[nsp] HSRP or switch issue?

[John Starta]

I'm currently experiencing an oddity with multicast traffic like HSRP that 
I'm looking for some ideas on. For simplicity the network design consists 
of 2 Cisco 3640 routers running HSRP between them connected to a single 
Extreme [Black Diamond] switch. Basically...

            extreme switch
              |       |
              |       |
             rtr1    rtr2

Normally everything works just fine, but periodically -- in time, not 
quantity -- HSRP indicates via the %HSRP-4-DUPADDR message that I have a 
duplicate [IP] address. (The quantity of the messages indicating the 
duplicate IP address ranges from half dozen to nearly a hundred. The time 
between messages closely matches the HSRP HELLO interval.)

When I receive these messages, on the active HSRP router for instance, they 
indicate the duplicate address as being the physical interface IP address 
of the active HSRP router with the source MAC address as the virtual MAC 
[address] of the active HSRP router. Receipt of these %HSRP-4-DUPADDR 
messages indicating the duplicate as itself suggests an issue with 
multicast -- a loop of sorts whereby the switch copies the multicast 
announcement [back] to the same switch port it originated. Keep in mind 
that there are no interface or HSRP state changes so the messages probably 
aren't coming from the standby HSRP router. (Especially since the indicated 
duplicate IP address is that of the physical interface on the active HSRP 
router, not the virtual IP.)

I did some poking around on Extreme's web site and they indicate an issue 
with HSRP in an earlier version of code, but that is/was fixed in the 
version being used.

Have anybody run into this before? Ideas regarding cause? I don't have 
access to the switch since it belongs to the customer.

.,