0

Let's say, that I receive the prefix 10.10.10.0/28 from two different iBGP peers:

root@r1> show route 10.10.10.3

inet.0: 939819 destinations, 2781064 routes (939795 active, 0 holddown, 44 hidden)
+ = Active Route, - = Last Active, * = Both

10.10.10.0/28      *[BGP/170] 00:00:52, localpref 210, from 192.168.0.6
                      AS path: I, validation-state: unverified
                    >  to 192.168.77.54 via ge-0/0/0.2
                    [BGP/170] 00:00:47, localpref 200, from 192.168.0.4
                      AS path: I, validation-state: unverified
                    >  to 192.168.77.52 via ge-0/0/0.2

root@r1>

The iBGP sessions are established between the loopback addresses and thus the addresses 192.168.0.6 and 192.168.0.4 seen above are the loopback addresses of iBGP peer routers. OSPF is used to resolve the protocol next hop of 192.168.0.6 to 192.168.77.54 and 192.168.0.4 to 192.168.77.52.

Now let's say that the 192.168.0.6 is no longer resolvable using the OSPF, but the iBGP session has not yet timed out:

root@r1> show route 10.10.10.3

inet.0: 939819 destinations, 2781063 routes (939794 active, 0 holddown, 45 hidden)
+ = Active Route, - = Last Active, * = Both

10.10.10.0/28      *[BGP/170] 00:00:06, localpref 210, from 192.168.0.6
                      AS path: I, validation-state: unverified
                      to Discard
                    [BGP/170] 00:17:04, localpref 200, from 192.168.0.4
                      AS path: I, validation-state: unverified
                    >  to 192.168.77.52 via ge-0/0/0.2

root@r1>

As seen above, the 192.168.0.6 is resolved to Discard. This happened because the r1 has an aggregated route of 192.168.0.0/16 with a discard as a next hop.

Using the example above, then I would prefer that the next hop of 10.10.10.0/28 learned from the 192.168.0.6 becomes unresolvable and thus the 10.10.10.0/28 prefix from 192.168.0.4 would be used.

I know that I could use BFD to drop the iBGP session (much) faster, but perhaps there is a way to avoid using aggregate route as a viable route for resolving BGP next-hops?

2
  • 1
    You can't have it both ways. You've told BGP to prefer the first route with the higher LP. As long as it's valid (discard is valid), that's the route that will be chosen. You could possibly play with timers to match OSPF, but that has its own problems.
    – Ron Trunk
    Commented Apr 4 at 18:10
  • @RonTrunk Thanks. In that case I'll use BFD for iBGP sessions as I need certain prefixes with higher LP.
    – Martin
    Commented Apr 5 at 11:14

0

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Browse other questions tagged or ask your own question.