04-10-2018 11:04 AM - edited 03-05-2019 10:15 AM
Hi everyone,
I have a dual HUB PFRv3 setup with a number of spokes. The WAN transport protocol is EIGRP. The routers are ASR-1001x's with the latest code (16.6.3).
Master Hub is Hub A (H1), Transit site is Hub B (H2). Each have two BRs.
There are two spoke sites, each with two BRs: Spoke A (S1) and Spoke B (S2).
Each are connected to the dual hubs.
EIGRP metric-wise, H1 is the preferred path for all traffic from the spokes. It is only when H1 has no active connections to the spoke does a spoke communicate via H2. This is done via changing the delays of the tunnels and cross interfaces between BRs at the Hubs and Spokes.
When traffic from S1 is destined to S2 when both H1 and H2 are available, it flows like this:
S1 -> H1 -> S2
During failover testing I disconnected S1 from H1. The traffic then flows like this:
S1 -> H2 -> S2
So far so good. However, when I return the connection between S1 and H1 the traffic classes continue to use H2 as their datapath.
It is only when I clear the traffic-classes at S1 that the TC's learn to use H1 as their Hub.
I checked the routing table including the PFR overrides, H1 is recognized as the preferred route. However, the traffic-classes never preempt back from H2 to H1.
Any ideas how this can be resolved, even if using a timer?
Thanks!
04-11-2018 01:25 PM
04-26-2018 10:17 AM
Hi,
Sorry for the delayed reply. I'll examine this issue shortly, some other issues are taking up my time at the moment.
I was wondering, do you have any idea if Spoke to Spoke PFRv3 is supported? I haven't found any documentation to suggest that PFRv3 works as designed between spoke to spoke, only spoke to hub or transit.
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