I seem to have run into a bug with the new –bridgeslaves= option. It would seem that if I tell the bridge device to use a virtual interface (like bond0) rather than a physical interface (em1/em2) that kickstart completely barfs on it. I have provided my network section below which works fine as long as i don’t enable all the bridge content.
When the installation starts off the creation of bond0 and all VLANs on bond0 go without issue. Jumping into TTY2 I can type ‘ip a’ and see all the bond0 device and bond0.VLANID devices no problem.
When I enable bridge content with –bridgeslaves=bond0 or –bridgeslaves=bond0.VLANID anaconda barfs
When I enable bridge content with –bridgeslaves=em1 or –bridgeslaves=em2 anaconda doesn’t have an issue.
Is this expected behaviour? I really don’t want to have to manually manipulate the ifcfg-bond* interfaces and create the ifcfg-BRIDGE* interfaces manually so I’d like this to work.
# Configure a bond over the slaves in failover mode network –device=bond0 –noipv6 –bootproto=dhcp –onboot=yes –bondslaves=em1,em2 –bondopts=mode