Common Exam Mistakes

Jun 6 2026

On this page

YAML manifest errors

Error message	Likely cause	Fix
`cannot be handled as a Pod`	Wrong or missing `apiVersion`	Set `apiVersion: v1` for Pod
`unknown field "xxx"`	Typo in field name	Check `kubectl explain <resource>`
`error parsing ...`	Invalid YAML syntax	Check indentation, colons, quotes
`no matches for kind "X" in version "Y"`	Wrong `apiVersion` for the kind	`kubectl api-resources` to find correct group

kubectl api-resources | grep -i <kind>
kubectl explain <resource>
kubectl explain pod                 # shows apiVersion at top

Common correct apiVersions:

Pod, Service, ConfigMap, Secret, ServiceAccount, Namespace → v1
Deployment, ReplicaSet, DaemonSet, StatefulSet              → apps/v1
Job, CronJob                                                → batch/v1
Ingress, NetworkPolicy                                      → networking.k8s.io/v1
RBAC Role/Binding, ClusterRole/Binding                     → rbac.authorization.k8s.io/v1
HPA                                                         → autoscaling/v2
StorageClass, PV, PVC                                       → storage.k8s.io/v1 / v1
ResourceQuota, LimitRange                                   → v1
PodDisruptionBudget                                         → policy/v1
HTTPRoute                                                   → gateway.networking.k8s.io/v1

Deployment crashing due to command/args typo

Pattern: kubectl get deploy shows 0/1 ready; kubectl describe pod shows CrashLoopBackOff.

kubectl get po -n <ns>
kubectl logs -n <ns> <pod>
kubectl logs -n <ns> <pod> --previous
kubectl get deployment <name> -n <ns> -o yaml | grep -A5 command
kubectl edit deployment -n <ns> <name>
# fix typo e.g. "sleeep 3600" → "sleep 3600"
kubectl rollout status deployment -n <ns> <name>

Common command/args bugs:

Typo: sleeep instead of sleep
Missing shell wrapper: sleep 3600 as command: [sleep, "3600"] works, but shell constructs need ["/bin/sh", "-c", "..."]
Wrong args order: command: ["3600", "sleep"]
args used when command needed (or vice versa): see containers.md

kubectl replace --force (pod in-place fix)

Pods are mostly immutable: can't kubectl edit most fields.

# Pattern 1: replace with force (deletes + recreates)
kubectl get pod <pod> -n <ns> -o yaml > pod-fix.yaml
vim pod-fix.yaml
kubectl replace -f pod-fix.yaml --force

# Pattern 2: delete + apply
kubectl delete pod <pod> -n <ns>
kubectl apply -f pod-fix.yaml

For Deployments/StatefulSets: kubectl edit or kubectl set; controller recreates pods
--force needed only for standalone Pods with immutable fields

Admission control mistakes

Error	Layer	Fix
`exceeded quota: compute-quota`	ResourceQuota admission	Lower requests/limits or delete objects; check `kubectl describe quota`
`must specify limits` / `must specify requests`	ResourceQuota + no LimitRange	Add resources to container or create LimitRange defaults
`maximum cpu usage per Container is`	LimitRange admission	Stay within `max`; check `kubectl describe limitrange`
`violates PodSecurity "restricted"`	Pod Security admission	Harden securityContext; use non-root image
`user cannot create` (no quota keyword)	RBAC	Fix Role/RoleBinding

kubectl apply -f pod.yaml --dry-run=server   # catch admission errors before apply
kubectl auth can-i create pods -n dev        # RBAC only: does NOT test admission

Exam gotchas

Can trigger a CronJob manually: kubectl create job manual --from=cronjob/<name>
ImagePullBackOff, CrashLoopBackOff, CreateContainerConfigError → all have distinct root causes; debug differently