Kubernetes health checks answer three different questions

Kubernetes separates startup, liveness and readiness checks because they trigger different actions. Startup checks give initialization time before other probes run. Liveness failures can restart a container, while readiness failures remove a Pod from matching Services’ traffic destinations. Understanding the distinction helps engineers choose whether an unhealthy application needs more time, less traffic or a restart.

Kubernetes Documentation

Why it matters

Useful when debugging deployment restart loops or Pods that appear running but do not receive traffic.

More context

Evergreen documentation; no original publication date is verified. Read the Container probes section, then compare each check in one deployment with the action it is supposed to trigger. Test slow startup and temporary unavailability separately.

Keep in mind

Probe results only reflect the checks you configure. Failure thresholds, timeouts and restart policy matter; this guide is not a diagnosis of a particular cluster.

Original read