Kubernetes PersistentVolumes are not deleted

[bgdnlp]

Describe the bug

I'm not entirely sure if this is considered a bug.

Once a PersistentVolume is created it seems to only be deleted from disk when OrbStack is closed. This leads to a situation where in order to start fresh one would have to quit Orbstack and start it again, or use the deprecated "Recycle" reclaim policy. It is not enough to run kubectl delete and then kubectl apply again, as the data on the persistent volume will still be there even if the reclaim policy for the volume is "Delete".

To Reproduce

1. Create a manifest containing a PV, PVC and a container mounting that volume. Set persistentVolumeReclaimPolicy to "Delete".
2. kubectl apply -f manifest.yaml the manifest.
3. Log in to the container, create a file: touch SOMERANDOMFILE.
4. kubectl delete -f manifest.yaml
5. kubectl apply -f manifest.yaml again.
6. Log in to the container again and ls -al.

Here's an example manifest using Wordpress:

--- # WP persistent volume
apiVersion: v1
kind: PersistentVolume
metadata:
  name: wp-pv
  namespace: wordpress
  labels:
    app: wordpress
    tier: app
spec:
  storageClassName: local-path
  persistentVolumeReclaimPolicy: Delete
  accessModes:
    - ReadWriteOnce
  capacity:
    storage: 1Gi
  hostPath:
    path: "/mnt/wp"

--- # WP persistent volume claim
apiVersion: v1
kind: PersistentVolumeClaim
metadata:
  name: wp-pvc
  namespace: wordpress
  labels:
    app: wordpress
    tier: app
spec:
  storageClassName: local-path
  resources:
    requests:
      storage: 1Gi
  accessModes:
    - ReadWriteOnce
  volumeName: wp-pv

--- # WP service
apiVersion: v1
kind: Service
metadata:
  name: wordpress
  namespace: wordpress
  labels:
    app: wordpress
    tier: app
spec:
  ports:
    - port: 80
      targetPort: 80
  selector:
    app: wordpress
    tier: app
  type: LoadBalancer

--- # WP deployment
apiVersion: apps/v1
kind: Deployment
metadata:
  name: wordpress
  namespace: wordpress
  labels:
    app: wordpress
    tier: app
spec:
  replicas: 1
  selector:
    matchLabels:
      app: wordpress
      tier: app
  template:
    metadata:
      labels:
        app: wordpress
        tier: app
    spec:
      containers:
        - image: public.ecr.aws/docker/library/wordpress:latest
          name: wordpress
          env:
            - name: WORDPRESS_DB_HOST
              value: wp-db
            - name: WORDPRESS_DB_NAME
              value: wordpress
            - name: WORDPRESS_DB_USER
              value: wordpress
            - name: WORDPRESS_DB_PASSWORD
              value: DBPass
          ports:
            - containerPort: 80
              name: wordpress
          volumeMounts:
            - name: wp-www
              mountPath: /var/www/html
          resources:
            requests:
              cpu: 100m
              memory: 256Mi
            limits:
              cpu: 500m
              memory: 512Mi
      volumes:
        - name: wp-www
          persistentVolumeClaim:
            claimName: wp-pvc

Expected behavior

I would expect the volume to be deleted so that a fresh apply would be completely fresh. In my case the WordPress DB once initialized means I have to quit Orbstack before re-applying while testing. Or use Recycle as reclaim policy, which is deprecated.

Diagnostic report (REQUIRED)

OrbStack info:
Version: 1.10.3
Commit: 2b5dd5f580d80a3d2494b7b40dde2ef46813cfc5 (v1.10.3)

System info:
macOS: 15.3.2 (24D81)
CPU: arm64, 10 cores
CPU model: Apple M1 Pro
Model: MacBookPro18,1
Memory: 16 GiB

Full report: https://orbstack.dev/_admin/diag/orbstack-diagreport_2025-03-24T09-23-53.871650Z.zip

Screenshots and additional context (optional)

No response

[slinorb]

This seems to be a bug with local-path-provisioner which does not honor Delete persistentVolumeReclaimPolicy. To work around the issue you can use PVC only without binding to a specific PV. Just remove PV manifest and volumeName attribute in the PVC manifest.