ext3 corrupted

[Tyler Aviss]

I've had good luck with this method as well. I've managed to stick
hard-drives in sealed freezer-bags from both laptops and desktops,
then get them running long enough to pull the data off. Sometimes the
cold is enough to unstick the bearings and get the suckers runnings,
but as soon as it's up make sure to copy all the data because it
usually doesn't last long.

I use either
dd if=/dev/hda of=/path/to/imagefile
(add blocksize and other arguements as needed, I generally go with "bs=1k")

or if you can get a good shot at the filesystem:
tar -cf - * | tar -xf - -C /path/to/new/location

On 7/23/07, Allen Taylor <tlug-G8usDCtqe957Ar2qsurDTA at public.gmane.org> wrote:
> On Mon, Jul 23, 2007 at 09:34:33AM -0400, Madison Kelly wrote:
> > Lennart Sorensen wrote:
> > >On Sat, Jul 21, 2007 at 05:37:19PM -0400, Marc Lijour wrote:
> > >>I have a drive that got corrupted (just before it started to send SMART
> > >>alerts). I can't mount it and I get this error message:
> > >>
> > >>EXT3-fs error (device sda12): ext3_check_descriptors: Block bitmap for
> > >>group 896 not in group (block 16777317)!
> > >>
> > >>I can't mount it as ext2 either, and I can't mount sda either. auto as
> > >>type does not work either.
> > >>
> > >>Running fsck and accepting all actions with a 'y' seems to have no
> > >>effect. I can't mount the disk and when I get to do fsck again I get the
> > >>same actions.
> > >>
> > >>dd works.
> > >>
> > >>Do you know a way to get the data back?
> > >>
> > >>Thanks a lot (!)
> > >
> > >Well you could try fsck with the -b 8193 option to tell it to use an
> > >alternate superblock.  Might work.
> > >
> > >If the drive is actually going bad, certainly the advice of letting it
> > >cool off is a good one.  I actually managed to recover all the data from
> > >a failed drive for my sister that refused to even try to boot, by trying
> > >it out right after getting home in the middle of winter while the
> > >computer was probably at about -10C.  Read everything perfectly and got
> > >all the data of it and onto new dists (I replaced it with raid1).
> > >Didn't even get a single read error, it just ran perfectly.  I have read
> > >that a lot of recovery people will actually freeze the drive before
> > >trying to read it since it apparently works quite well in many cases
> > >without having to do anything complicated.
> >
> > Just be careful to do the post-freeze DR in a low humidity environment!
> > Condensation could make things worse, if you are not careful. :)
>
> To address the humidity issue, I place the drive in a "zip lock" bag
> before putting it in the freezer. The next day I only open the zip lock
> bag enough to attach power and data cable and run that way.
>
> I've done this a few times with good results about half the time. The
> other half I suspect would be only be recoverable by the above mentioned
> experts.  YMMV
>
> Allen
>
> --
> The Toronto Linux Users Group.      Meetings: http://gtalug.org/
> TLUG requests: Linux topics, No HTML, wrap text below 80 columns
> How to UNSUBSCRIBE: http://gtalug.org/wiki/Mailing_lists
>
--
The Toronto Linux Users Group.      Meetings: http://gtalug.org/
TLUG requests: Linux topics, No HTML, wrap text below 80 columns
How to UNSUBSCRIBE: http://gtalug.org/wiki/Mailing_lists