Wouldn't that imply starting hald before mounting the disks? man, 27.09.2004 kl. 20.27 skrev Ola Thoresen: > Mon, 27 Sep 2004 at 16:05 GMT Elliot Lee <sopwith@xxxxxxxxxx> wrote > > On Sun, 26 Sep 2004, Arjan van de Ven wrote: > > > >> > >> > to ensure that the random number would be virtually unique which means > >> > long numbers ;-). > >> > >> well but ext2/3 do have UUID's; there's no reason we can't do > >> mount-by-UUID as well. > > > > We do have mount-by-UUID: > > > > mount UUID=b04a5cd1-3f15-45d3-8c25-3ef732eb3d82 /mnt/point > > > I think it's easier to remember hda1 or hdc5 than > b04a5cd1-3f15-45d3-8c25-3ef732eb3d82 ... > > But how about some kind of "fallback" to mount by uid if duplicate > labels are found? > > At install time or any time the admin wants, a script is run, writing > something like /etc/sysconfig/mountpoints with > <LABEL> <UUID> > > / b04a5cd1-3f15-45d3-8c25-3ef732eb3d82 > /boot 0f59430f-d62e-4de8-b553-7760822293ca > > > This info is _only_ checked if there are duplicate labels found. > > That way you can still have the easy to remember LABEL=/ in fstab, but > still avoid problems if more than one partition has the same label. > > But this start to sound like something hald or udev (or whatever is > responsible for this kind of actions) could do better? > > > Rgds. > > Ola Thoresen > > >
