help with tar on SGI 1/4" tape drive

Dave Olson olson at anchor.esd.sgi.com
Sat Jan 19 05:51:20 AEST 1991


In <1991Jan17.195800.14456 at portia.Stanford.EDU> dhinds at elaine15.stanford.edu (David Hinds) writes:
| In article <1991Jan17.003628.17045 at odin.corp.sgi.com> olson at newmedia.esd.sgi.com (Dave Olson) writes:
| >|    I have a tape that has been overwritten.  The user had created a tar
| >|    archive tape previously.  He created a new directory
| >|    and then typed "tar cv" instead of "tar xv", so he has destroyed at
| >|    least the first 400 blocks of the archive.  Is there any way to extract
| >|    the rest of the archive?
| >| 
| >Sorry, but the QIC drive specs allow no way to get past the EOD.  You
| >are out of luck (it isn't an issue with the tape driver, but rather
| >with the drive firmware and hardware).
| 
| Well, there is a way to get past the EOD, but I'm not sure how much good it
| will do you.  I did this recently when I accidentally overwrote the header
| of a 'bru' archive.  I just overwrote the EOD I wanted to get past with a
| short archive, but ejected the tape before the archive finished.

This is the key.  If you ejected the tape, the tape drive never created
the EOD indication.  Therefore it simply looks like a bad data block
(or possibly a media error, depending on how rough you were :) ).

| When I then re-read the tape, 'bru' read my new archive header and the truncated
| part of that archive, then registered a hard tape error, and re-synchronized
| with the original archive I was trying to recover.  I recovered the rest of
| the tape without any problems.  The reason I don't think this will help you
| is that 'tar' can't recover from this kind of error in a tape.

Sure it will.  The 'e' option of tar is intended for exactly this kind
of problem ('corrupted' data on the tape.  However, it won't help for
the more common case of the original posting, since in that case the
drive did write the EOD indicator.
--

	Dave Olson

Life would be so much easier if we could just look at the source code.



More information about the Comp.sys.sgi mailing list