This is the network architecture I have, in the attachment. There are clients at CE1 which need to speak to the clients int CE2. For this communication, they need to take the MPLS path but in case the MPLS path fails then it should choose the Metro Ethernet path. I have mentioned the routing protocols running on each link in the diagram.
Can someone tell me what will be the most efficient way to redistribute routes avoiding loops and still maintain redundancy?
Are you the Customer and Own the CE1 ? or you have control on PE Router.
If you are customer, since you have only 1 Link, so you need to depend on Provider to support for your requiremet ?
If this is LAB then there are number of ways to do ? but if Live network, i do not belive any SP still running RIP ?
So please clarify so we can suggest better.
Use the OSPF Down Bit and Route tagging to avoid loops when redistributing routes.
in your network diagram you mention several protocols OSPF 11 on CE1 , OSPF 33 on CE1 to PE1 link. OSPF 99 as IGP in the MPLS upper part of the network. You mention MPLS (here we can suppose to use LDP for label distribution)
Then you mention usage of RIPv2 over the metro ethernet link WITHOUT using MPLS.
So a basic question arires here: what type of service are you using to provide connectivity on the MPLS part of the network?
If you are using an MPLS L3 VPN you need also MP BGP to be run on PE1 and PE2 in addition to LDP on PE1, P1,P2, PE2 routers.
However, if this is the case the metro ethernet path does not qualify as an acceptable backup path because MPLS is not enabled over it and MPLS L3 VPN requires a working MPLS forwarding plane on all paths to make MPLS L3 VPN connectivity to work.
So if your service is a Layer 3 VPN as the used names CE1, PE1 and CE2, PE2 would imply you should enable MPLS and LDP also on the bottom part of the network the metro ethernet portion.
In this case the primary path would be used because OSPF AD 110 is better then RIPv2 AD 120.
You need to redstribute OSPF 33 into MP BGP AF vrf VPNA and you need to enable AF VPNv4 betweeen PE1 and PE2 loopbacks.
MP BGP should be redistributed into OSPF 33 on PE1 and PE2 the DOWN bit option allow to avoid any routing loop issue.
If you are not using an MPLS L3 VPN service You need to perform redistribution of:
OSPF 33 into OSPF 99
OSPF 33 into RIPv2
on PE1, PE2 and you should use route-tags .
Please note that RIPv2 tags are 16 bits integer and OSPF route tags are 32 bit integers this means you cannot pass an OSPF route tag to RIPv2.
Mutual redistribution may be avoided if PE1, PE2 can send a default route to CE1, CE2 respectively in OSPF 33.
This approach may be feasible or not.
Clarify if you are using an MPLS L3 VPN service or not. As I have explained above your network diagram is not ready for this service.
On the other end the MPLS L3 VPN service would simplify the routing redistribution issues by just adding MP BGP and taking advantage of OSPF Down bit feature.
Hope to help