I wrote the mail below, changed the subject and then came back to
review.... something caught my eye, see ***** below.
On Wed, Aug 06, 2008 at 08:32:04AM -0400, Michael Habashy wrote:
> Okay - thanks for your reply..sorry about top posting..i am in
> gmail..and it does what it wants.
> So let me answer your questions:
> On Tue, Aug 5, 2008 at 10:57 PM, Andrew Sackville-West
> >why do you think this? What evidence tells you this? have you looked
> > at /proc/mdstat?
> i looked into /proc/mdstat -- it has inactive raids.
> I have tried to re-correct the problem by removing and adding to the raid.
> it has not helped. I get : Block bitmap for group 1664 not in group
> EXT3-fs: group descriptors corrupted!
***** what the heck is ext3 doing in the picture at this point? Is
this after you've tried to build the array and mount it? Are you
running a degraded array that also has some fs corruption on it?
you've got to try to get network up on this machine so you can
get us real information. If you can get ahold of a live-cd
(knoppix) then you could use that to work on it. It'll be much
more comfortable than working from busybox... now back to the
previously scheduled response.
when you look at /proc/mdstat, it will show you which drives are
md1 : active raid5 hde5 hdc5 hda5 hdg5
14650944 blocks level 5, 64k chunk, algorithm 2 [4/4] [UUUU]
this is part of my mdstat. note the four U's at the end, that means
those disks are up. are you able to get information like this or has
the array completely failed?
> Mounting /dev/mapper/vg_main-lv_root on /root failed: Invalid arguement.
> then..there is a whole mess of failed scripts.
okay. if the raid has failed, then the vg won't be there to mount, so
that makes sense.
> the system boots..it just goes into Busybox.
> It never mounts any of the filesystems. the /boot filesystem seems to
> be okay..because i create a directory and mount it there.
> and i am able to see all the other kernels and grub directory too.
have you tried booting from older kernels yet? any luck there? I'm
hoping that what you're seeing is a problem with the kernel and
rolling back to an older one will help...
> the issue comes when i try to mount my /dev/md1 which has / and /home
> and all the other filesystems.
> that has the lvm2 volume groups.
did you say at some point that you *can* mount the filesystems when
you boot the rescue cd? Or does it fail to build the array there as
> I think i need to get the error "Group descriptor corrupted" error
> fixed, yes?
> how do i get past that?
don't know. If just one drive in the array has failed, you can bring
it back by adding another drive to the machine, and telling mdadm
about it... something like:
mdadm --fail /dev/hda(partition that's dead) --remove
/dev/hda(partition that's dead) --add /dev/hd(disk you've added)
but that's only a guess.
If the array has failed completely (for raid 5 this means two disks
have failed), then you're sunk, so far as I know. This is really
beyond my expertise. I'm changing the subject of this thread in the
hopes it will bring more expert eyes. Be prepared to provide
*specific* information, even if it means you have to laboriously
hand-copy text from the console...
> i am able to boot into rescue mode and execute commands.
> Please be more specific iwth the commands you want me to execute.
> To UNSUBSCRIBE, email to debian-user-REQUEST@lists.debian.org
> with a subject of "unsubscribe". Trouble? Contact firstname.lastname@example.org