Comment 5 for bug 1721787

Goutham, thanks for looking into this and for your comments!

Concerning performance: I'd think the time to run two commands (and copy the UUID in between)
will probably be longer than the increased runtime for sieving through all shares.
Also, I'd expect the number of shares in Manila to be small compared to the number of
volumes in Cinder (in our deployment the ratio is currently ~1:100), and Cinder finds
a volume by name in ~1 sec (sieving through our +5000 active volumes). So, I'd opt for
admin convenience (and consistency) here :)

As for collisions: as this whole issue is limited to admins, I think handling the
collisions is ok. As an admin looking through all projects you should not be surprised
to have several shares with the same name.

Anyway, all this is of course not a big problem. As you mentioned, the main thing may be
to establish consistency across projects.