Kubernetes liveness and readiness probes: how to handle unhealthy pods

Two Different Questions About Health

Kubernetes uses two probes to monitor containers. Liveness: is this container still working, or is it deadlocked? If the liveness probe fails, Kubernetes restarts the container. Readiness: is this container ready to accept traffic? If the readiness probe fails, Kubernetes removes the Pod from the Service's endpoints — no traffic is sent to it — but it is not restarted.

Configuring HTTP Probes

containers:
- name: api
  image: my-api:2.0
  livenessProbe:
    httpGet:
      path: /healthz
      port: 8080
    initialDelaySeconds: 10   # wait before first check
    periodSeconds: 10          # check every 10s
    failureThreshold: 3        # restart after 3 consecutive failures
  readinessProbe:
    httpGet:
      path: /ready
      port: 8080
    initialDelaySeconds: 5
    periodSeconds: 5
    failureThreshold: 2

Exec and TCP Probes

# Exec probe: run a command inside the container
livenessProbe:
  exec:
    command: ["sh", "-c", "pg_isready -U postgres"]
  periodSeconds: 15

# TCP probe: check if a port is open
readinessProbe:
  tcpSocket:
    port: 6379
  periodSeconds: 10

Use a startup probe for slow-starting containers (JVM apps, legacy monoliths) to give them time to boot before liveness checks begin. Set startupProbe.failureThreshold high enough to cover your worst-case startup time.