Why swift uses devices (/dev/sdb1) not UUIDs?

asked 2016-02-05 06:18:04 -0600

rozie gravatar image

I wonder why in every example I found ( http://docs.openstack.org/kilo/instal... ) partitions are added to fstab using device notation, not UUIDs. Since a few years UUIDs are default way in Linux distributions.

I believe using UUIDs would be more reliabe, as there is no chance for device order change if mounted in such way. I'm just not sure if this is intentional as an example (I believe there should be notice in documentation then to use UUIDs), is there any good reason for this or it was just oversight?

I'm not talking about mountpoints (/srv/node/sdb1) or ring topology, just devices in fstab.

edit retag flag offensive close merge delete