Running mpls over VRF interfaces

[ghellai ferhat]

Hello, I wnat to run mpls over interfaces associated to vrf, so i have two adjascent 7200 routers, the interfaces between them are in vrf, once mpls configured, I can see the ldp adjacency built and routes are exchanged between the routers by ospf vrf XXX, but they are not labled.

I tried to get the interface in one router off the vrf and I saw routes are locally labled but no outgoing labels.

This is the configuration

----------------                                                    ----------------

| R1 7200  | --------------------------------------------------| R2 7200  |

---------------- f0/0                                        f0/0 ----------------

R1:

     ip vrf IP4

          rd 4:4

     int lo 0

          ip vrf for IP4

          ip add 1.1.1.1 255.255.255.255

     int f0/0

          ip vrf for IP4

          ip add 192.168.1.1 255.255.255.252

          mpls ip

          no sh

     router osp 1 vrf IP4

          net 0.0.0.0 255.255.255.255 a 0

R2:

     ip vrf IP4

          rd 4:4

     int lo 0

          ip vrf for IP4

          ip add 2.2.2.2 255.255.255.255

     int f0!/0

          ip vrf for IP4

          ip add 192.168.1.2 255.255.255.252

          mpls ip

          no sh

     router osp 1 vrf IP4

          net 0.0.0.0 255.255.255.255 a 0

Show ip route vrf IP4

     C 1.1.1.1/32

     C 192.168.1.0/32

     O 2.2.2.2/32

Show mpls ldp nei vrf IP4

Peer LDP Identifier: 2.2.2.2:0
    TCP connection: 2.2.2.2:646 - 1.1.1.1:65530
    Graceful Restart: No
    State: Oper; Msgs sent/rcvd: 46/43
    Up time: 00:31:21
    LDP Discovery Sources: 
  POS 0/2/0/0
    Addresses bound to this peer:
     2.2.2.2     192.168.1.2    

Show mpls forwarding-table vrf IP4

     "Nothing"

Does this mean that mpls can't be run under vrf interfaces?

Thanks.

[Nagendra Kumar Nainar]

Hi,

AFAIK, this is supported. What do you see in LDP binding table?.

Can you get the below,

show mpls ldp binding vrf IP4

show mpls for vrf IP4

-Nagendra

[ghellai ferhat]

Hi, thank you for your answer

this is the output of the colmmands you asked for:

R1#sh mpls forwarding-table vrf ip4

Local  Outgoing      Prefix            Bytes Label   Outgoing   Next Hop  

Label  Label or VC   or Tunnel Id      Switched      interface            

As you see there no output

but

R1#sh mpls ldp bindings vrf ip4

  LIB not enabled

I dont see what's LIB and how to enable it

Thank you.

[Harold Ritter]

Hi,

Please make sure CEF is enabled.

Regards

[ghellai ferhat]

Hi, yes cef is enabled but still get nothing.

I made a lab in gns3 with two routers, each one has a loopback interface and routing protocol,

in the first palce a made all interfaces and routing protocol in vrf xyz, I analyzed the ldp packets between them when builtin the ldp session and i compared them with the second lab in which i removed th vrf form routing protocole and interfaces, i saw labels in routes but the wireshark capture are not diffrent with those under vrf.

Is this doccumented somewhere because i found nothing in the net?

[Harold Ritter]

Hi Ghellai,

It is probably due to the version of code you are running then. What IOS do you run on this router.

Why do you need to run MPLS over a VRF interface? The only scenario where this would be required is Carrier Supporting Carrier (CSC). Is this what you are trying to do?

Regards

[ghellai ferhat]

HI, Harold.

Well I tried both c7200-adventerprisek9-mz.124-24.T and c3660-telcoentk9-mz.124-13b images in my lab and the result is the same.

Actually, I'm trying to implement ipv6 over MPLS the hardwre is like this

--------Site 1 --------------------------------------||---------------------------------------Site 2----------------

Cat 6509-e ---> Cat 3750-e--->7206-----||------------>7206---> Cat 3750-e--->Cat 6509-e

the Cat 3750-E in both sites don't support mpls, so we went to create link between 7200RTR and 6509 to become ospf neighbors to became like this

--------Site 1 ----------------||------------------Site 2----------------

Cat 6509-e --->7206-----||------------>7206---> Cat 6509-e

but without interfering with the global routing tables in place (important production traffic) this is why i tried to creat a vrf

Any ideas?

