts: Avoiding Pitfalls
