Symptom

There is a 2-3 second multicast traffic drop after the "vpc role preempt" is issued on the VPC secondary router for L3 multicast core routes.

Conditions

The L3 core multicast routes have reached steady state and the user issues "vpc role preempt" command. This causes the VPC primary and the secondary routers to switch roles. Since the roles are switched, PIM protocol does RPF source election. This election is a multi step process. a) The current primary needs to receive information from the VPC infra to move to secondary b) The current secondary needs to move to primary and notify all clients including PIM. c) PIM uses CFS messages to exchange RPF metrics d) PIM processes these CFS messages and notifies MFDM to reprogram the multicast routes. Any delay in the above steps will lead to one of the two things. a) the new vpc primary router updating the hardware before the secondary removes them. This causes duplicate traffic. b) The new vpc secondary router updates the hardware before the primary adds them. This causes traffic loss as described in this bug. It is to be noted that this problem is seen only for the multicast sources in the L3 CORE. If the multicast sources ingress on a VPC vlan then this problem is not seen.

Workaround

There are no workarounds for this issue. It is possible that the customer can (re)design the network in such a way that the RPF interfaces for multicast sources in a VPC environment to be VPC vlans SVIs.

Further Problem Description