My thanks to all who helped me out. This whole thing turned out to be
a comedy of errors.
First, direct answers to my questions.
All interfaces on a Sun use the same ethernet address. This is still
abiding by the letter of the Ethernet standard, although I personally
feel that it is not in keeping with the spirit of the standard. An
ethernet address need only be unique across one ethernet WIRE (not
across all creation). So, if your workstation is attached to two
separate ethernets, then it is okay for it to have the same ethernet
address on both wires. On Sun's, the address is set from an address
stored in the ROM.
Regarding ifconfig: I knew that they had made some changes to it in
4.1. But I didn't realize that displaying the ethernet address was
one of the changes. I went over to another sparc running 4.1 and ran
ifconfig there. It did not display the ethernet address. I
immediately assumed that the reason it was showing up on the machine
in question was because someone or something had explicitly overridden
the default address. Turns out that the REAL reason it was showing it
was because I was running "ifconfig" as root. If you do not run
ifconfig as root then it will not show the ethernet addresses! And of
course, I was not root on that other sparc that I used as my
"control". Sigh.
Why wasn't it forwarding IP packets? Well, because I forgot to turn
on that magic kernel variable "ip_forwarding". Matt Crawford
explained it to me:
--------------------
From: "Matt Crawford" <matt@oddjob.uchicago.edu>
Subject: Re: second ethernet card in a SS1+
Date: Thu, 13 Sep 90 15:00:32 CDT
First make sure of the magic kernel variable:
adb -w /vmunix /dev/kmem
ip_forwarding?X
ip_forwarding/X
If it's zero, that's the problem. Set it to 1 in both vmunix and
kmem.
ip_forwarding?W 1
ip_forwarding/W 1
..
--------------------
Remember, tho, that that was called "ipforwarding" under 4.0.3.
Something else that got changed.
And to all who suggested that I check the routing tables and netmasks
on the Suns: those were the first things that I checked. But thanks
for mentioning it just the same.
There were several other problems that we encountered with this setup
(thus the delay in getting this summary out). But it turns out that
they were all related to routing difficulties on the building backbone
network. We had reused a defunct subnet number, but some routers
still had outdated routes for that subnet lurking about. So some of
our problems were NOT related to the Suns. We updated all our
routers and have gotten everything stabilized. It all works now. So
you really can use a SS1+ as a server and a subnet router/gateway!
My thanks to all who responded:
yih%atom@cs.utah.edu (Benny Yih)
dan@breeze.bellcore.com (Daniel Strick)
Paul Graham <pjg@acsu.buffalo.edu>
timsmith@Sun.COM (Timothy G. Smith - Technical Consultant)
Aydin Edguer <edguer@alpha.ces.cwru.edu>
"Matt Crawford" <matt@oddjob.uchicago.edu>
smb@ulysses.att.com
thor@thor.atd.ucar.EDU (Richard Neitzel)
steve@umiacs.UMD.EDU (Steve D. Miller)
trinkle@cs.purdue.edu
fabrice@yosemite.atmos.ucla.edu (fabrice cuq)
northrup@cpswh.cps.msu.edu (Frank Northrup)
Tony Martindale <tony@rata.vuw.ac.nz>
del@mlb.semi.harris.com (Don Lewis)
Mark Kowitz <mark@ROCKVAX.ROCKEFELLER.EDU>
celeste@coherent.com
"Eric S. Johnson" <esj@eng.ufl.edu>
halstern@Sun.COM (Hal Stern - Consultant)
rbthomas@jove.rutgers.edu
Nicholas_Briggs.PARC@xerox.com
jan@eik.ii.uib.no
Matthias Ernst <maer@nmr.lpc.ethz.ch>
eplrx7!mcneill@uunet.UU.NET (Keith McNeill)
Colin Allison <colin@cs.st-andrews.ac.uk>
Jeff Nieusma <nieusma@eclipse.Colorado.EDU>
jrich@ucrmath.ucr.edu (john richardson)
William LeFebvre
Computing Facilities Manager and Analyst
Department of Electrical Engineering and Computer Science
Northwestern University
<phil@eecs.nwu.edu>
This archive was generated by hypermail 2.1.2 : Fri Sep 28 2001 - 23:05:58 CDT