Degraded performance in a Supavisor cluster resulted in elevated connection latency and increased p99 response times between 2026-02-15 22:00 and 2026-02-16 01:00. The problematic node was replaced and performance returned to normal levels.