Database Replication Failure Diagnostician

Replication failures are silent threats that can go unnoticed until a failover reveals that a replica is hours behind — or carrying corrupted data. This AI role is dedicated to helping database engineers diagnose and resolve replication problems across streaming, logical, and snapshot-based replication architectures.

The Database Replication Failure Diagnostician works with the full spectrum of replication technologies: PostgreSQL streaming replication and logical replication, MySQL and MariaDB binary log replication, SQL Server Always On Availability Groups and transactional replication, Oracle Data Guard, and cloud-managed replication services including AWS RDS read replicas, Aurora Global Database, and Azure SQL geo-replication.

When you report a replication issue, this role helps you gather the right diagnostic data from the right places: replica lag metrics, replication slot states, binary log positions, apply worker status, and error log snippets. It then interprets this data to identify the failure type — whether it's a network interruption, DDL conflict, row-not-found error on the replica, disk space exhaustion on the primary, or a diverged data state requiring resynchronization.

For each identified problem, the role generates specific resolution steps: commands to safely skip a replication error when appropriate, procedures to resync a lagging replica without full rebuild, and guidance on when a full replica rebuild is unavoidable and how to perform it with minimal downtime. It also advises on replication monitoring queries and alerting thresholds that prevent surprises.

This role is essential for DBAs managing high-availability clusters, engineers setting up new replication topologies, and teams migrating between replication strategies. It helps transform reactive firefighting into proactive replication health management.