What is the failure mode for Serverless Aurora Postgress if AirByte CDC fails or falls behind in consumption of the WAL?

The knowledge sources only partially cover this, so the answer is limited to what’s documented. For Aurora PostgreSQL specifically, the docs call out a key limitation: Aurora has a CDC caching layer (

rds.logical_wal_cache

) that is incompatible with Airbyte’s CDC. To use Airbyte CDC reliably, that cache must be disabled by setting

rds.logical_wal_cache = 0

in the parameter group.[Postgres troubleshooting] For Postgres CDC in general, when Airbyte “falls behind” and the WAL segments it needs are no longer available (because Postgres/Aurora has purged or archived them), the connector detects that the saved offset is before the replication slot’s

confirmed_flush_lsn

. In that case: • If Invalid CDC Position Behavior is set to Fail sync, the sync fails with an error like: > Saved offset is before replication slot's confirmed lsn. Please reset the connection, and then increase WAL retention and/or increase sync frequency to prevent this from happening in the future.[Pg CDC issue; Postgres troubleshooting] • If it is set to Re-sync data, Airbyte automatically triggers a full refresh / re-snapshot from scratch when it detects that condition.[Pg CDC issue; CDC best practices] So, the effective failure modes when CDC falls behind WAL retention are: 1. Sync fails and stops (manual reset required) — with Fail sync. 2. Silent full re-snapshot of the source (potentially expensive) — with Re-sync data.[CDC best practices] The docs do not describe any Aurora-specific additional failure behavior beyond the incompatibility of the CDC cache layer; they recommend disabling that cache to avoid CDC issues.[Postgres troubleshooting] 👍 0 button 👎 0 button ℹ️ Tip: Mention @kapa.ai in the thread for follow-ups or DM the app for 1-to-1 help.