RHEL3 NFS based kickstart dies with 'install exited abnormally -- received signal 11'

[Oktay Altunergil]

Hello,

I am having a problem with kickstarting using NFS on Enterprise 3.0 
(update 1). I will try to describe the problem in detail.

Our setup uses NFS for both the kickstart config file and the rpm 
packages. We have /usr/local/export exported with ro,no_root_squash. 
This directory contains the directories 'config' and 'redhat'. Our 
kickstart configuration is in config/ks.cfg. Inside the redhat directory 
we have everything in the binary CDs copied so it looks like 
/usr/local/export/redhat/RedHat.

What's happening is, everything is working fine and the ks.cfg file is 
being loaded without a problem. This is an indication that there's 
nothing wrong with mounting NFS since it's the same export as the 
'redhat' directory. After this when it comes to mounting the rpms 
directory, i see in the logs that it's trying to NFS mount the correct 
IP and correct exported directory. Before this has a chance to complete 
though, I'm getting 'installed exited abnormally -- received signal 11' 
on the main install screen.  Interestingly, after the signal 11, the box 
keeps requesting authentication to mount the redhat directory for a few 
more minutes as seen on the server logs.

I've done a few things to rule out certain possibilities. First of all, 
we tried to do a CD install, getting the packages from NFS using the 
same directory. This works without a problem. If we switch from NFS to 
FTP or HTTP during kickstart this works fine too so there is no problem 
with permissions or the way the directories are structured. While doing 
a kickstart install via HTTP/FTP I can also mount and access the redhat 
directory manually when the shell comes up on tty2.

I would appreciate any help on this. I am including some relevant config 
information below.

Thank you.

Oktay Altunergil

Here's what our exports file looks like: 
/usr/local/export 10.2.2.0/255.255.255.0(ro,no_root_squash) 
 
And the directory structure: 
 
[root at RHESkickstart redhat]# pwd 
/usr/local/export/redhat 
[root at RHESkickstart redhat]# ls 
RedHat 
[root at RHESkickstart redhat]# 
 
And finally the ks.cfg file: 
 
#Generated by Kickstart Configurator 
lang en_US 
langsupport en_US 
keyboard us 
mouse none 
skipx 
network --device=eth1 --bootproto=dhcp 
rootpw --iscrypted $1$OXVsBg$2oB3379kaAWGzLCHGm7xJ. 
timezone --utc America/New_York 
#reboot 
text 
install 
nfs --server=10.2.2.1 --dir=/usr/local/export/redhat 
#url --url http://10.2.2.1/redhat 
## the above works. but we need NFS 
# I know this isn't a scsi module, but I'm cheating 
#device scsi ext3 
bootloader 
clearpart --all --initlabel 
part / --fstype ext3 --size 1 --grow --ondisk sda 
part  swap --size 1024 --ondisk sda 
auth  --useshadow  --enablemd5 
firewall --disabled 
#Do not configure the X Window System 
skipx 
%packages 
@Base 
@Mail Server 
@Text-based Internet 
#@SQL Database Server 
@Web Server 
@DNS Name Server 
@Editors 
@Emacs 
@Development Tools 
@Legacy Software Development 
@Kernel Development 
#@Unix Ops 
                                                                                

%post 
# add admin user 
/usr/sbin/useradd admin 
chfn -f 'Admin user' admin 
/usr/sbin/usermod -p '$1$OXVsBg$2oB3379kaAWGzLCHGm7xJ.' admin 
# copy over the config 
/etc/init.d/portmap start 
mount 10.2.2.1:/usr/local/export /mnt/ 
cp /mnt/config/network /etc/sysconfig/network 
cp /mnt/config/ifcfg-* /etc/sysconfig/network-scripts/ 
cp /mnt/config/resolv.conf /etc/ 
cp /mnt/config/hosts /etc/ 
echo "Protocol 2" >> /etc/ssh/sshd_config 
echo "41 */6 * * * /usr/sbin/ntpdate -s ntp.datapipe.net" >> /etc/crontab 
echo "SV:123456:respawn:/command/svscanboot" >>/etc/inittab 
# remove things that I don't want started up 
chkconfig --del nfslock 
chkconfig --del portmap 
chkconfig --del autofs 
chkconfig --del rhnsd 
chkconfig --del kudzu 
umount /mnt