Protocol Overview
EIGRP activities start at EIGRP start time by sending initial hello messages through the active interfaces of the underlying router node. The EIGRP node discovers neighbor EIGRP modules when they send hello messages, and the module begins to build its topology and routing tables accordingly. In an effort to discover the entire network, the EIGRP nodes then exchange topology and update messages with their new-found neighbors. When all update messages have been processed and there are no triggered update messages left, the EIGRP network has converged (that is, all routers have discovered all of the IP subnetworks in the entire network). At this point, the EIGRP process enters into steady state.
In steady state, periodic hello messages are used by EIGRP nodes to detect changes in neighbor node status. If an EIGRP node does not receive a hello message from a neighbor for a specific amount of time, the EIGRP process considers that neighbor to be down or unreachable. If the failed node recovers, then its EIGRP process will start sending periodic hello messages again, and these hello messages alert its neighbors that it has recovered.
In case of topology changes, the EIGRP process tries to recover quickly by using its topology table to find an alternate route. If the topology table does not yield a feasible alternate route for some subnetworks, the EIGRP process starts DUAL, which queries neighbor nodes’ topology tables for an alternate route. Execution of DUAL causes query and reply message exchanges between EIGRP processes, as well as the exchange of update messages. The EIGRP network reaches convergence again when there are no active query, reply or update messages in the network. Any subnetworks to which DUAL could not find a route become unreachable.