This section describes preparations for using BOOTP/DHCP. For BOOTP, you can use the native software on the boot host or the software included in the NCDware distribution.

If you are using DHCP to provide network information, the terminal configuration is the same as for BOOTP, and the terminal must have Boot Monitor version 2.8 or later.

If you are not using ncdinstall to prepare hosts and terminals for BOOTP/DHCP service, you need to perform the tasks described in the following subsections. The commands and steps may vary from those outlined in this section depending on your host, your BOOTP/DHCP implementation, and your network. These instructions are guidelines; they are not precise procedures.

NCD terminals send BOOTP/DHCP requests by default; consequently, terminal configuration is unnecessary unless you want to change the order in which the terminal sends requests for network information or specify a second or third source. The default order is first DHCP, then BOOTP, then RARP.

Note:

Information provided by BOOTP (for example, the subnet mask) is passed to the X server and may be stored in NVRAM, overwriting previous values.

The BOOTP protocol is implemented through daemon programs, such as bootpd(8), and a database file, such as /etc/bootptab. Hosts configured to use NCD's dynamic IP address allocation for terminals use an additional database file, /etc/bootptab.cfg.

For specific instructions on using BOOTP, refer to the following sources:

For Information About:	Refer to:
NCD's BOOTP implementation	The bootpd man page in the NCDware distribution
The native BOOTP already installed on your host	Your host's documentation
BOOTP/DHCP options recognized by NCD terminals	Table 3-1
Dynamic IP address allocation using BOOTP	The bootpd (8) and bootptab.cfg (5) man pages in the NCDware distribution and the System Administrator's Guide.

For instructions on using DHCP, refer to your host system's documentation.

Making Sure that BOOTP Is Enabled

To verify that BOOTP is enabled on your boot host, check the following:

• To find out if bootpd is running, enter a ps(1) command. For example:
  

  
  # ps -axc | grep bootp 
  

  
  On some systems, the command is:
  

  
  # ps -ef | grep bootp 
  

• Host operating systems based on the 4.3 BSD UNIX operating system (such as SunOS) require an entry in the /etc/inetd.conf file for BOOTP. Often, these entries exist, but are rendered ineffective by a comment symbol (#) at the beginning of the entry. If this is the case, remove the #. For example, for SunOS:
  

  bootps dgram udp wait root /usr/etc/bootpd bootpd
  

• Make sure that TCP/IP ports are reserved for the bootpd server and client processes in the /etc/services file.
  
  The usual entries in /etc/services are:
  

  
  bootps          67/udp
  bootpc          68/udp
  

  
  Remove any comment symbols (#) at the beginning of these entries.
  

• If you make any changes in the /etc/inetd.conf or /etc/services file, restart the inetd daemon.
  
  To restart the daemon, find the inetd process ID and send a hangup signal as shown in the following example. This causes the daemon to read the configuration file and use the new information. For example:
  

  
  # ps -axc | grep inetd
  17601 ?  I 0:12 inetd
  # kill -HUP 17601 
  

  
  On some systems, the command for finding the process ID is:
  

  
  # ps -ef | grep inetd 
  

If you do not use ncdinstall to add terminals to the network, you must add BOOTP options for each terminal to the /etc/bootptab database file on the boot host.

bootptab Format and Options

For BOOTP configuration on your boot host, see your vendor documentation. The normal rules for /etc/bootptab file entries are:

• A colon (:) indicates the end of a field, and a backslash (\) indicates that the entry is continued on the next line.
  

• Spaces are not permitted between the characters on a line.
  

• Fields consist of a tag followed by an equals sign (=) and a value. Each tag identifies a unique parameter.
  

• The Ethernet address has no internal punctuation, such as periods or dashes.
  

The typical file structure is one or more template entries, containing information common to all terminals or a group of terminals, followed by individual entries, each containing information about a specific terminal:

# Template entry
template.name:\
          tag=value:\
           .
           .
           .
           tag=value
# Entry for an NCD terminal
hostname:\
     tc=template.name:\
            tag=value:\
            .
            .
            .
            tag=value

The tags recognized by NCD terminals and supported by the version of bootpd provided in the NCDware distribution are listed in Table 3-1. The table lists the normal two-letter tag name or a tag number and a description of the information provided by the tag. Although some versions of bootpd may provide additional information, only information listed in the table is used by NCD terminals.

