mpls forwarding table has entries via interfaces not participating in mpls / ldp

Couple of questions here

1. What is rather perplexing to me about the mpls forwarding-table is that one path from R5 to R4 is via a path with interfaces configured for mpls; s 0/1/0.  The other path in the table is via 0/0/0.  Now why would an mpls table display a path that is not configured for mpls on R5 for both R4 and R6 ??       
            
Rack1R5#sho mpls forwarding-table
Local  Outgoing      Prefix            Bytes Label   Outgoing   Next Hop   
Label  Label or VC   or Tunnel Id      Switched      interface             
16     16            150.1.6.6/32      0             Se0/1/0    point2point 
      No Label      150.1.6.6/32      0             Se0/0/0    155.1.0.4  

17     Pop Label     150.1.4.4/32      0             Se0/1/0    point2point
       No Label      150.1.4.4/32      0             Se0/0/0    155.1.0.4 

Rack1R5#sho ip cef 150.1.6.6
150.1.6.6/32
  nexthop 155.1.0.4 Serial0/0/0
  nexthop 155.1.45.4 Serial0/1/0 label 16

Rack1R5#sho mpls int
Interface              IP            Tunnel   BGP Static Operational
Serial0/1/0            Yes (ldp)     No       No  No     Yes      

Rack1R6#sho mpls int
Interface              IP            Tunnel   BGP Static Operational
FastEthernet0/0.146    Yes (ldp)     No       No  No     Yes       

Rack1R4#sho mpls int  - VIA AUTOCONFIG FROM 14.2
Interface              IP            Tunnel   BGP Static Operational
FastEthernet0/0        Yes (ldp)     No       No  No     Yes        
FastEthernet0/1        Yes (ldp)     No       No  No     Yes        
Serial0/1/0            Yes (ldp)     No       No  No     Yes        
Serial0/0/0.1          Yes (ldp)     No       No  No     Yes 

2.  How does CEF operate when two paths are in the forwarding table, yet only one is label switched?  CEF operates on a "per desitnation" hash.  Which path is chosen?  It doesn't look good. I'm wondering how to control the layer 2 path selection since the interface mpls ip configuration does seem to be.

Looks hazardous:

Rack1R5#trace 150.1.6.6

Type escape sequence to abort.
Tracing the route to 150.1.6.6

  1 155.1.0.4 28 msec
    155.1.45.4 [MPLS: Label 16 Exp 0] 48 msec
    155.1.0.4 28 msec
  2 155.1.146.6 20 msec *  16 msec

Comments

  • Hi Tammy,

    Me again. [:D]

    I was wondering if you have the Advanced Technology Course or the Open lecture Product that INE offers.  They are both excellent tools! If you do you should take a look at the the lecture for MPLS and Brian McGahan did.  I think it will help with some of your confusion.  It certainly did mine!


    If you don't have it I would suggest getting it.  But I'll try to address some of your questions.  I'm sure if I'm wrong some one will correct me.


    Reqarding the MPLS forwarding table:  It's not that there is mixture of mpls and non mpls interfaces, as much as it is a list of instructions on what to do with the labels as they are received and transmitted.  Keep in mind that labels stack.  We can either add, remove (pop) a label or removed all labels (No Labels).  So if mpls traffic is to be sent out s0/0/0 to 155.1.0.4 (sited in your example) then all labels are removed.

     Next:

    CEF switching is the only IP switching mode that you can use to label
    packets. When a router receives an IP packet a lookup is done as an ip lookup. When a router receives a labeled
    packet, the lookup is done in the LFIB of the router. The router knows
    that it receives a labeled packet or an IP packet by looking at the
    protocol field in the Layer 2 header. If a packet is forwarded by
    either CEF (IP lookup) or by the LFIB (label
    lookup), the packet can leave the router either labeled or unlabeled (meaning via a standard ip routed path or a ip mpls path).  The point being that CEF is used to construct both tables employed in either lookup methodology therefore it should be safe (in my opinion) to assume that consistencies exist in the engines of both methods.

    Again I'm not an expert, so anyone correct me if I'm wrong.

    Terry

     

     

  • Hi Terry -

    I do have the CoD and have watched the MPLS track.  Maybe now that I've been working on the core labs for mpls I should go back and pick up the neuances. 

    Many thanks for taking the time to respond.  You did a fantastic job with the semantics.  I greatly appreicate your attention to detail.

  • Thanks,

    You just ask good questions.

    It's great that you have the COD!  The best thing you can do with that product's MPLS section (in my opinion) is follow along with Brian as he does his thing.  He produced the MPLS lecture on Dynamips and it's one of the few lectures where he doesn't go the speed of light. Then the best thing you can do after that is "go off roading". Make your own scenarios, and make them work or find out why you can't. Once you've done that go back and watch the lecture again.  I bet you will be amazed a how much you pick up that you missed the first time.  Then Check MPLS off your task list and move on!

    Regarding the Volume one workbook.  Keep in mind it goes deeper and broader than the blueprint.  I like that fact, but in the beginning at least focus on what is on the blueprint.  Master that and move on from there.

    Terry

     

Sign In or Register to comment.