When successor is down

Unanswered Question
Nov 21st, 2009

Suppose I have below successor and feasible successor to the network (

P, 1 successors, FD is 768
                          via (768/256), Serial0
                          via (1536/512), Serial1

Ok, but I could not understand why when successor is down the feasible successor when take place the FD for successors is still without any change, not replace it with FD of feasible successor, (why 768 not be 1536).

After successors down.


P, 1 successors, FD is 768
                          via (1536/512), Serial1

I have this problem too.
0 votes
  • 1
  • 2
  • 3
  • 4
  • 5
Overall Rating: 0 (0 ratings)
Mohamed Sobair Sun, 11/22/2009 - 01:58


The (Reported distance) and (Feasible distance) of a path is not changed. This even if the path was a feasible successor and becomes a successor.



Peter Paluch Sun, 11/22/2009 - 15:43


The feasible distance in your case has not changed because it is always a record of the historically minimal distance to a particular destination. A feasible distance is set once a route goes from active to passive state. Since that moment, the feasible distance will decrease if also the real distance decreases, or it will stay at its current value if the current distance grows and, at the same time, there is still some feasible successor available that provides the next available shortest path (i.e. when it is not necessary to again go to active state with that route). Note that this behavior is indeed equivalent to maintaining the record of the minimal distance to the destination.

A feasible distance may be increased (i.e. reset) only when a route goes from active to passive state. The "history" (as in the "historically minimal distance") starts and ends with the route going from active to passive state.

The logic behind this is, roughly speaking, that the diffusing computation algorithm itself makes sure that when a route goes from active to passive state, it is guaranteedly loop-free. A subsequent direct decrease in route distance signals a topology change but creation of a loop would never lead to a distance decrease, therefore, the topology change is completely trustworthy. That is why a router is always allowed to decrease the feasible distance and use the neighbor that provides the lowest distance without going to active state. If, however, a distance increase is detected, then a preliminary test is done: Is the neighbor closer to the destination than I have ever been? This is basically what the feasible condition says. If the neighbor passes this check then he again cannot be on a routing loop and thus he is usable, without us changing the feasible distance (note that we need it for further tests in future). If he does not pass it, then the route will go to active state in which case a new diffusing computation is started that creates a new loop-free path and after that, the feasible distance can be reset and again set to a higher value so that the neighbor can pass the feasibility check.

Best regards,



This Discussion