Path: utzoo!utgpu!news-server.csri.toronto.edu!cs.utexas.edu!uunet!mcsun!ukc!icdoc!sot-ecs!tjc
From: tjc@ecs.soton.ac.uk (Tim Chown)
Newsgroups: comp.dcom.lans
Subject: Re: Confused Retix bridge?
Message-ID: <4934@ecs.soton.ac.uk>
Date: 31 Oct 90 14:35:12 GMT
References: <1990Oct30.210914.23429@msuinfo.cl.msu.edu>
Organization: University of Southampton, UK
Lines: 29

In <1990Oct30.210914.23429@msuinfo.cl.msu.edu> tdd@convex.cl.msu.edu (Thomas D. Davis) writes:

>We have a number of Retix 2255M bridges on our campus ethernet (we have DEC
>and Crosscomm bridges as well) that seem to have intermittent problems.  The
>symptom that we've experienced which points to bridge trouble is basically
>the loss of connectivity (NFS block read errors, ping failures) to machines
>on the far side of the bridge.  After checking out as many variables as we
>have the capability to test, we have pointed an accusing finger at these
>bridges.  It "feels" like the bridge is getting its forwarding tables upset,
>although I am definitely going on a hunch here.  Like everyone else, our net
>traffic has increased substantially over the past months which may be a
>contributing factor (shouldn't bridges keep up with a loaded network?).

>Has anyone else had problems like this?  I know of no way to tell for sure
>that the bridge is to blame other than to swap it out with another and see
>if the problem recurs (yuck!).  How do other people debug such problems?

We have two 2265Ms which had similar problems.  What appears to be
happening is that they lock up on certain combinations of short/long
packets or CRC errors, which can occur.  Retix announced in April that
such bridges *must* be upgraded to firmware revision E (which has 
corrections for these bugs) while we in fact only had revision D.
It may well be the same for your 2255Ms.

A handy tip here is to get a portable ether analyser, which can be
found for under #1000 in the UK.  Well worth the expense, I feel.

Cheers,
	Tim.