Design for failover and high availability

Design for failover and high availability | BeforeMerge Rules | BeforeMerge

Why this matters

Managed databases fail — hardware, maintenance, and zone outages happen. A standby with automatic failover keeps you available, but only if clients reconnect to the promoted endpoint. Use the provider's HA endpoint and short-lived connections.

# Bad: single instance, app pins one IP and never retries
 
# Good: HA writer endpoint + reconnect-on-failover
connect(writerEndpoint)  // provider promotes standby transparently
retryOnConnectionError()

Why This Matters

Related Rules

Catch this automatically on every PR

Why this matters