On Thu, May 19, 2005 at 04:36:06PM +0200, Antonio Bravo wrote:
> I have put a SS10 in service for hhtpd.
> I took a Quantum Fireball SE8.4S (~8 gb) I had around and which was 
> probably dd zero-ized, as I often do with old drives I take out of 
> service and left on the side.
> 
> I connect the drive inside the box, probe-scsi it ok and boot 
> NetBSD-2.0.2.Disklabel shows no label (so indeed I had probably zero-ed 
> in the past) but can't go further, and spits instead a
> "no space left" and stop the system back to the <ok> prom
> 
> ??!
> 
> So I double checked the jumpers/physical condition and all seems
> fine.  Reboot the NetBSD cdrom, and still the same crash.  I got a try
> to an OpenBSD 3.7 which didn't crash but couldn't use the disk.  I
> booted a Linux installer, went into "s" option of the linux fdisk to
> write a Sun table to the disk, created some ext2 filesystems, mounted
> them just fine and did some read/writes without any trouble.  Back to
> NetBSD with that Sun table, disklabel was able this time to create
> slices, newfs them and run the rest of the installation.  I guess
> Solaris format could had been able to do the job too with its 'format'
> utility, but I haven't tried it (lazy) nor SunOS and NeXT.
> 
> I wonder if the issue was due to the combination of that Quantum drive 
> and NetBSD disklabel, or just some limitation of disklabel itself.
> I guess there were some threads in the past, around 1.6, about disklabel 
>  dumbness.
> 
> Conclusion: keep a sparc/linux handy before installing NetBSD: it can 
> save the day ??

Which is funny, cuz when I first installed a (new, out of the box) 30gb
HD in my Sparcbook, Solaris couldn't label the drive, don't remember
the error I got, its been nearly 2 years, but once I labeled it under
NetBSD/sparc, Solaris was able to handle it fine.  Guess its just a
matter of having enough tools for the job.

-- 
Michael Parson
mparson@bl.org