Frederick Bruckman writes:
> On Sat, 26 Jun 1999, John Valdes wrote:
> 
> > You'll need to use a separate dump command for each filesystem, eg:
> > 
> >       dump -0 -u -B 4096000 -f /dev/nrst0 /
> >       dump -0 -u -B 4096000 -f /dev/nrst0 /usr
> >       dump -0 -u -B 4096000 -f /dev/nrst0 /var
> >       dump -0 -u -B 4096000 -f /dev/nrst0 /home
> 
> It's even easier than that. Since /dev/nrst0 is the default, 

The man page sez /dev/rst0 is the default:

  FILES
       /dev/rst0       default tape unit to dump to

but doing a "strings /sbin/dump | grep rst0" (don't have the source
loaded at the moment... ;) ) shows that /dev/nrst0 is the default...

> 	dump -0uB 4096000 / /usr /var /home
> 
> does the same thing.

Does that create a single dump image containing all four filesystems,
or four individual dump images (ie, are there filemarks on tape
between the dumps for each fs)?

> I have mixed feelings about multiple files on one
> tape, though. On the one hand, it's economical of tape; on the other,
> it will take just about as long, in this example, to restore /home as
> it does to restore all four.

Hmm... it depends on your tape drive, but most tape drives (the
Exabyte 8200 being an exception, I believe) can space forward (and
backward) between file marks quickly.  So to restore /home, one can
quickly fast forward to the 4 file (from the beginning of tape,
'restore -s 4 -r' or 'mt fsf 3 ; restore -r') w/o having to read the
previous dump images.

> Um... No. "disklabel -r ..." writes a disklabel, but the kernel never
> reads it.

If one has a disk used only by NetBSD, does this mean that it still
has to be partitioned w/ a MacOS disk utility?

John