01-27-2012 10:23 AM - edited 03-07-2019 04:35 AM
Hi -
I have to following topology...
A-InternetEdge-A-Core-A-WANEdge-MPLS-B-WANEdge-B-Core-B-InternetEdge
I've got mutual redistribution between BGP and OSPF working but when I down a WANEdge inteface and bring it back up the routing table doesn't restore the routes as I expect.
When all is working correctly Site A routes are learned via BGP and traffic flows from Site A to Site B over the MPLS network. When I fail site A WAN link OSPF routes point towards the internet correctly. When the WAN interface is restored the BGP routes are not restored and traffic continues flow towards the internet.
Hoping someone can see something that I'm not. I've posted one A-WANEdge config, routing table and BGP table.
The route(s) in question are 10.24.101.0/24.
Thanks
Solved! Go to Solution.
01-29-2012 11:56 AM
Hi,
I think this might be a "race" problem discussed here several times.
Imagine following scenario:
The 10.24.101.0/24 had been received by OSPF.
Redistributed to BGP on your router.
Then the same prefix is received via BGP.
You would expect the BGP received prefix to override the OSPF one.
But as the OSPF prefix had been redistributed to BGP on your router, it's present in th BGP table already with the weight attribute equal to 32768.
So it beats the same prefix received via BGP.
The solution might be using a route-map while redistributing OSPF to BGP which would decrease the weight of the redistributed prefixes to 0 (and possibly also decrease the local preference, if necessary).
HTH,
Milan
01-29-2012 11:56 AM
Hi,
I think this might be a "race" problem discussed here several times.
Imagine following scenario:
The 10.24.101.0/24 had been received by OSPF.
Redistributed to BGP on your router.
Then the same prefix is received via BGP.
You would expect the BGP received prefix to override the OSPF one.
But as the OSPF prefix had been redistributed to BGP on your router, it's present in th BGP table already with the weight attribute equal to 32768.
So it beats the same prefix received via BGP.
The solution might be using a route-map while redistributing OSPF to BGP which would decrease the weight of the redistributed prefixes to 0 (and possibly also decrease the local preference, if necessary).
HTH,
Milan
01-30-2012 08:16 AM
milan
Thanks so much. That's exactly what the problem was. I found this link shortly after my post and created a route map to adjust the weight for the different networks.
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