Use dynamically provisioned disks for stateful applications - Container Service for Kubernetes

When containers fail, stateful application data stored in container storage can be lost or become unreliable. Persistent storage addresses this risk. This topic explains how to use dynamically provisioned disks to implement persistent storage.

How it works

Create a PVC that references a StorageClass.
Deploy your application with a volume that references the PVC.
The provisioner automatically creates a disk and binds it to the PVC as a PV.

Each ACK cluster includes a provisioner by default.

Prerequisites

An ACK managed cluster is created.
kubectl is connected to the cluster.
A disk provisioner is installed in the cluster.

Default StorageClasses

ACK creates four default StorageClasses at cluster initialization. They only work for single-zone clusters. For multi-zone clusters, create a custom StorageClass.

StorageClass name	Disk type	Description
`alicloud-disk-common`	Basic disk	Standard baseline storage
`alicloud-disk-efficiency`	Enhanced SSD (ESSD)	Balanced performance and cost
`alicloud-disk-ssd`	SSD	High-performance storage
`alicloud-disk-available`	High availability	Creates an SSD if available; falls back to ultra disk.

Create a custom StorageClass

Create a custom StorageClass if defaults do not meet your needs, for example in multi-zone clusters.

Create a file named storageclass.yaml:

    kind: StorageClass
    apiVersion: storage.k8s.io/v1
    metadata:
      name: alicloud-disk-ssd-hangzhou-b
    provisioner: alicloud/disk
    reclaimPolicy: Retain
    parameters:
      type: cloud_ssd
      regionId: cn-hangzhou
      zoneId: cn-hangzhou-b
      fstype: "ext4"
      readonly: "false"

Create the StorageClass:
```
    kubectl create -f storageclass.yaml
```

StorageClass parameters

Parameter	Description
`provisioner`	The disk provisioner plugin. Set to `alicloud/disk`.
`reclaimPolicy`	Disk behavior on PVC deletion. Valid values: `Delete`, `Retain`. Default: `Delete`.
`type`	Disk type. Valid values: `cloud`, `cloud_efficiency`, `cloud_ssd`, `available`.
`regionId`	Region of the disk. Must match the cluster region. Optional.
`zoneId`	Disk zone. Optional. Single-zone cluster: the zone ID. Multi-zone cluster: comma-separated IDs (for example, `cn-hangzhou-a,cn-hangzhou-b,cn-hangzhou-c`).
`fstype`	File system type. Optional. Default: `ext4`.
`readonly`	Whether the disk is read-only. Optional. Valid values: `true`, `false`. Default: `false`.
`encrypted`	Whether to encrypt the disk. Optional. Valid values: `true`, `false`. Default: `false`.

Note If you set reclaimPolicy to Delete, the disk is automatically deleted when you delete the PVC. The disk data cannot be recovered.

Create a PVC

Create a file named pvc-ssd.yaml:

    kind: PersistentVolumeClaim
    apiVersion: v1
    metadata:
      name: disk-ssd
    spec:
      accessModes:
        - ReadWriteOnce
      storageClassName: alicloud-disk-ssd-hangzhou-b
      resources:
        requests:
          storage: 20Gi

Create the PVC:
```
    kubectl create -f pvc-ssd.yaml
```

Create a Deployment

Create a file named pvc-dynamic.yaml:

    apiVersion: apps/v1
    kind: Deployment
    metadata:
      name: nginx-dynamic
      labels:
        app: nginx
    spec:
      selector:
        matchLabels:
          app: nginx
      template:
        metadata:
          labels:
            app: nginx
        spec:
          containers:
          - name: nginx
            image: nginx
            volumeMounts:
              - name: disk-pvc
                mountPath: "/data"
          volumes:
            - name: disk-pvc
              persistentVolumeClaim:
                claimName: disk-ssd

Create the Deployment:

    kubectl create -f nginx-dynamic.yaml

Verify persistent storage

Confirm the disk is mounted and data persists across pod restarts.

Confirm the disk is mounted

Find the pod name:

    kubectl get pod | grep dynamic

Expected output:

    nginx-dynamic-5c74594ccb-zl9pf     2/2     Running     0          3m

Verify the disk is mounted at /data:

    kubectl exec nginx-dynamic-5c74594ccb-zl9pf -- df | grep data

Expected output:

    /dev/vdh        20511312    45080  20449848   1% /data

List files in /data:

    kubectl exec nginx-dynamic-5c74594ccb-zl9pf -- ls /data

Expected output:

    lost+found

Write test data and verify persistence

Create a test file on the disk:

    kubectl exec nginx-dynamic-5c74594ccb-zl9pf -- touch /data/dynamic

Confirm the file exists:

    kubectl exec nginx-dynamic-5c74594ccb-zl9pf -- ls /data

Expected output:

    dynamic
    lost+found

Delete the pod to trigger a replacement:

    kubectl delete pod nginx-dynamic-5c74594ccb-zl9pf

Expected output:

    pod "nginx-dynamic-5c74594ccb-zl9pf" deleted

In a separate terminal, watch the pod lifecycle:

    kubectl get pod -w -l app=nginx

Expected output:

    NAME                               READY   STATUS    RESTARTS   AGE
    nginx-dynamic-5c74594ccb-zl9pf     2/2     Running   0          6m48s
    nginx-dynamic-5c74594ccb-zl9pf   2/2   Terminating   0     7m32s
    nginx-dynamic-5c74594ccb-45sd4   0/2   Pending   0     0s
    nginx-dynamic-5c74594ccb-45sd4   0/2   Pending   0     0s
    nginx-dynamic-5c74594ccb-45sd4   0/2   Init:0/1   0     0s
    nginx-dynamic-5c74594ccb-zl9pf   0/2   Terminating   0     7m32s
    nginx-dynamic-5c74594ccb-zl9pf   0/2   Terminating   0     7m33s
    nginx-dynamic-5c74594ccb-zl9pf   0/2   Terminating   0     7m33s
    nginx-dynamic-5c74594ccb-45sd4   0/2   PodInitializing   0     5s
    nginx-dynamic-5c74594ccb-45sd4   2/2   Running   0     22s

Get the name of the new pod:

    kubectl get pod

Expected output:

    NAME                               READY   STATUS      RESTARTS   AGE
    nginx-dynamic-5c74594ccb-45sd4     2/2     Running     0          2m

Verify the test file still exists in the new pod:
```
    kubectl exec nginx-dynamic-5c74594ccb-45sd4 -- ls /data
```
Expected output:
```
    dynamic
    lost+found
```
The dynamic file persists across pod recreation, confirming the disk provides persistent storage.