[SOLVED] VPLS MTU problems on almost directly connected RBs

RouterOS Forwarding Protocols
1

Hi,

I have a lab setup as follow:

CCR1072 <==10gB (MTU 9212)==> ARISTA 7050S <==LACP (MTU 9212)==> NETONIX WS-12-250-DC <==GBE (MTU 4074)==> RB2011

I’m connecting the RB2011 to the CCR via VPLS for testing purposes and I get all packets over 1488 bytes dropped, inside the tunnel.

[dmayan@HQ] > /ping 10.7.7.2 interface=vpls.core2 size=1500
  SEQ HOST                                     SIZE TTL TIME  STATUS                                                                               
    0 10.7.7.2                                                timeout                                                                              
    1 10.7.7.2                                                timeout                                                                              
    2 10.7.7.2                                                timeout                                                                              
    sent=3 received=0 packet-loss=100%
  
[dmayan@HQ] > /ping 10.7.7.2 interface=vpls.core2 size=1488
  SEQ HOST                                     SIZE TTL TIME  STATUS                                                                               
    0 10.7.7.2                                 1488  64 2ms  
    1 10.7.7.2                                 1488  64 0ms  
    2 10.7.7.2                                 1488  64 0ms  
    sent=3 received=3 packet-loss=0% min-rtt=0ms avg-rtt=0ms max-rtt=2ms

All L2MTU on the RBs are set to the maximum allowed by the interfaces (4074 on the 2011, and 9212 on the CCR). Jumbo frames are enabled on both switches, 9212 bytes.

MPLS MTU is set to 1600 on both RBs.

ROS 6.40.8 on the 2011
ROS 6.39.3 on the CCR1072

Is there something I’m not understanding about VPLS or may I have a misconfiguration somewhere?

Many thanks!

2

Arista and Netonix are pingable with mtu over 1488?

3

Hi,

Yes. I can ping the CCR with no-fragment on 1500 packet size, and both switches, without problems (Outside of the VPLS tunnel)

##ARISTA
[dmayan@HQ] > /ping 10.20.22.249 size=1500 do-not-fragment 
  SEQ HOST                                     SIZE TTL TIME  STATUS                                                                               
    0 10.20.22.249                             1500  61 2ms  
    1 10.20.22.249                             1500  61 1ms  
    2 10.20.22.249                             1500  61 1ms  
    3 10.20.22.249                             1500  61 1ms  
    sent=4 received=4 packet-loss=0% min-rtt=1ms avg-rtt=1ms max-rtt=2ms 

##NETONIX
[dmayan@HQ] > /ping 10.20.22.251 size=1500 do-not-fragment   
  SEQ HOST                                     SIZE TTL TIME  STATUS                                                                               
    0 10.20.22.251                             1500  61 53ms 
    1 10.20.22.251                             1500  61 3ms  
    2 10.20.22.251                             1500  61 3ms  
    3 10.20.22.251                             1500  61 1ms  
    sent=4 received=4 packet-loss=0% min-rtt=1ms avg-rtt=15ms max-rtt=53ms 

##CCR
[dmayan@HQ] > /ping 10.21.22.1 size=1500 do-not-fragment 
  SEQ HOST                                     SIZE TTL TIME  STATUS                                                                               
    0 10.21.22.1                               1500  64 1ms  
    1 10.21.22.1                               1500  64 0ms  
    2 10.21.22.1                               1500  64 0ms  
    3 10.21.22.1                               1500  64 37ms 
    sent=4 received=4 packet-loss=0% min-rtt=0ms avg-rtt=9ms max-rtt=37ms

Thanks again

4

Well, doing further checkings, we found that the Netonixs are not honoring the displayed MTU. Connecting only via the Arista, the tunnel began to work as expected.

We are getting in touch with Netonix to see how we can solve the issue.

Many thanks!

5

Hi,
did you get any feedback from Netonix?