GNOME Bugzilla – Bug 756547
System libvirt imported VM fails to start due to net config
Last modified: 2016-09-20 08:15:55 UTC
Simple vm created using defaults in virt-manager on qemu:///system Gnome Boxes offers to import system brokered vm. "Failed to start" error message pops up. Terminal outputs: Unable to start domain: Network not found: no network with matching name 'default' This is the same when the vm is already running.
Disregard that last sentence. The VM being run in that case is the system instance. I was unaware that boxes actually clones the system VMs to a local path instead of utilising the existing ones.
I was sure I fixed this already. Maybe in my dreams. :) I'll look into it..
This patch I just pushed should fix your main issue but it won't fix the issue of bridge networking not available on your machine. i-e expect networking to be very slow. If you don't see a network by the name 'virbr0' with ifconfig, it's likely a bug in your distro packaging. On Fedora you need libvirt-daemon-driver-network package, which AFAIK is installed as part of Boxes deps. commit: 30c2d638f8dc6605a6fc958ba1b605cdc03b2d0c vm-configurator: Always update network config Instead of only updating network config for existing domains if bridge networking is available and existing networking in the box is not already bridge network, let's always do so. This fixes the issue of VM not starting if bridge networking is no longer available (this has been known to happen due to setup/packaging issues) and existing network is set to either bridge or some other non-working configuration (e.g 'default' when importing from system libvirt).