jrhue
asked on
How to detect ethernet path failure and clear dynamic routes on that path on cisco router
I have routers that are interconnected via two independent fast-ethernet paths for redundancy and load sharing (Per-Destination).
I have used the ip sla and track features in the IOS to auto-fail a default route based on reachability of the far end ethernet interface IP address and it works very well.
However, I have dynamic routes advertised on both ends that need to see each other. When one path goes down, the default route fails over within 3 to 5 seconds. But the dynamic routes ( using RIPv2) don't fail -over for the full 180 seconds time-out that is the RIP route time-out timer.
Is there a way to signal / remove RIP on the failed path interface so the dynamic routes get pushed over to the redundant path quickly? The RIB should have both routes and the fail-over should happen as quickly as it takes the RIB to update to the FIB.
I know if I could shutdown the interface on the failed path, this would force a routing update on the RIB and FIB. But then I could no longer monitor the path to see when it comes back up.
Use VLANs on the path maybe? One VLAN to carry traffic and another strictly for monitoring?
I could shutdown the traffic VLAN forcing the routing re-set while leaving the monitoring VLAN up to detect when the path comes back up
Any ideas?
I have no idea on how to or if an interface can be shut down via a track object
I have used the ip sla and track features in the IOS to auto-fail a default route based on reachability of the far end ethernet interface IP address and it works very well.
However, I have dynamic routes advertised on both ends that need to see each other. When one path goes down, the default route fails over within 3 to 5 seconds. But the dynamic routes ( using RIPv2) don't fail -over for the full 180 seconds time-out that is the RIP route time-out timer.
Is there a way to signal / remove RIP on the failed path interface so the dynamic routes get pushed over to the redundant path quickly? The RIB should have both routes and the fail-over should happen as quickly as it takes the RIB to update to the FIB.
I know if I could shutdown the interface on the failed path, this would force a routing update on the RIB and FIB. But then I could no longer monitor the path to see when it comes back up.
Use VLANs on the path maybe? One VLAN to carry traffic and another strictly for monitoring?
I could shutdown the traffic VLAN forcing the routing re-set while leaving the monitoring VLAN up to detect when the path comes back up
Any ideas?
I have no idea on how to or if an interface can be shut down via a track object
ASKER
What would the converging time be with EIGRP assuming the physical interface did not go down ( due to it being ethernet and it traveleing through a few switches before reaching the other end.)?
ASKER CERTIFIED SOLUTION
membership
This solution is only available to members.
To access this solution, you must be a member of Experts Exchange.
ASKER
OK. I did not know you could configure the routing timers on a per interface basis.
So if I use default-route-originate on my router's connected to our Direct Internet Connections (statically routed -- not using BGP ) and set the hello-interval to say between 1 and 3 seconds, I should be able to see a fail-over occur in about 4 to 10 seconds.
My only other concern would be stability of using non-static default routes and the CPU surge on approx 2400 routes being re-converged to the redundant ethernet link. ( I am currently using 7500 series routers with RSP4's, VIP2, VIP4 with dCEF running.
I have used OSPF and RIP2 but never EIGRP. Are there any gotchas / caveats I should look out for with this protocol? I am going to set it up on my lab network and experiment later today.
So if I use default-route-originate on my router's connected to our Direct Internet Connections (statically routed -- not using BGP ) and set the hello-interval to say between 1 and 3 seconds, I should be able to see a fail-over occur in about 4 to 10 seconds.
My only other concern would be stability of using non-static default routes and the CPU surge on approx 2400 routes being re-converged to the redundant ethernet link. ( I am currently using 7500 series routers with RSP4's, VIP2, VIP4 with dCEF running.
I have used OSPF and RIP2 but never EIGRP. Are there any gotchas / caveats I should look out for with this protocol? I am going to set it up on my lab network and experiment later today.
with these two, you will not only have the inherent fast convergence, but you can further fine-tune their behavior to suit your environment. If you need further help wth these, yu know where to post.....