MPLS without OSPF?

I’m having trouble with MPLS forwarding without setting up OSPF. I’m distributing routes with BGP, and I’m unable to establish LDP session (targeted, or auto-discovered). Is this supposed to be working?

Nevermind. It was switch’s fault. Didn’t pass LDP (which is strange, since it didn’t have any problem with other multicast)

Anyway, on to another problem:

From wiki:
“As LDP distributes labels for active routes, essential requirement is properly configured IP routing. LDP by default distributes labels for active IGP routes (that is - connected, static, and routing protocol learned routes, except BGP).”

How does one change this “default”, so I can distribute labels for routes learned from BGP? (eBGP namely - I’m using individual AS numbers per router.)

Eh, after fighting for few hours with this, adding local-bindings by hand etc., I still couldn’t make it work.
Labels for VPLS tunnel are assigned, but it doesn’t enter running state. Loopback addresses get announced with impl-null label, and even when overriden with local-bindings ‘add’, it doesn’t help VPLS tunnel to make it work.

Any ideas?

setup your bgp with confederation. Then use the address families vpn4 (and l2vpn-cisco)

Didn’t work. Still impl-null labels assigned and VPLS doesn’t start. :frowning:

OK, problem “solved”.

“Thank you for suggestion, we will consider adding it in future.”

So… If like me, you have a fully BGP routed core network, no MPLS for you :confused:

Im confused. Using bgp for distribution of L2VPN works perfectly for me.

Could you do a /mpls export and /routing export from some router that only FORWARDS the MPLS traffic?

Any chance for those exports, or some update from mikrotik on per-instance option to distribute labels for BGP-learned routes?

Anyone get any further updates on this one? Re-checked in test lab and still could not see a way to force BGP learned routes into the MPLS forwarding-table. Would love to get rid of other IGP and just use BGP for static VPLS tunnels.