Solved: Understanding EIGRP Topolgy Table

jpl861 · ‎11-22-2007

Hi,

I just have an inquiry about the EIGRP topology table. I'm trying to filter routes from our one of our site. Here's the initial EIGRP topolgy table before applying the prefix-list blocking 10.0.0.0/8 prefix.

At first, the primary route is via ATM1/0.308 and other routes are via other ATM.

After applying the prefix-list and distribute-list that block 10.0.0.0/8 network in ATM1/0.308, the primary route became Fa0/0

I understand that Fa0/0 has the best metric to 10.0.0.0/8 after applying the configuration but why did it not appear in the topology table prior to applying the distribute list.

I also understand that Reported Distance should be lower than the Feasible Distance to become a Feasible successor but why is it that routes to ATM1/0.105 and ATM1/0.106 is in the topology table prior to the prefix-list and distribute list configuration. The reported distance of those routes are 2516992 which is greater than the ATM1/0.308's 2514432. So it did not become an FS. But why is it that the route to Fa0/0 is not there where 2517248 is also greater than the Successor metric.

Thanks.

John

Richard Burts · ‎11-22-2007

John

We do not know enough about the topology of your network to be sure. But I wonder if the neighbor at 10.199.32.2 on the Fa0/0 interface was learning the route to 10.0.0.0/8 from this router before the distribute list was applied. If so then split horizon would prevent advertising that route back on the Fa0/0 interface. After the distribute list is applied then the neighbor is no longer learning the route from this router and is able to advertise the route over the Fa0/0 interface. Perhaps you can remove the distribute list and then check on 10.199.32.2 to see how and where it is learning the route?

HTH

Rick

HTH

Rick

View solution in original post

royalblues · ‎11-22-2007

John,

Can you post a brief network topology of your network?

The FD of a link does not change even if there is a change in the link metrics provided the route on the device never became active.

Narayan

Richard Burts · ‎11-22-2007

John

We do not know enough about the topology of your network to be sure. But I wonder if the neighbor at 10.199.32.2 on the Fa0/0 interface was learning the route to 10.0.0.0/8 from this router before the distribute list was applied. If so then split horizon would prevent advertising that route back on the Fa0/0 interface. After the distribute list is applied then the neighbor is no longer learning the route from this router and is able to advertise the route over the Fa0/0 interface. Perhaps you can remove the distribute list and then check on 10.199.32.2 to see how and where it is learning the route?

HTH

Rick

HTH

Rick

jpl861 · ‎11-22-2007

Hi Rick,

I just figured it out. Yes, 10.199.32.2 learned the route from this router and split-horizon took place. I forgot the split-horizon rule while I was troubleshooting. Well anyway, thank you so much for replying and at least I know that my suspicion is correct. Split-horizon. :)

Thanks guys.

John

Richard Burts · ‎11-22-2007

John

I am glad that my response was helpful in resolving your question. Thank you for using the rating system to indicate that your question was resolved (and thanks for the rating). It makes the forum more useful when people can read a question and can know that they will read a response that helped resolve the question.

I encourage you to continue your participation in the forum.

HTH

Rick

HTH

Rick

jpl861 · ‎11-22-2007

Hi Rick,

One quick question, I'm getting kinda confused right now or I'm just confusing myself. For example, I have 3 interfaces, ATM, FR, and FE. All of those 3 interfaces received route to 10.0.0.0/8 network but the best one is the ATM, will it still advertise 10.0.0.0/8 to the FR and FE interface even though it also received that prefix on that interface?

Regards,

John

s.arunkumar · ‎11-22-2007

yes i think so,but the composite metric will be infinity,so the device connected to FR and FE wont accept it...

:)

Richard Burts · ‎11-23-2007

John

I do not believe that this is correct answer. If EIGRP receives advertisement of 10.0.0.0/8 on 3 interfaces it will choose which ever route has the best metric. Split horizon says that EIGRP will not advertise the route out the interface of the best route. But I believe that EIGRP will advertise its best route out all of its other interfaces.

HTH

Rick

HTH

Rick

royalblues · ‎11-23-2007

I think you can get a confirmation by doing the finite state machine debug

debug eigrp fsm

Narayan

jpl861 · ‎11-23-2007

Ei Guys,

Thanks for all your replies. I think it still advertises the route. Rick is right, the split-horizon rule applies only to the best route. Here's the scenario.

RouterA's best route to 10.0.0.0/0 is via ATM2/0.308. Router B's best route to 10.0.0.0/8 is via Fa0/0. Router A and B is connected by ATM PVC.

I checked the EIGRP topology table of both routers and the total delay was correct. Router A advertised 10.0.0.0/8 to the PVC to Router B but at the same time, Router A has a route to 10.0.0.0/8 received from Router B. The computed total delay was also correct.

Does it make sense? Or these are just old routes?

Thanks,

John

Richard Burts · ‎11-23-2007

John

It makes sense to me.

HTH

Rick

HTH

Rick

s.arunkumar · ‎11-25-2007

Interesting,I understand that the split horizon rule is applied in case of subinterfaces of frame relay etc..i made the above statement considering when interfaces were different,it behaves with poison reverse.

ref:: http://www.cisco.com/warp/public/103/eigrp-toc.html#splithorizon

anyway,thanks all to make it more clear.. :)

jpl861 · ‎11-26-2007

Thanks to all of your replies guys. =)

Really appreciate it.

John

ruwhite · ‎11-26-2007

"Split horizon says that EIGRP will not advertise the route out the interface of the best route."

This is the correct answer.... EIGRP actually runs split horizon against the routes installed in the routing table, not the topology table entries. So, if you have 10 equal cost paths, and only 8 are installed in the table, it will only split horizon the 8 paths installed in the routing table. This is the primary reason I always look at the number of alternate/equal cost paths in an EIGRP network when consider the probable stability over the long term.

HTH

:-)

Russ