Link to home
Start Free TrialLog in
Avatar of cparker
cparker

asked on

New Internal Disk for Sparc-4

Trying to install a Quantum Viking 2.3 Narrow-Fast SCSI as
new internal disk in Sun Sparc4.  Formats, install is okay.
Boot fails due to "short read" and sometimes "FILE is not
and ELF executable".  I have run low level disk checks, and
the disk is physically okay, and according to Sun the drive
is compatable.  Any help would be greatly appreciated.
Avatar of cmarquis
cmarquis

Is this the only disk you have on the system? Is there a cdrom attached to this system? Did you install the OS from scratch onto this disk? How did you format and install (Install disk or OS?)? If you could clarify these things for me I'm sure we can remedy this.
Avatar of cparker

ASKER

System had a 1.0 GB internal that got corrupted.  The new drive is the only internal drive.  I have an external cdrom that I use
to boot from and installed Solaris 2.5 from.  I used the normal
sun openwin based install.  Changed partitions so that everything
other than swap was under one partition.  Install goes fine, I
reboot the machine when its complete, and the kernel fails to
load, with the previously mentioned errors.  I boot to single
user off the cd, mount the drive, run checksums, etc. on the
files in question and they all appear normal.  I've called both
Quantum and Sun and they shrug their sholders and say that it
should work.  But obviously it doesn't.  :)
Avatar of cparker

ASKER

More info.  Prom revision is 2.24.  Prom rev less than 2.5 has
problems with boot partition larger than 1 GB but I get the same problems when I give / 500 MB and the rest to /usr.

Hi...

I looked at you guys profiles and I feel a bit of a dope suggesting the following, as I'm almost sure you're aware of it.
 
It sounds like you haven't set the prom variable, "boot-device" for your new drive. As a work around you could try this..

setenv boot-device /sbus@1,f8000000/esp@0,800000/sd

But substitute the relevent sbus@, esp@ and sd information, which you can get from cd'ing first to "/" then do an "ls" and "cd" into the sbus directory, then the esp, then the sd directory.
A "pwd" at the bottom level here will show you the pathname to use for your prom variable. Then type boot, if this works then don't forget to go back to the prom level and type "reset" to keep your variable. I have used this workaroun myself, especially when my startup disk is not what it should be, ie. target 3.

But I guess you already know about this.

regards
Avatar of cparker

ASKER

I've had to do that on other machines when it can't find the
proper boot disk, but in this case there is only one internal
disk, it is finding it, but is not able to load the kernel
from it.  No external scsi devices are attached, so only valid
scsi-id is 3, which is the internal disk.  

It looks like a memory alignment problem ( ie, the "short read"
errors ), but that is a little beyond the scope of my
knowledge at this point.

It is nice to know that somebody has read this...  :)
Hello me again...
I'm getting back to you because we had the same problem on an internal startup disk, today infact (23-10-97), that was OK until we added an external tape device to the external, built-in scsi port. Our supplier says that it was a termination problem due to older sparcs (ours is a Sparc10) having termination ON by default. They say this is changeable with jumpers on the motherboard. My thought is, If your disc has termination ON, then turn it off and let the sparc handle the termination.

This is a little beyond my scope, but I thought I'd pass on what I've been told by our suppliers.

regards

PS could anyone let me know if I've been misled, I.Y.C.O.
No root file system may be larger than 2 GB on Solaris. So if this is the root fisk you have a prblem. You gonna have to partition it.

Avatar of cparker

ASKER

As I stated...  
I have changed partitions so that the root partition is
500 MB, and the rest is distributed.

Still no luck.  Shipped it out to get repaired, they
replaced the "SEC" chip, it booted okay once, now is
back to the same problem.  I'm about to give up all hope.

:(  Arrrgghhh!

Thanks in advance...
Hi...
As this is the second drive to fail, its got to be the scsi setup. I have seen short read errors when scsi protocols have not been negotiated properly. I've researched, and after Solaris 2.3 (2.4 and above) The OS will negotiate this correclty, but as you've got an older Sparc, maybe you've got to set the scsi_options yourself in /etc/system. If you think I'm on the right track here, I will look this up in the Red Porsche book at the office or ask around some more.

regards
Avatar of cparker

ASKER

hmmmm, that could be.

if you're willing to look it up, I'm willing to try it...  :)

another thing i had to do, is since this is a new drive model,
was to create my own format.dat, so having to add some settings
to the /etc/system could well be a solution...  i have figured
out to turn off tagged mode queuing... :)

regards,
cparker

ASKER CERTIFIED SOLUTION
Avatar of red031197
red031197

Link to home
membership
This solution is only available to members.
To access this solution, you must be a member of Experts Exchange.
Start Free Trial