> > What would you suggest as a smart approach to volume management?
>
> What matters isnt "which physical disk" its "where is the file system that
> used to be /home". So UUID's are ideal for this (volume labelling).
If the concept of the UUID is similar to NT's "signatures" in the MBR's of
each physical disk, I think it is a good idea. As long as we don't do some
of the silly things that NT seems to do, like assign drive letters to the
first primary partition on each drive, and then go back and do everything
else.
> > > And wherehasitgone is a tool to walk the disk tree and find a volume by uuid
> > > and/or maybe ask LDAP/NIS maps to find it via NFS
I wonder what performance considerations we have for walking the local
disks looking for UUID's as opposed to untarring the devfs thing which is
presumably stored in a single location.
> I've no idea. That depends how you add such items to mount. ext2fs includes a
> libuuid and all e2fs file systems have uuids. DOS fs has a serial number on
> each disk, etc.
The actual fs has a UUID, as opposed to the partition in which it resides?
If we take a partition table approach, then we aren't being biased towards
ext2....
If I am not missing something obvious in this discussion, then my
question is:
Is there room for taking a partition table approach to this form of "disk
management?" Without clobbering the info that other OS' store (such as
NT)?
One more question: How universally unique are the UUID's? If I copy the
image of a fs from one system to another, what are the chances of a
"clash?"
/***************************** Eric McDonald ************************
Mail: emcdonal@cps.cmich.edu; Eric.A.McDonald@cmich.edu; \
Eric.McDonald@usa.net
Finger: emcdonal@mail.cps.cmich.edu (when it is up)
POTS: Home: V[/F/D]: 517-772-9209; Work: V: 517-773-1201
HTTP: [HTML docs under revision]
"Numquam periculum sine periculo vincitur." (Publilius Syrus)
"Never is danger conquered without danger."
********************************************************************/
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.altern.org/andrebalsa/doc/lkml-faq.html