LanGuardian Network Booting


Console Configuration

The LanGuardian console port should be configured as described in this document.

Telling the machine to boot from the network

A LanGuardian can be interrupted during the autoboot process and told to boot from the network. Or, the machine can be configured to autoboot from the network.

In either case, the LanGuardian needs to have the IP address for the boot interface and the server machine that it will boot from configured into the PROM environment. After this configuration is performed, the machine must be powercycled.

Here is an example of what the relevant sections of the PROM configuration look like:

   Secure [Encrypted] Network Load = N 
   Network Boot Mode: TFTP 
   Ethernet Port 0 IP Address = 192.168.215.32
   Ethernet Port 0 SERVER Address = 192.168.215.21
   Configuration File = /tftpboot/lg2.boot 

After this configuration is established in the machine, it can be network booted by entering the nb command into the PROM monitor program.

Contents of the Configuration File

The Configuration File that the LanGuardian uses is an ASCII file that has a a simple format. The file contents are divided up into lines, and each line is either a comment line, or a specification line. Comment lines start with a hash (#) character, and go to the end of the line. Specification lines contain: a keyword, an equals sign, and then the data value.

A sample configuration file looks like this:

#
# Boot description for 'nohost.example.com'
#
hostname = nohost.example.com
kernel = /lg2/rtmx
# the hostnames must not be fully qualified for "root" or "swap" directories
root = server:/spare/root/nohost
swap = server:/spare/swap/nohost
# if "true", this will cause the RB_SINGLE bit to be set in the howto
# options passed to the kernel, otherwise cleared
single-user = true

These values are used by the LanGuardian's PROM to finish the network booting of the Operating System kernel. As noted in the comments of the example configuration file, the hostnames associated with the root and swap directives must not be fully qualified hostnames.

How a LanGuardian Network Boots

The exact procedure that a LanGuardian uses to network boot is relatively simple to follow, but somewhat more complex than many other machine's network booting procedures.

  1. The LanGuardian will use ARP to establish the MAC address of the server that is configured into the PROM environment.
  2. The LanGuardian will TFTP an file called /etc/hosts from the server. It will use this file to lookup any IP addresses of the servers used later in booting.
  3. The LanGuardian will TFTP the Configuration File specified in the PROM environment.
  4. The LanGuardian will parse the configuration file, and TFTP the specified by the kernel entry.
  5. The LanGuardian will then construct an internal data structure from the remainder of the configuration file. This data structure is made available to the kernel. Finally, the PROM attempts to start the kernel.
  6. The Operating System kernel will then typically NFS mount a root directory and swap file, but the kernel is free to implement other methods of obtaining a root filesystem (e.g. TFTP loading of a RAM filesystem) and any binaries that it needs to operate.

Notes

If a LanGuardian is used in an environment where an ethernet switch connects it to other machines, it can useful to adjust the Auto boot timeout value in the PROM environment to a longer value than the default of 10 seconds. Some ethernet switches have relatively lengthly autonegotiation times, and if the LanGuardian attempts to ARP for the server's IP address before the ethernet switch is done negotiating, the switch may drop the ARP request. By increasing the timeout value, a value may be selected that is long enough to avoid this problem.

Last Updated: $Date: 2003/01/14 02:18:37 $