Path: utzoo!utgpu!utstat!jarvis.csri.toronto.edu!mailrus!tut.cis.ohio-state.edu!ucbvax!decwrl!sun!pitstop!sundc!seismo!uunet!mcvax!ukc!strath-cs!glasgow!bru-cc!linda
From: linda@cc.brunel.ac.uk (Linda Birmingham)
Newsgroups: comp.unix.questions
Subject: Re: reading dump tape with bad spots
Message-ID: <643@Terra.cc.brunel.ac.uk>
Date: 21 Feb 89 14:28:29 GMT
References: <1084@wpg.UUCP> <472@avsd.UUCP> <1086@wpg.UUCP>
Reply-To: linda@cc.brunel.ac.uk (Linda Birmingham)
Organization: Brunel University, Uxbridge, UK
Lines: 37

In article <1086@wpg.UUCP> russ@wpg.UUCP (Russell Lawrence) writes:
>> In article <1084@wpg.UUCP> I wrote:
>> >Does anyone know of any PD dump/dumpdir/restor programs persistent 
>> >enough to skip over bad spots on tape?  
>
>In article <472@avsd.UUCP>, childers@avsd.UUCP (Richard Childers) writes:
Advice on validation of dumps.
Having had tape errors on a couple of tapes recently, we are becoming
a little paranoid about dumps too.

I thought about using dd as a validation method but we found that that
read a particular tape error as end of file so to all intents and purposes
the tape looked ok, except it only read 10 blocks %-}.

I was also advised to restore the last file from a dump tape.
That sounds a more rigorous validation to me.

>I'm still hoping that someone will suggest a recovery procedure that will
>bail me out.  Nevertheless, the experience has convinced me that I 
>should abandon dump/restor and opt instead for ctar or pax.... given
>their ability to skip over bad spots.
Sounds interesting.
>
>If I knew more about dump headers, I suppose I could hack something out 
>that could look for the appropriate file on the tape (ie inode number) 
>and retrieve the data.  Any comments?  
I think that's a good idea Russell, let me know when you've done it :-)
Someone somewhere MUST have already done it though.

BTW we also have a Pyramid 9820 with a Fujitsu tape drive.

Linda.


-- 
Brunel University, Uxbridge, Middlesex, England.
janet: linda@uk.ac.brunel.cc |  :-)
uucp:...ukc!cc.brunel!linda  |