BGP NextHop issue

mounir.mohamed · ‎02-23-2006

Dear everybody,

Kindly advice about the below issue

There is a router R1 in my network using static route to reach specific subnet in another router R2 because there is no routing protocols running between each other, the static route like 1.x.x.0 x.x.x.192 2.2.2.2, R1 is running IBGP with a R-Reflector and it's shall advertise network 1.x.x.0 via bgp to the RR but the strange point is that R1 sent network 1.x.x.0/26 to RR with next-hope 2.2.2.2!!! how is that??

R1 Config

Neighbor 192.168.99.3 remote-as 65530

Neighbor 192.168.99.3 update-source loopback 0

Neighbor 192.168.99.3 ebgp 2

no auto-summary

no synchronization

network 1.x.x.0 mask x.x.x.192

The output from sh ip bg is

*> 1.x.x.0/26 2.2.2.2 0 32768 i

There is no problem from the RR side because the rest of clients working well

Best Regards,

Mounir Mohamed

pkhatri · ‎02-23-2006

Hi Mounir,

That's normal behaviour. When you inject local routes into iBGP, they use the same next-hop that the routing table uses for that route.

In order to prevent this, try the following bit of config on R1:

neighbor next-hop-self

That will set the next-hop to be R1's loopback address.

Hope that helps - pls rate the post if it does.

Paresh

mheusinger · ‎02-23-2006

Hello Mounir,

this is, what I would expect. The BGP update has a next hop field, which is set to the next hop found in the IP routing table. This is what the "network" command should do.

In case you want to override it use next-hop-self or redistribute static route-map Sta2BGP and set origin code and bgp next hop to a different value.

Hope this helps! Please rate all posts.

Regards, Martin

mounir.mohamed · ‎02-23-2006

Dear Martin,

First Thanks for your replay,

Actualy i'm already using next-hope-self with the neighbor ip address as workaround for that issue but did u face this problem before or do u know why the next-hope missing in the update?

Best Regards,

Mounir Mohamed

pkhatri · ‎02-23-2006

Hi Mounir,

As I indicated earlier, this is not a problem. This is correct behaviour as per RFC4271 Section 5.1.3 (1):

"When announcing a locally-originated route to an internal peer, the BGP speaker SHOULD use the interface address of the router through which the announced network is reachable for the speaker as the NEXT_HOP."

As you can see, the RFC mandates that the next-hop should not be changed in such cases..

Hope that helps - pls rate the post if it does.

Paresh