[Mohammed Imran Khan]

Hello Ghellai,

I am not sure if the feature is supported in this particular code. It doesnt even work for me either. But, what I can say is rather than using ospf, use bgp with 2 different AS numbers (ebgp), and remove mpls ip from the interfaces. This will work.

R1#

router bgp 1

no synchronization

bgp router-id 1.1.1.1

bgp log-neighbor-changes

no auto-summary

!

address-family ipv4 vrf IP4

  neighbor 192.168.1.2 remote-as 2

  neighbor 192.168.1.2 activate

  neighbor 192.168.1.2 send-label

  no synchronization

  network 1.1.1.1 mask 255.255.255.255

R1#show mpls forwarding-table vrf IP4

Local  Outgoing      Prefix            Bytes Label   Outgoing   Next Hop   

Label  Label or VC   or Tunnel Id      Switched      interface             

100    107           2.2.2.2/32[V]     0             Et0/0      192.168.1.2

102    Pop Label     1.1.1.1/32[V]     0             aggregate/IP4            

R1#

R2#

router bgp 2

no synchronization

bgp router-id 2.2.2.2

bgp log-neighbor-changes

no auto-summary

!

address-family ipv4 vrf IP4

  neighbor 192.168.1.1 remote-as 1

  neighbor 192.168.1.1 activate

  neighbor 192.168.1.1 send-label

  no synchronization

  network 2.2.2.2 mask 255.255.255.255

exit-address-family

R2#show mpls forwarding-table vrf IP4

Local  Outgoing      Prefix            Bytes Label   Outgoing   Next Hop   

Label  Label or VC   or Tunnel Id      Switched      interface             

104    Pop Label     192.168.1.0/30[V] 0             aggregate/IP4            

107    Pop Label     2.2.2.2/32[V]     0             aggregate/IP4 

[ghellai ferhat]

Hello Mohammed, this looks like a good idea to start, I just want to know why not ibgp but ebgp ? what does this give us?

Secondly, the reason we went to work with mpls is to use this mpls domain as transient domain for the ipv6 vrf mpls vpnv6, can we replace in this case with bgp?

Thanks.

[Harold Ritter]

Hi Ghellai,

ipv6 over mpls (6pe) in a vrf context is not supported in the version you mentioned not in current IOS versions.

You should be able to deploy ipv6 alonside ipv4 without using mpls nor putting it in a vrf. Given the limited size of the depicted network, my recommendation would be to deploy ipv6 natively. ipv6 will not interfere any more with your ipv4 traffic as if you were to deploy it in a vrf as ipv6 has it own routing table.

Regards

[ghellai ferhat]

Hi Harold, do you mean dual stack since ipv4 connectivity is already in place?

Nonetheless it seems the best idea to bypass all the limitations we've found

Even if implimenting MPLS, we went to make our wan equipment ready to multiple vpnv4 and vpnv6 custumers deployment.

Thanks

[Harold Ritter]

Hi Ghellai,

Yes, I meant dual stack. Can you elaborate on the limitations you've found?

Running ipv6 over MPLS is not a bad idea but trying to run ipv6 over MPLS inside a vrf context is.

Regards

[ghellai ferhat]

Hi Harlod

Here are the hardware conf and the limitations we've found:

PC&SRV --> 3750  -->  6500-E   -->  3750-E  --> 7200   --> 7200  -->  3750-E   -->  6500-E   -->  3750-E  --> PC&SRV

                                  FWSM                                                                            FWSM

3750-E doesn't support ipv6 vrf address-family

Layer 3 FWSM doesn't support DHCPv6 relay

Layer 2 FWSM doesn't support IPv6

3750-E also doesn't support MPLS

MPLS can't be run under vrf

It looks like our configuration will be like this

PC&SRV -->  6500-E   ---->  6500-E   -->  PC&SRV

It's just disappointing.

I will try running dual stack in our wan equipment, Thank you very match

[Harold Ritter]

Hi Ghellai,

Thanks for the feedback. I will convey that information to the respective product management teams.

Regards