Chuck Cox writes:
> This sounds like it's related to what Cisco calls the "native" VLAN on a
> trunk.
>
> When you configure a switch port for 802.1q tagging (which Cisco boxes still
> call trunking) you can specify which VLANs are allowed on the trunk (default
> is
> all) and what the "native" VLAN for the trunk should be (default is 1).
That's correct. In the 802.1Q-2005 standard, that's called the PVID.
> Any untagged frames entering the switch from this trunk are assumed to belong
> to
> the "native" VLAN and will be forwarded accordingly to other ports which are
> members of that VLAN. Frames being written out that trunk port which came
> from
> the "native" VLAN are sent untagged (all others are tagged with the VLAN
> number
> that they came from).
Correct.
> I think Solaris attempts to do something similar. You can attach ce0 to a
> switch port configured for trunking, then plumb ce0, ce123000, and ce456000.
> ce0 sends untagged frames onto the trunk, which the Cisco switch would
> forward
> into whatever VLAN is set as the "native" VLAN for the trunk. ce123000 and
> ce456000 would send tagged frames for VLANs 12 3and 456 respectively.
It's similar, but not exactly the same. There's a crucial detail
here.
In the 802.1Q-2005 standard, there's one PVID that's assigned by the
administrator, and that defaults to the number 1. In a standards-
conforming implementation, you can send (by default) packets with VLAN
tags set to 2 through 4094, but you can't send tag 1 -- because that's
the PVID, and those packets go untagged.
In OpenSolaris, though, there's no PVID at all. You can plumb up
interfaces on the untagged "ce0" interface *and* all 1-4094 VLAN tags
simultaneously. That's a situation that the standard doesn't allow.
We get into this situation because we never defined an interface for
that "native VLAN" -- aka PVID.
> However, last time I tested this in a lab, it did not work as expected.
> IIRC, I
> never saw any traffic coming in on ce0 even when I sent traffic from another
> system hanging off the same switch whose port was configured as a member of
> the
> "native" VLAN...
That sounds like a separate problem unrelated to both the VLAN PPA
hack discussion and the original PVID message. I've certainly plumbed
up multiple VLANs on Cassini ('ce') interfaces and used them without
trouble. If you've run into some sort of trouble with them, you
should file a bug and have that investigated.
--
James Carlson, Solaris Networking <[EMAIL PROTECTED]>
Sun Microsystems / 35 Network Drive 71.232W Vox +1 781 442 2084
MS UBUR02-212 / Burlington MA 01803-2757 42.496N Fax +1 781 442 1677
_______________________________________________
networking-discuss mailing list
[email protected]
