Hi Tom,
I followed your guide on the topic of passing through disks on XCP-ng as per the embedded video’s instructions. It worked and I could access my passed through disk (which I will be referring to as “disk A”). Unfortunately however, upon, passing disks through via the use of a symbolic link (eg: Disk A “/dev/sde” to the “/srv/pass_drives”) as you indicated, a serious problem arose:
After a hypervisor system reboot or after a temporary disk A reinsertion, the desired disk A changes from being identified as “/dev/sde” to, for instance, “/dev/sdg”, whereas other disks may then take on the place, name and role of “/dev/sde”. This leads Xen Orchestra or XCP-ng’s “Pass_Drives” storage repository to think there’s either no disk, or worse, points to the a different disk B that, after a reboot, has now taken the location of “/dev/sde”.
This is a major nuisance and security risk.
I think I know what the solution could be. That said, I don’t know how to implement it.
The ideal scenario would be that disk A be passed through based on said disk A’s UUID so that the association between the symbolic link (“srv/pass_drives/”) and the desired disk A itself persists after a restart.
If possible, how can this be achieved?
Thanks a million, Tom! No words describe your contributions. You’re a godsend.