[OpenAFS] Re: help, salvaged volume won't come back online, is it corrupt? [trimmed log]

Marcus Watts mdw@umich.edu
Wed, 13 Sep 2006 03:38:12 -0400

Adam Megacz <megacz@cs.berkeley.edu> writes:
> To: openafs-info@openafs.org
> From: Adam Megacz <megacz@cs.berkeley.edu>
> Subject: [OpenAFS] Re: help, salvaged volume won't come back online, is it corrupt?
>  [trimmed log]
> Sender: openafs-info-admin@openafs.org
> Errors-To: openafs-info-admin@openafs.org
> X-BeenThere: openafs-info@openafs.org
> X-Mailman-Version: 2.0.4
> Precedence: bulk
> Date: Wed, 13 Sep 2006 00:13:48 -0700
> John Koyle <jkoyle@koyle.org> writes:
> > That's the problem I'm having and it appears to be due to the root
> > vnode being missing or corrupt.  Apparantly the bos salvager cannot
> > fix this problem.  From the research I've done, it doesn't appear to
> > be possible (not easily anyway).
> Even if the root vnode is missing, I'd be quite happy with it
> reattaching whatever vnodes are floating around to some "bit bucket"
> directory the way ext2 puts stuff in lost+found.  Or even just dumping
> every *file* to a flat directory full of files I can grep through to
> find the handful that I need.
>   - a
> -- 
> PGP/GPG: 5C9F F366 C9CF 2145 E770  B1B8 EFB1 462D A146 C380
> _______________________________________________
> OpenAFS-info mailing list
> OpenAFS-info@openafs.org
> https://lists.openafs.org/mailman/listinfo/openafs-info

Uh -- I wouldn't swear to it, but, yes, lost+found has existed
since 7th edition Unix.  And, at least as late as some flavor
of sunos, if you roached the root inode, you were still SOL.
I remember a fairly gruesome experience once where I ran
newfs on a scratch partition of the same size and copied over
the first bit of data merely in order to get the root inode back.
I don't recommend this particular form of data recovery.
I'm actually hoping this was just a bad dream.  I also remember
piecing back together a large /etc/passwd+/etc/shadow pair of files
that somebody accidently rm'd.  That was painful too.
And then there was the disk controller which had an off-by
one half track write error, on hot days in the sun...
You know, you're bringing back some really bad memories...

How would you manage acl's in an AFS lost+found ?
Keep in mind that in a large environment, you do not want
your AFS administrators to be in the business of doing
fine grained data management.

You might want to try these 2 tools:
These will crawl through the local storage directly,
with some understanding of afs semantics.  volinfo will
print out what it finds; voldump will attempt to produce
a vos dump.  Assuming the latter actually contains interesting
contents (without the root vnode, I wouldn't count on this),
you may be able to use afs tools to munge your vos dump.
If you're really clever, you may be able to fabricate
a new root vnode and splice it into your afs dump,
although I would count this a fairly desperate measure.