Liveness and Readiness Probes

Kubernetes probes are periodic health checks run by the kubelet against each container. The liveness probe determines whether a container is alive: if it fails consecutively (failureThreshold times), the kubelet kills the container and restarts it according to the Pod's restartPolicy. Liveness probes detect deadlocks and infinite loops — scenarios where the process is still running but no longer doing useful work.

The readiness probe determines whether a container is ready to serve traffic: if it fails, the Pod's IP is removed from all matching Service endpoints, stopping new requests from being routed to it until the probe recovers. This enables graceful handling of startup time, dependency availability, and temporary overload — the Pod stays alive but is temporarily taken out of rotation. A third probe type, the startup probe, delays liveness and readiness checks until initial startup completes, preventing premature restarts of slow-starting applications.

Probes can use three mechanisms: httpGet (HTTP GET to a path and port), tcpSocket (TCP connection to a port), or exec (run a command inside the container). Configuration parameters — initialDelaySeconds, periodSeconds, timeoutSeconds, successThreshold, failureThreshold — must be carefully tuned to the actual startup and response time characteristics of the application to avoid false positives that cause unnecessary restarts.