The answer is "maybe". I neglected to give the configuration details,
which in fact seem to matter, so here they are for the record:
Sun Sparc 2 with 48 M running SunOS 4.1.2 plus DBE 1.2
and Sybase 4.8
Several disk drives (Wren 8 and 9) and one Exabyte 8500.
We do not use a third party device driver for the 8500.
At least one person said flatly it was just not possible. One gave me
a reference to a commmercial data recovery service.
According to some respondents, it seems that it is possible to skip past
the logical EOT by using mt fsr or mt fsf with an 8200 but NOT an
8500. I tried anyway and can confirm that indeed there is no way to
skip past Logical EOT for the above configuration at least that I could
find by doing mt fsf or mt fsr.
Some respondents also suggested turning off the drive after initiating
a write at logical EOT. Then one can mt fsf ahead from there after
restoring power to the drive or possibly using programs like gnutar,
badtar or dd to move ahead from that point. I did not try this as the
system that we have the drive connected to is too critical and the data
turned out to be of less value than I had realized to bother removing
the drive to a non-critical system--I certainly would have tried it if
the data had been more important or if I could have moved the drive
without much disruptio--I wish I had tried it so that I could make this
summary complete in my own mind.
One person suggested turning the drive off after sufficient time to
ensure the LEOT mark has been overwritten but I suspect that the
correct thing to do is to turn the drive off immediately after writing
begins since the LEOT must consist of an EOF mark and something else
and the writing begins after the EOF mark so as soon as it starts
writing it has overlayed at least part of whatever the LEOT mark is and
that is enough to prevent the driver from seeing it as an LEOT mark.
Besides you want to overwrite as little as possible of the data
previously on the tape. In my case the data of interest would have
been so far down the tape that this would not have been an issue.
It also seems clear from this that at least on an Exabyte 8500 LEOT is
NOT two EOF marks. Does anyone know exactly what it is? Incidentally,
I have learned that EOF marks on an exabyte are as long as 2 MB of data
so be careful when writing lots of files to a single tape, you can
easily use up the whole tape with "EOF" marks if you have lots of small
files.
There was also a similar question for 1/4 inch cartridge tapes asked
earlier in the year for which a summary was posted with the subject
line:
>Subject: SUMMARY: Reading past the eom on a 1/4" cartridge tape
If you ever need to do this for a quarter inch cartridge drive then
please use that summary since it is detailed and apparently based on
real experience.
Incidentally what I call LEOT (Logical End of Tape) everyone else seems
to call EOM (End of Media) but I can't bring myself to call it this,
since it is not really the physical end of media.
Thanks very much to:
hoogs@alc.com (Tim Hagajen) <-- He phoned so I probably have this wrong
Martin Frost <me@CS.Stanford.EDU>
Postmaster <Piete.Brooks@cl.cam.ac.uk>
sven@alkestis.mpim-bonn.mpg.de (Sven Maurmann)
louis@andataco.com
glawrence@ucsd.edu (Gabriel Lawrence)
raoul@Athena.MIT.EDU
bchivers@smiley.mitre.org (Brent Chivers)
Birger.Wathne@vest.sdata.no (Birger A. Wathne)
joef@VFL.Paramax.COM
Gerald Justice
Unix Systems Manager (still 8-)
Phone: (604) 363-0055 PDT Fax: (604) 363-0045 Telex: 049-7295
Internet: justice@dao.nrc.ca Mail: Dominion Astrophysical Observatory
National Research Council (Canada)
BITNET: justice@nrcdao 5071 W. Saanich Road
Victoria, B.C. CANADA V8X 4M6
---------------------------------------------------------------------------------
Original question:
>From justice Fri Jun 11 02:29:29 1993
>Subject: Can one move past the logical EOT on an 8mm tape?
>
>I have some 8mm tapes that have been recycled overwriting what was once on
>them and thought to be worthless, now that I know otherwise I want to
>try looking at what might still be there from the previous writes.
>Fortunately for me the older data was much bigger and was written as separate
>files and the file I want is one of the last ones on the tape. The newer
>data is much shorter so theoretically the data is still there on the tape
>but the double EOF (or whatever it is that actually marks the logical EOT
>on an exabyte--anyone know what it actually is?) won't let mt or dd skip
>past it. There has to be a way after all these are computers!
---------------------------------------------------------------------------------
Responses:
>Date: Fri, 11 Jun 93 12:49:02 -0700
>From: Martin Frost <me@CS.Stanford.EDU>
>
>If the 8mm tape is written in 8200 format, you should be able to
>'mt fsf NNN' past the logical EOT (at least on an 8200). If it
>is in 8500 format, you may be out of luck.
>Date: Fri, 11 Jun 93 20:59:32 +0100
>From: Postmaster <Piete.Brooks@cl.cam.ac.uk>
>
>* Go to the new EOT.
>* The start writing the tape.
>* Then while it's writing, ZAP THE TAPE DRIVE !
>* This mean that the "end of tape" marker isn't written.
