Soon after I wrote the Designing Fast Converging BGP Networks article, one of my regular readers sent me an interesting problem: BGP sessions would be lost in his (IS-IS based) core network if he would use fall-over on IBGP neighbors and the BGP router would have a primary and a backup path to the IBGP neighbor.
We’ve identified the source of the problem (delayed SPF run) and I decided to document it. It took “a while” (and a kind nudge from Mateusz following a discussion on Cisco-NSP mailing list) to find time to set up the test lab and document the behavior. You can find the (undesired) results in the Aggressive BGP fall-over behavior in the CT3 wiki.
Just in case someone from Cisco wants to put this on a wish list: please add a configurable delay to the neighbor fall-over command like you did with the bgp nexthop trigger delay command.