In article <20061031011356.76a65900.jklowden@schemamania.org>,
James K. Lowden <netbsd-help@NetBSD.org> wrote:
>Hello, 
>
>I didn't get any replies to my "manual upgrade" procedure, so I went ahead
>and booted the 3.0 install kernel. sysinst did a pretty OK job, with a
>couple of exceptions.  
>
>I'm sure I said "use existing bootblocks" but I noticed a message that the
>bootblocks were being replaced, and sure enough the boot(8) banner is
>dated 2006.  That wouldn't be so bad except
>
>	The new bootblocks don't work.  They don't boot any kernels -- generic,
>install, 2.0 -- on my box.  
>
>Fortunately, I have a second bootable device on that machine, a 128 MB
>SanDisk.  I was able to use the boot utility on that drive to boot the 3.0
>kernel.  All is well, mostly.  
>Now, what to do?  Do I replace both /boot and the primary bootblock? 
>Seems like I should, because that's the combination that booted the
>machine.  Do I have to boot up in the 2.0 kernel to do that, or can I use
>the 3.0 installboot(8)?  
>
>Assuming I can do this from the machine as it's booted now (in 3.0), this
>looks like the procedure:
>
>1.  cp /BACKUP/boot /
>2.  installboot -v -o timeout=5,console=com0,speed=9600 \
>	/dev/rwd0a /BACKUP/usr/mdec/bootxx_ffsv1
>
>Just checking on the filesystem.  Is this ffsv1?  
>
>$ disklabel wd0 | grep a:    
> a:  58341024        63     4.2BSD   1024  8192 46600  # (Cyl.      0*- 
>57878*)
>

You can use dumpfs | grep magic to see if it is UFS1 or UFS2.

christos