[j-nsp] Partition/Format new HD

[Kevin Oberman]

> Date: Fri, 21 Aug 2009 16:46:40 -0500
> From: Richard A Steenbergen <ras at e-gerbil.net>
> 
> On Fri, Aug 21, 2009 at 12:26:10PM -0700, Kevin Oberman wrote:
> > Yes, but it was the dd(1) that fixed the real problem. The disk was
> > pre-formatted for either FAT or NTFS and that resulted in a partition
> > table on the drive that FreeBSD (JunOS) could not work with. The dd(1)
> > command blanked the partition table on the drive so the 'request system
> > partition hard-drive' command could do the job.
> > 
> > I believe that the sequence of things (at the FreeBSD level) is:
> > Check for /dev/ad1s1
> > If it is not found, fdisk to create it.
> > bsdlabel to partition the slice
> 
> Back in the day I remember having to do a completely manual fdisk and
> bsdlabel, complete with manually calculating all the sizes and offsets
> for the slices when the drive size changed (*), whenever I had to
> install a new drive. Recently I tried installing 9.3 from install-media
> onto a completely non-standard sized drive with some pre-existing
> Windows partitions even, and was completely surprised to find that all
> the install scripts Just Worked (tm). Go Juniper.

I remember doing this for a long time on BSD 4.2 systems before JunOS
existed. That did not make it fun. Calculate every value at least
twice...more often when you didn't get the same answer both times.
 
We're stating to sound old, RAS.

> (*) Who else remembers having to boot their Juniper RE-2.0's into dos to
> flash the bios from 0.9 to 1.2 to work around the old award bios bug 
> that blew up when you put in > 32gb drives? Now THAT was a pain in the 
> ass. :)

That one I was fortunate enough to miss. Still all Cisco back then.
-- 
R. Kevin Oberman, Network Engineer
Energy Sciences Network (ESnet)
Ernest O. Lawrence Berkeley National Laboratory (Berkeley Lab)
E-mail: oberman at es.net			Phone: +1 510 486-8634
Key fingerprint:059B 2DDF 031C 9BA3 14A4  EADA 927D EBB3 987B 3751