HSRP and EIGRP

[aasson]

We have two routers (A & B) configured in HSRP on a VLAN and speaking EIGRP on that VLAN. The EIGRP routes are announced by both routers via their own IP address on the LAN and not via the "stanby IP address".

Is there any way to force the announces to have the

stanby IP address as source address ?

In other words, is it possible to see (in the routing table of the neighbor routers) the routes announced by A end B via the standy IP address instead of the IP address of the single A or B router ?

Thank You,

Alessandro Asson

[Richard Burts]

I am not sure why you would want the advertisement to have a source address of the standby address - what problem would it solve ? And I can think of several reasons why it is better to have it sourced from the local address: it is likely that some of the external destinations have a better metric through one router than the other. As long as the advertisement is sourced from the local address we know which is the best router to forward the data. If the advertisement were sourced from the standby it would become ambiguous which router was best for a particular destination.

EIGRP associates routes in its table with the particular neighbor from whom they were learned. If that neighbor fails, EIGRP needs to purge the routes learned from that neighbor. If the advertisements were sourced from the standby address it would become ambiguous which router had generated the advertisement and if router A fails and router B takes over the standby address, which routes should be adjusted in the topology table?

[aasson]

Yes, you are right. So perhaps it is no more useful the HSRP configuration: both routers advertise their networks (including the default net) and moreover the traffic is load-balanced between the two routers!

Correct ??

Thank You for your answer, best regards.

[tom.neteng]

Richard,

I have recently run into an issue with EIGRP is distributing interface addresses as opposed to the VIP on an SVI setup with HSRP.  The HSRP SVI is trunked, along with many other VLANs on a VPC to a distribution switch.  From there, we have routers plugged into that switch, along with many hosts.  While this setup does not affect any hosts, the resulting asymmetric traffic flows to the routers (and the hosts behind them) running CBAC IP INSPECT cause connections to fail when return traffic comes from the core that the router did not initially send the traffic through.  Intermittent network failures began to occur between 25% and 75% of the time, depending on which core received the traffic and which core returned the traffic.

I am temporarily using a work around, but ultimately, the fix for us is going to be to statically set the VIP to be the default gateway or adjust the metric on the EIGRP route to core 2 to be lower unless I can figure out how to get EIGRP to distribute only the VIP.

Is it possible advertise only the VIP from both cores and not the interface IPs, and how does one go about doing that, @Richard Burts?

Tom

[Richard Burts]

Tom

I do not have enough information to fully understand your issue. But I believe that I do understand the immediate question well enough to provide this in response: you ask "Is it possible advertise only the VIP from both cores and not the interface IPs".

- For one thing in EIGRP the advertisement is for routes (or prefixes) and the source address is part of the data in the advertisement but not the subject of the advertisement. (admittedly a picky point but one I hope is worth making)

- I do not know of any way to get EIGRP to use the VIP as the source address instead of the interface address. And if you did get EIGRP to specify the VIP as the source address I believe that several problems would be created as explained in my previous response in the thread.

- I do not understand your topology well enough to give good advice, but in respect to this statement ("the fix for us is going to be to statically set the VIP to be the default gateway ") I would think that having the VIP as the default gateway was generally viewed as a Best Practice. And in respect of this statement (" adjust the metric on the EIGRP route to core 2 to be lower") I would suggest that using an offset list (perhaps applied outbound on the originating router or perhaps more likely applied inbound on the receiving router) would allow you to make advertisements from one core more preferable than advertisements from the other core. 

HTH

Rick