EIGRP Topology Table question - please assist?

/usr/usr Member Posts: 1,768 ■■■□□□□□□□
in CCNP
Banging my head against the wall over something I feel shouldn't be that difficult...I don't know if I'm overlooking something simple (which is very likely), or what. I've attached screenshots below, but let me explain: When I run "sh ip eigrp topology all-links" from R3, I expect to see two routes to the 10.1.1.0 /24 networking hanging off R1. One route to R1 directly, one route through R2. However, I only see the route from R1. I understand R3 not adding the route to the topology table due to the horrible FD because my bandwidth on that link is so low, but am I wrong for expecting to see it being advertised from R2 using the "all links" command? When I shut down FastEthernet 0/0 on R1...R3 will learn the route from R2 and all is well. I just can't figure out why the hell I can't see it from R3 while all links are up. Am I missing something here? I feel like an idiot.
· Share on FacebookShare on Twitter

Comments

  • xXErebuSxXErebuS Member Posts: 230
    Make the switch a Layer 3 switch with each link having its own subnet....
    · Share on FacebookShare on Twitter
  • /usr/usr Member Posts: 1,768 ■■■□□□□□□□
    This is just a lab from the ROUTE lab manual that I setup in GNS3. I'm not worried about correcting the issue by changing the topology, I am primarily just curious if this is what I am supposed to be seeing and if so, why?
    · Share on FacebookShare on Twitter
  • xXErebuSxXErebuS Member Posts: 230
    I can barley see topology my guess would be split horizon
    · Share on FacebookShare on Twitter
  • networker050184networker050184 Mod Posts: 11,962 Mod
    What is R2's best route to reach 10.1.1.0/24? I assume you are hitting split horizon.


    Edit
    Beat me to it!

    One other thing to keep in mind is EIGRP is distance vector and only passes the best routes, hence the 'routing by rumor' description. Routers are not aware of the entire topology in a distance vector routing protocol like they would be in a link state routing protocol.
    An expert is a man who has made all the mistakes which can be made.
    · Share on FacebookShare on Twitter
  • /usr/usr Member Posts: 1,768 ■■■□□□□□□□
    R2 learns of 10.1.1.0 /24 from R1 via it's Serial interface and it's FastEthernet interface.

    How would Split Horizon prevent R3 from learning about 10.1.1.0 /24 from R2, if R2 received that route on it's serial interface?

    R3 has a neighbor relationship with R1 and is learning the route from there, in the screenshots. Only when I shut down FastEthernet0/0 on R1 will R3 learn the 10.1.1.0 route from R2...
    · Share on FacebookShare on Twitter
  • networker050184networker050184 Mod Posts: 11,962 Mod
    R2's best route is through R1. Distance vector protocols only advertise their best routes. So when R2 goes to create it's update for R1 it sees that the route for 10.1.1.0/24 is out the same interface it is about to send the update on. So no update sent.

    When you shutdown the R1-R3 link now R2 has its best route on the serial link. Split horizon is no longer an issue in sending the update to R3.

    Now that I look at it, it would seem R3 would also hit the split horizon rule before sending the route to R2. I'll have to look at it deeper....

    Yep split horizon as I originally thought. Discussed here not long ago.

    http://www.techexams.net/forums/ccnp/88604-ccnp-student-lab-manual-question-about-eigrp.html
    An expert is a man who has made all the mistakes which can be made.
    · Share on FacebookShare on Twitter
  • xXErebuSxXErebuS Member Posts: 230
    networker050184 wrote: »
    R2's best route is through R1. Distance vector protocols only advertise their best routes. So when R2 goes to create it's update for R1 it sees that the route for 10.1.1.0/24 is out the same interface it is about to send the update on. So no update sent.

    When you shutdown the R1-R3 link now R2 has its best route on the serial link. Split horizon is no longer an issue in sending the update to R3.

    Now that I look at it, it would seem R3 would also hit the split horizon rule before sending the route to R2. I'll have to look at it deeper....


    R3 will always hit split horizon rule except for its networks since it only 1 link. When you shut down the interface what happens is R3 no longer sees the R1 route and the lloop is out of picture
    · Share on FacebookShare on Twitter
  • /usr/usr Member Posts: 1,768 ■■■□□□□□□□
    Ugh, man...I should have caught that but it truthfully never even entered my mind.

    Because R2 knows about 10.1.1.0/24 off it's FastEthernet interface, it's not going to send that same route back out that interface to it's neighbors, even if it itself has another path to get there over it's Serial interface...

    I think I read too far into the first lab in that book - it never mentioned looking at the topology table, I was just poking around.

    Thanks for taking the time to look at this guys. I'm fairly certain I'll be posting more as I start to lab all this stuff up, haha
    · Share on FacebookShare on Twitter
Sign In or Register to comment.