cancel
Showing results for 
Search instead for 
Did you mean: 
cancel
242
Views
0
Helpful
1
Replies
Highlighted
Beginner

delays in route updates when redistributing bgp into ospf

I have a question about redistribution between routing protocols.  We run OSPF internally at our datacenters but our branches connect via an MPLS WAN that requires BGP routing.  For that reason we are redistributing between BGP and OSPF at our datacenter edge router.  We have a second datacenter edge router with a DMVPN backup connection to that same branch.  Since we didn't want to run multiple routing protocols at the branches, the DMVPN also uses BGP, and that edge router also redistributes the routes into OSPF at our datacenter. 

If a branch office's MPLS circuit fails, my understanding is that it will take the value of the bgp hold timer in seconds for our MPLS provider to remove the route from their edge router.  Then there's the advertisement timer between our MPLS provider and our datacenter edge router.  The branch hold timer is currently 90 seconds and the advertisement timer is 30 seconds, so that means it should be 2 minutes after a branch circuit failure that our MPLS edge router at the datacenter learns that it no longer has a BGP route to that branch office, allowing the DMVPN to take over.

My problem is that it's taking significantly longer - something like 7 minutes - for the OSPF route pointing to the DMVPN backup connection to get installed at the datacenter.  So my question is: is there some kind of timer or polling interval that determines how often routes being redistributed into OSPF are being checked for updates?  Does OSPF learn about the BGP route change immediately or does it wait for some interval of time?  Anything else that might introduce additional delay?

 

Thanks

-Mathew Rouch

 

1 REPLY 1
Highlighted
Cisco Employee

Hi Matthew,

 

Have you verified that BGP is in fact losing the route when you would expect it to and it is redistribution that is failing?

 

I would monitor the BGP table and OSPF DB on the DMVPN hub when a spoke route is lost to see what is happening and to check the timings.

 

As soon as BGP loses the route it should be "un-redistributed" from OSPF and OSPF should re-run SPF to find the new best path.