In this post we are exploring how to use PowerScale Container Storage Interface (CSI) driver. PowerScale is a highly scalable storage solution for unstructured data, supporting file and object storage.
Dell’s CSI drivers, including PowerScale CSI, are developed as a standard for exposing storage systems to containerized workloads. This way, the Kubernetes volume layer becomes truly extensible. It's really a way to combine the best of both worlds.
A good starting point is the PowerScale CSI GitHub repository. It provides all documentation and required software. Now, let's look closer at how to install and utilize the PowerScale CSI driver...
1. Preparation
Initially download PowerScale CSI repository. We need to prepare two yaml files ahead of the actiual installation. values.yaml contains CSI driver configuration parameters and secret.yaml includes PowerScale access details. You can find sample files in the repository.
https://github.com/dell/csi-powerscale/blob/main/helm/csi-isilon/values.yaml
https://github.com/dell/csi-powerscale/blob/main/samples/secret/secret.yaml
For ease of use this is the secret.yaml in my lab environment.
[ root @ DXB ] # cat secret.yaml isilonClusters : - clusterName : "powerscale" username : " root " password : "SecretPassword123!" endpoint : " 192.168.1.222 " endpoint Port : 8080 isDefault : true skipCertificateValidation : true isiPath : "/ifs/data/csi"
Another pre-requisite is the have Kubernetes snapshot custom resource definitions (CRDs) and Common Snapshot Controller installed. This might be already present in your cluster but if the installation is required you can follow the below steps.
Download the repository from https://github.com/kubernetes-csi/external-snapshotter
Snapshot CRDs Install:
kubectl kustomize client/config/crd | kubectl create -f -
Snapshot Controller Install:
kubectl -n namespace kustomize deploy/kubernetes/snapshot-controller | kubectl create -f -
2. Installation
First we need to create a namespace and secret utilizing our secret.yaml.
[ root @ DXB ~ ] # kubectl create namespace powerscale namespace/powerscale created
[ root @ DXB ~ ] # kubectl create secret generic isilon-creds -n powerscale --from-file=config=secret.yaml -o yaml --dry-run=client | kubectl apply -f - secret/isilon-creds created
If you want to validate the PowerScale API server’s certificates, follow these steps: https://dell.github.io/csm-docs/docs/csidriver/installation/helm/isilon/#certificate-validation-for-onefs-rest-api-calls
If not, create an empty secret using the following command as an empty secret must be created for the successful installation of CSI Driver for Dell PowerScale.
[ root @ DXB ~ ] # kubectl create -f empty-secret.yaml secret/isilon-certs-0 created
Now we can finally proceed to the actual CSI driver installation. I’m using helm in my example but an offline installer is also available.
[ root @ DXB ~ ] # ./csi-install.sh --namespace powerscale --values values.yaml ----------------------------------------------------- > Installing CSI Driver: csi-isilon on 1.20 ----------------------------------------------------- > Checking to see if CSI Driver is already installed ----------------------------------------------------- > Verifying Kubernetes and driver configuration ----------------------------------------------------- |- Kubernetes Version: 1.20 | |- Driver: csi-isilon | |- Verifying Kubernetes version |--> Verifying minimum Kubernetes version Success |--> Verifying maximum Kubernetes version Success | |- Verifying that required namespaces have been created Success | |- Verifying that required secrets have been created Success | |- Verifying that optional secrets have been created Success | |- Verifying alpha snapshot resources |--> Verifying that alpha snapshot CRDs are not installed Success | |- Verifying snapshot support |-- > Verifying that snapshot CRDs are available Success |-- > Verifying that the snapshot controller is available Success | |- Verifying helm version Success | |- Verifying helm values version Success ----------------------------------------------------- > Verification Complete Success ----------------------------------------------------- - Installing Driver Success |-- > Waiting for Deployment isilon - controller to be ready Success |-- > Waiting for DaemonSet isilon - node to be ready Success ----------------------------------------------------- > Operation complete -----------------------------------------------------Verify if pods are running as expected
[ root @ DXB ~ ] # kubectl get pods -n powerscale NAME READY STATUS RESTARTS AGE isilon - controller - 55c4d86 6/6 Running 58 246d isilon - node - cmf4f 2/2 Running 22 246d
3. Using PowerScale CSI Drivers
We can now create a new persistent volume claim and deploy an application using it.
[ root @ DXB sample_files ] # cat pvcnew.yaml apiVersion : v1 kind : PersistentVolume Claim metadata : name : powerscalepvc1 spec : accessModes : - ReadWriteOnce resources : requests : storage : 50Gi storageClassName : isilon - plain [ root @ DXB ] # kubectl create -f pvcnew.yaml persistentvolumeclaim/powerscalepvcl createdIf we now go to the File System Explorer as part of PowerScale's OneFS GUI we can see the created volume
Let's proceed with deploying a NGINX instance to consume our PVC.
[ root @ DXB ] # cat nginx.yaml apiVersion : vl kind : Pod metadata : name : nginx-pv-pod spec : containers : - name : task-pv-container image : nginx image PullPolicy : Never ports : - container Port : 80 name : "http-server " volumeMounts : - mount Path : " /usr/share/nginx /html " name : task-pv-storage volumes : - name : task-pv-storage persistentVolumeClaim : claimName : powerscalepvc1 [ root @ DXB ] # kubectl create -f nginx.yamlIf we review the pod details, we can see that the volume from our PVC was mounted sucessfully.
[ root @ DXB ] # kubectl describe pod nginx-pv-pod -n default ----- Removed some output for better readability ----- Volumes : task - pv - storage : Type : PersistentVolumeClaim ( a reference to a PersistentVolumeClaim in the same namespace ) ClaimName : powerscalepvc1 ReadOnly : false default - token - xrbxk : Type : Secret ( a volume populated by a Secret ) SecretName : default - token - xrbxk Optional : false QoS Class : BestEffort Node - Selectors : none Tolerations : node.kubernetes.io/not-ready : NoExecute op - Exists for 300s node.kubernetes.io/unreachable : NoExecute op - Exists for 300s Events : Successfully assigned default / nginx - pv - p AttachVolume.Attach succeeded for volume Container image "nginx " already present Created container task - pv- container kubelet Started container task - pv - container
3. Snapshots
The advantage of CSI is that array native features can be exposed to containers. In this example we are using PowerScale SnapshotIQ snapshots for our PVC.
[ root @ DXB ] # cat snap.yaml apiVersion : snapshot.storage.k8s.io/vl kind : Volume Snapshot metadata : name : pvcsnap namespace : default spec : volumeSnapshotClassName : isilon-snapclass source : persistentVolume ClaimName : powerscalepvc1
[ root @ DXB ] # kubectl create -f snap.yaml volumesnapshot.snapshot.storage.k8s.io/pvcsnap created
[ root @ DXB ] # kubectl get volume snapshot ----- Removed some output for better readability ----- NAME READYTOUSE SOURCEPVC SNAPSHOTCLASS pvcsnap true powerscalepvc1 isilon-snapclassWe can now as well see the snapshot in the File System Explorer and in the SnapshotIQ dashboard.
[ root @ DXB ] # kubectl delete volumesnapshot pvcsnap volumesnapshot.snapshot.storage.k8s.io "pvcsnap " deletedI hope you enjoyed this quick guide on how to install and use the PowerScale CSI driver. For further examples on how to utilizes the PowerScale CSI driver, have a look at the GitHub repository.