>* Then you just read ..... bingo !
>Date: Fri, 11 Jun 93 22:56:22 +0200
>From: sven@alkestis.mpim-bonn.mpg.de (Sven Maurmann)
>
>it is a little bit tricky to get past the EOM mark of an Exabyte, but
>there is at least a heuristic method. Instead of trying to move for-
>ward by number of files you may try to go record-wise (fsb); this some-
>times fools the controller after a few tries with moves forwards and
>backwards, and you then can access all the data behindd the EOM.
>This is the method by which recently I recovered all of a broken tape,
>so it may be worth a try.
>
>Sven Maurmann
>(Systems administrator at Max-Planck-Institute for Mathematics,
> Bonn -- Germany)
>Date: Fri, 11 Jun 93 15:02:08 PDT
>From: louis@andataco.com
>
>I don't know, but a friend of mine specializes in this sort of
>problem:
>
>Computer Conversions 619 693 1697
>
>That's a US phone number. Being a parochial sort, I don't know how to
>do international phone calls.
>
>Moderation in all things, except love.
>Louis M. Brune ANDATACO
>louis@andataco.com 9550 Waples Street
>619-453-9191 x171 San Diego, CA. 92121
>From: raoul@Athena.MIT.EDU
>Date: Sat, 12 Jun 93 12:37:28 -0400
>
>Hmmm: I've had no problem with "mt -f fsf", after I've already gotten
>to the EOT. EOT is apparently some variation of the magtape double
>EOF marks, so the fsf should get you past it.
>
> Nico Garcia
> raoul@athena.mit.edu
>Date: Sun, 13 Jun 93 20:41:46 EDT
>From: bchivers@smiley.mitre.org (Brent Chivers)
>
>Yes! But not necessarily with any control over what's going on.
>
>I used to do "mt eom" before starting my incremental backups.
>After noting some bizzare behavior, I discovered that if the
>tape was already at EOT when I gave the "mt eom" command, it
>might advance further. Needless to say, this made it very
>difficult to read the data later.
>
>I wrote a script that did "mt asf" commands increasingly
>further into the tape before trying to read, and I did
>find some of my backup sets way out in never-never land.
>________________________________________________________________________
>
> Brent Chivers Mail Stop Z213
> Systems Engineer The MITRE Corporation
> bchivers@mitre.org 7525 Colshire Drive
> (703) 883-6734 McLean, VA 22102
>Date: Mon, 14 Jun 93 08:48:35 +0200
>From: Birger.Wathne@vest.sdata.no (Birger A. Wathne)
>
>On some drives you can use
>mt eom
>dd if=some.big.file of=/dev/nrst9
>Wait until you are shure you are past EOT.
>Turn off tape streamer.
>
>This should get past the first EOT. If the tape has been reused several times,
>there may be many EOT marks around. You could make a guess, and dd a file
>big enough to get past all the EOT's, then use the procedure above to erase
>the last one.
>
>I know this procedure works on my QIC drive. I have never tried it on the
>Exabyte. The file I wanted was the first one on the tape... I used something
>called xtar or tarx. A tar clone that was able to resync on damaged
>archives. It worked!
>
>
>Birger
>Date: Mon, 14 Jun 93 10:40:17 EDT
>From: joef@VFL.Paramax.COM
>
>
>>From sun-managers in January of this year, see below.
[ DELETED to avoid two copies in the summary archives ]
> Might work with
>an 8mm drive. However, if the drive is in compressed mode
>(I.E., an Exabyte 8500 operating compressed), I believe they
>write indexing info up front on the tape, and that would
>probably null out this approach.
>
>Best of luck
>
>
>Joseph M. Fedock joef@vfl.paramax.com
>Paramax Systems Co. voice (215) 648-2573
>Electronic Systems voice (net2) 323-2573
>Valley Forge Labs pager (800) 502-6723 (SE PA only)
>70 E. Swedesford Road pager (215) 578-9513 (all others)
>P.O. Box 517 fax (215) 648-2281
>Paoli, PA 19301
>From: johnb@blas.cis.mcmaster.ca (John Benjamins)
>Date: Tue, 15 Jun 1993 01:00:38 -0400
>
>hmmmm. one thing to try is to go to the EOT (ie. the double EOF, or
>whatever it is). then start a tar command, and turn the tape drive
>off just as it starts writing. this is not for the squeamish!-)
>
>this should get rid of the EOT marker, now the problem is one of
>getting a program that will read over the garbage that is now on the
>tape at that point. i believe there is a program called badtar that
>might help (never used it myself). my first choice would be to use
>gnutar. you might also coerce dd into doing the skipping as well.
>
>hope this helps.
>
>--
> // E. John Benjamins -- <johnb@edge.cis.mcmaster.ca>
>// Calvin: "Verbing weirds language."
>\\ Hobbes: "Maybe we can eventually make language a complete impediment
> \\ to understanding."
This archive was generated by hypermail 2.1.2 : Fri Sep 28 2001 - 23:07:57 CDT