[CentOS] Kickstart stopped working

Mon Jun 18 19:20:18 UTC 2007
Alfred von Campe <alfred at von-campe.com>

I've been using kickstart for a while now, but today I tried to  
kickstart a system and it's just not working at all.  I can't figure  
out why it's hanging.

I booted from the CentOS 4.5 CD #1 and typed "linux ks=http:// 
centosmirror/ks/sys20.cfg".  If I check my web server's log, I can  
see that sys20.cfg was accessed.  This file begins with the following  
lines:

   install
   url --url http://centosmirror/

Looking at the text in alternate console 3 (or was it 4), I can see  
that it got an IP address via DHCP, got the config file mentioned  
above, set the hostname as specified in the config file, and then  
ends with the following messages:

   trying to mount CD device hdc
   mntloop loop0 on /mnt/runtime as /mnt/source/CentOS/base/ 
stage2.img fd is 21
   transferring http://centosmirror//./CentOS/base/product.img to a fd
   transferring http://centosmirror//./disc1/CentOS/base/product.img  
to a fd

A long time (5 minutes?) passed between the last two lines, and then  
the entire things repeats (with some additional output lines that I  
didn't bother to write down).

First, I don't have a product.img file in my base directory.  In a  
recent thread on this list about doing an ftp install, I remember  
reading that copying stage2.img to product.img helped, so I created a  
hard link between the two.  Second, looking at the webserver logs,  
there is no mention of product.img or stage2.img.  Finally, there is  
no alternate console #2, so I can't get a shell prompt to poke around  
a little more.

I've done dozens of kickstarts in the past using this server, but I  
don't remember if I've successfully done a kickstart install since  
4.5 came out.  Are there some known issues with kickstart in 4.5?  I  
rsync my local repo every night from one of the public mirrors.

Alfred