Assume we have a network of nodes connected via point to point links, and a source S sends a message that will be broadcast to all nodes using Reverse-Path-Flooding. Assume that routers do not keep track of broadcast messages they have seen earlier.
Assume also that routing tables change frequently, however, when a node changes its next-hop neighbor towards S, it is always ensured that the spanning tree used in RPF is loop-free (i.e. all nodes remain connected to the tree and there are no loops, the integrity of the tree is preserved)
Show that even with the above restrictions, it is possible that a message traverses the network and never stops (well, it will stop when its time-to-live expires, but assume messages don't have a time-to-live limit)