On Thu, 02 Mar 2006 19:08:40 +0100
Martin <nakal@nurfuerspam.de> wrote:
>
> Hi,
>
> I've done some strange(?) experiments and I managed to get
> a panic of type "baddir" and just now another one that only
> rebooted my system instantly without writing a vmcore.
>
> I don't know if this is important. I could not find any
> information on how gvinum detects failed hard disks.
> That's why I tried something really simple.
>
> Steps to reproduce:
>
> 1)
> dd if=/dev/zero of=a bs=1024 count=256000
> dd if=/dev/zero of=b bs=1024 count=256000
> dd if=/dev/zero of=c bs=1024 count=256000
>
> 2)
> mdconfig -a -t vnode -u 0 -f a
> mdconfig -a -t vnode -u 1 -f b
> mdconfig -a -t vnode -u 2 -f c
>
> 3)
> gvinum create raid5.conf
>
> with raid5.conf:
> drive a device /dev/md0
> drive b device /dev/md1
> drive c device /dev/md2
> volume raid5
> plex org raid5 491k
> sd length 0 drive a
> sd length 0 drive b
> sd length 0 drive c
>
> 4)
> newfs -m 0 /dev/raid5
>
> 5)
> mount /dev/gvinum/raid5 /mnt/
> mkdir /mnt/aaa
>
> 6)
> dd if=/dev/zero of=c bs=1024 count=256000
>
> 7)
> cd /mnt/aaa
> touch a
>
> 8)
> Repeat the steps 6) and 7) if it does not
> panic at this point.
Looks similar to the one I reported a few days ago:
http://lists.freebsd.org/pipermail/freebsd-geom/2006-March/001095.html
regards
Marius