Pod crashes (exit code non-zero)
        │
        ▼
kubelet on node detects container stopped
        │
        ▼
kubelet attempts restart (respects restartPolicy)
        │
        ├── restartPolicy: Always  → Restart immediately
        │   (exponential backoff: 10s, 20s, 40s... → CrashLoopBackOff)
        │
        ├── restartPolicy: OnFailure → Restart only if exit code != 0
        │
        └── restartPolicy: Never   → Don't restart (use for Jobs)

If node dies entirely:
        │
        ▼
Node controller detects node NotReady (after ~40s)
        │
        ▼
After pod-eviction-timeout (default 5 min):
ReplicaSet controller reschedules pods to healthy nodes
        │
        ▼
Scheduler assigns pods to available nodes
        │
        ▼
kubelet on new node starts replacement pods

# CrashLoopBackOff means the container keeps crashing
# K8s uses exponential backoff to avoid hammering a broken service:
# Attempt 1: restart immediately
# Attempt 2: wait 10s
# Attempt 3: wait 20s
# Attempt 4: wait 40s
# Attempt 5: wait 80s
# Attempt 6+: wait 300s (5 min max)

# Debug CrashLoopBackOff
kubectl logs my-pod --previous         # Logs from the crashed container
kubectl describe pod my-pod            # Check events for crash reason
kubectl get pod my-pod -o jsonpath=\
  '{.status.containerStatuses[0].lastState.terminated.reason}'