XEN network-bridge issues

I have three network interfaces on my server. My server runs multiple XEN fully virtualized machines. I was using only the eth0 interface for the XEN machines by using the network-bridge script - specifically I did ./network-bridge start eth0 and everything worked beautifully, the script I guess creates a peth0 that is used by the XEN machines? Someone can fill me in on those details maybe?

A few nights ago there was an extended power outage and the server did a hard shutdown. The bridge for XEN was not properly stopped because of the crash, so when I went to start it again it told me that peth0 already exists. I tried ./network-bridge stop eth0 and then tried starting it again but the ghost peth0 exists and is preventing the script from working. As a work around I am using eth1, but need to determine how to clean up the half-baked bridge that exists on eth0.

Just a thought: try ./network-bridge stop
without the reference to the interface. That got rid of peth0 on my system. Does ./network-bridge status not show peth0?

On a similar issue I have no network access from inside the virtual machine. The virtual machine config refers to:
vif [mac=“xx:xx:xx:xx:xx:xx”,bridge=pan0,model=rtl8139]

It uses pan0 for the network bridge. Is this the same as peth0? Does it link to eth0 in the host or to nothing? ./network-bridge status shows pan0, not peth0.
This is openSUSE 11.1 with associated XEN.
In XP or Win2k3server there is only one device shown that is not working properly and that is the VMDP device that requires proprietary drivers from Novell.

Thanks
Aaron

Hey,
If you haven’t found it elsewhere, I have an idea of what was happening with your peth0 issue.

network-bridge renames your physical eth0 to peth0 then creates the bridge as eth0. After your crash it must have kept the peth0 as a reference to the physical device.

I’m not that clued up on linux, so I’m not sure how to fix it if it doesn’t show up in the network settings gui.

For those that follow…
For my problem, it was rather simple. The vm uses eth0 as the device, not xen-br0 or whatever as most guides seem to tell you. If you open the /etc/xen/scripts/network-bridge script in your favourite editor, it has lots of comments to tell you how it works.
By default there was no network script run, so it had to be changed to:
(network-script network-bridge)
With this changed, I edited the VM info file e.g. /etc/xen/vm/WinXP and put bridge=eth0 in the network line. Renewed the VM in the manager and voila!
Renew by # xm delete WinXP … then # xm new /etc/xen/vm/WinXP