From: Jian Gu (guxiaojian@gmail.com)
Date: Wed Mar 25 2009 - 19:42:15 ART
You mean Junipers are VPLS PEs, a core router does not see VPLS MACs ...
Anyhow, I don't get the problem, if CE is proxying ARP, it should give its
own MAC, how can a VPLS PE learns the same MAC from different port if each
CE has unique MAC?, VPLS PE sees duplicate MACs mostly when spanning tree
BPDUs are not tunnelled across MPLS core and you have a L2 loop.
On Wed, Mar 25, 2009 at 12:39 PM, Matt Bentley <mattdbentley@gmail.com>wrote:
> Hi GS:
>
>
>
> Quick question for you all. We are seeing some interesting behavior
> regarding proxy-arp and multipoint VPLS. We have many customers we provide
> VPLS services to.
>
>
> Our Access Rings are L2 circuits provided by a variety of carriers
> terminated by either Nortel/Tasman or RAD pairs on either end
>
> Our distribution rings are Extreme Switches
>
> Our core consists of Junipers
>
>
> For multipoint VPLS, we are seeing some occasions where one CE will go
> down,
> and the other CE devices in that VPLS will proxy for the switch that has
> gone down. As soon as our Junipers see that a MAC is being answered for
> out
> of two separate ports , they generate duplicate MAC errors
>
> -I would think this is expected behavior for Junipers??
>
> How are proxy-arp requests different at the structural level from normal
> ARP
> requests? If there were some flag we could filter proxy-arps with at our
> edge for, it would prevent this problem.
>
> Aside from having customers disable proxy-arp, can anyone think of a viable
> workaround?
>
> Thanks,
>
>
> Blogs and organic groups at http://www.ccie.net
>
> _______________________________________________________________________
> Subscription information may be found at:
> http://www.groupstudy.com/list/CCIELab.html
Blogs and organic groups at http://www.ccie.net
This archive was generated by hypermail 2.1.4 : Mon Apr 06 2009 - 06:44:07 ART