05-09-2012 06:50 AM - edited 03-04-2019 04:18 PM
Running into an issue with new routing setup. What I am trying to accomplish is to have the routes learned via eBGP and redistributed into EIGRP to be preferred.
Setup is BGP over MPLS , neighbors are the carriers routers; EIGRP running internally.
Here is the redistribution point.
router eigrp 1
network 10.255.254.0 0.0.0.255
redistribute connected
redistribute bgp 65011 metric 1000000 10 255 1 1500
!
router bgp 65011
bgp log-neighbor-changes
network 172.16.255.1 mask 255.255.255.255
redistribute static
neighbor x.x.x.x remote-as 65000
default-information originate
Alot of info out there , not sure what applies and is the best way
Thanks in advance
05-09-2012 07:18 AM
Do you only have 1 exit point? If so, I can't see a reason why you'd want to prefer the redistributed BGP route in EIGRP if it's the only route out of the network. You should just be able to leave the defaults unless you're learning from somewhere else. Can you post a topology?
05-09-2012 07:27 AM
Yes it is being learned from EIGRP. Let me expand - currently we point to point links with VPN backup all running eigrp 1. The point to points are being replaced with MPLS running BGP. What I need to accomplish is to have the BGP links be preferred. When I cut over a site to BGP the traffic was looped ; remote site sending via BGP but the return traffic was via eigrp
05-09-2012 07:20 AM
Hello Gizri,
if I understand you would like to have EIGRP routes coming from BGP redistribution into EIGRP preferred over EIGRP internal routes.
I guess you have a L3 VPN service but you still have a direct link between different sites
By default EIGRP internal routes have admin distance 90 and external routes have admin distance 70.
You can alter this behaviour with distance eigrp command under router eigrp configuration.
see
router eigrp 1
distance eigrp 90 80
!
this could do the job but you would need to do this on all routers in the site.
Also you could make less attractive routes coming on the direct link between sites using offset list
see
you can increase with offset list the delay component on received routes on the link between sites increasing in this way the EIGRP metric ( delay is cumulative in EIGRP metric calculation)
Hope to help
Giuseppe
05-10-2012 01:31 AM
Hi Giuseppe,
Could the backdoor feature of BGP be another alternative to solve the problem ?
Regards.
Alain
05-10-2012 03:05 AM
Hello Alain,
at first glance I would say no, as the problem here is that the OP would like to use L3 MPLS VPN PE-CE links as primary links and not as secondary. He had p2p direct links so before he/she was able to prefer primary paths over VPN backups because routes were all of the same type EIGRP internal routes compared to EIGRP internal routes.
On the CE nodes eBGP routes are preferred over EIGRP internal routes by default for their lower AD (20 < 90), the problem arises if the VPN backup ( p2p GRE over IPSec or DMVPN I guess) is terminated on a different router.
If so internal routers in the site will compare EIGRP internal routes coming from VPN with EIGRP external routes that are the result of CE redistribution of PE-CE eBGP into EIGRP domain.
In a case like this the best solution may be the introduction of an additional EIGRP process on routers that terminates the VPN with appropriate redistribution to/from primary EIGRP process. In this way routes coming from VPN will be external routes as the routes coming from MPLS L3 VPN and are comparable. At this point just tuning the seed metric will make MPLS L3 VPN derived routes preferred over the backup VPN routes.
Another option (if allowed by IP addressing plan) would be the use of EIGRP summarization on VPN backup links in order to have more specific routes learned from MPLS L3 VPN PE-CE eBGP session.
To be noted if the protocol was OSPF and OSPF was also the PE-CE protocol the sham link feature allows to handle the backdoor direct link between sites.
Hope to help
Giuseppe
05-10-2012 05:09 AM
We had this exact issue a few weeks ago we solved by a combination of changing the metric in the bgp--eigrp redistribution statement and some selective route maps. I can post an example later if you need. Note that we are using one router for the MPLs/bgp network and a separate router at the head-end for the VPN/eigrp network.
Sent from Cisco Technical Support iPad App
05-10-2012 07:23 AM
Jeff
Exactly our setup, I would like to see your configs.
Thanks
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