Skip to content
vJailbreak

Migrating RDM Disk

This guide walks you through the steps required to migrate a VM with RDM (Raw Device Mapping) disks using RDM disks are mostly used of windows machine.

RDM disk migration is only supported for PCD version >= July 2025 (2025.7). For multipath support (connecting to SAN array), PCD version >= October 2025 (2025.10) is required.

Prerequisites

Before you begin, ensure the following:

  1. RDM disk is attached to the Windows machine.
  2. vjailbreak is deployed in your cluster.
  3. PCD Requirements:
    • Minimum version: July 2025 (2025.7).
    • For multipath support (connecting to SAN array): October 2025 (2025.10) - includes default libvirt and QEMU packages
    • Check PCD release notes for exact version names/numbers
    • Volume type must have multi-attach support enabled in OpenStack.
  4. All required fields (like cinderBackendPool and volumeType) are available from your OpenstackCreds.

You can fetch these values using:

Terminal window
openstack volume backend pool list
openstack volume type list

Or by describing the Openstack credentials:

Terminal window
kubectl describe openstackcreds -n migration-system

On Vmware

  • Add the following annotation to the VMware Notes field for the VM:

    VJB_RDM:Hard Disk:volumeRef:"source-name"="abac111"

    Replace Hard Disk with the RDM disk name and abac111 with the actual source ID.

  • To obtain the source-id,source-name or source details, you can run the following command against SAN Array:

    Terminal window
    openstack block storage volume manageable list SAN_Array_reference --os-volume-api-version 3.8

Migration Steps

1. Verify RDM Disk Resource

Check if the RDM disk resource is created in Kubernetes:

Terminal window
kubectl get rdmdisk -n migration-system

2. Fetch RDM Disk Details

For each VM to be migrated, list its details:

Terminal window
kubectl describe vmwaremachine <vm-name> -n migration-system

This will show the RDM disk identifiers.

3. Detach RDM Disk and Power Off VM in Vmware

Since VMware does not allow snapshots of a VM with attached RDM disks, you must:

  • Power off the VM to be migrated.
  • Detach the RDM disk from the VM.

Steps to detach RDM disks in vmware

  1. For each VM go to the Edit Settings, click on the cross icon near the RDM disks, keep “Delete files from storage” unchecked.
  2. For each VM go to the Edit Settings,click on Remove the SCSI controller used by these disks, this will be in Physical sharing mode.

Detach RDM Disk in vmware

This ensures the snapshot and migration can proceed without errors.

4. Patch RDM Disk With Required Fields

Edit the RDM disk to add cinderBackendPool and volumeType. Example:

Terminal window
kubectl patch rdmdisk <name_of_rdmdisk_resource> -n migration-system -p '{"spec":{"openstackVolumeRef":{"cinderBackendPool":"backendpool_name","volumeType":"volume_type"}}}' --type=merge

5. Create MigrationPlan

Finally, create a migration plan using the CLI.
Follow the detailed CLI steps here:

Migrating Using CLI and Kubectl

6. Wait For Disk To Become Available

Confirm that the disk is in Available state:

Terminal window
kubectl get rdmdisk <disk-id> -n migration-system -o yaml

Look for:

status:
  phase: Available

7. Ensure all VM’s of cluster is migrated

  1. Check RDM disk is available as volume in PCD or openstack

  2. Ensure all VM’s of cluster is migrated

  3. Power on all VM’s together

Rollback plan - if migration fails

  1. Delete VMs created in PCD or openstack

  2. Delete managed volume openstack volume delete volumeid --remote

  3. Attach RDM disk in VMware to powered off VM’s

    • Add the reference VMDK disks

    • Add New Device > Existing Hard Disk. This will add the disk as New Hard disk.

    • Change the controller of this hard disk to “New SCSI Controller” which we created in firs step.

    Repeat the process for all the RDM disk.

Re attach RDM disk on failure

  1. Power on all the VM’s on Vmware