Re: Worth attempting an Ultra 30 install with 1.1???u

Re: Worth attempting an Ultra 30 install with 1.1???u

[Jakub Jelinek]

> 
> 
> Subject says it all.  Is 1.1 (from the mirrors) ready to be installed via TFTP
> on an Ultra 30, or should I wait for awhile?

Unless you want to do a CD-boot, which will probably fail for you, go ahead
with the installation, everything should be fine.

Cheers,
    Jakub
___________________________________________________________________
Jakub Jelinek | jj@sunsite.mff.cuni.cz | http://sunsite.mff.cuni.cz
Administrator of SunSITE Czech Republic, MFF, Charles University
___________________________________________________________________
Ultralinux - first 64bit OS to take full power of the UltraSparc
Linux version 2.1.128 on a sparc64 machine (498.80 BogoMips).
___________________________________________________________________

Re: Worth attempting an Ultra 30 install with 1.1???u

[Rick Richardson]

Jakub Jelinek writes...
> > 
> > 
> > Subject says it all.  Is 1.1 (from the mirrors) ready to be installed via TFTP
> > on an Ultra 30, or should I wait for awhile?
> 
> Unless you want to do a CD-boot, which will probably fail for you, go ahead
> with the installation, everything should be fine.

I just tried to load UP1.1 on an Ultra 30.  This particular ultra
30 has a broken onboard ethernet controller, so we have a scsi/hme
card in it.

The tftpboot (boot /pci@1f,4000/pci@4/SUNW,hme@0,1) works (yippee,
first time we've gotten even this far on an Ultra 30) and we get to
the install screens.

But thats it.  The ethernet driver keeps complaining that "Link down,
Cable problem?" when it tries to access out local NFS mirror.  But of course
there is no cable problem, since the tftpboot process worked fine.

-Rick

-- 
Rick "The Hack" Richardson  rick@dgii.com http://RickRichardson.freeservers.com

Re: Worth attempting an Ultra 30 install with 1.1???u

[Jakub Jelinek]

> 
> Jakub Jelinek writes...
> > > 
> > > 
> > > Subject says it all.  Is 1.1 (from the mirrors) ready to be installed via TFTP
> > > on an Ultra 30, or should I wait for awhile?
> > 
> > Unless you want to do a CD-boot, which will probably fail for you, go ahead
> > with the installation, everything should be fine.
> 
> I just tried to load UP1.1 on an Ultra 30.  This particular ultra
> 30 has a broken onboard ethernet controller, so we have a scsi/hme
> card in it.
> 
> The tftpboot (boot /pci@1f,4000/pci@4/SUNW,hme@0,1) works (yippee,
> first time we've gotten even this far on an Ultra 30) and we get to
> the install screens.
> 
> But thats it.  The ethernet driver keeps complaining that "Link down,
> Cable problem?" when it tries to access out local NFS mirror.  But of course
> there is no cable problem, since the tftpboot process worked fine.

Aha, that's the problem. Of course, the link is down, when you claim eth0 is
broken. But I'm not sure what to do, as it would require quite large surgery
to the installation program (which has hardcoded eth0), and solving it
genericaly would require adding suport for multiple different ethernet
modules during install and stuff like that.
But AFAIK, Solaris also uses le0 for network installations and does not
offer you the possibility to select another ethernet. That's not an excuse
for us, but actually the reasoning behind it is that situations like you
have are pretty rare.
Other options for you include e.g. booting the tftpboot image and selecting
local CD-ROM.
After installation, you can obviously use another network interface instead
of eth0...

Cheers,
    Jakub
___________________________________________________________________
Jakub Jelinek | jj@sunsite.mff.cuni.cz | http://sunsite.mff.cuni.cz
Administrator of SunSITE Czech Republic, MFF, Charles University
___________________________________________________________________
Ultralinux - first 64bit OS to take full power of the UltraSparc
Linux version 2.1.128 on a sparc64 machine (498.80 BogoMips).
___________________________________________________________________

Re: Worth attempting an Ultra 30 install with 1.1???u

[Rick Richardson]

OK, here's the status of this UP1.1 install on the Ultra 30.  This
particular box has a broken onboard ethernet controller, so we used
the one on the SCSI/HME PCI card to TFTP boot it.

The machines internal SCSI hard drive (with Solaris on it) was also
removed, an an external drive (with terminators enabled) was connected.
NOTE: I sure hope that the SUN SCSI controller is smart enough to enable
termination for the now empty internal drive slot, and disable termination
on the now connected external SCSI bus.

The first problem I ran into was that "fdisk" would bus error when
trying to init the hard disk.  I fixed that by going to a shell
prompt, making the /dev/sda device node by hand, and then doing a
"cat /dev/zero > /dev/sda".  After that, fdisk was happy and I could
lay down a a new disk lable and partitions.

I then had some trouble accessing the CD-ROM.  So I did a tftboot
again.  This time I couldn't get to the hard disk, but the CDROM
was there.  Various messages were coming out of the SCSI device
driver, but I didn't write them down.  After about the third
tftpboot, the SCSI driver loaded and both the hard drive and
CDROM were visable.

I then installed the software without incident from the CDROM and got
all the way thru SILO.

Unfortuantely, at this point I cannot get the thing to boot from the
PROM monitor.  It just sits there.

I was able to use the tftprescue.img to boot the machine and mount
the hard disk.  Everything is there.

So, any ideas what my next move should be????

-Rick


-- 
Rick "The Hack" Richardson  rick@dgii.com http://RickRichardson.freeservers.com

Re: Worth attempting an Ultra 30 install with 1.1???u

[Rick Richardson]

Rick Richardson writes...
> 
> OK, here's the status of this UP1.1 install on the Ultra 30.  This
> particular box has a broken onboard ethernet controller, so we used
> the one on the SCSI/HME PCI card to TFTP boot it.
> 
> The machines internal SCSI hard drive (with Solaris on it) was also
> removed, an an external drive (with terminators enabled) was connected.
> NOTE: I sure hope that the SUN SCSI controller is smart enough to enable
> termination for the now empty internal drive slot, and disable termination
> on the now connected external SCSI bus.

<snip>

> Unfortuantely, at this point I cannot get the thing to boot from the
> PROM monitor.  It just sits there.

OK, we seemed to have found a way around this issue.  It turns out
that if I plug the internal SCSI hard drive back in and set the
external hard drive to SCSI ID 1 I can get this thing to boot UP1.1
when I type "boot disk1".  No idea why the PROM monitor won't let me
boot the external drive with "boot disk" without the internal one
installed.  But thats not an UP1.1 problem.

Then we hit the "overwritting initrd" problem.  I then moved the SIMMs
from the first set of four slots to the third set of four slots,
as suggested by Jakub.  What a weird solution to a bug.  Whatever.

So after doing that, UltraPengiun 1.1 is finally booting on the Ultra
30.  Wow, what a day!!!!

Now the only issue left is the network.  Since the onboard ethernet
controller is broken, we are using the PCI hme board.  Unfortunately,
UP1.1 sees and configures that controller OK (as eth1), and will send
packets out on it, but is absolutely deaf when it comes to hearing
replies such as ARPs.

This smacks of an interrupt problem to me.  If the hme driver
isn't seeing interrupts from the board, this would explain the
behavior.

UltraPeniguin is *so* close to usable now.  I'm psyched.

-Rick


-- 
Rick "The Hack" Richardson  rick@dgii.com http://RickRichardson.freeservers.com