Observability & Maintenance

A liveness probe detects a dead or hung container that is still running but cannot recover on its own. When it fails (beyond the failureThreshold), the kubelet kills and restarts the container according to the Pod's restartPolicy.

A readiness probe determines whether a container is ready to serve traffic. On failure the Pod's IP is removed from the Service's Endpoints/EndpointSlices, so it stops receiving traffic. The container is NOT restarted.

Liveness failure restarts the container; readiness failure stops traffic by removing the Pod from Service endpoints without restarting it.

A startup probe protects slow-starting containers. While it runs, liveness and readiness probes are disabled. Once it succeeds, the other probes take over. This prevents an aggressive liveness probe from killing an app that simply takes a long time to boot.

Multiply failureThreshold by periodSeconds to set the maximum allowed startup time. For example failureThreshold: 30 with periodSeconds: 10 gives the app up to 300 seconds to start before the kubelet kills it.

httpGet (HTTP request, success on 200-399), tcpSocket (TCP connection opens), exec (runs a command, exit code 0 = success), and grpc (gRPC health check on a port).