Why our liveness probe timed out even though /healthCheck never touched Redis

The alert looked like a dependency outage. Readiness and liveness on our Java service were failing in bursts — context deadline exceeded (Client.Timeout exceeded while awaiting headers) — and around t