For anyone using Xen, or seeking to, a bit of advice. Do not use the Xen
scripts to create the bridge if you go that route. You can do it, and
for many it will work fine. But the scripts are really nasty, and it
will not get rid of the bridge if you stop xen. Thus it does not clean
up after itself, which is bad. Also the xen scripts to create the bridge
can be problematic

At the recommendation of others, after which I myself also HIGHLY
recommend, use your distro's network management to create the bridge as
you normally would, if you wanted a bridge device. Then you can let xen
add devices to the bridge.

In /etc/xen/xend-config.sxp you will want to comment out the following
line that creates the bridge

#(network-script network-bridge)

This one is ok to keep and you will need for domU interfaces to be added
to the bridge.
(vif-script vif-bridge)

Xen tends to name the bridge as xenbr0, with the last being the number
of the bridge. If you have more than one, then it will be xenbr0,
xenbr1, xenbr2, etc. Though pretty sure thats specified in the domU
scripts, so naming it might be moot. Providing what ever the bridge is
called by your distro, you reference that in the domU config.

Here is an example for Gentoo, also using bonded interfaces in the
bridge. Very sexy and minimal config IMHO :)

config_eth0="null"
config_eth1="null"
config_eth2="null"

slaves_bond0="eth0 eth1"

config_bond0="null"
bridge_xenbr0="bond0"
config_xenbr0=( "192.168.0.10/24"
                "192.168.0.81/24"
)
routes_xenbr0="default gw 192.168.0.1"
brctl_xenbr0=(
        "setfd 0"
        "sethello 0"
        "stp off"
)

Then in a domU I do

vif = [ "mac=00:16:3E:00:00:00, bridge=xenbr0" ]

(not a real mac, but valid manufacturer part)

-- 
William L. Thomson Jr.
Obsidian-Studios, Inc.
http://www.obsidian-studios.com


---------------------------------------------------------------------
Archive      http://marc.info/?l=jaxlug-list&r=1&w=2
RSS Feed     http://www.mail-archive.com/[email protected]/maillist.xml
Unsubscribe  [email protected]

Reply via email to