future of fai

[Bruce Edge]

1st of all, great news!! 

My $.02 inline

[snip]

> I also think it's very difficult to merge all patches for
> setup_harddisks, that I received. So, some enhancements that are only
> requested by a limited number of users (almost exactly one) will not
> be applied. If I'm wrong, please tell me. Are uuid and labeled mount
> really so important ?

Perhaps post a list of submitted patches with a one line description and we can vote on them?

- should we switch the installkernel to devfs ? What parts are
  affected ? Can both be used at the same time ?

No, please. 
I'm working on getting FAI installs working with the veritas volume manager, which wraps the root with a virtualization driver, which will not support devfs.

- write a script that creates a bootable CD. I do not know if it's
  possible to detect the name of the CD-ROM without using a
  initrd. What if the computer has (several) IDE and SCSI CD-ROMs? The
  CD should be made from an existing nfsroot on a fai server. I think
  Niall Young will work on this.

I am working on this too now. I'm using Marc Scharfer's http://search.alphanet.ch/~schaefer/FAI/
as a starting point. Let me know if there's anything I can do here, coding, testing, whatever.

I think going to an initrd based FAI kernel would be a plus. You can support more devices in the kernel with an initrd. For example, the e1000 gigabit NIC driver only builds as a module, you have to load it from an initrd.
Requiring kernels to fit on a floppy isn't as important as it used to be. One can boot from CD now, as well as TFTP load everything using grub (for suported NICs), which is what we do.

- write a little success story of your experience using fai or answer
  the questionnaire

As soon as I get a chance to come up for air.

-Bruce