Storage migration with passthrough devices

Last published : Jul 06, 2026
With raw device mapped disks in an on-premise InfoScale cluster, IKE can consume these raw LUNs and spin up virtual machines in the OCP-V environment. This provides the same storage mapping in the InfoScale application VMs, allowing the on-premise cluster to be recreated. The source cluster is then imported as InfoScale application VMs on the current IKE cluster.

Pre-requisites

  • NNCP and NAD Configurations for InfoScale Application VMs
  • In case of disk-based I/O fencing in a virtualized cluster, virtual machines (VMs) accessing shared Raw Device Mapping (RDM) disks typically must reside on separate physical hypervisor hosts
  • SCSI3 PR should be enabled in OpenShift virtualisation config
  • SG_IO should work with VxVM backed volumes.
  • Vendor should be NAA-6 compliant
  • IKE Pre-requisites should be followed

Usage and VM tunables

To consume the RDM disks; static PVs with correct WWN should be created first. Assuming that InfoScaleCluster resource is created perform the following checks:
  1. IKE Cluster is in Running and Healthy state
  2. InfoScale block storage class is configured
  3. Storage class should reflect correct topology mapping in case of multiple InfoScaleClusters
  4. User have to compile the (RDM list) passthrough device list with the corresponding wwids
  5. Create static PVs and PVCs pair per wwid with following template
---
# Static PV consuming RDM disk
apiVersion: v1
kind: PersistentVolume
metadata:
  name: rdm-disk-1
spec:
  accessModes:
  - ReadWriteMany
  capacity:
    storage: 40Gi
  csi:
    driver: org.veritas.infoscale
    volumeAttributes:
      wwid: 600A098038314F36612B587549484D59    # wwid of RDM disk-1
    volumeHandle: clust_21432/wwid-600A098038314F36612B587549484D59 # mapped IKE cluster id
  persistentVolumeReclaimPolicy: Retain
  storageClassName: csi-infoscale-sc-block
  volumeMode: Block


---
# Linked PVC
apiVersion: v1
kind: PersistentVolumeClaim
metadata:
  name: pvc-rdm-disk-1
  namespace: openshift-virtualization-os-images # Change as per namespace of InfoScale VM
spec:
  accessModes:
  - ReadWriteMany
  resources:
    requests:
      storage: 40Gi
  storageClassName: csi-infoscale-sc-block
  volumeMode: Block
  volumeName: rdm-disk-1 # Should match static PV above
Following VM configuration and tunables are mandatory for InfoScale VM consuming RDM disk.
  1. Make sure the error policy is marked as report.
    - disk:
       bus: scsi
       errorPolicy: report
     name: rdm-disk1
  2. There is an additional network required in which the underlying physical interface should be wired with NodeNetworkConfigurationPolicy and additional NetworkAttachmentDefination. Visit RedHat documentation for creating and configuring above resources.
  3. Configure the VM to use the additional network and RDM-backed PVCs before starting the InfoScale application workload.Sample VM yaml as below.
    apiVersion: kubevirt.io/v1
     kind: VirtualMachine
     metadata:
     name: infoscale-vcs-sample-vm
         # Must match the namespace of pvc-rdm-disk-1
     namespace: openshift-virtualization-os-images 
     spec:
     runStrategy: Running
     template:
         spec:
         architecture: amd64
         domain:
             devices:
             disks:
             # 1. OS Disk (Example)
             - name: rootdisk
                 bootOrder: 1
                 disk:
                 bus: virtio

           # 2. InfoScale RDM Attachment
           - name: rdm-disk-1-mapping
             lun:
               bus: scsi
               reservation: true
             shareable: true
             errorPolicy: report  
       volumes:
       # 1. OS Volume (Example)
       - name: rootdisk
         containerDisk:
           image: quay.io/kubevirt/rhel9-container-disk-demo:latest

       # 2. InfoScale PVC Attachment
       - name: rdm-disk-1-mapping
         persistentVolumeClaim:
           claimName: pvc-rdm-disk-1

Limitations

With IKE-9.2 release, there are certain limitations for InfoScale VMs consuming these RDM disks with SCSI-3 PR enabled.

CSI operations

Not all CSI operation supported since static PV linked to DMP device is involved. so the resize, snapshot or cloning such resource is prohibited. Hence virtual machine clone or snapshot operation will not be applicable here.

Cold migrations

Virtual machines with SCSI-3 PR disks, by default are not 'LiveMigratable', hence only cold migrations of such VMs are possible.

Upgrade

IKE software upgrade or OpenShift upgrade these VMs should be restarted on another node. IKE software upgrade will be paused if these VMs are in running state, highlighting that the workload should be moved to another node by user.

Hot plugging

Persistent Reservation disks cannot be hot plugged into running VMs, so the virtctl add-vol OR doing the same operation from UI / VM patching is not supported. VMs needs to be restarted for performing hot plugging.

Unplanned node restarts

Unplanned node reboots can cause un-deterministic behavior based on VM configuration. There might be failed VMIs that needs additional cleanup post movement of InfoScale VMs. User can remove the respective finaliser from the PVC if VMI deletion stuck due to application protection gate.