Description of problem: On systems where eth0 is not the primary interface, the installation of a guest OS will fail due to the fact that no network driver can be found. This is caused by the fact that the xen bridge device is named based on the primary ethX device, and xenguest-install is hardcoded to use senbr0. On my system eth0 is not plugged in, and eth2 is the primary device. The xen bridge is therefore named xenbr2, and xenguest-install can't see it. Version-Release number of selected component (if applicable): [root@localhost ~]# rpm -qa | grep xen python-xeninst-0.92.0-1 kernel-xen-2.6.17-1.2614.fc6 kernel-xen-2.6.17-1.2517.fc6 xen-libs-3.0.2-29 xen-3.0.2-29 How reproducible: Always Steps to Reproduce: 1. Boot up the system with eth0 unplugged and another ehtX device plugged in 2. Try to install via xenguest-install or virt-manager 3. guest OS Installer will fail to bring up network device Actual results: guest OS Installer will fail to bring up network device Expected results: Guest OS installer should find the network device Additional info: To workaround, use can modify /etc/xen/xend-config.sxp as follows: Change the network-script line to: (network-script 'network-bridge bridge=xenbr0') Reboot.
You can pass another bridge by using '-b foo' where foo is the other bridge you want to use. Not sure if this is really auto-detectable.
But what if you are using the GUI (virt-manager) - can't pass the option with that i don't think. Thanks, Sam
virt-manager can also set the appropriate bits. The problem is that asking is almost definitely wrong (as no one is going to be able to answer this) so we really need to figure out a way to either auto-detect this or fix it so that we don't have to
python-virtinst will now automatically detect the correct primary bridge device based on the defualt route.