> This was working until recently. There are two possibilities here.
> 
> 1) The system may have developed a hardware problem.
> 
> 2) The dump script is flawed and the flaw is causing a problem now
>    because the partitions have exceeded the size I specify with the
>    dump parameters.

Don't think it's 2), see below ...

> b=126
> d=108000	# original Exabyte uses 54000, Eliant is double that
> s=8500		# for 160m tape (note that 160*6000/112=8571)
> 		# standard 112m uses 6000
> f=/dev/nrst0
> dump 0ubdsf $b $d $s $f /

I've been using "-d 54000 -s 6000 -b 56" on an 8500 and now an 8505 without 
problems.  I recall reading somewhere that 56 was the optimum block size for
Exabyte.  Not sure if it applies to newer drives.  Of course, once you have
done one dump, the amount of tape left will be less than these figures, but
that should be fine - all I think you really need to do is to stop dump from
thinking the tape is shorter than it is.

> > dump 0ubdsf 126 108000 8500 /dev/nrst0 /

> >   DUMP: estimated 62012 tape blocks on 0.01 tape(s).

> >   DUMP: DUMP IS DONE
> > dump 0ubdsf 126 108000 8500 /dev/nrst0 /usr

> >   DUMP: estimated 1224630 tape blocks on 0.18 tape(s).

> >   DUMP: End of tape detected

The first dump is fine and uses a small fraction of the tape.  The second
dump estimate is for less than one tape.  However, dump gets an EOM from
the drive, so it prompts for a new tape.

When I had problems with my 8500, the tape would just stop, but I don't know
how the Eliant would behave.  I take it you've tried different tapes and also
cleaning the heads?

J

-- 
                    My other computer also runs NetBSD
                          http://www.netbsd.org/