Kubernetes storage

apiVersion: v1
kind: PersistentVolumeClaim
metadata:
  name: data
spec:
  accessModes: ["ReadWriteOnce"]
  resources:
    requests:
      storage: 10Gi

​

How it works

1. You declare what you need in Kubernetes — a PersistentVolumeClaim with a size, applied with kubectl (App Hub apps create their own). You never touch the dashboard.
2. The cluster’s storage driver calls the platform and a Volume of exactly that size is created under your account. Billing starts; the claim is usually Bound within seconds.
3. The disk attaches to the right worker automatically. Wherever Kubernetes schedules your pod, the volume is attached to that node, formatted on first use, and mounted into the pod.
4. It follows the pod. If the node is replaced or the pod moves, the volume detaches and re-attaches to the new node — data intact, typically in seconds.
5. Grow a volume any time — edit the claim’s requested size (grow only): the disk and filesystem expand live, no pod restart, and billing moves to the new size.
   kubectl patch pvc data -p '{"spec":{"resources":{"requests":{"storage":"20Gi"}}}}'
   
6. Deleting the claim deletes the volume and billing stops.

• Pay for exactly what you claim — $0.08/GB/month, billed hourly, and every GB you pay for is a GB you can use. Storage is never part of the cluster price; it starts billing when a PVC is created and stops when it’s deleted.
• Volumes follow pods. When a pod moves to another node, its volume detaches and re-attaches automatically — data intact.
• Visible in your dashboard. Each PVC appears in your Volumes list, named after its cluster. It can only be managed through Kubernetes — deleting it from the dashboard is blocked while the PVC exists.
• Deleted with the PVC (reclaim policy Delete), and cluster deletion removes any remaining volumes — nothing keeps billing after teardown.
• ReadWriteOnce access (one node at a time, like AWS EBS and DigitalOcean block storage), 10 GB–10 TB per volume, ext4 by default, online expansion supported (grow only).
• Worker root disks are ephemeral — anything a pod writes outside a PVC is lost when the node is replaced or scaled away.
• Apps from App Hub that need persistence (PostgreSQL, WordPress, NextCloud, …) work out of the box — their PVCs land on raff-block.