BGP Next-Hop-Self

[kim]

Hi guys

Thanks for an awesome resource!

A quick question related to BGP and NEXT-HOP-SELF.

We have a classic WAN setup with two separately running border switches. They run EBGP towards a firewall setup and IBGP between each other. There is no NEXT-HOP-SELF configured on the IBGP peers. Yet routes seen on the left IBGP peer still have the address of the right IBGP peer as their next hop. This is for routes learned from the right EBGP peering.

Could someone explain under which circumstances BGP behaves like this? Shouldn't the IBGP part avoid changing the nexthop address?

Cheers and thanks!

/Kim

[Leonid Voronkin]

Please show us the proofs. Show run from both border routers and show ip bgp from left peer.

IBGP should keep next-hop by default.

[Georg Pauwen]

Hello,

not sure what exactly your setup looks like, but do the routers advertise the networks that link them to their eBGP neighbors ? In that case, no 'next hop self' would be needed.

If possible, post the configs of both iBGP routers.

[Martin L]

if your 2 eBGP connect and get same routes from single ISP, maybe what you see is the result of eBGP  preferring routes over iBGP.

IBGP have next-hop by default but not eBGP.