Table 3-1 BOOTP/DHCP Tags Used by NCD Terminals

A fragment of an example bootptab file follows, with comments identifying the fields.

# Template entry
global:\		Template entry name
	gw=192.43.153.1:\	Gateway address
	sm=255.255.255.000:\	Subnet mask
	hd=/tftpboot:\	X server file directory
	ht=ether	Interface name
# Entry for an individual NCD terminal
ncd1:\		NCD terminal's hostname
	tc=global:\	Include the template
	ha=0000a70015d5:\	Ethernet address
	ip=192.43.153.224:\	IP address
	bf=Xncdxpl	X server filename

Adding Terminal Entries to the bootptab File

The required and optional entries in the bootptab file are:

• One or more template, or global, entries for the information common to all terminals or groups of terminals (optional)
  

• An entry for each terminal containing at least the terminal's Ethernet address and IP address or its Token-Ring address and IP address
  
  If you do not supply the address and the terminal cannot determine its address through other means (using RARP or reading the address from NVRAM) the terminal will be unable to boot.
  
  

  Note:

  If the terminal has a TRP board, the current Token-Ring address (also called the active address) is different from the built-in Token-Ring address (also called the static address). When configuring BOOTP, use the current address.
  
  Both addresses are printed on the sticker on the terminal base and on the packing box. The built-in address always begins with 00:00:a7 (for example, 00:00:a7:11:2a:4b). The current address is a bit-reversed version of the built-in address (for example, 00:00:e5:88:54:d2).
  
  The current address is displayed by the Boot Monitor during booting. Both the current address and the built-in address are displayed in Statistics -> Show Version. The current address is also displayed in Statistics -> Show Statistics -> Network Interfaces -> Interfaces Table and in Statistics -> Show Statistics -> Token-Ring -> Interface Table.

  
  

• If you use BOOTP/DHCP to specify the X server to be downloaded by the terminal, add the X server filename (and the X server directory name, if not specified in the global, or template, entry).
  
  If you do not specify the X server in the /etc/bootptab file, the Boot Monitor uses the default X server download sequence. For information about this sequence, see Chapter 4, Booting-X Server Loading.
  
  If you specify an X server file in /etc/bootptab and bootpd cannot locate the file, the terminal cannot download an X server. In that situation, the Boot Monitor displays its prompt and waits for a manual boot command. Syntax errors in the file may also prevent the terminal from booting.
  

• If the terminal is booting through a gateway (that is, servers are installed on a host located on a remote network), make sure the gateway to the boot host and subnet mask (if used) are specified in the bootptab file. See the next section for more information about booting through a gateway.
  

For booting through a gateway, the device serving as the gateway and the bootptab file on the boot host must be properly configured.

This section provides an example setup (illustrated in Figure 3-1) for a Cisco router, which is a commonly used gateway device.

The IP addresses of the gateway's Ethernet interfaces are:

192.43.157.2	Ethernet Interface 0-for the remote network where the boot host is located.
192.43.156.2	Ethernet Interface 1-for the local network where the terminals are located.

The IP address of the boot host is 192.43.157.40.

The IP address of the NCD terminal is 192.43.156.171. Its Ethernet address is 0000a70023a3.

On the router, set the helper address to the address of the remote boot host by entering the following command:

ip helper-address 192.43.157.40

On the boot host, the bootptab file contains the following entries for this example:

# Template entry - every host uses this information
global.dummy:\
	:sm=255.255.255.0:\	Subnet mask
	:hd=/tftpboot/Xncd.4.2.0:	X server file directory
# Entries for each subnet
subnet.one:\		Include the template and specify the gateway address
:tc=global.dummy:gw=192.43.157.2:
subnet.two:\		Include the template and specify the gateway address
:tc=global.dummy:gw=192.43.156.2:
# Entry for an NCD terminal
ncdhmx1: tc=subnet.two:\		Include the template
	:ht=ethernet:\	Network interface type
	:ha=0000a70023a3:\	Terminal Ethernet address
	:ip=192.43.156.171:\	Terminal IP address
	:bf=Xncdhmx	X server filename