EIGRP and Static Routes backup

[nikalleyne]

Hi All,

I have a situation where I have a functioning EIGRP topology across DMVPN. However, the way the traffic is flowing is not the way we prefer it to flow. Our infrastructure uses MPLS and DMVPN. All traffic is currently going over the DMVPN. We wish for it to go across the MPLS. I tried adding a static route with an administrative distance of 50 for my internal routes with the next hop as the MPLS PE. While the traffic is now flowing though the MPLS, if I unplug the cable from the PE device, the EIGRP route does not go back into the routing table (and cross the DMVPN) as a result no traffic is reaching their required destination.

Can someone tell me what may be wrong or what am I missing?

Thanks

[dgroscost]

You can try to use a floating static route but that may not be best design. What routing protocol are you using for the MPLS routes? Is it EIGRP as well?

[Vivek Ganapathi · Accepted solution]

Hello,

As per my understanding, you have multiple sites connected via a DMVPN with EIGRP running over it. You even have an MPLS cloud connecting multiple sites now & wish to pass the traffic via MPLS & use DMVPN as the backup solution.

I understand that you have a static route pointing to your PE. if yes, then it won't help. Any failure in the MPLS cloud will have no effect on your static route (it would never get away from the routing table & continue to stay there). That's the reason you are not having your EIGRP route getting into the routing table. Two ways to solve.

1) Run a dynamic routing protocol between your MPLS PE & yourself. Probably BGP.

or

2) Use IP SLA tracking. IP SLA will track a particular object & all the routes pertaining to the track will be withdrawn from the routing table if the object is not responding (ICMP Probe failure etc).

Hope it helps. Do let me know if you need further clarity.

[nikalleyne]

Hi Vivek let me create a lab environment based on your suggestion and will update you shortly.

Thanks

[nikalleyne]

Hey Vivek,

Thanks again, I sat down and thought about what you said in addition, to testing it in a lab and your suggestion is absolutely correct. I would probably go the sla route.

Thanks again

[Vivek Ganapathi]

Good to hear. IP SLA is a great feature indeed ! One more suggestion, create a loopback IP on a router to perform an ICMP probe (IP SLA). You must always have a static route to this loopback IP via your primary path because the failback must happen as soon as the primary link is back online. The reason why i emphasize on a loopback IP is you don't require this to be reachable when the primary path is down. If you are using loopback IP's as your management IP's on your routers / switches in your organization, then simply create one more loopback only for IP SLA.

Thanks to let me know you were successful !

Regards

Vivek

[bava_ccna]

Hi Vivek,

I am little bit confused with your Loopback concept for IP SLA, can you elaborate a little bit with config example..

thanks in adv

-bava

[Vivek Ganapathi]

First thing, the IP which you poll must have a static route via the primary path always. Without this, you would never achieve a failback (because IP SLA wouldn't know if the primary path has come back online). As said this, now if you were to poll a management IP of a remote end device & your primary path fails, the device will not be reachable as the IP has a static route via primary path. So, what happened?? You lost your management connectivity to the box.

Now, assume if you had a loopback IP configured (which is not a management IP of the remote end device) & your primary path fails, you would still retain your access to the device via it's management IP on the backup path.

Hope it's clear. I don't have a config handy as i m out on a vacation. Once i'm back i will put it up.