Path: utzoo!utgpu!news-server.csri.toronto.edu!rpi!usc!apple!agate!dog.ee.lbl.gov!elf.ee.lbl.gov!torek
From: torek@elf.ee.lbl.gov (Chris Torek)
Newsgroups: comp.unix.questions
Subject: Re: *** Accidently overwrote V.IMP data tape. Please help ***
Message-ID: <12989@dog.ee.lbl.gov>
Date: 9 May 91 12:31:58 GMT
References: <48701@ut-emx.uucp>
Reply-To: torek@elf.ee.lbl.gov (Chris Torek)
Organization: Lawrence Berkeley Laboratory, Berkeley
Lines: 39
X-Local-Date: Thu, 9 May 91 05:31:58 PDT

In article <48701@ut-emx.uucp> muhammad@chaos.utexas.edu
(Muhammad Pervez) writes:
>We have accidently overwritten our *very* important data tapes!
>[They were] high density 150MB tapes ( 3M DC6150 ).

[A DC6150 is a Quarter Inch Cartridge tape, which tells most of the
story.  You have a SCSI QIC tape drive.]

This question comes up astoundingly often.

There is no easy way to read the tape.  SCSI QIC tape devices always
refuse to let you advance past the point where they believe the tape
ends.  This is not under software control.  (There may be QIC tape
controllers with firmware that does not have this `feature'.  If so, no
one has ever mentioned it here.)

QIC devices usually use a full-width erase head, so that when you write
on track zero it wipes out all the rest of the data.  Thus, if you are
using a 4-track format and have written 1/4th of the tape, none of the
original bits are there anymore.

If not much of the tape has been erased, the remaining tracks still
retain their bits, but they are hard to read.  Probably the easiest
method to read any given bit is to use magnetic developer and a
microscope.  Another possibility is to splice out the overwritten
portion of the tape.  Whether your tape device will be able to handle
this is another question.

>	Thirdly what is the best way to append new files to the
>existing files on the tape.

The safest way to append to a tape is to use a second tape.  Never
write on a tape containing useful data until that data has been
duplicated elsewhere.  (Staging the data on disk will also suffice,
though if the tape is a backup for that disk, you have gone from two
copies to one by doing so.)
-- 
In-Real-Life: Chris Torek, Lawrence Berkeley Lab CSE/EE (+1 415 486 5427)
Berkeley, CA		Domain:	torek@ee.lbl.gov