Folks,
For 7.1, I needed to configured SW1 and SW2 to support IPv6. In doing so, I caused a loop to form between R4 and R5 for all the prefixes directly redistributed into EIGRP on SW1, SW2 and R6. In order to solve this, I had to break the loop manually by shutting down one of the frame-relay interfaces on R4. Can this method be a valid fix to re-align the lab?
I find when digging into the problem, R5 received notification via EIGRP that SW1 and SW2 (and inevitably R6) were dropping their adjacencies because of the reboot. Therefore, since R4 loops the control plane back to R5 via OSPF, R5 decided to pick up the route with the higher distance (AD 200) and point itself to R4 for these prefixes. What is interesting though, is once it converges again, R5 doesn't natually point back to the correct location (even in the presence of lower AD routes.). Why is this?
Since this lab specifically asked us to only use distance to solve the original looping problem, we cannot add more resiliency to the loop and therefore we are vulnerable to this behavior if the topology reconverges.
Thanks
Mike