[j-nsp] Another SRX240 VPLS question - "trunking" multiple VLANs through single VPLS

[Tom Storey]

Hi all. Sorry for the noise on this topic, but Im getting my feet very
wet right now. :-)

Im passing on the "access port" idea from my previous email at the moment.

Right now Im trying to get a different configuration working, whereby
I assign multiple units of one interface in to a VPLS routing instance
and allow them to be trunked to other VPLS sites.

I had a previous configuration working fine whereby the whole
interface itself was assigned to the VPLS. Trunking worked great in
that instance, I could pass as many VLANs through as I wanted
seemingly.

Heres what I was doing:

interfaces {
   ge-0/0/12 {
       description "L2VPN test interface";
       encapsulation ethernet-vpls;
       unit 0 {
           family vpls;
       }
   }
}
routing-instances {
   VPLS-1 {
       instance-type vpls;
       interface ge-0/0/12.0;
       route-distinguisher 12345:2;
       vrf-target {
           import target:12345:2;
           export target:12345:2;
       }
       protocols {
           vpls {
               no-tunnel-services;
               site CORE {
                   site-identifier 1;
               }
               vpls-id 1;
           }
       }
   }
}

This config works fine.

Now what Im trying to do is, in order to allow VLANs to be aggregated
via one interface of the SRX and assign them at will to various L3VPN
and VPLS instances, as follows:

interfaces {
    ge-0/0/5 {
        description "Aggregation interface";
        vlan-tagging;
        mtu 1618;
        encapsulation flexible-ethernet-services;
        unit 10 {
            encapsulation vlan-vpls;
            vlan-id 10;
        }
        unit 30 {
            encapsulation vlan-vpls;
            vlan-id 30;
        }
        unit 40 {
            encapsulation vlan-vpls;
            vlan-id 40;
        }
        unit 50 {
            encapsulation vlan-vpls;
            vlan-id 50;
        }
        unit 60 {
            encapsulation vlan-vpls;
            vlan-id 60;
        }
    }
}
routing-instances {
    VPLS-1 {
        instance-type vpls;
        vlan-id all;
        interface ge-0/0/5.50;
        interface ge-0/0/5.60;
        route-distinguisher 12345:1;
        vrf-target {
            import target:12345:1;
            export target:12345:1;
        }
        protocols {
            vpls {
                no-tunnel-services;
                site CORE {
                    site-identifier 1;
                }
                vpls-id 1;
            }
        }
    }
}

Ive also tried removing "vlan-id all", and also replacing it with
something like "vlan-id 4000" for what I believe is referred to as
normalisation.

The problem is that, when I only have one logical interface assigned
to the VPLS, it works great. As soon as I add a second or more, it
just seems to flop.

With a single logical interface, if I run the command "show route
forwarding-table family vpls" I see a nice big list of MAC addresses
as I would expect. When I add the second+ logical ints, after a few
minutes (probably mac table aging) they all seem to disappear.

Everything Ive tried configuring to date is based on what examples I
can find online. Now, a lot of that is geared towards the bigger boys
toys routers like the M/MX series. Am I trying to do something that
the SRX series simply cant do?

Im trying my hardest to work this out on my own, but I would again be
greatly appreciative if anyone has any tips or pointers. I think Ive
been through just about every forum post, blog, and random note I can
find on this topic, I just cant seem to get it working.

Thanks!
Tom