CKA Storage Master Reference Link to heading

Domain 1 — Storage Link to heading

Access Modes Link to heading

Mode	Short	Meaning
ReadWriteOnce	RWO	One NODE read/write (multiple pods on same node OK)
ReadOnlyMany	ROX	Many nodes read-only
ReadWriteMany	RWX	Many nodes read/write

Trap: RWO = one node, NOT one pod. Deployment + Job can both mount RWO PVC if on same node.

PVC Stuck Pending — Access Mode Mismatch Link to heading

k get pv -o yaml          # check spec.accessModes
k get pvc -n <ns> -o yaml # check spec.accessModes
# PV must support what PVC requests
# Fix: edit PV to match PVC access mode (question says alter only PV)

PV Reclaim Policies Link to heading

Policy	What happens when PVC deleted
`Retain`	PV stays, data preserved
`Delete`	PV auto-deleted, data gone
`Recycle`	Deprecated

StatefulSet PVC Issues Link to heading

PVC Bound to Wrong Node (local-path provisioner) Link to heading

# Scale to 0 first (can't delete PVC while pod uses it)
k -n <ns> scale sts <name> --replicas=0

# Delete PVC (PV auto-deleted if reclaimPolicy: Delete)
k -n <ns> delete pvc <pvc-name>

# Verify PV gone
k get pv

# Scale back — StatefulSet recreates pod + PVC + PV on current node
k -n <ns> scale sts <name> --replicas=1

Key: volumeClaimTemplates causes StatefulSet to recreate PVCs automatically.

Trap: Must scale to 0 first — PVC deletion hangs if pod still mounts it.

PVC Data Migration Link to heading

Pattern: Job with Two PVCs Link to heading

apiVersion: batch/v1
kind: Job
metadata:
  name: migrator
  namespace: migration
spec:
  template:
    spec:
      restartPolicy: Never
      containers:
      - name: migrator
        image: bash:5
        command:
        - bash
        - -c
        - cp -r /old-data/* /new-data/
        volumeMounts:
        - name: old
          mountPath: /old-data
        - name: new
          mountPath: /new-data
      volumes:
      - name: old
        persistentVolumeClaim:
          claimName: app-data
      - name: new
        persistentVolumeClaim:
          claimName: app-data-retain

Switch Deployment to New PVC Link to heading

k edit deploy <name> -n <ns>
# Change claimName in volumes section

# Delete Job first, then old PVC (Job pod still mounts PVC — deletion hangs)
k delete job migrator -n migration
k delete pvc app-data -n migration

StorageClass Link to heading

apiVersion: storage.k8s.io/v1
kind: StorageClass
metadata:
  name: retain-storage
  annotations:
    storageclass.kubernetes.io/is-default-class: 'true'
provisioner: <provisioner>
reclaimPolicy: Retain
volumeBindingMode: WaitForFirstConsumer

HostPath Volumes Link to heading

volumes:
- name: vol
  hostPath:
    path: /configurator
volumeMounts:
- name: vol
  mountPath: /mount

Write to /mount/config in pod = writes to /configurator/config on node.

Trap: HostPath data persists on node even after pod is deleted. Must rm -rf manually.

Keep Pod Running After One-Shot Write Link to heading

command: ["bash", "-c", "echo <value> > /mount/config && sleep infinity"]

Volume Can’t Be Mounted (drain scenario) Link to heading

# 1. Drain node (evicts pod)
k drain <node> --ignore-daemonsets

# 2. Add toleration to StatefulSet for controlplane taint
k describe node controlplane | grep Taint    # get exact taint key
k edit sts <name> -n <ns>

# Add under spec.template.spec:
tolerations:
- key: node-role.kubernetes.io/control-plane
  operator: Exists
  effect: NoSchedule

# 3. If PVC still stuck (local to old node) — delete PVC
k scale sts <name> -n <ns> --replicas=0
k delete pvc <pvc-name> -n <ns>
k scale sts <name> -n <ns> --replicas=1