iBGP local preference conflict

[tedauction]

Hello, I have two adjacent iBGP VRFs, which each have a bordering eBGP VRF (diagram attached).

My topology includes a route for 10.0.0.0/8 which I am passing from each of the eBGP VRFs into the inside iBGP VRFs. I am preferring eBGP VRF route paths over the other by using AS prepending.

The problem is that by default, when the iBGP VRFs receive this 10.0.0.0/8 route, they forward this route to each other with the 'Local Preference' default attribute of 100 (expected iBGP behaviour). This is installed in their routing tables due to 'Local Preference' having a higher order of preference than AS prepending.

Is there any way, I can force the iBGP VRFs to instead recognise the 'AS prepend' attribute rather than the 'Local Preference' attribute - or is it not feasible to try to pass routes via iBGP and eBGP VRFs as per my topology ?

[MHM Cisco World]

There is nothing with LP,

Can i see

Show ip bgp 10.0.0.0?

[MHM Cisco World]

Cisco Best Path 
""Prefer the path with the shortest AS_PATH.""

NOW 
AS prepend x3 from Side 
AS prepend x2 from other Side (keep notice that iBGP not add AS-path)

so for iBGP vrf-A it better via iBGP than eBGP because it have shortest path.
so I think you make misconfig
instead in eBGP AS 65100 use x2 
in eBGP AS 65000 use x3

[Flavio Miranda]

Hi

 "...The problem is that by default, when the iBGP VRFs receive this 10.0.0.0/8 route, they forward this route to each other with the 'Local Preference' default attribute of 100 (expected iBGP behaviour). "

https://networkengineering.stackexchange.com/questions/44908/why-are-ibgp-routes-not-propagated-to-other-ibgp-neighbors 

This is wrong.  The expected behavior is not advertise routes leaned from iBGP peer as Loop prevention mechanism.

[Harold Ritter]

Hi @tedauction ,

The local preference is the same for all eBGP learnt routes (100). It should therefore not play a role in the best path selection.

Please provide the output for "show bgp vrf <vrf name> ipv4 uni 10.0.0.0/8" from both iBGP peers.

Regards,

[paul driver]

Hello

Set the LP attribute higher than current default (100) for those specific routes to take the egress path you wish.

Example:
ip prefix-list LP_pf permit 10.0.0.0/8 ge 32

route-map LP_rm permit 10
match ip address prefix-list LP_pf
set local-preferance 200

route-map LP_rm permit 99

router xxx
neigbour <Ebgp peer> route-map LP_rm IN