Setting up storage for OpenShift Virtualization

Prerequisites for ODF

Bare metal worker nodes with local NVME storage

Use flavors with the d suffix that include local NVME disks.

Worker node requirements

ODF requires at least 3 worker nodes for high availability and data redundancy.

Zone configuration

• OpenShift Virtualization deployments: Single-zone clusters are recommended to minimize storage latency for VM workloads. While multi-zone deployments provide better high availability, cross-zone storage replication can impact VM performance.
• General ODF deployments: At least 3 zones are recommended for high availability and disaster recovery.

Installing ODF from the console

1. Log in to your IBM Cloud account and navigate to your cluster.

2. Click Add-ons in the cluster navigation.

3. On the OpenShift Data Foundation card, click Install.

4. Select the Local Storage option.

   Remote block volume attachment is not supported for VPC bare metal nodes.

5. Choose one of the following discovery methods:

   Automatic discovery (recommended)

   Select Automatic discovery to automatically detect and use available disks on your bare metal worker nodes.

   Manual disk specification

   If you prefer to specify disks manually, follow these steps:

   1. Log in to each worker node to identify available disks.

      oc debug node/<node-name>
      

   2. Allow host binaries.

      chroot /host
      

   3. List available devices.

      lsblk
      

   4. Find the by-id for each disk.

      ls -l /dev/disk/by-id/
      

   5. Add the disk IDs in the OSD pod disk IDs parameter.

6. Configure additional parameters as needed:

   • Resource profile: Choose balanced or performance based on your workload
   • Encryption: Enable encryption with Hyper Protect Crypto Services if required
   • Number of OSD disks: Specify the number of disks per worker node

7. Click Install to deploy ODF.

For more information about ODF parameters, see OpenShift Data Foundation parameter reference.

Installing ODF from the CLI

1. Review the available add-on options.

   ibmcloud oc cluster addon options --addon openshift-data-foundation --version 4.20.0
   

2. Enable the OpenShift Data Foundation add-on with automatic disk discovery and virtualization support.

   ibmcloud oc cluster addon enable openshift-data-foundation \
     -c <cluster-name> \
     --version 4.20.0 \
     --param "odfDeploy=true" \
     --param "osdStorageClassName=localblock" \
     --param "autoDiscoverDevices=true" \
     --param "resourceProfile=performance" \
     --param "setDefaultStorageClassForVirtualization=true"
   

3. Verify the add-on is in a Ready state.

   oc get storagecluster -n openshift-storage
   

   Example output:

   NAME                 AGE   PHASE   EXTERNAL   CREATED AT             VERSION
   ocs-storagecluster   53m   Ready              2025-03-10T12:20:52Z   4.20.0
   

Verifying ODF installation

1. Check the storage cluster status.

   oc get storagecluster -n openshift-storage
   

2. Verify that storage classes are available.

   oc get storageclass | grep ocs
   

3. Confirm that ODF pods are running.

   oc get pods -n openshift-storage
   

For more information about deploying ODF, see Deploying OpenShift Data Foundation on VPC clusters.

Prerequisites for VPC File Storage

• Any supported bare metal flavor
• VPC infrastructure

Step 1: Disable VPC Block Storage as default

1. Edit the VPC Block CSI driver configmap.

   oc edit cm addon-vpc-block-csi-driver-configmap -n kube-system
   

2. Set IsStorageClassDefault to false.

   data:
     IsStorageClassDefault: "false"
   

3. Save and exit the editor.

Step 2: Install the VPC File CSI driver

Step 3: Set a default storage class

1. List available storage classes.

   oc get storageclass | grep vpc-file
   

2. Choose an appropriate storage class based on your workload:

   • ibmc-vpc-file-1000-iops: For moderate workloads (40 GB - 2000 IOPS)
   • ibmc-vpc-file-6000-iops: For higher workloads (100 GB - 6000 IOPS)
   • ibmc-vpc-file-metro-1000-iops: For multi-zone deployments

3. Edit the VPC File CSI driver configmap.

   oc edit cm addon-vpc-file-csi-driver-configmap -n kube-system
   

4. Set the SET_DEFAULT_STORAGE_CLASS field to your chosen storage class.

   data:
     SET_DEFAULT_STORAGE_CLASS: "ibmc-vpc-file-metro-1000-iops"
   

5. Save and exit the editor.

6. Wait a few minutes for the changes to take effect.

7. Verify the default storage class.

   oc get storageclass | grep "(default)"
   

Step 4: Configure storage profiles for VPC File Storage

After installing the OpenShift Virtualization Operator, you must configure storage profiles for VPC File Storage because the Container Disk Importer (CDI) does not automatically recognize VPC File Storage properties.

This step must be completed after installing the OpenShift Virtualization Operator.

for sp in $(oc get storageprofile.cdi.kubevirt.io -o json | jq -r '.items[] | select(.status.provisioner == "vpc.file.csi.ibm.io") | .metadata.name'); do
  oc patch storageprofile.cdi.kubevirt.io "$sp" --type=merge -p '{
    "spec": {
      "claimPropertySets": [
        {
          "accessModes": ["ReadWriteMany"],
          "volumeMode": "Filesystem"
        }
      ]
    }
  }'
done

Verifying VPC File Storage installation

1. Verify the VPC File CSI driver pods are running.

   oc get pods -n kube-system | grep vpc-file
   

2. Confirm the default storage class is set correctly.

   oc get storageclass | grep "(default)"
   

3. Test storage provisioning by creating a test PVC.

   apiVersion: v1
   kind: PersistentVolumeClaim
   metadata:
     name: test-pvc
   spec:
     accessModes:
     - ReadWriteMany
     resources:
       requests:
         storage: 10Gi
   

4. Verify the PVC is bound.

   oc get pvc test-pvc
   

5. Delete the test PVC.

   oc delete pvc test-pvc
   

ODF installation fails

If ODF installation fails, check the following:

• Verify that your bare metal nodes have local NVME storage
• Ensure you have at least 3 worker nodes
• Check that disks are unmounted and unformatted
• Review ODF operator logs: oc logs -n openshift-storage -l app=ocs-operator

VPC File Storage PVCs remain pending

If PVCs remain in a pending state:

• Verify the VPC File CSI driver add-on is installed and healthy
• Check that the default storage class is set correctly
• Ensure storage profiles are configured with RWX access mode
• Review CSI driver logs: oc logs -n kube-system -l app=ibm-vpc-file-csi-controller

Storage profiles not updated

If storage profiles are not showing the correct configuration:

• Ensure the OpenShift Virtualization Operator is installed
• Wait for storage profiles to be created automatically
• Manually patch storage profiles using the commands provided
• Verify the provisioner name matches vpc.file.csi.ibm.io