Use volume snapshots with Azure Container Storage

Azure Container Storage is a cloud-based volume management, deployment, and orchestration service built for containers. This article shows how to take a point-in-time snapshot of a persistent volume and restore it with a new persistent volume claim (PVC).

Limitations

Volume snapshots aren't supported when you use ephemeral disk (local NVMe) as backing storage.

Prerequisites

• If you don't have an Azure subscription, create a free account before you begin.

• This article requires Azure CLI version v2.83.0 or later. For more information, see How to install the Azure CLI. Disable extensions such as aks-preview if issues occur. Install or update extensions as needed:

  • az extension add --upgrade --name k8s-extension
  • az extension add --upgrade --name elastic-san (Elastic SAN only)

• You need the Kubernetes command-line client, kubectl. It's already installed if you're using Azure Cloud Shell. You can install it locally by running the az aks install-cli command.

• Check if your target region is supported in Azure Container Storage regions.

• This article assumes you installed Azure Container Storage on your Azure Kubernetes Service (AKS) cluster and created a PVC using Azure Elastic SAN.

Create a volume snapshot class

A volume snapshot class defines snapshot settings. Follow these steps to create a volume snapshot class for Elastic SAN.

1. Create a YAML manifest file such as acstor-volumesnapshotclass.yaml.

2. Paste the following code. The volume snapshot class name value can be any value.

   apiVersion: snapshot.storage.k8s.io/v1
   kind: VolumeSnapshotClass
   metadata:
    name: elasticsan-snapshot-class
   driver: san.csi.azure.com
   deletionPolicy: Delete
   

3. Apply the manifest to create the volume snapshot class.

   kubectl apply -f acstor-volumesnapshotclass.yaml
   

   When creation is complete, you see a message like:

   volumesnapshotclass.snapshot.storage.k8s.io/elasticsan-snapshot-class created
   

   You can also run kubectl get volumesnapshotclass to confirm creation. You should see output similar to this example:

   NAME DRIVER DELETIONPOLICY AGE
   elasticsan-snapshot-class san.csi.azure.com Delete 11s
   

Create a volume snapshot

Next, create a snapshot of an existing PVC and apply the volume snapshot class you created in the previous step.

1. Create a YAML manifest file such as acstor-volumesnapshot.yaml.

2. Paste the following code. Set volumeSnapshotClassName to the name of the volume snapshot class you created in the previous step. Set persistentVolumeClaimName to the name of the PVC you want to snapshot. The volume snapshot name value can be any value.

   apiVersion: snapshot.storage.k8s.io/v1
   kind: VolumeSnapshot
   metadata:
    name: elasticsan-volume-snapshot
   spec:
    volumeSnapshotClassName: elasticsan-snapshot-class
    source:
    persistentVolumeClaimName: elasticsanpvc
   

3. Apply the manifest to create the volume snapshot.

   kubectl apply -f acstor-volumesnapshot.yaml
   

   When creation is complete, you see a message like:

   volumesnapshot.snapshot.storage.k8s.io/elasticsan-volume-snapshot created
   

   You can also run kubectl get volumesnapshot to check that the volume snapshot is created. If READYTOUSE is true, you can move on to the next step.

Create a restored persistent volume claim

Create a new PVC that uses the volume snapshot as a data source.

1. Create a YAML manifest file such as acstor-pvc-restored.yaml.

2. Paste the following code. The storageClassName must match the StorageClass you used when creating the original PVC. For the data source name value, use the name of the volume snapshot you created in the previous step. The PVC metadata.name value can be any value.

   apiVersion: v1
   kind: PersistentVolumeClaim
   metadata:
    name: pvc-elasticsan-snapshot-restored
   spec:
    accessModes:
    - ReadWriteOnce
    storageClassName: azuresan-csi
    resources:
    requests:
    storage: 100Gi
    dataSource:
    name: elasticsan-volume-snapshot
    kind: VolumeSnapshot
    apiGroup: snapshot.storage.k8s.io
   

3. Apply the manifest to create the PVC.

   kubectl apply -f acstor-pvc-restored.yaml
   

   When creation is complete, you see a message like:

   persistentvolumeclaim/pvc-elasticsan-snapshot-restored created
   

   You can also run kubectl describe pvc pvc-elasticsan-snapshot-restored to check that the persistent volume is created. The status shows Pending with a message that it is waiting for the first consumer to be created before binding.

Delete the original pod (optional)

Before you create a new pod, you can delete the original pod that you created the snapshot from.

1. Run kubectl get pods to list the pods. Make sure you delete the correct pod.
2. To delete the pod, run kubectl delete pod <pod-name>.

Create a new pod using the restored snapshot

Create a new pod using the restored PVC. Create the pod using Flexible I/O Tester (fio) for benchmarking and workload simulation, and specify a mount path for the persistent volume.

1. Create a YAML manifest file such as fio-pod-restore.yaml.

2. Paste the following code. The persistent volume claim claimName should be the name of the restored PVC you created. The pod metadata.name value can be any value.

   apiVersion: v1
   kind: Pod
   metadata:
    name: fio-restore
   spec:
    volumes:
    - name: iscsi-volume
    persistentVolumeClaim:
    claimName: pvc-elasticsan-snapshot-restored
    containers:
    - name: fio
    image: nixery.dev/shell/fio
    volumeMounts:
    - mountPath: /volume
    name: iscsi-volume
   

3. Apply the manifest to deploy the pod.

   kubectl apply -f fio-pod-restore.yaml
   

   You should see output similar to this example:

   pod/fio-restore created
   

4. Check that the pod is running and the PVC is bound:

   kubectl describe pod fio-restore
   kubectl describe pvc pvc-elasticsan-snapshot-restored
   

5. Check fio testing to see its current status:

   kubectl exec -it fio-restore -- fio --name=benchtest --size=800m --filename=/volume/test --direct=1 --rw=randrw --ioengine=libaio --bs=4k --iodepth=16 --numjobs=8 --time_based --runtime=60
   

You now have a pod that uses the restored PVC and can use it for your Kubernetes workloads.

See also

• What is Azure Container Storage?
• Use Azure Container Storage with local NVMe