[linux-lvm] A way to bypass LVM and extract the raw data off?

[Holger Grothe]

On Wed, Jan 09, 2002 at 07:01:37AM -0800, IpSo wrote:
> Well, I gave this a shot and everything seemed promising until I mounted the
> loop back file.
> 
> dd if=/dev/hda bs=512 skip=4670127 count=35461377 of=/big/backup_hda6.bin
                             ^^^^^^^
> 
> -rw-r--r--    1 root     root     17105949184 Jan  8 22:05 backup_hda6.bin
> 
> File size seems about right. 
> 
> [root at ipso big]# mount -o loop ./backup_hda6.bin /test -t reiserfs
> mount: wrong fs type, bad option, bad superblock on /dev/loop0,
>        or too many mounted file systems
> 
> Understandable...
> 
> [root at ipso big]# reiserfsck --rebuild-sb ./backup_hda6.bin

IMHO the wrong way. To my experience the mount error message indicates
a "bad filesystem structure", which means you skipped a wrong number of 
blocks when using dd.

> Uhh, what happened to all the data? And theres no lost+found directory either.

Due to the "bad filesystem structure" the filesystem is _more_ or less
initialised by the reiserfsck command. reiserfs has no lost+found directory.

> > > Possible extended partition at offset(249mb)
> > >    Possible partition(Linux LVM physical volume), size(17315mb),
> > offset(2280mb)
> > >       type: 142(0x8E)(Linux LVM physical volume)
> > >       size: 17315mb #s(35461377) s(4670127-40131503)
                                         ^^^^^^^
If I'am right this means sectors #4670127 to #40131503 (35461377 sectors)
are used for the volume. Therfore the dd command should skip only the
first 4670126 sectors:
  dd if=/dev/hda bs=512 skip=4670126 count=35461377 of=/big/backup_hda6.bin
                             ^^^^^^^
Please try.

Holger
-- 
Holger Grothe  (Email: grothe at mathematik.tu-darmstadt.de)
Fachbereich Mathematik, TU Darmstadt