SCO ODT 3.0 - installation with 'boot' and 'root fs'?

[Stefan]

Without installation diskettes, but with
'boot' and 'root filesystem'
(created from existing SCO ODT 3.0 before removed),
trying to get access to installation CD.
After the process was finished, there was root prompt available.
I created, and activated, one Unix partition occupying whole drive.
Adapter (AHA 1540/1542 compatible), HD, and CD-ROM are all SCSI.
BTLD diskette for the controller was created in any case,
although the diskette was not bootable.

Could someone explain exact steps, from this point,
in order to get access to CD and start full installation.

Thank you.

[Bela Lubkin]

Stefan wrote:

> Without installation diskettes, but with
> 'boot' and 'root filesystem'
> (created from existing SCO ODT 3.0 before removed),
> trying to get access to installation CD.
> After the process was finished, there was root prompt available.
> I created, and activated, one Unix partition occupying whole drive.
> Adapter (AHA 1540/1542 compatible), HD, and CD-ROM are all SCSI.
> BTLD diskette for the controller was created in any case,
> although the diskette was not bootable.
>
> Could someone explain exact steps, from this point,
> in order to get access to CD and start full installation.

It wasn't designed to be installed that way. The installation scripts
are on the N1, N2 and M1 diskettes that you don't have. You need to use
those; and furthermore they must be a correctly matched set -- all four
pieces (N1, N2, M1, and CD) must be from the same release and media set.

>Bela<

[Stefan]

Bela Lubkin <belal@sco.com> wrote in message news:<20030824010512.GG24551@sco.com>...

> Stefan wrote:
>

> > Without installation diskettes, but with
> > 'boot' and 'root filesystem'
> > (created from existing SCO ODT 3.0 before removed),
> > trying to get access to installation CD.
> > After the process was finished, there was root prompt available.
> > I created, and activated, one Unix partition occupying whole drive.
> > Adapter (AHA 1540/1542 compatible), HD, and CD-ROM are all SCSI.
> > BTLD diskette for the controller was created in any case,
> > although the diskette was not bootable.
> >
> > Could someone explain exact steps, from this point,
> > in order to get access to CD and start full installation.

>
> It wasn't designed to be installed that way. The installation scripts
> are on the N1, N2 and M1 diskettes that you don't have. You need to use
> those; and furthermore they must be a correctly matched set -- all four
> pieces (N1, N2, M1, and CD) must be from the same release and media set.
>

> >Bela<

I understand that what I was trying to do is not standard installaiton
procedure, although this is a kind of recovery process as well.
In any case, under given circumstances, there was no alternative.

The post is for anyone who can help in resolving problems,
but particularly for Tony's attention, if he is available.

As mentioned initially, after employing 'boot' and 'root fs'
there was root prompt available.
Beforehand was created a DOS partition
(although I think it was irrelevant issue).
Now rest of hard drive was occupied by Unix, and activated.
At that moment, with very restrictive set of tools, trying to make
hard drive bootable, and, the most important, getting installation CD
in operation.

Division table was created and installed, with the following
distribution:
hd0root EAFS no 0 0 489820
swap Non FS no 1 489821 504820
Not Used no 2 - -
Not Used no 3 - -
Not Used no 4 - -
Not Used no 5 - -
Not Used no 6 504821 505855
hd0 WHOLE DISK no 7 0 505855
(all defaults, no /u due to small hd size).

Then, mounted the hard disk:
# mount /dev/hd0root /mnt
No problem, but operation to create lost+found
was skipped, because 'mkdir' was missing from available toolset.

Next was to re-write the boot tracks:
# dd if=/etc/hdboot0 of=/dev/hd0a
# dd if=/etc/hdboot1 of=/dev/hd0a bs=1k seek=1
# dparam -w
Operations were completed without any system complaints.

At that moment I checked some config files using (only available
editor) ed.
mnttab had one entry (/dev/hd0root/mnt), and there was no space
between fields, and some strange trailing characters (\n?).
The same file, on UnixWare 7 available to me, was perfectly formatted.
Anyway I did not modified the file, because 'mount' returned correct
response.
/etc/default/filesys was manually created with only one entry for
/dev/root.

In regard to boot process, 'boot' diskette as a default used following
string:
'fd(60)unix root=fd(60) swap=fd(60) swplo=2680 nswap=200'.
But file /etc/default/boot, created during booting process,
had different content:
DEFBOOTSTR=hd(40)unix Srom=ad(0,1,0)
AUTOBOOT=YES
FSCKFIX=YES
MULTIUSER=YES
PANICBOOT=NO
MAPKEY=YES
SERIAL8=YES.
Firsty, I changed on CD-ROM SCSI ID from '1' to '5'
(SCO installation recommendation),
and accordingly, Srom configuration in the previous file.
I also changed hd(40)unix to hd(60)unix.

After 'reboot'ing, and changing in BIOS boot sequence,
printed error message was:
boot not found
Cannot open
Stage 1 boot failure: error loading hd(40)/boot.

After that I changed, just to try, in /etc/default/boot hd(60)unix
to original hd(40)unix,
but error message during booting attempt was identical.

I ask for an expert advice what can be done here.

Additionally, I am interested in 'inside' look at the procedures,
from development point, included on M1 installation diskette
(used to transfer control to ODT 3.0 CD-ROM during installation
process).

