This is unreleased documentation for SUSE® Virtualization v1.8 (Dev).

Storage Migration

Storage migration allows you to move virtual machine volume data between different StorageClasses. This is useful for operations that require you to change the underlying storage, such as the following:

  • Migrating from replicated (SUSE Storage) to local (LVM) storage for better performance

  • Migrating from local to replicated storage for higher data availability

SUSE Virtualization supports two modes of storage migration:

  • Online storage migration: Migrates volumes while the virtual machine is running. The guest operating system is unaware of the underlying disk movement.

  • Offline data migration: Clones volume data to a new StorageClass when the volume is not attached to any running virtual machine.

Prerequisites

  • The target StorageClass already exists in the cluster.

  • The target storage backend has sufficient capacity to hold the volume data.

  • For online migration, the virtual machine is live-migratable and running.

Online storage migration

Use online storage migration when a virtual machine must remain running while you switch its StorageClass. The process utilizes KubeVirt’s volume migration mechanism to live-migrate the virtual machine while seamlessly copying volume data to the target destination in the background.

Online storage migration steps

  1. Create a new PVC using the target StorageClass. The PVC’s capacity must be equal to or greater than the source volume.

  2. On the SUSE Virtualization UI, go to the Virtual Machines screen.

  3. Locate the target virtual machine, and then select ⋮ → Migrate Storage.

  4. Select the volume you want to migrate and the target PVC.

  5. Confirm the migration.

    SUSE Virtualization begins copying volume data in the background while the virtual machine continues running.

  6. Monitor the progress on the virtual machine details screen.

Canceling an online storage migration

  1. On the SUSE Virtualization UI, go to the Virtual Machines screen.

  2. Locate the target virtual machine, and then select ⋮ → Cancel Storage Migration.

    The controller aborts the ongoing migration and the virtual machine continues to run with the original volume. No data is lost.

  3. (Optional) Delete the target PVC.

    The PVC is not automatically deleted. To avoid orphaned resources, manually delete the PVC if it is no longer required.

Online storage migration limitations

The following limitations are inherited from KubeVirt’s volume migration:

  • The virtual machine must be migrated to a different node during storage migration. Same-node storage migration is not supported.

  • Shareable disks are not supported due to data consistency concerns with multiple writers.

  • Hotplugged disks cannot be migrated.

  • virtiofs disks are not supported as they lack live-migration capability.

  • LUN disks cannot be migrated when the destination StorageClass differs from the source.

  • Migrating between two ReadWriteOnce PVCs backed by local storage is only permitted when the PVCs reside on different nodes.

  • Only PersistentVolumeClaim and DataVolume sources are supported.

Offline data migration

Use offline migration for volumes that are not attached to a running virtual machine. This process utilizes the Containerized Data Importer (CDI) to clone data to a new PVC with the target StorageClass.

Offline data migration steps

  1. Verify that the volume is not attached to any running virtual machine.

  2. On the SUSE Virtualization UI, go to the Virtual Machines screen.

  3. Locate the target virtual machine, and then select ⋮ → Data Migration.

  4. Specify the target volume and the target StorageClass.

  5. Click Confirm.

    SUSE Virtualization creates a CDI DataVolume resource that clones the data from the source volume.

  6. Monitor the progress on the Volumes screen (DataVolume status).

Once the migration is completed, the new volume appears in the Volumes list. You can then replace it in a virtual machine.

Offline data migration limitations

  • The source volume must be in Bound status and must not be attached to a running virtual machine.

  • The target volume name must not conflict with an existing PVC or DataVolume in the same namespace.

  • The target StorageClass must exist.

Troubleshooting

Online storage migration failure

Online migration can fail due to insufficient space on the target storage, network issues, node failures, and other reasons.

When a failure occurs, KubeVirt automatically enters a retry loop that continues until the migration succeeds or you explicitly cancel it. To determine the root cause of a failure, check the virtual machine’s status and conditions.

ManualRecoveryRequired condition

In some cases, a failed volume migration may set a ManualRecoveryRequired condition on the VirtualMachine resource. When this occurs, perform the following actions:

  1. Check the condition message for details about the failure.

  2. Resolve the underlying cause (for example, insufficient storage capacity or network connectivity issue).

  3. Clear the condition by following the instructions in the KubeVirt documentation.

Offline data migration failure

  1. Check the DataVolume resource’s status for error details.

    kubectl get dv -n <namespace> <target-volume-name> -o yaml`.

  2. Resolve the underlying cause (for example, insufficient storage capacity or incompatible StorageClass configurations).

  3. Delete the failed DataVolume resource and retry.