This document describes how to install the operating system on a cluster worker node. The following steps must be completed before attempting to clone. The boot order in BIOS must be set correctly. See configuring BIOS. The hard disks must already be partitioned. See partitioning hard disks. The node must be entered into dhcpd.conf correctly (the correct MAC addr must correspond with the hostname/slave-number you're about to attempt installing), see DHCP configuration. You should now change directories on medusa to /tftpboot/pxelinux.cfg. In this directory you'll find a script called pxesimlink (?). To make the magic of the network based kickstart install work, you need to run this command, giving as arguments the starting AND ending node numbers you want to install AS well as the config file (located in the same directory) that will install the OS. You would type "./pxesimlink 305 305 default.netinstall" to set up the server to allow node 305 to install as instructed by the default.netinstall config. See Kickstart doc's for more info. What you should see after you've executed the script, is some number of symlinks, named something link 81xxxxxx (which is the IP addr of the node, converted to hex and minus the "."'s that separate the octets of and IP addr) that link to the file default.netinstall. If all of the preceding conditions are met, you should be able to power on the node you want to "clone" and watch the automated install progress. No keyboard or video are required for the autoinstall to complete, but you can safely watch if you like. If you're interested, attach a monitor and you should see the Fedora RPM transaction be created, packages install, post-install script run. You will see the node reboot a few times, after the last of which you should see the console sitting at a login prompt. If there were any errors detected during the install, you should fine a log file at /tmp/install-error.log (?), you'll be notified on the console of the existence of that file. A brief overview of the process is: symlink made on master node slave node attempts pxe-boot receives a boot-image via dhcp/tftp slave node installs via kickstart node finishes install, deletes it's own symlink node reboots and voila! Please refer to the slave-node test checklist to see how we test the install to validate this installation.