Harry Smith wrote:
When this runs it accesses the CD drive but apparently doesn't read the ks.cfg file
> since it starts to prompt for setup parameters.
Hmm. Any unusual messages on the consoles? If you proceed and go to
prompt on Alt-F2, can you find your ks.cfg? Does it appear intact?
Check /proc/cmdline; is your ks argument present?
Try putting some commands in %pre that echo to one of the consoles. See
if they appear.
echo ' Fixing LILO...' >> /dev/console
rpm -i /mnt/source/gravRPMS/liloconf-1.0-grav1.noarch.rpm
# Install JRE & gravmsg last due to
# JRE RPM missing a package order...
echo 'Installing Java Runtime Environment...' >> /dev/console
rpm -i /mnt/source/gravRPMS/jre-1.3.1_02.i386.rpm
I get open failures on the rpm files (File not found). I know the installer can read
> the CD since it installs linux onto the platform. I have verified
that the files are
> there. I used the Alt-F2 window to verify that the /mnt/source is
valid and the files
> are there. Does the installer use a different mount point? Could
this problem be
> related to the first??
You didn't show us the %post line. You did use the --nochroot argument,
didn't you? Otherwise you will be chrooted into /mnt/sysimage.
The last time I saw this behavior was because of carriage returns.
Python deals with them properly when it parses the kickstart directives
and packages, but it passes the %pre an %post sections verbatim to the
shell, which *does not* treat carriage returns as whitespace. In your
case that would cause rpm to look for files like
/mnt/source/.../blah.rpm^M
which aren't present.
So check your ks.cfg for carriage returns with a command like
% od -t x1 ks.cfg |grep '0d'
