Hi all. We've been using FAI at the MIT Computer Science & Artificial Intelligence Lab to install servers and workstations since the early part of this year, and for the most part, it is very nice.
Because we support wildly varying hardware configurations, I'd like to do something to allow FAI to be slightly more flexible when it comes to figuring out which disks to install onto. The problem, as far as we're concerned, is that FAI basically seems to expect that we know in advance which disks will be partitioned and used as the installation target. We don't have that luxury. I've thought of defining classes for the most commonly used disks (e.g. the class SDA would be defined if /dev/sda existed, and a disk_config/SDA file would specify the partition layout for that disk). That seems too kludgy for my taste, and really isn't any more flexible than the existing setup. The docs say that I should use a hook to configure the disks and partitions if the existing disk configuration code is not usable. I think the right thing to do is to write a hook to synthesize a disk_config definition based on which disks are discovered. It would write this disk_config file out to e.g. /tmp/fai/disk_config/<hostname>, and then call setup_harddisks with '-c /tmp/fai/disk_config'. Does that seem like the right solution? Have people solved this problem in better or different ways? noah -- Noah Meyerhans System Administrator MIT Computer Science and Artificial Intelligence Laboratory
signature.asc
Description: Digital signature
