Better handle unrealized/unmanaged devices. See the upstream bugzilla for more details.
The rollback now restores the previous unmanaged status of a
device. For example, if you perform a checkpoint on eth0, then set
'nmcli device modify eth0 managed no', the rollback should make it
The branch also fixes the case in which a software device is
snapshotted and then removed. The rollback now should recreate the
Also, passing an empty list to the snapshot functionality is
equivalent to passing all devices (including unmanaged ones).
There are a lot possible scenarios and so testing everything is
probably difficult. Here is a list in non-increasing importance order
of what could be tested:
- checkpoint with no timeout, manual rollback
- checkpoint with timeout, automatic rollback
- checkpoint/rollback of multiple devices
- global checkpoint (all devices)
- check that the explicitly-unmanaged status of a device is restored
- check that software devices are recreated on rollback
- checkpoint flags 
Since the problem described in this bug report should be
resolved in a recent advisory, it has been closed with a
resolution of ERRATA.
For information on the advisory, and where to find the updated
files, follow the link below.
If the solution does not work for you, open a new bug report.