BGP - Multihop and Multipath -

[yhaas1]

Hi, 

hope you can help me. I want to do bgp with two C1111 Routers with two links between the routers.

First direct connected and secound via multihop. 

But I am not able to bring be secound link up. 

Setup like this: 

All Networks are /28. 

This is my BGP config: 

router bgp 100
bgp log-neighbor-changes
neighbor 10.31.4.18 remote-as 200
neighbor 10.31.4.18 fall-over bfd
neighbor 10.31.4.65 remote-as 200
neighbor 10.31.4.65 ebgp-multihop 4
neighbor 10.31.4.65 fall-over bfd
!
address-family ipv4
network 10.31.4.0 mask 255.255.255.240
network 10.31.4.16 mask 255.255.255.240
neighbor 10.31.4.18 activate
neighbor 10.31.4.65 activate
maximum-paths eibgp 5
exit-address-family

router bgp 200
bgp log-neighbor-changes
neighbor 10.31.4.17 remote-as 100
neighbor 10.31.4.17 fall-over bfd
neighbor 10.31.4.49 remote-as 100
neighbor 10.31.4.49 ebgp-multihop 4
neighbor 10.31.4.49 fall-over bfd
!
address-family ipv4
network 10.31.4.16 mask 255.255.255.240
network 10.31.4.32 mask 255.255.255.240
neighbor 10.31.4.17 activate
neighbor 10.31.4.17 soft-reconfiguration inbound
neighbor 10.31.4.49 activate
neighbor 10.31.4.49 soft-reconfiguration inbound
maximum-paths eibgp 5
exit-address-family

Maybe there is only a simple trick I don't see... 

Thanks! 

[Harold Ritter]

Hi @yhaas1 ,

Can you post a "show bgp ipv4 uni summary" output?

Regards,

[MHM Cisco World]

bgp with fall-over BFD have two 
single hop command "fall-over bfd"
multi-hop command "fall-over bfd multi-hop"

[MHM Cisco World]

the same BGP peer "which identify  by router-id" must have one single BGP session to any other Peer, 
here you config two session between same two peer. 
this config is wrong, you need to config any IGP between the two peer and use loopback as update source and loopback for other peer as neighbor. 
this achieve the two link IGP one BGP session between two Peer.

[vencislav.metev]

Hi,

Can yoy try to remove:

ip route static bfd 10.31.4.49 10.31.4.65 unassociate

ip route static bfd 10.31.4.65 10.31.4.49 unassociate

 Regards,

Ventsi

[yhaas1]

I tried this but no change. 

[MHM Cisco World]

Freind you can not do two bgp session between two peer.

There must be one session, and to improve that remove direct connect link config bgp with bfd "dont forget multihop" and it will work.

[Harold Ritter]

@MHM Cisco World ,

You can certainly have more than one session between two BGP peers, as long as they do not use the same end points.

Regards,

[MHM Cisco World]

This make loop, and hence it is not possible.

RA send route to RB,

RB resend it to RA since it assume it different peer.

So if there are multi link then we run any igp or config static and config loopback in both peer.

This loopback will use as update source and as nieghor ip address.

[Harold Ritter]

Hi @MHM Cisco World ,

There are a few use cases, where I have used multiple BGP sessions between the same two routers in the past. This will not create a loop.

> RA send route to RB,

> RB resend it to RA since it assume it different peer.

RA rejects the route as the AS-PATH contains it own ASN.

Regards,

[MHM Cisco World]

I think you talk about multi session bgp with different address family not same address family, am I right?

But here he use only one address family ipv4.

[Harold Ritter]

Hi @MHM Cisco World ,

> I think you talk about multi session bgp with different address family not same address family, am I right?

No, I am talking about two BGP IPv4 unicast sessions between the same pair of routers. This works and I have used it for a few use cases in the past. This is will work without any issue.

Regards, 

[vencislav.metev]

Hi,

Can you try telnet from 10.31.4.65 to 10.31.4.49 on port 179?

Or vice versa.

Regards,

Ventsi

[MHM Cisco World]

do 
show bgp ipv4 unicast neighbor
see if the multisession is enable or disable.