API response times increased to ~800ms due to a connection pool exhaustion during a traffic spike. No data loss.
Resolved — Pool limits increased, latency back to normal.
Monitoring — Fix deployed, watching metrics.
Investigating — Elevated p99 latency detected.