Comparison
Liveness probevsReadiness probe
Liveness probe
the orchestrator asks whether the process is wedged and restarts the container when it stops answering.
A health check whose failure means "kill and restart this instance". It should test only whether the process itself is stuck, never whether dependencies are reachable. A liveness probe that checks the database will restart every instance in the fleet during a database incident, turning a degradation into an outage.
Full entry →Readiness probe
the instance says "not yet" while the cache warms, so the load balancer keeps traffic off it until it is genuinely able to serve.
A health check whose failure means "stop sending traffic", without restarting the instance. This is the probe that may legitimately consider dependencies, since removing one instance from rotation is recoverable. Conflating it with liveness is one of the most common Kubernetes configuration mistakes.
Full entry →