[NBLUG/talk] Recovering data from a failing hard disk

[Kyle Rankin]

On Tue, Aug 24, 2004 at 03:14:32PM -0700, Lincoln Peters wrote:
> On Mon, 2004-08-23 at 23:42, Lincoln Peters wrote:
> > Now it has gotten a bit silly:
> > 
> > # df -h
> > Filesystem            Size  Used Avail Use% Mounted on
> > [snip]
> > /dev/hdi1             128G   96G   33G  75% /old
> > /dev/hdg1             234G  126G  109G  54% /new
> > 
> > It would seem that the 96GB's of data that I tried to recover from /old
> > are taking up 126GB's on /new!  And "cpio" is *still* running!
> 
> Final count on /new was 170GB.  Oddly enough, I found that all but 11 of
> the files appeared to be intact (no read errors), and all but one of the
> remaining files passed an MD5SUM check against the old data.  The 12
> damaged files, fortunately, will be very easy to rebuild.
> 
> I still wonder why the same 94GB of data on /old (minus the 12 damaged
> files), although unmodified, occupy 170GB on the /new.  Although 65GB
> are still available on /new, so I'm under no pressure to figure this out
> any time soon.
> 
> ---
> Lincoln Peters
> <sampln at sbcglobal.net>
> 
> Any programming language is at its best before it is implemented and used.
> 

Sounds like possibly symlinks were copied as real files or something, so
you ended up having multiple copies of a file instead of symlinks to a
single copy (maybe the same thing with hard links too).

-- 
Kyle Rankin
NBLUG President
The North Bay Linux Users Group
http://nblug.org
IRC: greenfly at irc.freenode.net #nblug 
kyle at nblug.org