[SunRescue] (Help!) Problems with Solaris 8 and SPARCServer 1 000E? [Late entry]

Thu Jan 4 06:36:04 CST 2001

Al,
	I'm runnin' a 1000E, Solaris 7 Server, 1.7Gigs RAM, 8 NVRAM, 110 SSA
with about 12Gig in RAID disks, two 2.1Gig disks in an external 911 box, and
two 2.1 Gig drives inside.  docs.sun.com tells me the 1000E is supported in
Solaris 8
(http://docs.sun.com/ab2/coll.28.20/SPARCHW/@Ab2PageView/1987?Ab2Lang=C&Ab2E
nc=iso-8859-1)

I'm thinkin....Jim Lockwood hit it with, "The watchdog reset is a good
indication that you've got a hardware issue."

'Mechanical' problems is my read.  It does too much to be failed
electronics.  Do a probe-scsi at the ok prompt.  And a probe-scsi-all.  If
no 'big hints' there... 
	- check the pins on the chassis XDBus connectors -
I had pin1 in row A and pin1 in row C bent on the far upper right looking in
from the back.  You had to look...I mean REALLY LOOK to see they were bent.

	Everything worked! - except the internal SCSI stuff (like the
CDROM!).  I could load and boot from external stuff.  Even boot the SSA.
But the internal disks would not do squat.  They'd blink, and spin/start-up,
and report out their existence to a probe-scsi, but it would fail to
load/run repeatedly.  
	This box is supposed to tell you if something goes boink.  But you
can fool it into all sorts of unusual behavior if something isn't hooked up
'per the drawins.'
	This may get me in trouble...drop external termination unless you
have something hanging on it.  Active terminators are active.  And they
don't tell you if they are goofy.  Something attached?  Then terminate it
(like on the other connector on an external 911 case).  
	Board 0 (top one) is your 'normal' boot bus and is the SCSI bus that
the internal devices use and that the external plug supports.  So, it runs
from the CDROM/Tape back by the internal drives, across the controller card,
to the SCSI controller on Board0, and out to the attached devices through
the HD50 connector.  One bus.
  	Each remaining board then has its own separate SCSI bus.  Only the
top board supports the internals (I though the controller card had its own
SCSI controller - silly me - but it apparently controls only the XDBus which
is how it shares memory and processors).
	The FEH warns not to swap boards (1 to 0, etc), as there is some
kind of interaction with the controller card (in front above the drives) and
you loose the factory programming or something.  Pretty vague, I know, but
its there.  I ran Jim Birdsall's FAQ
(http://www.sunhelp.org/faq/sunref4.html  - see 501-1979) reset before I
found the bent pins.  Still works.  God bless Jim Birdsall.
	Last item: Memory should be installed symmetrically across all
boards - Bank0 on all installed boards, then bank1 on all boards, etc.
Since it goes in 128Meg increments, you might try 256 on each board (banks 0
and 1) and hold off on the last bank until you have another 128Meg.  
	Poke it and see if it squeals.

Good luck

DG

Dave Garten
6508 Rock Crystal Dr
Clifton, VA  20124
Office:  (703) 614-4616
Home:  (703) 222-9057
Personal email:  dgarten at nova.org

-----Original Message-----
From: Corda Albert J DLVA [mailto:CordaAJ at nswc.navy.mil]
Sent: Wednesday, January 03, 2001 2:45 PM
To: 'rescue at sunhelp.org'
Subject: [SunRescue] (Help!) Problems with Solaris 8 and SPARCServer
1000E?

I recently accquired a 1000E with 4 60 Mhz CPUs and
2 Hard disks (2.1 Gb ea.).  I have been trying to
install Solaris 8 for the past week, and its driving
me nuts (Actually, It doesn't have to drive very far
to get me there :-) Has anyone out there successfully
installed Solaris 8 on one of these beasts?

The system seems to be hanging at various random (non-
reproducable) places in the install. The only error
message is an occasional watchdog-timer reset. Most of
the time, it just stops dead.

The system originally had 2.5.1 on it, which seemed to
boot fine, although I couldn't really test it since
I didn't have the password.

I did a deja search, which turned up a very small amount of
info. The official sun docs site doesn't seem to have
any manuals online for the 1000(E)that I could find. :-(
(If anyone knows where I can find the service manual
for the 1000(E), please let me know!)

My current configuration consists of 2 CPU Boards with
2 60 Mhz modules ea., 640 Mb of RAM, two 2.1 Gb hard
disks, a TGX card and a fiberchannel card. Both external
SCSI buses are terminated. Prom Rev. is 2.31

Things I've tried:
	Memory test diags (all seems fine)
	Reseating the RAM and CPU modules.
	Swapping the two CPU boards (i.e. putting the
	lower board in the top slot. I did this as
	an experiment since I am assuming the installation
	software only uses CPU0)
	Tried using a different hard disk... no luck.

At this point, I'm kind of stuck...is this a Solaris 8
problem, or a problem with my hardware? any suggestions?

Perhaps I shouldn't even be tring to use Solaris 8... The
install is dog-slow (which probably indicates that 8 will
run dog-slow) Is there a recommended version of Solaris
for the 1000? (Linux is out, since a deja search indicated
that it seems to have a problem with SMP on the 1000) Perhaps
I should drop back to 2.5.1, 2.6 or 2.7?  Opinions are
welcome!

-Thanks in advance...
-al-
-acorda at geocities.com

_______________________________________________
Rescue maillist  -  Rescue at sunhelp.org
http://www.sunhelp.org/mailman/listinfo/rescue