-
The same mounting used to probe the subvolumes could be used to probe the qgroups.
-
The same mounting used during the libstorage-ng's commit phase to create the subvolumes could be used to modify the qgroups.
-
The libstorage-ng's actiongraph would contain the actions to modify the qgroups, no need to "enrich" it with additional compound actions in the Ruby layer.
-
Waiting/forcing for device synchronization is so far libstorage-ng's bussiness (calls to
udevadm settle
, etc). and is likely also needed for gqroups (rescan and sync).
The content of this section was moved to https://jira.suse.com/browse/SLE-16696. See the bullet points in the description there.