You run it once, boot the frontend and then run the script again. The second time you run it it takes note of the mac address that it saw from the first boot and binds this to a fixed ip address which it will always allocate to that FE. The ip address also determines which nfsroot gets mounted.
The entire root filesystem for the frontend is stored in /nfsroot/frontend_name, so it you need different drivers, you can set them up and it will only affect the frontend. The only problem can arise when you FE will not boot an image created from the BE install. In that case only do you need to do some complicated stuff to setup the image for the FE.
Things like vdpau settings are mostly stored in the database (against the fe hostname).
If you have trouble with your router interfering, change your router so that the FE is considered to have a "static" IP. Then the routrer wont try to DHCP it. I found that before I had done this, the boot would stall part way through, but I could get it to complete by issuing a broadcast ping (ping -b 10.0.0.255 - or theboradcast address for your home subnet).
The FE will always be given the same IP if you run the script twice, so sshing should be a snap. If you run it a third time it will allow you to boot another FE (and a fourth time to fix its IP address after you boot it and the BE sees its mac address).
|
