We are setting up a rocks cluster. We have one compute node directly connected via a cat5 ethernet cable. Rocks is installed on our main node, and we are trying to install a compute node. The node is showing on our main node's insert-ethers (which tells us that the dhcp request is being accepted and sent back) up when we boot, but the computer node acts like it's receiving the Kickstart file, yet it remains at 0%. The asterisk is not showing in insert-ethers on the main node, therefore telling us the Kickstart file isn't being sent over properly. Any ideas or solutions as of why the Kickstart isn't working properly will help us. Thank you greatly

Dunno anything about rocks but have some exprience with FAI for auto installs as well as some network booting. Lets see
DHCP working. What options do you shove the client through dhcp?
How do you ready the kickstart file? HTTP download, tftp? Is the filename correct? Do you need dns resolution on the node and do you shove it over with dhcp?
Did you test the method of providing the kickstart file from a "normal" client. Like can he download.

Is the kickstart file in the correct format? Does the client notice that?

@zhjim

I'm just trying to download it from the head node. It is being transferred over http. This is the installation process we are going through - http://www.rocksclusters.org/rocks-d...ute-nodes.html

We are not receiving the asterisk in parentheses in the insert ethers, which means that the Kickstart file isn't being downloaded yet. It acts like it is downloading the Kickstart file on the compute node, but it stays at zero percent. It tries about 10 times then gives up