redistribution multiple OSPF external 1 routes into MBGP+multipath

Unanswered Question
Mar 12th, 2008

I'am unable to configure multiple OSPF routes into MBGP+multipath redistribution.

I have two routes in routing table learned via OSPF:

O E1 x.y.80.0/20 [110/11] via 192.168.64.5, 02:53:33, GigabitEthernet6/2

[110/11] via 192.168.64.1, 02:53:33, GigabitEthernet5/2

but BGP is importing only one:

*> x.y.80.0/20 192.168.64.1 11 32768 ?

Is possible to configure this?

thx

ivast

I have this problem too.
0 votes
  • 1
  • 2
  • 3
  • 4
  • 5
Overall Rating: 0 (0 ratings)
Loading.
ebreniz Tue, 03/18/2008 - 12:06

If you decide to use one OSPF process, it would also eliminate the need for mutual redistribution between two routing protocol processes (BGP and OSPF).Hypothetically if the router did prefer a redistributed route over the source, you would wind up with a black hole since the redistributed route would point to a next-hop of itself...

http://www.cisco.com/en/US/docs/ios/12_0st/12_0st14/feature/guide/confmbgp.html

Ivast Wed, 03/19/2008 - 07:46

Thanks for hint. I really have needs for this "special" configuration. Because it is about redistribution ospf route into MBGP for inter vrf route leaking on one box, there is no possibility to black hole ...

marikakis Tue, 03/18/2008 - 12:50

Hello,

BGP by default will install only one best path in the routing table. Have you tried to issue the command "maximum-paths 2" under the BGP routing process?

Kind Regards,

M.

Ivast Wed, 03/19/2008 - 07:41

Thank you for the hint, but I definitively tried this. Problem is really with redistribution process, because there are multiple routes in ospf database and routing table, but only one of them is redistributed into BGP ( that mean only one is picked and inserted into bgp network table).

Giuseppe Larosa Wed, 03/19/2008 - 11:05

Hello,

the redistribution process works in this way: you have a prefix with two OSPF equal cost paths that is imported in the VRF BGP table.

However, what is important to notice is that until one of these two OSPF routes is alive in the local VRF the MBGP will send out its own advertisement so there is no need to have two MBGP advertisements: just one is enough.

From the point of view of a remote PE importing this prefix there is no need to know if there is one or two OSPF routes in the local VRF site.

MBGP multipath that can be enabled by using eibgp multipath command is useful when you have two local PE routers that advertise the prefix (using a different RD but same RT): if you enable this command on the remote PE you will be able to see two BGP routes installed in the remote PE VRF routing table and save convergence time in comparison to have to install a second BGP advertisement in case of a failure of the best BGP path.

Hope to help

Best regards

Giuseppe

amenichetti Thu, 07/18/2013 - 04:44

Hello,

I've the same problem: route leaking between two vrf with mp-bgp.

It's not a mpls implementation and bgp is used only for the leaking process.

DC-CORE1-1.2.1#sh ip route vrf SERVICES 172.20.140.0

Routing entry for 172.20.140.0/24

  Known via "ospf 2", distance 110, metric 11, type intra area

  Redistributing via bgp 65001

  Advertised by bgp 65001 route-map SKIP_PEERING_NET

  Last update from 100.100.15.2 on FastEthernet1/11, 00:49:14 ago

  Routing Descriptor Blocks:

  * 100.100.18.2, from 1.3.2.2, 00:49:14 ago, via FastEthernet1/14

      Route metric is 11, traffic share count is 1

    100.100.15.2, from 1.3.1.2, 00:49:14 ago, via FastEthernet1/11

      Route metric is 11, traffic share count is 1

DC-CORE1-1.2.1#sh ip route vrf EXTRANET 172.20.140.0

Routing entry for 172.20.140.0/24

  Known via "bgp 65001", distance 20, metric 11, type external

  Last update from 100.100.15.2 on FastEthernet1/11, 00:23:04 ago

  Routing Descriptor Blocks:

  * 100.100.15.2 (SERVICES), from 0.0.0.0, 00:23:04 ago, via FastEthernet1/11

      Route metric is 11, traffic share count is 1

      AS Hops 0

I need also the second route via 100.100.18.2 to be added in the vrf EXTRANET routing table.

Is it possibile to do this?

Thanks.

Regards,

AM

Actions

This Discussion