OSPF over L2 MPLS Routing Loop

[NETAD]

Hello, one of our clients have RIP running between their core switch and their WAN router and OSPF over L2 MPLS to interconnect the other offices. They're also redistributing rip into OSPF and OSPF into RIP with no route filtering. The issue they're facing is a routing loop when any of the WAN routers reboots for any reason and the loop won't go away until they issue clear ip route. They've set the priority to 1 on all remote offices and to 128 on the DC router but the DC isn't the DR. It' one of the remote offices. 

Does this sound like a routing loop caused by re-election process of the DR or more of the fact that they're doing the redistribution between OSPF and RIP? 

what do you recommend to fix this design?

I'm thinking to get rid of rip and just run one instance of OSPF between all the locations(P2P between the core switch and the wan router), or just run EIGRP. 

[Julio E. Moisa]

Hi

I have not faced this situation before, but have you tried to configure the OSPF interfaces involved into the L2MPLS as network point-to-point? The prority is used to indicate who will be the DR and BDR but if you are using OSPF between routers over L2MPLS, I don't see the reason to use it because it is basically a point to point. 

Hope it is useful

:-)

[NETAD]

I don’t think p2p is doable in this senario between the locations. If I was to change the network type it would be probably P2M but that comes with everything hving to route through the hub.

[Julio E. Moisa]

Hi

Are you using VPLS?

[NETAD]

I haven’t asked the client yet but I did read somewhere about vpls and ospf in broadcast mode since vpls doesn’t propogate broadcast like a real layer 2 network and might cause issues with the election prosess.

Ref https://ieoc.com/discussion/16073/