Subject: Re: Install over two disks + QIC tape problems
To: None <port-sparc@netbsd.org>
From: None <salvage@plethora.net>
List: port-sparc
Date: 01/15/2001 08:41:41
On "Mon, 15 Jan 2001 09:39:09 +0000", Chris.Smith@raytheon.co.uk wrote:
>I've managed to get NetBSD onto my Classic at last, but I'm
>going to trash it and start again now I know a bit more about
>the hardware and OS.  Can anyone answer a couple of evil
>questions please, so I can do it properly this time :)

Mmm.  Evil.

>I have two SCSI disks in this set up, one 205Mb internal and
>an external 1.05Gb disk.  Can I install NetBSD over the two
>disks as in NetBSD-1.5 it assumes you are only installing on
>one disk?

It is possible; I've done similar setups in the past.

> The OpenBSD installer allows labelling and writing
>mount-points of sd0 *and* sd1 before committing the
>installation.  Any ideas.

You'll need to go to the shell to manually do part of the setup.

> I'd ideally like this but
>disklabel and the installer doesn't like it:-
>
>sd0a /
>sd0b swap
>sd0e /tmp
>
>sd1a /usr
>sd1e /var
>sd1f /home
>
>FYI sd0 = 205Mb, sd1 = 1Gb

Looks okay.  How much RAM does your Classic have?  If
it has enough, I'd do:

sd0a /
sd0d /var
sd1a /usr
sd1b swap
sd1d /home
swap /tmp

>Can I drop out of the installer to edit a manual disklabel
>for example on both drives, write the fstab and then
>continue?

Something like that.  There's an option to get a shell in
one of the installer submenus[1].  The basic strategy, as
I recall, would be to create and edit the disklabels from
the shell, exit back into the installer, and do a reinstall.

I don't recall all the details anymore.  You may need to
mount everything (relative to /mnt) before exiting the
shell, or just create an fstab (in /tmp?) before exiting
the shell back into the installer, or some other
variation.

>Also, I have a Sun Archive QIC-150 tape that I might as well
>use to back the box up occasionally.  When it boots it
>reports:-
>
>st0: Archive......etc etc (SCSI/1)
>st0: rouge drive empty
>
>or something similar both when there is a tape and isnt a
>tape in it.  Using any `mt' command on any /dev/*st0 does
>absolutey squat and pops up with an IO error.  It says the
>tape is a SCSI-1 device - could this do it or is it a duff
>tape drive?  Rewinding a tape would be a massive step at
>the moment :P  FYI - At the ok prompt, using probe-scsi
>*does* show that the tape drive exists and is connected!

The dmesg output looks normal; I've got such a drive on
an IPC that I'm running.  As for the error on mt....
Is this the only drive?  If so, what does just `mt rew`
get you?  If that's giving you errors I'd be suspicious
of the tape(s).  I've been dealing with about a 14%
failure rate on new tapes, and the ones that aren't bad
don't necessarily last that long.

[1]  I don't recall which one, and I don't have a console
     in front of me to play with the installer.