[nycbug-talk] OpenBSD questions...

[Marco Scoffier]

On Tue, Jul 12, 2005 at 09:48:08PM +0000, Louis Bertrand wrote:
>
>I use raidframe but only for the non-root filesystems. I have a dedicated
>root disk for the kernel and / filesystem. It keeps things simpler for me.
>Probably not the answer you're looking for. Grab a beer and psych yourself
>for yet another reboot...
>

I did that also.  I had an ide that I would boot off of and then 3
scsi's would take over with the root and other partitions in a raid.
But the ide disk died perhaps with the heat.  I first noticed because
the clock was mysteriously set back to 1979??  After freaking that I had
been hacked cause I didn't upgrade to 3.7, kicking myself in the head
for rebooting a machine that no longer had a workable kernel, I forced
myself to figure out how to boot off any of the three 256m partitions on
the head of each disk in the raid array.  There is this excellent guide:

  http://argon18.com/raid_openbsd.html

which makes the whole process seem pretty easy.  But it got me to thinking
about which kernel is runnng when you 

  raidctl -A root raid0

after having set the option RAID_AUTOCONFIG ...  and of course the answer is in
the manpage: the kernel must reside outside the raid.

 -A root dev
   Make the RAID set auto-configurable, and also mark the set as be-
   ing eligible to contain the root partition.  A RAID set config-
   ured this way will override the use of the boot disk as the root
   device.  All components of the set must be of type RAID in the
   disklabel.  Note that the kernel being booted must currently re-
   side on a non-RAID set and, in order to have the root file system
   correctly mounted from it, the RAID set must have its `a' parti-
   tion (aka raid[0..n]a) set up.

No more reboots !  
I did yank a disk and test booting off the other two, tres cool...

-- 
Marco