Backing up and restoring RDM disks presents unusual challenges. The original backed-up RDM configuration might not apply, and is probably not appropriate, if users restore:
- A virtual machine to a different host or datastore.
- A virtual machine that was deleted, when its originally mapped RDM was also deleted, or the containing LUN was repurposed and rewritten.
- The RDM to a different virtual machine, even if that virtual machine is on the same host and datastore. Users might do this to access files on the disk, or to test a restore.
When performing a proxy backup of an RDM disk, you must present the same LUN ID to both the ESXi host and the proxy server. (This restriction does not apply to VMFS disk because the virtual disk library reads the VMFS header and matching UUID. But for RDM the host and proxy require the same LUN ID.)
Restoring RDM disks is appropriate if the original virtual machine’s VMX file and disk mapping is no longer available, but the LUN containing the RDM is still available. In this case, the RDM image on the LUN might still be valid, so it does not need to be restored. If this is true, do not make changes to the RDM configuration during your restore operations. Instead, complete the restore process in two phases:
- Restore the virtual machine configuration (VMX) and system disk. This restores the virtual machine, but does not restore the RDM.
- Add the RDM disk to the virtual machine. After doing so, you can complete normal restore operations on the RDM disk.
Alternatively, it is possible to create a virtual machine that can host the RDM disk and access its contents. After you create the virtual machine, restore its virtual machine configuration (VMX) from backup, and then restore any selected disks.