MP-BGP for multiple customers; can't get labels through

[scott.bridges]

Hello all,

CustomerRTR1 -- PE -- OrgRTR1 -- OrgRTR2 -- CustomerRTR2

I'm trying to get CustomerRTR1 pinging to CustomerRTR2.
PE has multiple Customer L3VPNs.
PE connects to OrgRTR1 and passes CustomerRTR1 traffic via routed subinterface port (no MP-BGP).

I put OrgRTR1 -- PE in vrf Cust1, rd 123:1, export as 123:1, import 123:2.
OrgRTR1 -- OrgRTR2 is over a GRE P2P. MP-BGP formed on tunnel interface. OSPF & LDP formed over Tunnel. Tunnel in Global.
I then VPNv4 between OrgRTR1 & OrgRTR2 to pass labels.
OrgRTR2 puts CustomerRTR2 routed port into vrf Cust1, rd 123:2, export as 123:2, import 123:1.

At least that's my plan. However, I cannot seem to get it working. What am I missing? Does the Tunnel overlay/underlay need to be in VRF?

I believe this is pretty much a carrier-supporting-carrier thing, but I'm not familiar with those methods. I just have a PE that's handing off multiple customers to me (subinterfaces). I need to accept those routes (OrgRTR1), then move them all to another router within my Org (OrgRTR2).

Any advice is appreciated

[Vasilii Mikhailovskii]

Hello.

From the description it looks like inter-AS option A (as PE hands over VRF as raw IP on subinterfaces), if so - you must be running BGP with PE on customer VRFs. Is this correct?

Also, as I understood,  you are building MPLS over GRE. I would suggest you need to verify if this is supported on the platform/release you are running.

You wrote that the setup didn't work for you. Could you elaborate a little more here? Do you have labelled path between the RTR1/RTR2 (that are technically PE devices)? Do you see VPNV4 exchanged labels?

So, could you provide relevant configuration a commands you verify operation? Could you ping from RTR1 to RTR2 inside VRF?

(show mpls for, sh mpls ldp dis, show mpls for vrf ..., sh bgp vpnv4 uni all summ, sh bgp vpnv4 un rd 123:1, sh bgp vpnv4 uni rd 123:2, sh ip cef vrf ... <remove destination> detail, ping vrf ... <remote PE>)