ATC: Back-to-Back VRF: OSPF Capability VRF


When setting up the Inter-AS VPN, we had to enable the OSPF capability VRF feature between the ASBRs but why do we have to do that of OSPF when each ASBR is considered a CE router by the neighboring ASBR... Let me put it in a diagram:



Is it in relationship to redistributing OSPF into BGP is PE1 and it being redistributed back into OSPF on ASBR1?



  • hudsonmbg2

    this means that you mus be using OSPF as the CE1-PE1 and/or PE2-CE2 routing protocols with the same process-id (domain id ) as the process you are using between the ASBRs.

    what happens is for O or O IA routes, OSPF uses the down bit as a method of preventing loops. So  if CE1 advertises OSPF routes to PE1, then PE1 redistributes them into MP-BGP, these routes will then be sent to ASBR1, which will in turn redistribute them back into OSPF. if the vrf-aware OSPF process on ASBR1 uses the same process-id as the vrf-aware OSPF process on PE1, then the routes will show as O IA  in the ASBR1 vrf routing table and the down bit will be set. When these OSPF routes are sent to the vrf-aware OSPF process of ASBR2, they will be dropped because this is the normal behavior of OSPF when using VRF.

    the only way to prevent this from happening is to either enable the OSPF capability vrf feature,  or  to make sure that the process ids are different (or manually set differnet domain ids) on PE1  for example, that way the routes will show as  External in ASBR1 vrf-aware OSPF process and the down bit will not be set when sent to ASBR2



Sign In or Register to comment.