On Nov 14, 2007 3:06 AM, Uncle Jim [email protected] wrote:
/sbin/route add -host 229.255.0.1 dev eth0
I'm not a multicast expert (or even multicast user) BUT my guess is that the proper command should be:
/sbin/route add -net 224.0.0.0 netmask 240.0.0.0 dev eth0
This is not to say that Monty's command wouldn't work but that this is the more general command for the 224.0.0.0/4 (224.0.0.0 - 239.255.255.255) multicast network.
But we explicitly do NOT want all multicast traffic to go to the other server. We ONLY want this one multicast IP to go there. Presumably the other server is set up the other way, and in doing this we've defined a multicast IP they can use to talk to each other, but there's nothing else on that subnet, so no other multicast will work there.
Put it in /etc/rc.local so it will always be run at boot.
Monty has spent too much time in the SCO world. On a RedHat system you should use /etc/rc.d/rc.local.
On my RHEL4 box, /etc/rc.local is a link to rc.d/rc.local, so it's all the same, and 5 characters less to (mis)type.
This should allow other multicast packets to stay on eth1.
I don't know why you would want other multicast packets to go to "The Internet" rather than your local 10. network
Because something else may need to use *routable* multicast (not 224/8, but any other multicast IPs), and sending ALL multicast traffic out eth0 where there is no routing to those other machines will definitely break that. Remember that the 10. network has exactly two servers and no routers. I'm trying to move the specific routing to the "non-default" interface, because I just don't know what else needs to use eth1 with multicast.
but your 169.254.0.0 traffic, which is defined as "LINKLOCAL", is still not on your "local" network. But since you don't use it, who cares?
169.254 is a separate UNICAST network that allows DHCP clients to function if they can't obtain a lease.
Your more general route should make any 224/8 multicast on eth1 fail. That might be a problem.