If I initiate a ping from a PE node to a remotely attached CE node on VRF-A using the following command:

ping 192.168.199.5 routing-table=VRF-A source-address=172.30.16.1

(the ip address 172.30.16.1 is assigned to a loopback bridge interface on the PE node that is part of the VRF-A address space)

The ping reaches the CE node properly, but has a source address of the interface it exited through (wrong VRF).

Is this intended? It would be *really* nice for troubleshooting purposes to initiate pings/traceroutes from a PE node (assuming it had an interface that was part of the VRF) into a VRF.