Kickstart and NFS

[Justin Moore]

Hello,
   Due to somewhat frequent disk failures at our cluster, we're working on
a way to quickly bring a node back up and configured when a replacement
disk comes in.  The stragegy is to use RedHat's kickstart install to get
the base system installed from a CD.  The post-installation script then
attempts to mount an NFS partition and run a series of scripts from that
partition.  Everything goes well until the host attempts to mount the NFS
partition.  The mount simply doesn't work.

   The ks.cfg file statically assigns a constant temp IP address which
will be blown away when the node reboots and obtains its IP through DHCP.
The NFS server has the IP address in the /etc/exports file, and syslog on
the server indicates a successful authenticated mount request from that
IP.  The distribution is 6.2 for x86 using kernel version 2.2.14.  Looking
into it more, I find that even a mount request from the command line
fails, but silently.  Even a simple insmod of the various nfs kernel
modules (sunrpc, lockd, nfs) fails in the same manner.

   The kickstart documentation is sadly outdated and full of errors.  I
have no problem submitting updates or patches to the project, but for now
I just want my nodes to come up properly and configured by placing a
floppy and CD in it and turning it on (I'm lazy like that).  I also
realize this problem is deeper than just kickstart, but I didn't see
anything on the RedHat-related lists or message boards about kernel module
problems in 6.2/2.2.14 systems.

   Has anyone else experienced this, and if so what can I do as a
work-around?

Thank you.
-jdm

----------------------------
Justin Moore
Assistant Centurion SysAdmin
University of Virginia
jdm2d at cs.virginia.edu
----------------------------