All things related to multicast

[Bill Owens <>]

I'm seeing odd behavior on my two Juniper M20s that multicast peer 
with Abilene. Overall they do multicast very well, and even seem 
resistant to the MSDP storms. But I am seeing some unexplained 
behavior on the BGP sessions with the two Abilene GSRs (NYCM and 
CLEV). The unicast routing is fine, but over time the number of 
multicast routes received across the peering session slowly 
decreases. I have confirmed with the NOC that the GSRs are still 
sending the full multicast table. A soft reset on the session has no 
effect, but after a full clear the entire table comes back.

Is anyone else seeing this problem, either with a Juniper or a Cisco? 
I suspect it is related to the MSDP storms; unfortunately, since they 
refuse to go away I can't really tell. . .

Bill.