Thank you very much.

[Bela Lubkin]

Stefan wrote:

> Bela Lubkin <belal@sco.com> wrote in message news:<20030824010512.GG24551@sco.com>...

> > Stefan wrote:
> >

> > > Without installation diskettes, but with
> > > 'boot' and 'root filesystem'
> > > (created from existing SCO ODT 3.0 before removed),
> > > trying to get access to installation CD.
> > > After the process was finished, there was root prompt available.
> > > I created, and activated, one Unix partition occupying whole drive.
> > > Adapter (AHA 1540/1542 compatible), HD, and CD-ROM are all SCSI.
> > > BTLD diskette for the controller was created in any case,
> > > although the diskette was not bootable.
> > >
> > > Could someone explain exact steps, from this point,
> > > in order to get access to CD and start full installation.

> >
> > It wasn't designed to be installed that way. The installation scripts
> > are on the N1, N2 and M1 diskettes that you don't have. You need to use
> > those; and furthermore they must be a correctly matched set -- all four
> > pieces (N1, N2, M1, and CD) must be from the same release and media set.

>
> I understand that what I was trying to do is not standard installaiton
> procedure, although this is a kind of recovery process as well.
> In any case, under given circumstances, there was no alternative.

You could buy a dozen brand new ODT 3.0 installation sets, with license,
at full original list price, for what it is going to cost you to do a
"recovery" this way...

> The post is for anyone who can help in resolving problems,
> but particularly for Tony's attention, if he is available.
>
> As mentioned initially, after employing 'boot' and 'root fs'
> there was root prompt available.
> Beforehand was created a DOS partition
> (although I think it was irrelevant issue).
> Now rest of hard drive was occupied by Unix, and activated.
> At that moment, with very restrictive set of tools, trying to make
> hard drive bootable, and, the most important, getting installation CD
> in operation.
>
> Division table was created and installed, with the following
> distribution:
> hd0root EAFS no 0 0 489820
> swap Non FS no 1 489821 504820
> Not Used no 2 - -
> Not Used no 3 - -
> Not Used no 4 - -
> Not Used no 5 - -
> Not Used no 6 504821 505855
> hd0 WHOLE DISK no 7 0 505855
> (all defaults, no /u due to small hd size).
>
> Then, mounted the hard disk:
> # mount /dev/hd0root /mnt
> No problem, but operation to create lost+found
> was skipped, because 'mkdir' was missing from available toolset.
>
> Next was to re-write the boot tracks:
> # dd if=/etc/hdboot0 of=/dev/hd0a
> # dd if=/etc/hdboot1 of=/dev/hd0a bs=1k seek=1
> # dparam -w
> Operations were completed without any system complaints.

Possibly if you had added:

dparam /dev/rhd00 `dparam /dev/rhd00`

it would have helped. That "stamps" the disk parameters. `dparam -w`
just writes /etc/masterboot (the OSR5 masterboot code) without stamping
parameters. Under certain circumstances, ISL uses different parameters
than get used when trying to boot off the hard disk, which can lead to
the "Stage 1 boot failure" error cascade you saw.

> At that moment I checked some config files using (only available
> editor) ed.
> mnttab had one entry (/dev/hd0root/mnt), and there was no space
> between fields, and some strange trailing characters (\n?).
> The same file, on UnixWare 7 available to me, was perfectly formatted.
> Anyway I did not modified the file, because 'mount' returned correct
> response.
> /etc/default/filesys was manually created with only one entry for
> /dev/root.
>
> In regard to boot process, 'boot' diskette as a default used following
> string:
> 'fd(60)unix root=fd(60) swap=fd(60) swplo=2680 nswap=200'.
> But file /etc/default/boot, created during booting process,
> had different content:
> DEFBOOTSTR=hd(40)unix Srom=ad(0,1,0)
> AUTOBOOT=YES
> FSCKFIX=YES
> MULTIUSER=YES
> PANICBOOT=NO
> MAPKEY=YES
> SERIAL8=YES.
> Firsty, I changed on CD-ROM SCSI ID from '1' to '5'
> (SCO installation recommendation),
> and accordingly, Srom configuration in the previous file.
> I also changed hd(40)unix to hd(60)unix.

That was an incorrect change. hd(40) means "division #0 of the active
partition on drive 0". hd(60) means "division #4 of an illegal
partition on drive 0". fd(60) means "floppy drive 0 in 1.44MB format".
You can't mix minor numbers from one driver to another.

> After 'reboot'ing, and changing in BIOS boot sequence,
> printed error message was:
> boot not found
> Cannot open
> Stage 1 boot failure: error loading hd(40)/boot.
>
> After that I changed, just to try, in /etc/default/boot hd(60)unix
> to original hd(40)unix,
> but error message during booting attempt was identical.

The error is coming from hdboot1, long before anything tries to read
/etc/default/boot. hdboot1 is trying to access the disk with different
geometry parameters than it was installed under.

> I ask for an expert advice what can be done here.
>
> Additionally, I am interested in 'inside' look at the procedures,
> from development point, included on M1 installation diskette
> (used to transfer control to ODT 3.0 CD-ROM during installation
> process).

You would have to decompose an M1 disk yourself to get that sort of
information. It contains catalogues of data to be found on the real
install media (CD-ROM or tape), plus installation scripts and stuff like
that.

>Bela<