multicast TTL settings

[Robert Olson]

Okay ... so it sounds like TTL-selection is the way to go. The experiment
we'll need to do is to first get full-duplex multicast to each site, then
measure the ttl required to each. It looks like mtrace gives us that info
-- we'll have to see if it is correct.

This sounds like a goal for next week's nanocruises (that is: get wide-area
multicast working right to the nodes participating).

--bob