MPLS LDP Neighbor Addresses Blank

RouterOS Forwarding Protocols
1

I have an odd issue that I’m hoping someone can help me with. The problem is that when I do an /mpls ldp neighbor print command the “Addresses” field is blank for some of the neighbors. I believe this is causing a problem with my VPLS connections.

Here’s my configuration:

/mpls
set propagate-ttl=no
/mpls interface
set [ find default=yes ] mpls-mtu=1526
/mpls ldp
set enabled=yes lsr-id=10.249.4.6 transport-address=10.249.4.6
/mpls ldp interface
add interface=ether1
add interface=Tunnel0
add interface=Tunnel1
add interface=sfp1

And here’s the result of the neighbor print command:

 /mpls ldp neigh print value-list where transport=10.249.4.8
         transport: 10.249.4.8
    send-targeted: yes
             peer: 10.249.4.8:0
  local-transport: 10.249.4.6
        addresses:

Other peers show up properly for the addresses field.

Not sure how best to troubleshoot this one and I’m hoping someone out there can help me solve this one.

I will keep working on it and if I find the solution I will post it for those out there that may be having the same problem.

Thanks.

2

For all those that have been looking at this I did figure out the issue. There was an MTU problem on the switch that connected the two segments. One side was set to 1500 when it was supposed to be set to 9000 (jumbo). I noticed this as there were a large number of input errors and giants seen on the port counters. Once I changed to the jumbo frames on the affected port everything started working properly.

So in conclusion, make sure your MTU is set properly when connected to other peers (especially when running MPLS over those links.)

3

Glad you figured it out…MTU can play havoc with an MPLS network. Two basic MTU rules to remember if you want to pass a 1500 byte frame in MPLS.

Minimum MTU for MPLS (for 1500 byte frame)

1526 (L2 and MPLS MTU) - untagged frame
1530 (L2 and MPLS MTU) - tagged frame

4

Should the MTU be set on the VPLS tunnels or on the ether interfaces, or both?

5

The VPLS MTU is the size of the MTU you want to hand off to your customer. The interface MTU should be set to accommodate the overhead of VPLS. You need at least 1530 to send an 802.1q tagged frame through a VPLS tunnel.

Take a look at this MUM presentation that I did and it has MTU recommendations for MPLS

https://mum.mikrotik.com//presentations/US16/presentation_3327_1462279781.pdf

6

Try and follow this scenario and try to help me with what I have found regarding your statement.

In the instance of going through another users equipment to link two ospf routers and lets say its interface limitation is 1500 for its mtu. And you are using it to pass your /30 subnet between two routers.

You can still have a VPLS mtu of 1500, and the customers equipment connected through the vpls tunnel will pass 1500 byte non fragmented packets. But you have to set the mpls interface on both routers to 1486 IIRC. Then the routers will cut the packets as needed, but the customers on the vpls tunnel will still think they are passing 1500 byte non fragmented packets. I think in the testing I have done that works. Optimally the size of the mpls interface should be higher, but if you are limited by mtu issues on paths, you have to address that in mpls interface, in order for the customers to still pass 1500 byte packets. Are your findings the same?

7

Can we have more info regarding :

1526 (L2 and MPLS MTU) - untagged frame
1530 (L2 and MPLS MTU) - tagged frame