Re: [Xen-users] Xen host guest bridging transparency issue

[Simon Hobson <linux@xxxxxxxxxxxxxxxx>]

Adam Goryachev <mailinglists@xxxxxxxxxxxxxxxxxxxxxx> wrote:

> What interface is p6p1 ? Seems to be a strange name for a network interface, 
> these are normally eth0 or similar by default...

Welcome to the world according to Poettering (yes, he of SystemD 'fame'). 
Apparently it's far far too complicated to change Udev rules if an interface is 
changed, instead we have to suffer interface names based on physical location - 
so some of them can be quite long, and just moving an interface (eg to a 
different slot or USB socket) will change it's identity.

I suspect some people have never worked on servers - where interface names 
appear in many places (manual network config, firewall, scripts such as link 
state and/or traffic monitoring, ...

> My guess, the dom0 is doing NAT on the incoming traffic, and sending it over 
> xenbr1 or something like that...

That would be my guess as well.

One other thing for the OP - use of the network script to configure the bridges 
is deprecated. Current advice is (and has been for quite some time) to set 
network-script to "dummy" and configured the bridges in the host OS. Debian, 
and it's derivatives, has had native support for bridge configuration in 
/etc/network/interfaces for some time now. You just need a stanza like :

auto ethext
iface ethext inet static
        bridge_ports pethext
        address a.b.c.d
        netmask 255.255.255.0
        gateway a.b.c.z
        bridge_stp off
        bridge_waitport 0
        bridge_fd 0

(note - I use Udev rules to rename the physical interfaces to things meaningful 
like "pethext" for the outside interface and so on)

This stanza will bring up a bridge named ethext, bind the interface pethext to 
it, and configure the specified address to the bridge. The last three lines 
disable STP and set the forwarding time to 0 - so there's no delay between 
adding an interface or bringing up a link and traffic flowing through it.


_______________________________________________
Xen-users mailing list
Xen-users@xxxxxxxxxxxxx
http://lists.xen.org/xen-users