Path: utzoo!utgpu!water!watmath!clyde!att!osu-cis!tut.cis.ohio-state.edu!bloom-beacon!bu-cs!purdue!decwrl!ucbvax!pasteur!ames!amdahl!pyramid!octopus!sjsumcs!hom
From: hom@sjsumcs.UUCP (Hom Bahmanyar)
Newsgroups: comp.protocols.tcp-ip
Subject: A host sending an ARP request to itself?
Keywords: ARP ethernet
Message-ID: <117@sjsumcs.UUCP>
Date: 7 Jul 88 21:34:34 GMT
Organization: San Jose State Univ., Math & Computer Science Dept., San Jose, CA
Lines: 32

We have an ethernet in our Math & Computer Science Department that has a
Sequent Balance, a Convergent Technologies Miniframe, and more than 50
PC compatibles on it.  The Sequent and the PC's work perfectly on the
network (for example with PC-NFS); however, the Convergent machine has
some problems with the network:  Commands such as telnet and ftp to or
from the Convergent machine always fail but other commands such as rwho
and ruptime from the Sequent show the Convergent host as being on the
network.  Also the Convergent host never responds to pings from the
Sequent or the PC's.  After we checked all the network setup files and
couldn't find anything wrong, I wrote a simple ethernet monitor program
for one of the PC's that receives all the packets directly from the PC's
3c501 ethernet controller and displays them on the screen.  To my surprise,
I saw that the Convergent host was sending an ARP request about every 5
seconds to itself to resolve its own IP address.  That is, in the ARP
packet there were:
	sender's physical addr:		Convergent host's ethernet addr
	sender's IP addr:		Convergent host's IP addr
	target's physical addr:		0 0 0 0 0 0
	target's IP addr:		Convergent host's IP addr

But why would the Convergent host want to resolve its own address when
it already knows its ethernet address?

Any help you can give me on this will be greatly appreciated.

----

Hom Bahmanyar	(408) 924-5107
San Jose State University, Math & Computer Science Department
San Jose, CA 95192-0103
...!{sun,ames,pyramid,hplabs}!sjsumcs!hom