If R1 is sending its distance to the destination, it won’t send the metric via the R1-R3-R4 path, because the next hop IP is that of the router (R3) from which it received the EIGRP update. It will however send the metric via the R1-R2-R3-R4 path, even though R3 is in that path! However, it is not the next hop IP, because that update was received from R2, so the split horizon rule is not being violated.
Yes, that’s absolutely it. Only if you dig deeper, like the way you and @davidilles are doing, will you learn the intricacies and the details of how these protocols work. And such intimate knowledge is especially useful in troubleshooting. That’s what makes the difference between a good and a great network engineer!
I’m not sure if the statement “If you lose the successor because of a link failure, EIGRP will copy/paste the feasible successor into the routing table” is 100% accurate.
If you have an additional path that failed the FC check, but has an overall lower distance to the destination as the FS, it will end up being picked as the successor, albeit after a diffusing computation cycle, assuming the router does not learn of any other paths with a lower CD.
When a successor route fails, EIGRP will look for an FS. If an FS exists, it will be promoted to the successor route immediately, without going through a DUAL process. This will happen regardless of whether or not there is another path with a lower distance that failed the FC check. By definition, the DUAL process will not be run if an FS exists. This is the fundamental mechanism that makes EIGRP have such a fast reconvergence.
Now I understand your statement, and how you came to this conclusion. However, let’s take a closer look at how EIGRP works when it involves successors, FSes, FC checks and the topology table. Let’s take a look at the following scenario:
Route 1 - This is the original successor.
Route 2 - This is the feasible successor that has been calculated by EIGRP
Route 3 - This is another route that has an overall lower distance to the destination but has failed the FC check.
Let’s say that Route 1 fails. EIGRP will look for an FS and immediately place that in the routing table. DUAL will not run.
EIGRP will not replace any current successor unless DUAL is triggered. For this reason, Route 3 will not be considered as a “better” route than Route 2. Only when DUAL is triggered will this happen.
In the meantime, a new FS may be determined as part of EIGRP’s normal operations. Determining an FS is not done during a DUAL computation (reconvergence) but is done in the background since it does not affect current stable routing conditions.
Route 3 will only become a successor if a DUAL computation is triggered. DUAL is triggered under the following circumstances:
a. Loss of the Successor route without an available Feasible Successor (FS).
b. Significant topology change with no immediate Feasible Successor.
c. Receiving a DUAL query from a neighbor.
d. Route metric change that invalidates the current Feasible Successor.
e. No Feasible Successor found during a route update.
So, Route 3 will not become the successor even after the failure of Route 1. It will only be evaluated as a candidate successor if and when the DUAL process is triggered. Does that make sense?
