PXE KickStart

PXE + Kickstart, Automating CentOS/RedHat Installs

Red Hat allows you to install the OS over the network using NFS, FTP or HTTP. If the hardware supports PXE (Pre-eXecution Environment) the NIC card will send out a broadcast request for DHCP information. The DHCP server provides the client with a IP address and other network infomation such as the TFTP server address (which provides the files necessary to start the installation) and the location of the files on the TFTP server. This is possible because of PXELINUX, which is part of the syslinux package.

Client BIOS (DHCP broadcast)-----> DHCP Server (Network info, tftp server and PXE file name)
Client BIOS (TFTP request for pxelinux.0)---> TFTP/PXE Server (supplies the pxelinux.0 file)
Running PXE (pxe request for kernel and Kickstart install)--------> TFTP/PXE Server (C0A80028 or default file supplies info)
Running Kernel (NFS KICKSTART)----------> NFS Server (supplies the kickstart config file)
Running Anaconda (NFS request for rpms)------------> NFS Server (supply RPMS)
Running Anaconda (NFS request post installtion)----> NFS Server (runs post install scripts)
Running Anaconda (reboot) 

The PXE boot process
• NIC requests DHCP information
• DHCP server provides bootloader name and IP of tftp server
• NIC uses tftp to fetch bootloader into RAM
• BIOS executes bootloader
• Bootloader uses tftp to find and retrieve configuration file
• Bootloader follows directives in file

n order to setup a PXE installation the following must be carried out:

• Install the necessary packages
• Configure the network (NFS, FTP, HTTP) server to export the installation tree (redhat packages, etc)
• Configure the DHCP server
• Configure the files on the tftp server necessary for PXE booting
• Configure which hosts are allowed to boot from the PXE configuration
• Configure the Kickstart file
• Boot and start the installation

You should already have a working kickstart server in place before setting up anything else in this post. For those that don't as a quick refresh you should have the following directory structure:

/var/www/pub
|-- CentOS
|-- images
`-- pxeboot
|-- isolinux
`-- isolinux.cfg
|-- kickstart
|-- repodata 

In the pxeboot folder should be vmlinuz and initrd.img files, and the kickstart folder should contain your kickstart file (test.cfg in our case). You can also refer to this earlier post to setup this up. Next you will need to setup a DHCP server first.

# yum -y install dhcp

# cp /usr/share/doc/dhcp-3.0.5/dhcpd.conf.sample /etc/dhcpd.conf

# vi /etc/dhcpd.conf

## /etc/dhcpd.conf file ##
ddns-update-style interim;
ignore client-updates;
authoritative;
allow booting;
allow bootp;

subnet 172.168.1.0 netmask 255.255.255.0 {
# default gateway
option routers    172.168.1.1;
option subnet-mask   255.255.255.0;
option domain-name   "mydomain.org";
option domain-name-servers 172.168.1.1;

# EST Time Zone
option time-offset   -18000;

# Client IP range
range dynamic-bootp 172.168.1.100 172.168.1.1.200;
default-lease-time 21600;
max-lease-time 43200;

# PXE Server IP
next-server 172.168.1.1;
filename "pxelinux.0";

}

## END FILE ## 

Now you will need to save the file and set the service to start on boot.

# chkconfig dhcpd on

# service dhcpd restart

Now your DHCP server should be setup and working properly. You can test this if you'd like by allowing a client to lease an ip address from the server to verify that it is working (run the dhclient command on any linux box). Next we will need to setup a TFTP server to server up the PXE file to clients. We will need to install the server and configure it run with xinetd service. Essentially all you need to do is change the "disable" option to "no".

# yum -y install tftp-server

# vi /etc/xinetd.d/tftp

## /etc/xinetd.d/tftp file ##

service tftp
{
   socket_type           = dgram
   protocol              = udp
   wait                  = yes
   user                  = root
   server                = /usr/sbin/in.tftpd
   server_args           = -s /tftpboot
   disable               = no
   per_source            = 11
   cps                   = 100 2
   flags                 = IPv4
}

## END FILE ## 

Save the file and restart the service for it to take effect:

# service xinetd restart

Next is going to be the install of syslinux which is required to allow the clients to actually PXE boot.

# yum -y install syslinux

Simple enough. Next we will need to create the TFTP directory layout for the clients to PXE boot from.

# cd /

# mkdir tftpboot

# cd tftpboot

# mkdir images

# mkdir pxelinux.cfg

# cp /usr/share/syslinux/menu.c32 .

# cp /usr/share/syslinux/pxelinux.0 .

* Some will have to use /usr/lib/syslinux

Now your directory structure should be in place with the required files. Last we will just copy over the kernel for the clients to use when booting.

# cd images

# cp /var/www/pub/images/pxeboot/vmlinuz .

# cp /var/www/pub/images/pxeboot/initrd.img .

Finally we just need to make the PXE file that directs the clients where you boot from.

# cd /tftpboot/pxelinux.cfg

# vi default

## /tftpboot/default ##

default menu.c32

prompt 0

timeout 10

MENU TITLE PXE Menu

LABEL CentOS 5.4 x32

MENU LABEL CentOS 5.4 x32

KERNEL images/vmlinuz

append initrd=images/initrd.img linux ks=http://172.168.1.1/pub/kickstart/test.cfg

## END FILE ##

Once you save and close this file you are done with the setup! There is one small change I forgot to mention...you will need to adjust your firewall settings for these new services.

# vi /etc/sysconfig/iptables

# -A RH-Firewall-1-INPUT -m state --state NEW -m tcp -p tcp --dport 67 -j ACCEPT

# -A RH-Firewall-1-INPUT -m state --state NEW -m tcp -p tcp --dport 68 -j ACCEPT

# -A RH-Firewall-1-INPUT -m udp -p udp --dport 69 -j ACCEPT

#service iptables restart

That should do it. Now if many of you haven't guessed by now I use the following addresses on my "lab" network to perform these test installs:

DHCP Server: 172.168.1.1
DNS Server: 172.168.1.1
PXE Server: 172.168.1.1
Clients: 172.168.1.100 - 172.168.1.200