[c-nsp] iBGP between PE and CE

[Pshem Kowalczyk]

Hi,

I have a small problem with a iBGP on PE-CE link. The setup's been
'inherited' but never caused us trouble until the requirements
recently changed slightly (it's a bit boring story). To cut it short,
we have the following setup:

(CE1)--ibgp--(PE1)--mpbgp--(P)--mpbgp--(PE2)--ibgp--(CE2)

(the PE2 and CE2 is the new addition).

Both CEs are route-reflector-clients of the PEs. PEs are 6503s. Routes
from CE1 get into the MPLS-cloud without any issues. The problem is,
since it's ibgp between CE1 and PE1 the next hop is not getting set
after the prefix enters the VRF. As a result of that PE2 refuses to
install the route (the next hop is still the IP address from the vrf
on PE1).

We have exactly the same setup with eBGP on the fringes and that
obviously works fine. The question is - how to make it go with that
setup (if it's possible at all)? I tried various techniques  to  get
it sorted (mainly by trying to manipulate the prefix using route map).
I have a lot of experience with bgp but not a lot with mpbgp, so I'm a
bit confused here. for normal ibgp I would just set up the 'next hop
self' on the session between PE and P, but I don't think it's going to
work on mpbgp (since ebgp already works fine I don't want to break
it).

Any hints how to fix the problem?

kind regards
Pshem