Pod Lifecycle

A Pod moves through phases such as Pending, Running, Succeeded, and Failed during its life.

Every Pod has a phase that summarizes where it is in its life. You see it in the STATUS column of kubectl get pods.

The phases

  • Pending — accepted by the cluster but not yet running (e.g. still pulling the image).
  • Running — bound to a node and at least one container is live.
  • Succeeded — all containers finished successfully and won't restart.
  • Failed — all containers stopped and at least one failed.
  • Unknown — the node stopped reporting.

Restart policy

The restartPolicy field controls what happens when a container exits: Always (default), OnFailure, or Never. A crashing container that keeps restarting shows the status CrashLoopBackOff — a very common thing to debug.

Example

Example · bash
# Watch a Pod change phase in real time
kubectl get pods -w

# Inspect events and container states for clues
kubectl describe pod nginx-pod

# Read logs of a crashing container's previous run
kubectl logs nginx-pod --previous

When to use it

  • An SRE watches pod phases in real time during a deployment rollout to confirm pods transition from Pending to Running without going into CrashLoopBackOff.
  • A developer adds a preStop hook to gracefully drain in-flight requests before a pod is terminated during a rolling update.
  • A platform team investigates an OOMKilled pod by checking its last lifecycle phase and resource usage to right-size memory limits.

More examples

Watch pod phase transitions

Watches pod status in real time to observe the lifecycle phases from Pending through ContainerCreating to Running.

Example · bash
kubectl get pods -w
# NAME   READY   STATUS              RESTARTS
# web    0/1     Pending             0
# web    0/1     ContainerCreating   0
# web    1/1     Running             0

Lifecycle hooks in pod spec

Adds postStart and preStop lifecycle hooks: postStart runs immediately after container start, preStop delays SIGTERM by 5 seconds.

Example · yaml
spec:
  containers:
    - name: web
      image: nginx:1.25
      lifecycle:
        postStart:
          exec:
            command: ["/bin/sh", "-c", "echo started > /tmp/started"]
        preStop:
          exec:
            command: ["/bin/sh", "-c", "sleep 5"]

Inspect failed pod reason

Extracts lifecycle failure details from describe output, here showing the pod was killed for exceeding its memory limit.

Example · bash
kubectl describe pod failed-pod | grep -A5 'State:\|Reason:\|Exit Code:'
# State: Terminated
# Reason: OOMKilled
# Exit Code: 137

Related

k8s pod manifestk8s probes

Discussion

  • Be the first to comment on this lesson.