We have successfully deployed HA across two CSR1000v routers in each of our Azure regions. Routing is set up on the Azure backend between regions using VNet Global Peering.
While highly unlikely, in the event of both routers going down in Region A or a carrier issue into that region, I would like to know if it's possible to fail over to use the routers in Region B automatically. Using eigrp delay we have set up each region to advertise the other regions routes with a worse metric, so inbound traffic would be automatically failed over. The problem we are having is figuring out how to automatically initiate the Route Table updates in Region A to use the routers in Region B for the return traffic.
