05-14-2009 08:21 AM - edited 03-04-2019 04:46 AM
I have an unreliable WAN link that drops out for ~15 a couple times a day. All my GRE tunnels drop at once causing a hard outage, shown in the log as new adjacency due to expired hold-timer. I increased the hello-interval and hold-timers and things are very stable. Is there a disadvantage to increasing the hello-interval and hold-timer?
ip hello-interval eigrp 10 60
ip hold-timer eigrp 10 180
05-14-2009 08:46 AM
Matthew
"Is there a disadvantage to increasing the hello-interval and hold-timer?"
Well not in your case as it seems to have fixed the problem :-)
The obvious thing is that if you increase the timers and there is a genuine outage then the routing protocol will take longer to recognise the failure and longer to converge.
Jon
05-14-2009 08:46 AM
Hello Matthew,
generally speaking the EIGRP convergence is increased with these values but however you get a more stable network.
It is a question of tradeoff, now those 15 seconds out of service intervals don't cause the GRE tunnels to change state, however they are still there and user traffic is dropped for 15 seconds.
Depending on traffic patterns (use and percentage of TCP sessions that can recover) this can be acceptable.
Hope to help
Giuseppe
05-14-2009 01:35 PM
Matthew:
"Is there a disadvantage to increasing the hello-interval and hold-timer?"
Yes, there is. You will be masking an outage that will disrupt communication.
Artificially stabilizing the control plane when the data plane is down hard because of a physical link failure buys you precious little in terms of operational integrity and application survivability.
What you should be doing is working on fixing the physical link.
HTH
Victor
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