In article <20061024031250.GA28668@run.galis.org>,
George Georgalis <george@galis.org> wrote:
>On Mon, Oct 23, 2006 at 04:42:49PM +0200, Geert Hendrickx wrote:
>>On Mon, Oct 23, 2006 at 10:33:24AM -0400, George Georgalis wrote:
>>> I'm pretty sure this has nothing to do with the disk errors that started
>>> this journey.  The system happens to be 3.1_RC2, but if needed, I could
>>> change this box easily enough.  What other commands are available to
>>> doctor the disk?
>>
>>Try to find your /var partition with scan_ffs(8) or scan_lfs(8) (new in
>>NetBSD 3.1).  When you've got that, you can recover your fdisk table and
>>disklabel(s) from /var/backup/work/.
>
>oh, this was a data disk, so the backups where on the working
>fine thank you /var partition. :) Actually I think this is the
>first disk I learned BSD fdisk/disklabel partitioning with; well I
>struggled for a while and came up with something that worked....
>but restoring these files is not going to restore the full d
>partition.
>
>I tried manually setting c and d
>#        size    offset     fstype [fsize bsize cpg/sgs]
> a: 390721968         0     4.2BSD      0     0     0  # (Cyl.      0 - 387620)
> c: 390721905        63     unused      0     0        # (Cyl.      0*- 387620)
> d: 390721968         0     unused      0     0        # (Cyl.      0 - 387620)
>
>
>but what's the point... when I made a: the sole partition.
>newfs -O2 wd1a works and I can mount it. I unmounted it
>and now I'm doing dd if=/dev/zero of=/dev/rwd1d bs=4096
>to verify media errors.
>
>Seems working. So really, what is the point of c and d partitions?

This makes the whole disk a netbsd disk. You usually should leave the first
63 sectors out for the mbr.

christos