06-12-2017 11:02 AM - edited 03-05-2019 08:41 AM
Hi
I have a dual 4500X VSS setup. I have one WAN cct on each chassis running BGP. I have found if I fail the active VSS chassis it takes a significant time for BGP to converge on the standby (now active) chassis. If I simply fail the each circuit but not the complete chassis the reconvergence via the surviving circuit it acceptable.
Will NSF resolve this problem? If so do the WAN BGP peers need to configure anything and what failover times should I expect?
As always, thank you for any input.
Stuart.
06-12-2017 06:14 PM
I'm betting if you lower the BGP timers you'll get a better result.
neighbor a.b.c.d timers 8 24
06-13-2017 01:10 PM
Thanks for your input Philip.
I already have the timers reduced with timers bgp 1 3 under the bgp process. I believe this applies to all neighbors. However, having read into the VSS active chassis failover process I believe the routing neighbors drop as the previous standby chassis takes over control. I think this is the issue I am seeing. I think the timers config assists when I lose a neighbor but not when I lose the active chassis.
Although I don't see a specific description of my setup in NSF documentation from what I can gather it may be solution.
Thanks, Stuart.
06-13-2017 03:45 PM
I've moved this discussion to the WAN/Routing Forum for greater visibility..
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