Hi All,
This has to be something simple....but it's really busting my chops. We have a PXE boot server that is used for initial installation of a number of operating systems and it works well. However the CentOS 5.1 x86_64 install is seriously broken.
We've made the PXE boot images available from "centos/5/os/x86_64/images/pxeboot/" in the tftp boot etc. We've rsynced our mirror with a local Tier 1 mirror (rsync://ftp.monash.edu.au::ftp/linux/CentOS/)
Next we fire off the installation using the following in /tftpboot/pxelinux.cfg/default:
label 4 kernel centos5/x86_64/vmlinuz append initrd=centos5/x86_64/initrd.img ramdisk_size=7000 ks=http://<insternal_install_server>/CentOS-5/dot/kickstart/ks.cfg.x86_64 ksdevice=eth0 language=en text
The installer starts, loads the kickstart script (attached), successfully verifies the installation media, checks the dependencies for the packages to be installed, formats the hard drive(s), then attempts to download the package: sysklogd-1.4.1-39.2.x86_64.rpm
At this point the installation will not continue because that package is not part of the installation! I've verified our internal mirror with 3 other Tier 1 rsync hosts and all agree, our images are up-to-date and the CORRECT sysklogd package is actually: sysklogd-1.4.1-40.el5.x86_64.rpm (Note the build is 40.el5, NOT 39.2).
I've tried creating a symbolic link from sysklogd-1.4.1-40.el5.x86_64.rpm -> sysklogd-1.4.1-39.2.x86_64.rpm, and sourcing a real package for sysklogd-1.4.1-39.2.x86_64.rpm. These last two attempts have still resulted in the installer claiming the RPM is corrupt, not available etc.
This is killing us :( The odd thing is the
James