Strange network congestion

[miller@GEM.cam.nist.gov (Bruce R. Miller)]

    Date: Tue, 11 Dec 90  09:10:00 EST
    From: delaney@xn.ll.mit.edu (John R. Delaney)

	... net congestion ...

We had a similar problem once; it affected XL's as well, but didn't
completely cripple them. 

The down  side of  the LispM  was that  I was  the only  one on my hall
noticing a degradation (others were Suns & SGI's).  Apparently it  takes
`less work' for these unices to figure out to ignore a packet?

The upside of the LispM was that using Peek I was able to determine that
there were a phenomenal number of malformed (i think) broadcast IP packets
coming in, hundreds  per  second.   When  I  called the (unix-oriented)
network manager and told  him what  I was  seeing he  was surprised
that I  had figured that much of the problem out without a `real'
network monitor. 

It turns out that they were trying to configure yellow pages on
an unnamed new Risc machine and booting.  Apparently it figured if it
hadn't got a reply within a microsecond, it better ask again!

Anyway, PEEK should tell you not only what kind and how many packets are
being recieved (apparently not from where), but should also tell you
what each machine is sending out.  Unless that latter is lying, you
should be able to isolate or rule out any lispm trashing the net.  Since
you said disconnecting the router didn't affect it, it must be something
local.   As mabry suggested, look for a trying-to-boot (or
trying-to-anything) machine on your local net.

good luck.

bruce
miller@cam.nist.gov