Giuseppe Larosa Tue, 09/29/2009 - 11:58
User Badges:
  • Super Silver, 17500 points or more
  • Hall of Fame,

    Founding Member

Hello Steve,

I don't think the device has a MAC address learning function in EoMPLS enabled subif:


when we tested it years ago we generated flows with destination MAC address = interface MAC address and those frames were carried over the EoMPLS link.


This happened on a GSR with 3 GE engine 2 linecard and IOS 12.0(23)S.


I think it just checks the vlan-id and the FCS for this reason it is only a point-to-point L2 transport service.


Hope to help

Giuseppe


Steve Rodrigue Tue, 09/29/2009 - 12:02
User Badges:

So, there is no equivalent to "show mac-address-table" for Subif and/or xconnect stuff???


It might exist something... Maybe diging into the cef table or something like that???

Giuseppe Larosa Tue, 09/29/2009 - 12:12
User Badges:
  • Super Silver, 17500 points or more
  • Hall of Fame,

    Founding Member

Hello Steve,

I did some search in a C3750 ME when using port based EoMPLS you can enable mac address learning:


Starting with Cisco IOS Release 12.2(44)SE, when you configure port-based EoMPLS on a Layer 3 interface, you can enter the mac address-table learning interface interface-id global configuration command to enable MAC address learning on the port. This ensures that bidirectional traffic can occur at line rate. However, when MAC address learning is enabled, duplicate MAC addresses across VLANs are not supported. An example would be running HSRP groups across the port-based EoMPLS session.


see


http://www.cisco.com/en/US/docs/switches/metro/catalyst3750m/software/release/12.2_46_se/configuration/guide/swmpls.html#wp1184930


But you are not doing port based EoMPLS


So the scenario is different on multilayer switches.


What platform and IOS are you using?


Hope to help

Giuseppe


Giuseppe Larosa Wed, 09/30/2009 - 06:46
User Badges:
  • Super Silver, 17500 points or more
  • Hall of Fame,

    Founding Member

Steve Rodrigue Mon, 10/19/2009 - 10:01
User Badges:

Effectively, this command is not possible (C7600):

mac address-table learning interface GiX/Y


:(


Back to start...

Giuseppe Larosa Mon, 10/19/2009 - 10:31
User Badges:
  • Super Silver, 17500 points or more
  • Hall of Fame,

    Founding Member

Hello Steve,

from another point of view this lack of address learning on EoMPLS provides more scalability:

there is no risk to fill the CAM table on the L2 PE node.


the price to pay is that you cannot monitor what MAC addresses are seen on the xconnect pseudowire.


Hope to help

Giuseppe




Steve Rodrigue Mon, 10/19/2009 - 10:58
User Badges:

I understand, but that would be great to at least have some way to see those mac-addresses. For example: while testing or to confirm that we do really see a specific mac at the end of the "pseudowire".

Actions

This Discussion