BGP, OSPF and Routing
Hi, I have two BIRD routers in two locations connected to the same CIX-LAN. Both establish BGP sessions to the route server of the CIX. Both routers do OSPF on their internal interfaces and see each other there (in area 0). How do I prevent the following situation: When I disable one of the BGP sessions to the route server the router still has direct routes into the CIX LAN because it learned them via OSPF from its partner. I would like to see the routes pointing to the router that still has the BGP sessions established (even if that means a hop more). Is that even possible? Do I really have to take down the interface to the CIX LAN? Regards -- Robert Sander Heinlein Support GmbH Schwedter Str. 8/9b, 10119 Berlin http://www.heinlein-support.de Tel: 030 / 405051-43 Fax: 030 / 405051-19 Zwangsangaben lt. §35a GmbHG: HRB 93818 B / Amtsgericht Berlin-Charlottenburg, Geschäftsführer: Peer Heinlein -- Sitz: Berlin
On Thu, Sep 15, 2016 at 08:31:36PM +0200, Robert Sander wrote:
Hi,
I have two BIRD routers in two locations connected to the same CIX-LAN.
Both establish BGP sessions to the route server of the CIX.
Both routers do OSPF on their internal interfaces and see each other there (in area 0).
How do I prevent the following situation:
When I disable one of the BGP sessions to the route server the router still has direct routes into the CIX LAN because it learned them via OSPF from its partner.
I would like to see the routes pointing to the router that still has the BGP sessions established (even if that means a hop more).
Hi First, i would suggest to not redistribute BGP routes through OSPF, but use OSPF only for internal topology and use IBGP to redistribute BGP routes. But the issue is similar in both cases - the route could be redistributed (by OSPF or IBGP) either with or without its next hop. When redistributed without the next hop, the intermediate router is used as a gateway (as you want). For IBGP, that could be enforced by option 'next hop self'. For OSPF, it is based on whether next hop is part of OSPF domain - so if OSPF is not enabled on CIX LAN iface, it should not be a problem. In both cases it can be enforced by export filter to internal protocol (OSPF ro IBGP) that removes regular next hop by setting 'dest' attribute to some special type, e.g.: dest = RTD_UNREACHABLE; -- Elen sila lumenn' omentielvo Ondrej 'Santiago' Zajicek (email: santiago@crfreenet.org) OpenPGP encrypted e-mails preferred (KeyID 0x11DEADC3, wwwkeys.pgp.net) "To err is human -- to blame it on a computer is even more so."
Am 15.09.2016 um 20:57 schrieb Ondrej Zajicek:
For OSPF, it is based on whether next hop is part of OSPF domain - so if OSPF is not enabled on CIX LAN iface, it should not be a problem.
You are right, there was a definition for interface "*" { stub; }; in area 0. Removing that has the expected effect. Regards -- Robert Sander Heinlein Support GmbH Schwedter Str. 8/9b, 10119 Berlin http://www.heinlein-support.de Tel: 030 / 405051-43 Fax: 030 / 405051-19 Zwangsangaben lt. §35a GmbHG: HRB 93818 B / Amtsgericht Berlin-Charlottenburg, Geschäftsführer: Peer Heinlein -- Sitz: Berlin
participants (2)
-
Ondrej Zajicek -
Robert Sander