Intermediate Part 4: Persistent Volumes & Storage – Saving Data¶

So far our nginx app is stateless : if a Pod dies or we scale to zero, all data inside it is lost.
Real apps need persistent data: databases, uploaded files, logs, user sessions, etc.

Kubernetes separates storage concerns from the app using Persistent Volumes (PV) and PersistentVolumeClaims (PVC).

Pizza chain analogy:
- Pods are the kitchens : temporary and replaceable
- Persistent Volumes are the walk-in fridges : the ingredients (data) stay even if the kitchen burns down and gets rebuilt
- A PersistentVolumeClaim is the kitchen manager saying: "I need a 100-liter fridge with fast access"

In Minikube, we use hostPath or local storage (simple for learning : not production-grade).

Goal for this section¶

Run a tiny file server (or database) that saves a file.
Delete the Pod → recreate it → file is still there.

We'll use a simple busybox Pod + PVC.

Step 1: Create a PersistentVolumeClaim (PVC)¶

The PVC is what your app requests: "I need 1 GB of storage."

Create pvc.yaml:

`apiVersion: v1 kind: PersistentVolumeClaimmetadata: name: my-pvcspec: accessModes: - ReadWriteOnce # Can be mounted by one Pod at a time resources: requests: `` storage: 1Gi # 1 GiB

Apply:

Bash

kubectl apply -f pvc.yaml

Check:

Bash

kubectl get pvc

You should see my-pvc with STATUS: Bound (Minikube auto-provisions a PV).

Step 2: Run a Pod that uses the PVC¶

Create storage-pod.yaml:

YAML

apiVersion: v1
kind: Pod
metadata:
  name: storage-demo
spec:
  containers:
  - name: storage-container
    image: busybox
    command: ["/bin/sh", "-c", "sleep 3600"]
    volumeMounts:
    - mountPath: /data
      name: storage-volume
  volumes:
  - name: storage-volume
    persistentVolumeClaim:
      claimName: my-pvc

Apply:

Bash

kubectl apply -f storage-pod.yaml

Wait until Ready:

Bash

kubectl get pods -w

Step 3: Write data to the volume¶

Exec into the Pod:

Bash

kubectl exec -it storage-demo -- sh

Inside the Pod shell:

sh

echo "Hello from persistent storage!" > /data/hello.txt
cat /data/hello.txt
exit

→ You wrote a file to the mounted volume.

Step 4: Delete & recreate the Pod – data survives¶

Delete the Pod:

Bash

kubectl delete pod storage-demo

Recreate it (same YAML):

Bash

kubectl apply -f storage-pod.yaml

Wait for Ready, then check the file:

Bash

kubectl exec -it storage-demo -- cat /data/hello.txt

→ "Hello from persistent storage!" is still there!

→ Proof: the data survived Pod deletion/recreation : the PVC kept it alive.

Step 5: Cleanup¶

Bash

kubectl delete pod storage-demo
kubectl delete pvc my-pvc

(The PV might stay in Minikube : harmless.)

Key Takeaways¶

• PVC = what your app requests ("give me storage")
• PV = the actual storage (Minikube auto-creates simple ones)
• Mount as a volume → data persists beyond Pod lifetime
• ReadWriteOnce = common for single-node or single-writer apps
• In production: use cloud PVs (AWS EBS, GCP PD, Azure Disk), CSI drivers, or distributed storage (Longhorn, Rook Ceph)

This is how databases (PostgreSQL, MySQL), file servers, or stateful apps run on Kubernetes.

→ Next: Part 5 – Perform a Rolling Update and Rollback