Installing kubernetes-csi/csi-driver-nfs on OpenShift with the Console

We have an existing way to do this with the cli but I wanted to show how we can also do it with the OpenShift Console

Create a HelmRepository

1. Log into the OpenShift console and change to the Developer Perspective

2. Click Helm on the left navigation bar

3. Change Project to kube-system

4. Create a new Helm Chart Repository

5. Leave it as a Namespaced scoped type

• Set Name: csi-driver-nfs
• Optionally set Display Name: csi-driver-nfs
• Set URL: https://raw.githubusercontent.com/kubernetes-csi/csi-driver-nfs/master/charts

6. Create the Repository
7. Click Helm again and Create a new Helm Release

8. Filter by Chart Repositories and select Csi Driver Nfs and select the Csi Driver Nfs chart card to install

9. Click Create

These notes were created using version csi-driver-nfs version 4.6.0

10. Change the Chart version to v4.6.0
11. With the YAML view of the release options we need to change a few settings:

controller:
...
  runOnControlPlane: false    # change to true
...
externalSnapshotter:
  customResourceDefinitions:
    enabled: true             # change to false
  enabled: false              # change to true

12. Click Create
13. All three circles will go dark blue as the Helm Release rolls out the containers to the cluster

We are deploying into the kube-system project because the csi-driver-nfs needs elevated permissions on the OpenShift cluster which are provided by default in that privileged project

Create a StorageClass

1. Proceed to the Administrator Perspective in the

2. Navigate to Storage -> StorageClasses and click Create StorageClass

3. Use the GUI to set:

• Name: nfs-csi
• Reclaim policy: Delete
• Volume binding mode: Immediate
• Provisioner: nfs.csi.k8s.io
• Additional parameters:
  • server: nfs-server.example.com
  • share: /example-dir
  • subDir: ${pvc.metadata.namespace}-${pvc.metadata.name}

If we need specific mount options like NFS v4 we can add a parameter:

• mountOptions: nvsvers=4.1

subDir is optional, by default the template for subdirectories is {nfs-server-address}#{sub-dir-name}#{share-name}

Set Default StorageClass

If we do not have one already we should set this StorageClass as default for the cluster

1. Click Actions -> Edit Annotations

2. Set a new annotation

• storageclass.kubernetes.io/is-default-class: true

3. Click Save

Now you can provision a new PersistentVolumeClaim!

Create a VolumeSnapshotClass

Snapshots are supported with csi-driver-nfs after creating a VolumeSnapshotClass. Creating snapshots and clones can be slow because a snapshot request simply runs tar -czf $DESTINATION_FILE $SOURCE_FOLDER. More information can be found in the community discussion. If you want fast/instantaneous snapshots and clones, please use your storage vendor's CSI driver (e.g. NetApp, Dell/EMC, Portworx, OpenShift Data Foundation, etc…)

A basic VolumeSnapshotClass only requires two pieces of information.

---
apiVersion: snapshot.storage.k8s.io/v1
kind: VolumeSnapshotClass
deletionPolicy: Delete
driver: nfs.csi.k8s.io  #Required to be nfs.csi.k8s.io
metadata:
  name: csi-driver-nfs  #Any name is OK

Virtual Machine snapshots have a 5 minute timeout by default. The GUI allows you to manually specify a longer time. Or you can create VM snapshots via YAML with failureDeadline set to a large number.

---
apiVersion: snapshot.kubevirt.io/v1alpha1
kind: VirtualMachineSnapshot
metadata:
  name: snapshot-with-60-minute-timeout
  namespace: my-amazing-namespace
spec:
  failureDeadline: 1h0m0s  # Make this long enough to allow tar to finish
  source:
    apiGroup: kubevirt.io
    kind: VirtualMachine
    name: my-amazing-vm