fall-over bfd multi-hop not supported

[sebastien3]

Hello,

Is there a trick when it is not possible to use BFD multihop and mapping for BGP (fall-over bfd multi-hop) ?

I have three routers that have a BGP session :

The connection from router 1.1.1.11 to the other two is not direct because there are optical transeivers on link.

For example the router 1.1.1.100 goes down, router 1.1.1.11 will not see the cut at the BGP level because it uses the loopback IP and BFD work only on interface 10.0.1.X.

Is a solution possible in your opinion ?

Thanks !

[MHM Cisco World]

neighbor fall-over route-map eBGP-multihop
!
route-map eBGP-multihop
match ip address prefix-list <neighbor Link>

this make BGP check the RIB for link and fall-over depend on it.

[sebastien3]

@MHM Cisco World 

I'm not sure, thinking about this configuration on router 1.1.1.11 ?

router bgp 500
bgp router-id 1.1.1.11
neighbor 1.1.1.100 remote-as 500
neighbor 1.1.1.100 description Link to R1
neighbor 1.1.1.100 fall-over route-map eBGP-multihop
!
route-map eBGP-multihop
match ip address prefix-list BFD
!
ip prefix-list BFD description IP OF R1
ip prefix-list BFD seq 5 permit 10.0.1.2/32

[paul driver]

Hello
You can applend BFD per interface via using a template on the interface ( in your case not supported ) or manually applying it on the interface>

Example:
int x/x
interval min-tx 500 min-rx 500 multiplier 5

router bgp xxx
neighbor x.x.x.x fall-over bfd multihop 

[sebastien3]

Hello@paul driver 

On interface I use bfd interval 750 min_rx 750 multiplier 3

The problem is that if for example between the router 1.1.1.100 and 1.1.1.11 there are two switches if one fails the port will not shut on the routers and the BGP convergence will not be instantaneous...