CKA — etcd Snapshot Restore

Restoring from an etcd Backup

Restoring etcd is more complex than backing up. You restore to a new data directory and then update the etcd configuration to use it.

Step 1 — Restore the Snapshot

ETCDCTL_API=3 etcdctl snapshot restore /opt/etcd-backup.db \
  --data-dir=/var/lib/etcd-restored

This creates a new data directory at /var/lib/etcd-restored with the restored data.

Important: Do NOT restore directly to /var/lib/etcd. Always restore to a new directory first.

Step 2 — Update the etcd Static Pod Manifest

Edit the etcd manifest to point to the new data directory:

sudo vi /etc/kubernetes/manifests/etcd.yaml

Make three changes:

1. Update the –data-dir flag:

# Before
    - --data-dir=/var/lib/etcd

# After
    - --data-dir=/var/lib/etcd-restored

2. Update the volumeMount mountPath:

    volumeMounts:
    - mountPath: /var/lib/etcd-restored   # <-- Update this
      name: etcd-data

3. Update the hostPath:

  volumes:
  - hostPath:
      path: /var/lib/etcd-restored        # <-- Update this
      type: DirectoryOrCreate
    name: etcd-data

Step 3 — Wait for etcd to Restart

The kubelet detects the manifest change and automatically restarts the etcd Pod. Wait 30-60 seconds:

# If kubectl works
kubectl get pods -n kube-system | grep etcd

# If kubectl is not responding, use crictl
sudo crictl ps | grep etcd

Step 4 — Verify the Restoration

kubectl get nodes
kubectl get pods --all-namespaces
kubectl get deployments --all-namespaces

Complete Restore Workflow

Step	Command	Purpose
1	etcdctl snapshot restore –data-dir=new-dir	Restore to new directory
2	Edit etcd.yaml –data-dir	Point etcd to new directory
3	Edit etcd.yaml volumeMount	Update mount path
4	Edit etcd.yaml hostPath	Update host path
5	Wait 30-60 seconds	kubelet restarts the Pod
6	kubectl get nodes	Verify cluster is functional

Common Restore Mistakes

Mistake	Consequence	Prevention
Restoring to /var/lib/etcd directly	May corrupt existing data	Always use a new directory
Forgetting to update volume mounts	etcd Pod cannot access data	Update all three locations
Not waiting for restart	kubectl commands fail temporarily	Wait 30-60 seconds
Wrong file permissions	etcd cannot read data	Ensure root ownership

Exam tip: This is a high-value exam question. Practice the full workflow until you can do it without looking at notes. Remember the three edits in etcd.yaml.

---

Next: Verify Backup ->