I'm currently working on a project where Azure Site Recovery will be used as a DR solution for a fully virtualized datacenter based on Hyper-V and managed by SCVMM.
The datacenter contains several Linux VMs, and most of them have several virtual disks; those disks can be a mix of IDE and SCSI (depending on how and when the VMs were created), and most Linux systems use old-style mount tables, referring to disks as /dev/hdX
or /dev/sdX
, instead of using GUIDs or volume labels.
We are concerned about what will happen to those VMs when failing them over to Azure; they will get a new disk (the temporary one) which will for sure screw up device numbering, and I'm also wondering if some IDE-to-SCSI conversion will take place, further complicating things.
Also, this is a disaster recovery solution, not a permanent migration: thus it will also be necessary to fail back all those VMs, after a disaster (or a planned DR test) happened and was risolved.
This should be relatively easy to manage on Windows systems (see here), but what about Linux systems, and especially Linux systems which use old-style mount tables?
Will the Azure agent (to be installed in guest systems prior to failing them over) take care of this, or is additional work required?