On Tue, Nov 28, 2006 at 03:35:16AM -0000, David Lord wrote:
> On 28 Nov 2006, at 6:57, Henry Nelson wrote:
> > field is somewhat of preference, and that there is no "best" way, _but
> > there is a "right" way_.  It would be nice to have the same guide for
> > fdisk, too.  Otherwise, I will spend another some weeks going through all
> > the possible permeations.  TIA)
> 
> I'm not so sure there is one right way as mostly there are several 
> different methods that will work.

You're absolutely right.  (My English suffers a bit.)

> > > Try 'fdisk -u wd0' which should take you through interactive mode, one of
> > 
> > This is where my troubles have begun, I would like to run fdisk on the
> > command line NON-interactively (use the -f option).
> > 
> > So far I have one example on how to do this: "/sbin/fdisk -v -v -f -u -0
> >  -s 169/63/2001825 wd1".  Somehow I managed to botch this, but I hope to
> > re-try Saturday night.  (Although I still don't understand why two "-v"s
> > would be necessary  --  the reason for wanting multiple real-life examples.)
> 
> I've just looked back through the thread and see you mention disk has 
> data on it and fdisk indicated it to have a dos partition.
> Did you try to mount the dos partition to see if any data was there?

It says DOS partition, but it sure acts and feels like NetBSD.  Anyway,
all of the data has been backed up on another disk and compared (diff).
No need to worry about the data (except that all the partitions on the
backup disk (wd0) are like 96-98% capacity.  When I have more time I will
zero out the whole disk (the new one, wd1) and start over.

> I also noted that your fdisk -s option doesn't correspond to the
> 12594897 of the dos partition. You didn't give the full fdisk output 

Right, that was another person's example.  For the particular disk of this
thread, I _thought_ the -s option should have been: "-s 169/63/12596850",
for the disk which probed as (`dmesg | grep wd1`):
wd1 at atabus0 drive 1: <ST36421A>
wd1: drive supports 16-sector PIO transfers, LBA addressing
wd1: 6150 MB, 13330 cyl, 15 head, 63 sec, 512 bytes/sect x 12596850 sectors

This information agrees with the manufacturer's specification.

Sorry about not posting all of the fdisk output:  For the record, it is:
===========================================================================
% fdisk wd1
Disk: /dev/rwd1d
NetBSD disklabel disk geometry:
cylinders: 13330, heads: 15, sectors/track: 63 (945 sectors/cylinder)
total sectors: 12596850

BIOS disk geometry:
cylinders: 784, heads: 255, sectors/track: 63 (16065 sectors/cylinder)
total sectors: 12594960

Partition table:
0: Primary DOS with 32 bit FAT - LBA (sysid 12)
    start 63, size 12594897 (6150 MB, Cyls 0-784), Active
1: <UNUSED>
2: <UNUSED>
3: <UNUSED>
Bootselector disabled.
===========================================================================

Ah-ha.  So fdisk uses the "BIOS disk geometry" and disklabel uses the
"NetBSD disklabel disk geometry".  I think this explains why I've been
having troubles.  I used the "NetBSD disklabel disk geometry" parameters
for fdisk. :(

If that is true, it really needs to be in the man pages for both fdisk
and disklabel.

> I don't have the procedure mapped out as there were too many 
> disasters and much backtracking along the way so very messy. I'll try 
> to write something up then post a link. I restored NetBSD 3.0 earlier 

Those "disasters and much backtracking along the way so very messy" is
exactly what is needed to gleen knowledge of how to troubleshoot.

> tonight so job completed except from size of /netbsd  I'd guess I've 

Congratulations!

-- 
henry nelson
  WWW_HOME=http://yuba(dot)ne(dot)jp/(tilde)home/