Anyone using arp-on-stp for particular use cases in production?
We have a number of RVIs that are OSPF enabled, which support next-hop reachability for iBGP sessions running on lo0. We are seeing some instances of iBGP flapping if we have a rstp TCN. I believe it is due to all the mac learning going on, the routing-engines on the ToR switches end up losing the BGP keepalive packets from the upstream peer.
I'm thinking I can use arp-on-stp to help minimize the impact of this. Although, the documentation is pretty spare - it says you must have RVIs to take advantage of this (which we do), but docs are totally silent on if we have to enable this feature on all physical interfaces that carry the vlans where the RVIs live?
I'd love to get rid of the RVIs and just have fixed layer 3 boundaries between ToR/Core, but we have a number of folks that need to span VLANs between switches, in a very heterogeneous environment (multi-tenant, some physical, variety of virtual). But alas, I can't figure out the config to make it work on EX series switches we are running as ToR.
No comments:
Post a Comment