Trouble with mpls vpn and P router

Hi,
I've configured a basic topology with 2 PE routers, 1 P router and a CE for every PE. The protocol running PE-CE is RIP configured for VPN, with redistribution into MP-BGP. The IGP in the core is OSPF. When I connect the PE's directly I get ping from CE to CE, but when I connect them through the P router I can't seem to get connectivity even though I do get the correct routes on both CE's, which means BGP is configured correctly and redistributing. When I check the mpls forwarding-table on the P router I see only labels towards the loopbacks of the PE's and no entry for the tag which I see on the PE's ( tag 20).
Does anyone have an idea how should I troubleshoot this issue? I've gone through the MPLS and BGP configuration for both the PE's and P and I've found nothing. I've even tried changing the source of the BGP updates but to no avail.
Thanks in advance for any assistance

Find more posts tagged with

Comments

SteveO86

I'd post the configs so we can take a look

nompanem

Your'e right. I'll post them next time. For now I've found the problem with my config, I forgot to make the loopbacks ospf point-to-point.
Thanks

SteveO86

Awesome!

It's always the small details that get you

rwestphal

There probably something wrong with your LDP configuration.

" When I check the mpls forwarding-table on the P router I see only labels towards the loopbacks of the PE's and no entry for the tag which I see on the PE's ( tag 20)."

This is _weird_ but it wouldn't be a problem if you changed the source of the BGP Updates to your loopback interfaces...

Forsaken_GA

rwestphal wrote: »

There probably something wrong with your LDP configuration.

" When I check the mpls forwarding-table on the P router I see only labels towards the loopbacks of the PE's and no entry for the tag which I see on the PE's ( tag 20)."

This is _weird_ but it wouldn't be a problem if you changed the source of the BGP Updates to your loopback interfaces...

It's not at all weird when using OSPF at the core IGP. The problem comes from the fact that there's a discrepancy between the routing table and the prefix the interface is actually configured with. LDP will advertise a label for the prefix on the interface (probably a /24 I'm guessing), but OSPF will advertise the loopback as a /32 by default. LDP finds no prefix in the routing table to match it to, so no label gets advertised.

So either change the network type to point-to-point to get the true prefix advertised into OSPF, or do smart IP management and actually address your loopbacks with /32 masks.

nompanem

As my previous post says, your completely right and that was the cause of the problem. Next time I'll be more carefull