MPLS Label not getting assigned

[ryel.dsouza]

I am having a issue with mpls label not getting assigned to vpnv4 routes. My lab setup is like

R6(CE) s1/3 ------à s1/3 R1 (PE) s1/2 -----------à s1/2 R3 (p) s1/3 ----------------à s1/3 R2 (p) s1/2-----à s1/1 R4 (PE) s1/2 --------à s1/2 R5 (CE).

The vpnv4 routes from R1 (PE) do appear on R4 (PE) but without any label binding. With i use the “sh mpls ldp  binding vrf ryel on R4 it says TIB not enable when i’m using ldp for  label distribution .

I have attached my config of all the routers in my lab kindly provide me your inputs.

(R5 & R6 config's r in R5.txt file)

[Giuseppe Larosa]

Hello Ryel,

you are using loopback interfaces with a non /32 mask.

the problem should be at primary label level

OSPF advertises a non /32 loopback as a /32

add

int loop x

ip ospf network point-to-point

this makes OSPF to advertise the actual subnet mask OR use /32 loopbacks that is the best practice.

on the loopbacks used as BGP endpoints

also check that the same loopback is chosen as router-id by: OSPF, BGP.

that is you should see

sh mpls forwarding 4.4.4.4

as untagged now.

Hope to help

Giuseppe

[ryel.dsouza]

i have used the ip ospf network point-to-point on all the loopbacks and also have R1 using lo 0 (1.1.1.1)  & R4 using lo 0 (4.4.4.4) for the router id in ospf and BGP. The bgp peering is also done on the two ip's

I still do not receives mpls label for the vpnv4 routes coming from R1 on R4

[Giuseppe Larosa]

Hello Ryel,

add under

address-family ipv4 vrf ryel

red connected

let's see if in this way there is a change it is a way yo help the PE.

may  you post

sh ip bgp vpnv4 all label

Hope to help

Giuseppe

[ryel.dsouza]

Sorry for the late reply. The redistributed command did not help. Apart from that. It seems that labels are assigned in the cef tabel of R4 but they don't appear in the mpls for warding tabel. Kidly help me understand why ???

R4#sh ip cef vrf ryel 6.6.6.6
6.6.6.0/24, version 47, epoch 0, cached adjacency to Serial1/1
0 packets, 0 bytes
  tag information set
    local tag: VPN-route-head
    fast tag rewrite with Se1/1, point2point, tags imposed: {23 18}
  via 1.1.1.1, 0 dependencies, recursive
    next hop 24.0.0.1, Serial1/1 via 1.1.1.0/24
    valid cached adjacency
    tag rewrite with Se1/1, point2point, tags imposed: {23 18}

R4#sh ip cef vrf ryel 66.66.66.66
66.66.66.0/24, version 48, epoch 0, cached adjacency to Serial1/1
0 packets, 0 bytes
  tag information set
    local tag: VPN-route-head
    fast tag rewrite with Se1/1, point2point, tags imposed: {23 21}
  via 1.1.1.1, 0 dependencies, recursive
    next hop 24.0.0.1, Serial1/1 via 1.1.1.0/24
    valid cached adjacency
    tag rewrite with Se1/1, point2point, tags imposed: {23 21}

R4#sh mpls for vrf ryel
Local  Outgoing    Prefix            Bytes tag  Outgoing   Next Hop   
tag    tag or VC   or Tunnel Id      switched   interface             
21     Untagged    5.5.5.0/24[V]     0          Se1/2      point2point 
22     Aggregate   45.0.0.0/30[V]    4424              

Thanks in Advance.              

[Giuseppe Larosa]

Hello Ryel,

now you should be able to ping in VRF you can check this with

ping vrf ryel 6.6.6.6 from R4

to be noted that what you see is correct: the sh ip cef vrf ryel 6.6.6.6 or sh ip cef vrf ryel 66.66.66.66 is showing us the label stack used in the forwarding plane.

fast tag rewrite with Se1/1, point2point, tags imposed: {23 18}

Here label 23 is the label to be used to reach R1's loopback.

You can check this with

sh mpls forwarding 1.1.1.1 on R4.

the inner label 18 is the VPN label for prefix 6.6.6.6 you can check this with

sh ip bgp vpnv4 all labels

the sh mpls forwarding vrf ryel shows only the labels that are part of the VRF and that are assigned to prefixes learned locally, connected or by OSPF process.

That is you cannot expect to see the labels for remote VRF prefixes here.

As you can see a different inner label is associated to prefix 66.66.66.66

Sorry if I haven't got this point first.

To be noted the behavior of this show commands is IOS dependent. You can find IOS images that don't show the complete label stack in sh ip cef vrf output.

Other changes to sh mpls forwarding vrf have been made in very recent IOS images for the adoption of the new MPLS MLF.

Hope to help

Giuseppe