I just read the detailed explanation from ATS about today's network outage
that was posted to the IT Exchange list, and how the triggering event was a
loop somewhere.  Last week here in the Kellogg Center, we had the same thing
happen in miniature -- somebody (who was not me, by the way) innocently
created a loop in an office, and brought down the whole building for a
while.  ATS had to send somebody out to diagnose it and yank the offending
cable (which they did quickly).
 Now I thought we had Spanning Tree Protocol for this kind of thing.  My
understanding is that any modern network switch (maybe not the low-end
models from Netgear, but anything that a large university is going to have
in the network closet) is going to implement spanning tree, and will detect
a loop.  Instead of bringing down the network, the offending port is
disabled, and maybe a light blinks somewhere.
 So am I completely backward here?  Admittedly, I haven't set up anything
bigger than a home network in a few years, but I thought this was a solved
problem.  How, in 2010, are we still susceptible to such a common mistake?
As a colleague said to me today, I guess I know what to do when I feel like
going home early.

Thanks,

--
Carl Raymond
Software Developer
University Outreach & Engagement
Michigan State University
Kellogg Center, Garden Level
East Lansing, MI 48824-1022

[log in to unmask]
(517) 353-8977
http://outreach.msu.edu/