I'm having problems with multicast traffic in an environment using HSRP.
I have a MSFC2 in a 6500/Sup2, and RSFC in a 5500/Sup3G configured in a domain with 60 vlans (60 subnets in a Class B, with approx 5000 clients). The two router cards are configured with HSRP on all 60 vlans, with the MSFC2 having a higher priority, and hence active for all vlans. The 5500/6500 are at opposite corners of a square, with 3508's at the other two corners, joined together with gigabit dot1q fibre trunks. There is also a direct connect dot1q trunk between the 6500/5500 across the square. The redundant links are controlled by STP, active on all vlans. VTP pruning is active.
All interfaces on both routers are configured for ip pim sparse-dense-mode and ip cgmp.
Microsoft Media Player 7 crashes between 120 and 180 seconds after starting to view a multicast video stream. The error message (catastrophic failure 8000ffff) indicates that it has recieved duplicate multicast packets.
A multicast ghost PC disk imaging between subnets/vlans will take the 5500 to 100% backplane usage, and it will forget to transmit traffic to the RSFC on port 15/1, causing the HSRP to start flapping. Sometimes it even locks the RSFC up.
The video stream problem is resolved by removing the ip pim/ip cgmp commands from the standby router. This means I will not have multicast support on HSRP fallover, which defeats some of the purpose of having HSRP.
The fact that it normally crashes after at least 120 seconds, rather than the first video packet implies it has more to do with the IGMP/CGMP traffic than the data stream.
Removing the ip pim/ip cgmp from the HSRP does NOT fix the Ghost multicast storm. The multicast activity is about 10 to 20mbit, 1000 pps.
There also appears to be multicast traffic on the STP blocked links. I've disabled them, and it makes no difference.
Anyone have any clues? I've also a problem with multicast pruning on my other 13 routers, but for this exercise they were multicast isolated.
