SUMMARY: Strange Network Interface Problem

From: Alex Finkel (afinkel@pfn.com)
Date: Thu Jan 16 1997 - 10:22:53 CST


I got many helpful replies on this one and the fix turns out to be very
simple. It's actually more of a work around than a fix.

My Original Question (truncated):
---------------------
I have a Sparc 20 with a single 60 Mhz CPU/1 MB supercache, and 192 MB
physical ram. Using a serial console and running Solaris 2.4.

The last two times I have rebooted this system, the system would start fine
and the console login prompt would appear and allow logins on the console.
The problem is that no one was able to telnet into the machine from
anywhere on the LAN *Until* I logged in on the console and sent out a
single ping.
---------------------

I had checked the FAQ, the archives, and SunSolve without success, but I
may not have used the best keywords to describe the search. I did make an
effort though.

Much thanks to the following people:
------------------------------------
moreilly@fsaia.qld.gov.au (Mark J O'reilly)
Stephen Harris <sweh@mpn.com>
Bob Woodward <bobw@kramer.filmworks.com>
Davin Milun <milun@cs.Buffalo.EDU>
David Kinney <dkinney@gti.net>
"F. David Sinn" <dsinn@dsinn.seanet.com>
"Liew.Chee.Wah" <cwliew@bass.com.my>
Robin Marquis <rmarquis@ordsvy.gov.uk>
Jochen Bern <bern@penthesilea.uni-trier.de>

Summaries:
-------------
I won't include everyone's replies as the actual workaround is very simple.
 Mark J. O'reilly described having the same problem and sent me a very
thorough reply. Thanks Mark!

This is a known bug (bug id 1195862) in Solaris 2.4 and only seems to
affect Sparc 5's and 20's. Looks like the OS has trouble deciding whether
to use the AUI or the Twisted-Pair port for the interface and sending out a
ping gives it that kick in the pants it needs to start working.

There is a point patch released, 102729-01, that simply does what others
who didn't mention the patch suggested to do. It adds an outbound ping
somewhere in the rc scripts. In lieu of the patch, adding a ping to the
startup scripts is an adequate workaround.

Other suggestions had to do with routing problems and hub problems. I was
pretty sure it was neither of these as this was the only system on the
segment that had the problem. Also, the subnetting has not been changed in
quite some time. This was a recent problem.

Having the Legato Networker server on this machine was probably what kept
the problem from showing up before, as Networker probably tries to contact
it's clients on startup, thus generating outbound network traffic. When I
moved the server software off the machine, the problem appeared.

I will apply the patch, or add the ping command to /etc/rc2.d/S69inet.

Thanks everyone!

- Alex

--
Alex Finkel       | 26 Landsdowne St.      |    Macintosh, Jr.
Network Manager   | University Park at MIT | 
PfN, Incorporated | Cambridge, MA  02139   |  The power to crush
afinkel@pfn.com   | http://www.pfn.com/    |    the other kids.



This archive was generated by hypermail 2.1.2 : Fri Sep 28 2001 - 23:11:43 CDT