[afnog] ad0 FAILURE

[Phil Regnauld]

On Mon, Oct 03, 2005 at 07:17:37AM -0700, Randy Bush wrote:
> >> ad0: FAILURE - READ_MUL status=59<READY, DSC,  DRQ, ERROR> error=40<UNCORRECTABLE> LBA= 1505695
> >> I googled it and all points to hard disk failure.
> > Nope, it's a hardddisk error.
> 
> and what is the functional difference to the sysadmin?

    The result when fsck destroys your disk because it thinks
    it's correcting software errors.

> emmanuel, you could have a drive problem or possibly a controller
> problem.  or both, if it's a bad day :-)/2
> 
> can you put the drive in another freebsd machine and fsck the 
> partitions on it?

    That is a relatively bad idea if the disk has got hardware errors on it.

    The recommended procedure is:

    Run fsck, preferably with "-n", don't write any changes yet -- see
      if you can actually get it to recognize the filesystem

    If it looks like too many inodes will be cleared:

    - Find a spare disk at least as big as the partition you are trying
      to recover:

    - dd if=/dev/ad0? of=/mnt/my/spare/disk/ad0?.img bs=32k conv=notrunc

    - mdconfig -a -t vnode -f /mnt/my/spare/disk/ad0?.img -u 0

    Run fsck on the resulting /dev/md0? partition.