dump and TK50

[George Robbins]

In article <1990Jul25.110248.14366 at squirrel.mh.nl> Johan Vromans <jv at mh.nl> writes:
> I have noticed that dump(8) does not utilize TK50 tapes as it should.
> 
> A TK50 can hold about 90Mb of (formatted) data. To dump 200Mb of disk,
> dump(8) uses 3.24 tapes. This reates about 60Mb per tape.
> When I supply dump with a huge tape length argument, say 13200, it
> needs only about 2.5 tapes to dump the same amount of data. However,
> dump(8) calculates that it needs 0.3 tapes...

Because of the streaming technique used on the TK50, the exact capacity
varies according to the system performance.  Any time the system isn't
ready, the TK50 wastes a little bit of tape.  If it still isn't ready,
it will backspace, stop and wait, so the tape isn't wasted, but it runs
real slow.

If you accept the default Ultrix tape sizes then it will use a "safe"
figure that will "never" hit the end of the tape.  However, since Ultrix 3.x
there is automatic end-of-tape sensing that you can use instead of
letting dump play it safe.  What I do is specify 10 times the nominal
length to force dump to rely on end-of-tape sensing, but still be able
to make some sense out of the estimated number of tapes needed.

-- 
George Robbins - now working for,     uucp:   {uunet|pyramid|rutgers}!cbmvax!grr
but no way officially representing:   domain: grr at cbmvax.commodore.com
Commodore, Engineering Department     phone:  215-431-9349 (only by moonlite)