Right. If I understand correctly, and I'm not sure that I do, the segment
between the Cisco and the Linux box is still a full Ethernet repeater
domain and could have something like a hub between them with other devices
hanging off that. Packets still need to reach that repeater domain and be
able to enter and exit that repeater domain. One of the devices connected
to it (either the Cisco or the Linux box) needs to be active to provide
that connectivity. Thus, it's quite conceivable, depending on how the
priorities are set up, that the Linux box is active while the Cisco is
not.

With some ASCII art:

A-Linux1-B ---DomX--- A-Cisco-B ---DomY--- A-Linux2-B -|
|                                                      |
--------------------DomZ--------------------------------

Where "A-Linux1-B" means the box Linux1, with ports A and B, with port A
connected to repeater domain Z and B connected to repeater domain X.

While there aren't any other boxes shown here, you have to consider that
each of the repepater domains, X, Y, and Z have other nodes attached in
them to repeaters. Thus, this could lead to the following result:

Linux1 is the root bridge
Linux1-A is forwarding
Linux1-B is forwarding
Cisco-A is not forwarding
Cisco-B is not forwarding
Linux2-A is forwarding
Linux2-B is forwarding

A node on DomX can thus communicate to a node on DomY thusly:
DomX -> Linux1 -> DomZ -> Linux2 -> DomY

I think Troopy expects Linux1-A to be non-forwarding, but this would cut
off connectivity to DomY.

-- Dave

________________________________

        From: [EMAIL PROTECTED]
[mailto:[EMAIL PROTECTED] On Behalf Of Nick Davey
        Sent: Wednesday, December 05, 2007 10:42 AM
        To: [EMAIL PROTECTED]
        Cc: Vyatta-Users
        Subject: Re: [Vyatta-users] Spanning-tree


        I think I understood you. In a layer 2 triangle the root ports
will all be forwarding as it's the STP root. Both devices will be
forwarding on the port connected to the root, as they are root ports. This
leaves two ports in question, the ports connecting the linux machine to
the Cisco router. If you do a show spanning-tree for that interface on the
Cisco router you should see a blocked port on the interface connecting it
to the Linux box . On the Linux box the port connected to the router will
be the designated port for that segment, and will be in forwarding mode. I
have a Cisco 2811 and some 2950s, I'll set them up shortly and send in the
output, but this conforms to what I'm seeing in STP docs I referenced.
This should not be a loop regardless as the port on the Cisco router is
blocking, and by it's very nature not forwarding traffic. Is this actually
creating a loop?


        On Dec 5, 2007 11:09 AM, Troopy . <[EMAIL PROTECTED]> wrote:



                HEllo

                Thanks for your answer but i think you didn't understand
me.

                If you have a layer 2 triangle:
                2 (non-root stp) machines (1 Cisco + 1 Linux)have 1
blocking and 1 forwarding ports
                1 (root stp) machine (Linux) has 2 forwarding ports.
                You can try it by yourself, the case study is very simple,
you just
                 need one cisco router and 2 linux.


                The thing is on the non-root Linux machine, there are 2
forwarding ports which is NOT normal. This creates a loop.
                I think something special has to be changed on Linux ...

                Regards

                TRoopy






                ---------- Original Message
----------------------------------
                From: "Nick Davey" <[EMAIL PROTECTED]>
                Date:  Wed, 5 Dec 2007 10:01:40 -0500

                >>From what I've seen both ends of the redundant link
won't be blocking, one
                >will be forwarding on one end, and one will be blocking.
If you refer to the
                >original spanning tree diagram drawing in this article:
                > http://en.wikipedia.org/wiki/Spanning_tree_protocol
<http://en.wikipedia.org/wiki/Spanning_tree_protocol>  diagram 5 you'll
see one
                >end of the redundant link is set to designated
forwarding, and one is
                >blocking. I believe this is what you're describing? I
think this is due to
                >spanning tree's requirement that one port on a segment
must be chosen for a
                >forwarding, or designated role. Hope that helps.
                >
                >Nick
                >
                >On Dec 4, 2007 12:32 PM, Troopy . < [EMAIL PROTECTED]
<mailto:[EMAIL PROTECTED]> > wrote:
                >
                >>
                >> Hello,
                >>
                >> I tried to test the spanning tree on Vyatta.
                >>
                >> There is one problem with Vyatta and more precisely the
Linux plateform.
                >>
                >> We made a triangle with two Linux/Vyatta router and a
Cisco router.
                >> One of the Linux is set as the  root STP.
                >>
                >> The Cisco router has a correct behaviour because it has

                >>  a blocking and forwarding port
                >> The root Linux stp has a correct behaviour because it
has
                >>  two forwarding ports
                >> The problem is on the non-root stp Linux. It has
                >>  2 forwardind ports instead of 1 blocking and one
fowarding port.
                >>
                >> I think something has to be changed on the Linux level
but what?
                >>
                >> Regards
                >>
                >> TRoopy
                >>
                >>
                >>
                >>
                >>
                >> ______________________________________________________
                >> Désirez vous une adresse éléctronique @suisse.com?
                >> Visitez la Suisse virtuelle sur http://www.suisse.com
<http://www.suisse.com>
                >>
                >> _______________________________________________
                >> Vyatta-users mailing list
                >> Vyatta-users@mailman.vyatta.com
                >> http://mailman.vyatta.com/mailman/listinfo/vyatta-users
                >>
                >
                >



                ______________________________________________________
                Désirez vous une adresse éléctronique @suisse.com?
                Visitez la Suisse virtuelle sur http://www.suisse.com





_______________________________________________
Vyatta-users mailing list
Vyatta-users@mailman.vyatta.com
http://mailman.vyatta.com/mailman/listinfo/vyatta-users

Reply via email to