Tape overwritten

Mike Ewan mike at raven.uss.tek.com
Thu Jan 31 04:16:56 AEST 1991


In article <1355 at geovision.UUCP>, terry at geovision.uucp (Terry McGonigal) writes:
|> In <1991Jan25.045736.12356 at tkou02.enet.dec.com>
|> diamond at jit345.swstokyo.dec.com (Norman Diamond) writes:
|> 
|> > In article <1991Jan24.213649.2804 at cc.ic.ac.uk> pjc at cc.ic.ac.uk writes:
|> > 
|> > >   TK50 that has had the front of a tar file overwritten by another
|> > >   much shorter one. Can I get at any of the data of the original file
|> > >   on the tape?
|> > >   I am willing to consider manually winding on the TK50 If I have to.
|> > >   MT seems unable to get past the EOF and the first partial block on
|> > >   the tape.
|> > 
|> > The default device used by the mt command is likely to rewind after
|> > obeying your instructions, so you might not know if it has gotten past
|> > the EOF.  Try:
|> > 
|> [ suggestions omitted ]
|> 
|> My experiences with this problem would suggest that it's not software
|> EOF marks that are the trouble here, but rather a hardware EOT mark that
|> the TKxx driver writes to the tape when it's finished writing to it.
|> 
|> Some poking around in man pages and .h's lead me to beleive that there is
|> a command you can give the drive to both write and erase these marks, but
|> I could'nt fingure out how to get the commands through the ioctl interface.
|> It does not, however seem to be possible to get the drive to space past
|> or over one of these marks, the drive `thinks' the tape ends there and gives
|> only i/o errors (I think, it's been a while) if you try to move past it.
|> 
[stuff deleted]

The TK driver sets an error condition on EOF that must be cleared before you
can read the next file.  Simply closing the device and opening it again will
clear the error.  If it's an EOT problem then:
 "mt -f /dev/rmt0h eotdis; mt -f /dev/nrmt0h fsf 2" may work to get past
the EOT mark.

Mike

-- 
 Michael Ewan    (503)627-6468      Internet:  mike at raven.USS.TEK.COM
 Unix Systems Support                   UUCP:  ...!uunet!raven.uss.tek.com!mike
 Tektronix, Inc.                  Compuserve:  73747,2304
"Fig Newton: The force required to accelerate a fig 39.37 inches/sec."--J. Hart



More information about the Comp.unix.ultrix mailing list