OSPF - Area STUB multiple exit points

Hello friends,

I am reviewing the following excersise:


I have a doubt regarding that. In the exercise they cover the problem of the ABR's loosing the connection to the area 1. But the excercise doesn't cover the failure of the connection from the ABR to area 0.

I mean, if the ABR is isolated and cannot reach the area 0, but it still has connectivity to the area 1, I have realized that in this situation, the ABR is still advertising a default route to the area 1, but the ABR is isolated, so some of the routers will lost its connectivity due to that.

Is there any solution to solve this problem in scenarios with two or more exit points in a stub or nssa area??

Thank you very much!







  • To fix this blackholing, which means to make that router NOT inject a default route, you need to make sure that the router is not an ABR anymore, which means that router needs to have no interfaces in area 0 in the UP/UP state.

  • Hello Cristian and many thanks for your response!


    I was using a GNS3 topology to simulate that, and I wasn't able to achieve the solution.


    My problem was that I was trying to "shutdown" the port on the remote end, and the ABR was connected to a switch. Because of that, the interface remained in "UP/UP" state and the ABR continued injecting the default route.


    If I shut down the port locally on the ABR, the default route stops being advertised.


    Thank you very much!

  • I am also curious on this one.  I've setup this basic network below.  


    Basically routers 123 have their loopbacks in Area 0.  Router 1 also has a second loopback of that is being redistributed into Area 0.  Routers 2 and 3 have links in Area 1, which is an NSSA.  Router 2 is sending default-information-originate as a Type 7, but is also sending all of the Type-3 LSAs into the area.  Router 3 is not advertising any Type-3 summary LSAs except for the default route, as seen below.  


    So essentially, to get to, Router 4 is choosing Router 3's default route (which is a Type 3) instead of Router 2's default route (Type-7).

    If you kill the connection between Router 1 and SW1, Router 3 has to wait for the OSPF dead time to expire before it realizes the neighbor is down... so you get the (.......) pings while waiting the 40 seconds.  That being said, even once the dead timer expires - since R2 still has its loopback in Area 0, it doesn't stop sending the default summary route into Area 1, and hence, Router 4 continues to attempt to send its traffic towards R2, except the pings start returning as U.U.U.U..etc.


    My question is - in real-world deployments, would this mean that best practice would be to either:

    A.) Only have 1 interface on an ABR actually in area 0, and keep the loopbacks in a different area - OR

    B.) Have some kind of connection in Area 0 between R2/R3?

  • Hello Aruncie,

    I will try to answer you but I am not the best expert in the topic :)

    OSPF Area 0 should be continuos so, when you break the link between R1 and SW1, area 0 becomes discontiguos so extrange things could happen!

    In real life: Yes, you should have a link between R2 and R3, if you don't have it, you should consider having only the link between R1 and R3 in area 0 and consider using another technique like BFD or something like that to detect the link failure quickly.

    If you have a loopback in area 0 as you said, the problem is that this interface is going to be "up" "up" forever, and because of that you are going to advertise the default to the NSSA forever too. As you said, your R4 prefer R3 route, so when the link between R1 and R3 fails, R4 continues sending traffic to R3, but R3 has no way to send the packet back to R1, so you see this U.U.U sequence in the pings...


    Hope this helps.

    Best regards!



Sign In or Register to comment.