Equal Cost Uplinks Killing Rendevouz Point

mrashby · ‎01-14-2009

All,

I have a couple 6509 access switches with two routed uplinks, one goes to Dist 1 and the other to Dist 2. My access-switch keeps dropping the RP for my multicast stream. I noticed that my EIGRP routed uplinks are equal cost and when I shut one down the RP mapping shows up again. Can anyone tell me first why is this happening and second how to fix it. Obviously I don't want to have to shut down one of my uplinks in order for multicasting to work. Thanks.

Mario

Giuseppe Larosa · ‎01-14-2009

Hello Mario,

two equal cost paths can be a problem for RPF check.

Actually, PIM uses the unicast routing table but it is not able to take advantage of parallel links.

From the point of view of RPF check only one of the two links is the expected link where to receive multicast traffic from a given source.

the RPF check should be done on sources of multicast traffic not on RP mapping but if you are using autoRP mapping information is carried inside packets with destination 224.0.1.39 or 224.0.1.40.

So may be are these packets there are dropped and then the RP mapping is lost.

You can:

use bootstrap protocol or manual RP instead

or you can:

disable multicast on physical uplinks and to set up a GRE tunnel where you enable ip pim sparse-dense-mode

Hope to help

Giuseppe