I was doing some testing with eigrp based on the attached diagram
Metric weights 0 0 0 1 0 0 command was used under EIGRP so that only delay influences the metric. The metrics were configured as follows
R1 - R2 delay 1
R2 - R4 delay 1
R1 - R3 delay 2
R3 - R4 delay 1
R4 loop - delay 1
This resulted in R1 choosing the path via R2 (FD=768) for the loopback and the path via R3(1024) as a FS as expected
Now i shut down the interface at R2 connecting to R1
Since the FS via R3 was available, the path was immediately chosen in the routing table. The FD didn't get updated as the route never went active.
I manipulated the metric a bit further by changing the delay between R1-R2 to 3
Then i unshut the interface at R2 towards R1
Now the path via R2 was selected as FS (FD=1280). If now i shut down the interface connecting to R3, the FD would still not be updated as it can find as FS in the topology.
If the metric changes again in such a manner that the delay between R3-R4 becomes 2, then the path via R3 is not chosen as FS (as AD=FD) and hence would result in a Dual computation of the route in the event of failure eventhough the FS condition would have been satisfied had the FD been updated earlier
Nothing except clearing the EIGRP process at R1 seems to update the FD and reflect the true vlaue
So does it make sense to update the FD when FS route is installed in the RT
Happy New year to all
Narayan
