| Step-by-Step Installation
This page is intended to be a quick cook book approach to getting the installation services started. Other help screens have more detail on starting/testing individual services
- Begin on your server, assuming you have installed a RedHat gnome or kde workstation install
(other installations probably work too, but we haven't tested all of them)
- install the inetd or xinetd (RedHat 7.0 and later) RPM (e.g rpm -Uvh inetd*rpm)
- install the nfs rpm (nfs-utils), create the /etc/export entries for LUI, and start nfs (refer to the "Installing LUI and required services" link from the main LUI page)
- download LUI, install it using rpm -ivh {rpm file name} (LUI will install in /usr/local/lui-{version}, by default).
- create the /tftpboot directory (if it does not exist)
- if you are not using diskettes for the clients:
- install the dhcp rpm, create an /etc/dhcpd.conf file (follow sample provided with LUI) and start dhcpd
- install and start tftp-hpa if not using xinetd (refer to the link "Installing LUI and required services" from the main LUI page)
- extract pxelinux from the syslinux-1.48 or later tar (refer to link from the main LUI page)
- copy pxelinux.bin (from the syslinux tar) to /tftpboot/pxelinux.bin
- copy a network boot kernel, bzImage, to /tftpboot/bzImage. LUI provides a tool, mkluikernel, to build a suitable kernel for most users, it is in the samples directory. If you find the kernel created is not acceptable, create your own using the .config file distributed with LUI as a starting point.
- if you are using diskette boot for the clients:
- get the bootp rpm and install it (from RedHat 5.2 or earlier)
- get the etherboot package (refer to link) and follow the instructions for creating boot diskettes
- copy a tagged kernel to /tftpboot/vmlinuz. You can use the mkluikernel tool in the LUI samples directory to generate a bzImage, and then use mknbi to created a tagged kernel.
- If you are doing an rpm install of your nodes, copy all the rpms you require for your nodes to /tftpboot/rpm.
(e.g. rm /tftpboot/rpm/*486* /tftpboot/rpm/*586* /tftpboot/rpm/*686* )
Note that if there are multiple rpm files for the same rpm on different architectures, you will either need to specify the architecture portion of the filename in the rpm list or remove those files that aren't for your desired architecture from the rpm directory.
(e.g. mkdir /tftpboot/rpm;cp /mnt/cdrom/RedHat/RPMS/* /tftpboot/rpm)
- You are now ready to start running LUI commands to define servers, clients and resources. The most simple way to define your machines and resources is to try the graphical user interface to LUI, aka, 'glui'. Just enter 'glui' from the command line, and it's pretty straightforward from there. Remember to define a single server machine, one or more client machines, and a set of installation resources. A typical minimal installation will have a file system resource for root (/), a file system resource for /boot, a set of rpms in the rpm resource, usually a source resource for /etc/shadow, and a disk resource that describes the layout of the disk partition table. Define resource & machine groups to make the job easier.
There are sample scripts provided with LUI that show examples of the command line interface. Use:
- sample.script1 if you are doing a tarball install
- sample.script2 if you are doing an rpm install
- sample.script3 if you are doing an rpm install with NFS mounted filesystems
- The most tricky part of using LUI is getting the disktable right. Refer to the README.disk file, or the 3 sample disktables (sample.disktable1,2,3) for examples
|