Hello Ondrej, bird-users. Thanks for your explanation. hereby the output of the "show route 40.0.40.0/21 all" on both routers. (ip's scrambled) The second transit has one AS hop extra (5) where the first transit has only 4 AS hops. which makes is valid to me, to let router2 choose the patch via router1. router1# BIRD 1.2.3 ready. bird> show route 40.0.40.0/21 all 40.0.40.0/21 via <Transit1_BGPRTR1> on eth5 [Transit1_BGPRTR1_IPv4 2010-06-04 11:26:29] * (100) [AS4249i] Type: BGP unicast univ BGP.origin: IGP BGP.as_path: 24785 3356 7332 4249 BGP.next_hop: <Transit1_BGPRTR1> BGP.med: 40231 BGP.local_pref: 100 via <Transit1_BGPRTR2> on eth5 [Transit1_BGPRTR2_IPv4 2010-06-04 11:26:39] (100) [AS4249i] Type: BGP unicast univ BGP.origin: IGP BGP.as_path: 24785 3356 7332 4249 BGP.next_hop: <Transit1_BGPRTR2> BGP.med: 40231 BGP.local_pref: 100 bird> router2# BIRD 1.2.3 ready. bird> show route 40.0.40.0/21 all 40.0.40.0/21 via <gw1_router1> on eth1 [iBGP_IPv4_1 2010-06-08 13:39:27] * (100) [AS4249i] Type: BGP unicast univ BGP.origin: IGP BGP.as_path: 24785 3356 7332 4249 BGP.next_hop: <gw1_router1> BGP.med: 40231 BGP.local_pref: 100 via <gw1_router2> on eth0 [iBGP_IPv4_2 2010-06-08 13:28:18] (100) [AS4249i] Type: BGP unicast univ BGP.origin: IGP BGP.as_path: 24785 3356 7332 4249 BGP.next_hop: <gw1_router2> BGP.med: 40231 BGP.local_pref: 100 via <Transit2_BGPRTR1> on eth5 [Transit2_BGPRTR1_IPv4 2010-06-04 11:26:30] (100) [AS4249i] Type: BGP unicast univ BGP.origin: IGP BGP.as_path: 38930 1299 3356 7332 4249 BGP.next_hop: <Transit2_BGPRTR1> BGP.local_pref: 100 bird> Regards, On Tue, 8 Jun 2010, Ondrej Zajicek wrote:
On Tue, Jun 08, 2010 at 02:15:07PM +0200, Arjan Filius wrote:
Hello bird-users,
Running bird 1.2.3-ppa ubuntu 10.04,x64, on 2 hosts, internally connected with 2 lines, and running 2 iBGP sessions, both routers have their own GlobalTransit provider
the iBGP protocols have no filters on both routers
What is noticed (for example with network 40.0.40.0/21): router1: bird> show route 40.0.40.0/21 40.0.40.0/21 via <Transit1_BGPRTR1> on eth5 [Transit1_BGPRTR1_IPv4 2010-06-04 11:26:29] * (100) [AS4249i] via <Transit1_BGPRTR2> on eth5 [Transit1_BGPRTR2_IPv4 2010-06-04 11:26:39] (100) [AS4249i] bird>
router2: bird> show route 40.0.40.0/21 40.0.40.0/21 via <gw1_router1> on eth1 [iBGP_IPv4_1 2010-06-08 13:39:27] * (100) [AS4249i] via <gw2_router1> on eth0 [iBGP_IPv4_2 2010-06-08 13:28:18] (100) [AS4249i] via <Transit2_BGPRTR1> on eth5 [Transit1_BGPRTR1_IPv4 2010-06-04 11:26:30] (100) [AS4249i] bird>
so router1 isn't showing it's route option to a transit on router2 (via 2 iBGP sessions)
Is this correct, end if yes why?
It is correct because in router2, iBGP route received from router1 is primary and therefore it is not exported back back to router1 (and only the primary route is exported).
But the question is why internal route is primary (usually, external routes are primary). Could you send the output of 'show route 40.0.40.0/21 all' on router2?
-- Arjan Filius mailto:iafilius@xs4all.nl