Kubernetes Pod Diagnostic Decision Tree & Exit Codes

                                  [Pod Unhealthy]
                                         │
        ┌────────────────────────────────┼────────────────────────────────┐
        ▼                                ▼                                ▼
[Exit Code 137]                  [Exit Code 1 / 2]                [Exit Code 143]
  OOMKilled                        Application Error                Graceful SIGTERM
  - Pod exceeded memory limit      - Uncaught exception/panic       - Teardown initiated by K8s
  - Action: Check container limits - Action: kubectl logs --previous - Action: Check terminationGracePeriod
                                         │
        ┌────────────────────────────────┴────────────────────────────────┐
        ▼                                                                 ▼
[Pending Status]                                                 [ImagePullBackOff]
  - Insufficient CPU/Memory on node pool                           - Registry authentication failure
  - PVC bound to different availability zone                       - Image tag does not exist
  - Action: kubectl describe pod | grep -A 5 Events               - Action: verify imagePullSecrets

Exit Code Reference Table

Exit Code	Signal / Name	Root Cause	Immediate Action
137	SIGKILL	OOMKilled (Memory limit exceeded) or container killed by host kernel.	Run kubectl describe pod, inspect memory limits, collect memory profile.
139	SIGSEGV	Segmentation fault in native binary or C-extension.	Inspect core dump, check shared library architecture (x86 vs ARM).
143	SIGTERM	Graceful termination signal sent during pod eviction or rollout.	Normal during deploy; if unexpected, check node autoscaler / spot termination.
1	General Error	Application syntax error, uncaught exception, or missing env var.	Check kubectl logs <pod> --previous.
2	Misuse of Shell	Missing command executable or invalid shell arguments in command: [].	Verify entrypoint binary exists in the container image.