Reading an EIGRP topology-table entry with three paths, how do you pick the successor and feasible successors, and which reach the routing table?
answer
- lowest total distance first
- feasible distance is the yardstick
- neighbour's reported distance, strictly lower
- ties install together
basics
~20 sThe successor is the path with the lowest computed distance, which sets the feasible distance; any other path whose reported distance is strictly below that feasible distance is a feasible successor. Only successors, including equal-cost ones, reach the routing table.
solid answer
~50 sTake one prefix with paths via A, B and C. First find the lowest **computed distance** (CD), the total through each neighbour: that path is the **successor**, and in a stable network its CD is the **feasible distance** (FD). Then compare every other neighbour's **reported distance** (RD), its own distance to the prefix, against the FD: if RD is **strictly less** than FD, that neighbour is a **feasible successor**; if it is equal or higher, it is neither and is merely recorded. Only the successor goes into the routing table. If two paths tie on the lowest CD, both are successors and both are installed, up to the implementation's limit on parallel paths. A path can have a lower CD than a feasible successor and still not be one, because the test looks at RD, not CD.
go deeper
Recall the three distances: reported distance from the neighbour, computed distance through it, and one feasible distance per destination, and that only the successor is installed.
Walk the entry in order: lowest computed distance gives the successor and the feasible distance, then each reported distance strictly below it marks a feasible successor.
Spot the traps under pressure: a cheaper path that fails the test, a stale feasible distance after a metric rise, equal-cost ties against an implementation's path limit.
Use the table to reason about design: where topology and metrics guarantee a feasible successor exists, failover stays local; where they do not, every failure becomes a query across the domain.
## The entry you are reading An EIGRP router keeps a **topology table**: for every destination, every neighbour that advertised it, with two distances per neighbour and one per destination. EIGRP's public specification is Informational RFC 7868, which defines them: - **Reported distance (RD)**: the neighbour's own distance to the destination, as it advertised it. - **Computed distance (CD)**: this router's total distance through that neighbour, built from the neighbour's RD and the cost of the link to it. - **Feasible distance (FD)**: the lowest distance this router has known for the destination since the route last went from active to passive. One per destination. Here is one prefix, 10.1.1.0/24, on router R1, which has three neighbours advertising it. Metrics are in abstract units so the arithmetic stays visible; real EIGRP metrics are much larger numbers built from bandwidth and delay. | Path | Neighbour's RD | R1's CD | |---|---|---| | via A | 1,000 | 2,000 | | via B | 1,500 | 3,000 | | via C | 2,500 | 3,500 | Assume the route has been stable, so the FD equals the best CD, 2,000. ## The walk, step by step 1. **Find the least-cost path.** The lowest CD is 2,000, via A. 2. **Name the successor.** RFC 7868 defines a successor as a neighbour that meets the feasibility condition and provides the least-cost path. A's RD of 1,000 is below the FD of 2,000, so A is the successor. 3. **Test every other neighbour's RD against the FD.** This is the **feasibility condition**, which DUAL owns; the table just supplies the numbers. - B reports 1,500. 1,500 < 2,000, so B is a **feasible successor**: B is closer to the destination than R1 is, so B cannot be routing through R1. - C reports 2,500. 2,500 is not below 2,000, so C is **neither**. C might be upstream of R1, reaching 10.1.1.0/24 back through R1, so its path cannot be proven loop-free. 4. **Install.** Only A goes into the routing table, with a metric of 2,000. B waits in the topology table as the ready backup. C is recorded but not used while the route is passive. ## Ties and the path limit Add a fourth neighbour D with RD 800 and CD 2,000. D ties A on the lowest CD and passes the feasibility condition, so **both** A and D are successors; RFC 7868 says feasible successors providing the least-total-cost path are also called successors. Both are installed and the router forwards over both, splitting flows between equal-cost next hops as any IP router does. How many equal-cost paths a router will install is an **implementation choice**, not a number RFC 7868 sets. ## The traps in this reading - **Strictly less.** A neighbour whose RD equals the FD is not a feasible successor. The condition is RD < FD, never RD <= FD. - **RD, not CD.** C's total of 3,500 is the reason people think it is a "worse backup". The real reason it is excluded is its RD. Move the numbers so C reports 2,500 but its CD is 2,600, cheaper than B's 3,000, and C is still not a feasible successor while B still is. - **FD is a history, not a live value.** RFC 7868 stresses that FD is the lowest distance since the route last became passive and is not necessarily the current best. If A's link cost later rises so its CD becomes 2,400 without the route going active, the FD stays 2,000 and the test keeps using 2,000. - **Neither is not discarded.** C's entry stays in the topology table. If A and B both disappear, DUAL recomputes using every neighbour's current distance, including C's. - **Feasible successors are not installed.** B does not enter the routing table unless A fails, or an implementation's unequal-cost load balancing explicitly admits it. ## What a strong answer adds A strong candidate reads the table in the right order: lowest CD first, then the FD, then each RD against the FD. They say plainly that the successor is installed and the feasible successor is not, that ties produce several successors, that the path limit is an implementation's, and that a path's place in the table depends on its neighbour's reported distance rather than on how cheap it looks from here. They mention, in a single clause, that when no feasible successor exists the route goes active and DUAL queries the neighbours, without wandering into that process.
- In the EIGRP table above, what happens if the path via A fails?B is a feasible successor, so R1 promotes it at once: B becomes the successor and its path, CD 3,000, enters the routing table without R1 querying any neighbour. C is still not usable, because its RD of 2,500 was never below the FD of 2,000.
- Why does an EIGRP feasible successor's reported distance have to be strictly lower than the feasible distance rather than equal to it?A neighbour reporting exactly R1's own best distance might be reaching the destination through R1 at equal cost, so its path cannot be proven loop-free. Only a neighbour strictly closer to the destination than R1 has ever been since the route last became passive is guaranteed to be downstream.
- How many equal-cost EIGRP successors can be installed for one prefix?As many as the implementation's parallel-path limit allows. RFC 7868 does not fix the number; routers set a default and let the operator raise it. Paths beyond the limit stay in the topology table as successors that are not installed.
saying these in an interview costs you the question
- A neighbour whose reported distance equals the feasible distance is a feasible successor.
- Feasible successors are chosen by comparing computed distances, the totals through each neighbour.
- The second-cheapest path is always the feasible successor.
- Feasible distance is recalculated to the current best metric on every change.
- Paths that are not feasible successors are deleted from the topology table.