[ale] How to fix a crashed disk with fsck or debugfs

[Gary Maltzen]

Then it's a question of how much time it will take to restore the drive
versus how much time it will take reloading software. Unfortunately I only
know of Win/DOS-based recovery tools and I don't know how well they will
recover Linux partition types.

If you have a sector editor, you can do this manually - I used to do it with
Norton DM (DiskMaint?) in "Maintenance" mode.

The first primary partition (sda1) usually starts at cylinder 0, head 1,
sector 1 - from that you can determine the length of the partition.
Assuming, for example, that the partition is 100 cylinders in length, the
next partition would start at cylinder 100, head 0, sector 1. This
frequently is an extended (type 5) partition so the 0-sector of the
partition is a partition table describing (1) the first logical partition
and (2) the rest of the drive.

Chances are if sda1 is primary and sda2 is extended, recovering the master
partition table (in MBR) may suffice to get you back online.

(On a perverse day, I'd just suggest reloading from your latest backup ;-)

> no fdisk /dev/sda does not show the partitioning scheme.... <with tears in
> my eyes>
>
> -----Original Message-----
>
> Based on your description so far, you may be SOL.
>
> Question: does 'fdisk /dev/sda' properly show the partitioning scheme?
> (I'm wondering if the MBR is intact)
>
> >Here is the problem, had a pwoer failure and fcsk reported errors. It got
> >the ide drives up and working but not the scsi. Here are the details;
> >
> >sda is 8746MB which is a IBM DNES 309170W SCSI (1115cyl 255head 63sec)
> >- sda6 is /m01 6087MB
> >- sda7 is /home 1655MB
> >
> >Somehow I lost power and crashed the drives. Tried to run
> >fsck -A -V ; echo == $? ==
> >got ==8== which is defined as an OPERATOR ERROR
> >
> >Then tried fsck -t ext2 -b 8193 /dev/sda6 and got
> >'bad magic number in super-block when trying to open
/dev/sda6...superblock
> >does not describe a correct ext2 filesystem"
> >
> >Tried superblocks at 16385 and 24577 same message...
> >
> >Also tried dumbe2fs /dev/sda6 and got the same bad magic number stuff...


--
To unsubscribe: mail majordomo at ale.org with "unsubscribe ale" in message body.