Lester Vecsey wrote:
When I ping the multicast address, i.e. the default gmond multicast address of 239.2.11.71, on a network of say 20 gmonds, I get back one successful result and then 19 marked as "DUP!"
When you ping a multicast address, you should get a response from every host listening on that address. So that's correct.
Is this correct or does it suggest that the cisco 4000 series catalyst I have the machines plugged into should be tweaked further to manage multicast traffic on a vlan more efficiently?
See below.
This seems to be the case because I can run tcpdump or another sniffer on a machine on the same vlan not even runnnig gmond, and it can still get all the multicast packets that are being sent around.
A host should only receive multicast packets after it "joins" that IP (starts listening on it). I just verified this locally. So yes, you've got a problem.
I brought this up previously on the list and the idea of unconfigured switches got some mention by Steves write-up that was posted recently, but I could really use a definite answer on this and/or exact commands to reconfigure things :)
I can't really help you with specific commands to configure your Cisco gear. We use Foundry (www.foundrynet.com) everywhere.
Fortunately, due to the economic downturn, CCNEs are in abundance. You could probably get the exact answer at your local Starbucks...