From: Brian Hescock (bhescock@xxxxxxxxx)
Date: Sun Oct 21 2001 - 12:18:21 GMT-3
The previous discussion about multicast made me think about a problem I
see occasionally and I thought I'd pass it along. Don't use anything in
the 224.0.0.x range for a multicast address. It will work fine if the
source and destination are in the same vlan (unless you're using one of
the reserved addresses, such as 224.0.0.10 for eigrp, which would
probably wouldn't be a good thing to do... ;-). The reason it doesn't
work when routing multicast is the 224.0.0.x is a "link-local" address,
it never gets forwarded off the local segment, you will never get ip
multicast for 224.0.0.x to work across a router unless you bridged it
(haven't tried it but it should work).
Most people wouldn't use 224.0.0.x but I see it happen occasionally and
wanted to help save some people the grief of troubleshooting the problem
if you used that range of addresses by mistake. Another common problem
in production networks is many multicast servers have a default ttl of 1
and, since one of the first things a router does is decrement the ttl by
one, the packets get dropped at the router. The solution is to increase
the ttl of the multicast server to be at least one higher than the
number of hops to the furtherest multicast receiver.
Brian
p.s. in case anyone was wondering, the previous e-mail with my comment
about mutlicast being on by default was referencing a discussion about
NLSP, not ip multicast (the part about NLSP was further down in the
e-mail thread).
This archive was generated by hypermail 2.1.4 : Thu Jun 20 2002 - 22:33:22 GMT-3