No info in MPLS forwarding-table

[Chanuka D]

Hi Guys, 
I configured basic MPLS configuration on our core routers and border router. I can see ldp neighbor relationship by using show mpls ldp neighbors command in border router.

But when see mpls forwading-table in border router not even one byte is label switched. Note that labels are exchanged and assigned to routes. OSPF is running as IGP. Core router is label switching according to the forwarding table. What could cause border router to avoid label switching ? 

Thanks.

[Harold Ritter]

Hi Chanuka,

Bear in mind that the ingress PE will not label switch the traffic. It will use the FIB rather than the LFIB to perform the label imposition function on the outgoing packets. The P router will use the LFIB to perform the label swapping function. The egress PE will use the FIB to forward the traffic in a 1 label stack scenario, as the single label will have been popped by default a the penultimate hop router (PHR). So what you are seeing seems to be the proper behavior.

Regards,

[Chanuka D]

Hi Harold,
Thank you very much for your response. we have a another router at one of our edges. In mpls forwarding-table I can see outgoing labels. But 0 bytes label switched. I'm confused why it's not doing label switching. LDP relationship is there. Any idea what's happening ? 

Thanks.
Chanuka

[Harold Ritter]

Hi Chanuka,

The "show mpls forwarding" command shows the content of the LFIB. The reason the traffic is not being labeled switched is that the traffic arrives at the egress PE unlabeled and in that case the egress PE does a FIB lookup rather than the LFIB lookup to forward the traffic, hence the counters being set to 0 in the LFIB.

If you had services configured on your PEs (l2vpn, l3vpn, 6pe, etc) or if the PE is being used as a transit MPLS nodes (P router), the traffic would arrive labeled, the LFIB would be looked up and the the LFIB counters would be incremented.

Regards,

[Chanuka D]

Hi Harold,


Thank you again. Please note that currently we don't have any VPN configurations. Just basic MPLS configuration. Our cisco border router which is facing our upstream, connected to two core routers and all our edges are connected to that two core routers. we two total 8 edge routers. OSPF is running as IGP. According to my understanding any traffic coming from our upstream should assign a label by MPLS and forward to core routers by border router unless destinations are directly connected to core routers. (PHP). But my question is although there are LDP neighbors I can't see anything on forwarding table. When I do a trace route from a looking glass, I can't see MPLS label on border router. But I can see core routers labels. I hope you will understand my concern. Kindly correct me If I'm wrong.

Thank you.
Chanu

[Harold Ritter]

Hi Chanu,

Let me try to clarify.

- Ingress PE does a FIB lookup and does the label imposition.

- P router(s) does a LFIB lookup and does label switching. If P router is the penultimate hop, it also does the label disposition. If there is no services, there is only one label in the label stack and after label disposition the outgoing packet forwarded towards the egress PE is IP, not MPLS.

- Egress PE does a LFIB lookup if incoming packet is MPLS and FIB lookup if incoming packet is IP.

Egre

[Chanuka D]

Hi  Harold,

I aware about the things that  you clarified. Just like you mentioned as the first point, our border router should impose a label and forward it to core routers. But I can't see anything on forwarding table. Is it normal ? core routers are popping label as they are directly connected to LERs. 

Thank you.

[Harold Ritter · Accepted solution]

Chanuka,

The counters in the LFIB are only incremented for traffic that is label switched. On the ingress PE, label imposition is performed, but that is not label switching, hence the counters in the LFIB not being incremented on the ingress PE.

Regards,

[Chanuka D]

Hi Harold, 

Thank you very much for clarifying me. I was confused about this operation. However I have one more question. How can I verify label imposition and php  is being done in a router ?? Which debug command I should use ? I tried few different mpls debugging commands, but no output. 

Thank you
Chanu

[Harold Ritter]

Hi Chanuka,

For the label imposition part, you can check in CEF and see if there is a label or labels assignes to a given prefix. This doesn't necessarily mean that this router is a PE. The same router can be a PE for some flows and a P for other flows. As far as the PHP is concerned, you can check in LDP whether an implicit null label is received for a given prefix or you can check the output of the "show mpls forwarding-table" command  and see which prefixes have an outgoing action of "pop label".

Regards,

[Chanuka D]

Hi  Harold,

Thank you very much for your support and appreciated. It helped me a lot to understand how MPLS works. Thanks again.