March 25 thread "Qgroups are not applied when snapshotting a subvol?"
Short answer is:
btrfs subvolume snapshot -i 1/1
The problem being that any user can make snapshots and thus escape the
quota by omitting -i. However for containers, presumably it's some
daemon behind the scene that is
Just some food for thought: there's already a tag that correctly assigns
filesystem objects to users. It is called owner(ship). Instead of making
qgroups repeat ownership logic, why not base qgroup assignments on
ownership itself? (At least on per-subvolume basis.)
--
With Best Regards,
Hi guys,
If a qgroup is created for a btrfs subvolume /some/path and limits are set and
a new btrfs subvolume /some/path/bla is created it does not inherit the parent
subvolume's /some/path qgroup and limits. The only way to achieve something
similar is to create a common "parent" qgroup and