As shown in the above linked diagram, when the client (vnet A) needs to ping server (vnet C), is it possible to force that traffic to go through the two appliances inside the Hub vnet (vnet B)?
UDR1 forces traffic to go through Virtual Appliance Router 1, and UDR2 forces traffic out from Router 1 to Router 2, once traffic comes out of Router 2, it will arrive at server, this is done via peering itself.
UDR3 and UDR4 are for return traffic.
tcpdump on Router 1 shows traffic coming in, and then Router 1 forwards that traffic out to Router 2, but there is no traffic seen on Router 2, despite UDR 2.
However, pinging from client to server does work, this means Azure is doing the delivery, which causes the traffic to bypass Router 2.
Is multiple virtual appliances supported in a peered vnet environment? It is supported when all traffic is within the same vnet, this is tested working.