05-23-2011 10:08 AM - edited 03-07-2019 12:38 AM
Dear All
i am trying to do a basic mpls vpn
all routers are running ospf 1 in the backbone
and ospf 1 And 2 on the CE routers
everything is fine till i try to put redistribution on PE routers has below and i get the messages below any help i would by gratful for yours Jimmy
Solved! Go to Solution.
05-25-2011 12:23 AM
Hi,
Let me quoute the text from "MPLS and VPN Architectures by Guichard and Pepelnjak" page 178.
... a mechanism is needed so that the PE-router can distinguish which routes belong to which VRF. This is accomplished through association of a particular customer interface to a specific VRF (per the standard MPLS/VPN model that we have already described) and to a particular OSPF process.
A separate OSPF process is necessary for each VRF that will receive VPN routes via OSPF. This is different than the procedures used when either RIP Version 2 or BGP are deployed across the PE-to-CE link because these protocols can run different routing contexts within the same process. Due to the complexity of OSPF and the associated topology database, this option is not currently available, so a different OSPF process (with a different process-ID) is required per VRF.
HTH.
PS: Please don't forget to mark your question answered
BR,
Aliou
05-23-2011 01:00 PM
Hi,
Can you post sh ip vrf and sh ip vrf int commands output.
Regards.
Alain.
05-23-2011 04:18 PM
05-24-2011 02:08 AM
Hi Blusox123,
Ensure to use a separate OSPF process for each VRF and for the global routing table.
BR,
Aliou
05-24-2011 04:45 AM
like tamassini suggested you should have a different pid for the global routing table and each vrf.
Regards.
Alain.
05-24-2011 05:46 PM
Dear Cadet and Tamassina
Thank you for your help i changed the vrf process numbers to ospf 3 and 4 and it worked i bought
and it was fine for doing rip and eigrp but i could not work out what was wrong in OSPF
i just read the instructions and asked why the lab would not work in GNS3
I Am still waiting for a reply from configure terminal
if you had not spoted it i would be still on the lab
probably a silly question why do you need different ospf processes for the vrf and routeing table
yours
Jimmy
05-25-2011 12:23 AM
Hi,
Let me quoute the text from "MPLS and VPN Architectures by Guichard and Pepelnjak" page 178.
... a mechanism is needed so that the PE-router can distinguish which routes belong to which VRF. This is accomplished through association of a particular customer interface to a specific VRF (per the standard MPLS/VPN model that we have already described) and to a particular OSPF process.
A separate OSPF process is necessary for each VRF that will receive VPN routes via OSPF. This is different than the procedures used when either RIP Version 2 or BGP are deployed across the PE-to-CE link because these protocols can run different routing contexts within the same process. Due to the complexity of OSPF and the associated topology database, this option is not currently available, so a different OSPF process (with a different process-ID) is required per VRF.
HTH.
PS: Please don't forget to mark your question answered
BR,
Aliou
05-25-2011 03:11 AM
Dear Tamassina
again thank you for clearing things up for me
yours
Jimmy
Discover and save your favorite ideas. Come back to expert answers, step-by-step guides, recent topics, and more.
New here? Get started with these tips. How to use Community New member guide