Put your logo here!

TOC PREV NEXT INDEX



8

Supporting a Centralized System - Generic UNIX

This chapter contains procedures for setting up the necessary operating system support files. Use the Workbook (Chapter 2) to determine the procedures needed.

Your environment and the features you enable dictate the operating system tasks. You must be root to modify the host files.

Note: This chapter is based on a Sun Sparcstation 2. Both the Berkeley (ps -aux) and System V (ps -eaf) ps commands are shown.

Enabling NFS Access

This section describes the steps necessary to set up the host if NFS is used as the boot_method or file_access_method.

Edit the /etc/exports file. Add each of the file systems to be exported at boot time, along with any options. The file systems must be exported to be accessible to network computers. The exported systems are specified with an nfs_table command in the xp.cnf file, or in the nfs.tbl file. An example xp.cnf file entry:
nfs_table "montana:/tekxp" "/tekxp"  "8192"
An example nfs.tbl entry:
montana:/tekxp /tekxp 8192
/etc/exports entry, shown with the read only option (-ro):
/tekxp -ro
  1. To enable the mount point immediately, without booting, run:
    # /usr/etc/exportfs -a [-o options]
  2. Use /usr/etc/exportfs without any parameters to verify the exports.
  3. Edit /etc/hosts and add the network computer name and address. For example:
    128.07.60.30 portland
  4. Some hosts require an explicit update to the arp table to add new entries:
    # arp -s ether portland 080011010045
    # arp -s ether corvallis 080011010046
    The ether switch indicates an ethernet address. arp can be run with a
    -f filename option, where filename is a file of network computer entries to set.
  5. Use the appropriate ps command to see if nfsd is running:
    # ps -aux
    # ps -eaf
    root 88 1 0 Jul 27 ? 0:01 nfsd
    1. If nfsd is not running, type:
      # /usr/etc/nfsd 8 &
    2. If nfsd is running, use the kill command with the PID to restart the daemon (the PID is 88 in the preceding ps example):
      # kill -HUP PID

Enabling TFTP

This section describes the steps necessary to set up tftp if it is used as the boot_method or file_access_method.This is the default boot method for network computers not equipped with TDEnet, Flash Memory or ROM options. For details see the tftp man page on your host, or in /tekxp/man.

    1. The /etc/inetd.conf or /usr/etc/inetd.conf file is read at startup, and should contain the tftp command line. Verify that the tftp startup command is in the /inetd.conf file. Add the line if it does not exist in your inetd.conf file. To verify the command:
      # grep tftp /etc/inetd.conf
      tftp dgram udp wait root /usr/etc/in.tftpd in.tftpd
      Your startup could be different. Check the man page for specifics. Here is another example:
      tftp dgram udp wait nobody /usr/etc/tftpd tftpd
    2. If /etc/inetd.conf contains a tftp entry with the -s (secure tftp) option:
      tftp dgram udp wait root /usr/etc/in.tftpd in.tftpd -s /tekxp
      both the boot_directory and install_directory used during the media installation would have to be within the secure path, in this case, /tekxp. Refer to the following section for information on secure tftp.
    3. Verify that in.tftpd, the tftp binary resides where it is specified in inetd.conf. For the preceding examples, tftp would be located in /usr/etc. If there is a mismatch between the actual location and the specified location, either move the binary or change the /etc/inetd.conf file.
    4. If you edited inetd.conf, restart the inetd daemon. Use the appropriate ps command to find the process ID number (PID) for inetd. In this example the PID is 112.
      # ps -aux | grep inetd | sed /grep/d
      # ps -eaf
      root 112 1 0 Jul 27 ? 0:01 /usr/etc/inet/inetd
      1. If inetd is running, use the kill command with the PID to restart the inetd daemon.
        # kill -HUP PID
      2. If inetd is not running, either reboot the host or type:
        # inetd
Note: The tftp binary from the installation media requires a /usr/tmp directory to perform the file transfers. If this directory does not exist, it must be created.

Using Secure TFTP

If you use secure tftp, there are various things to consider before you begin:

Note: You must supply the initial slash (/) in the path that you specify (e.g., you cannot use boot/os.900 in the previous example).

Enabling BOOTP

Bootstrap protocol, bootp, is the recommended way to establish communications from the host to the network computer in an internet protocol environment. Bootp obtains booting data from the bootptab file. With the proper information stored in the bootptab file, the network computer can find its own name and IP address and boot from the network without any intervention, even for a first time boot.

Note: If bootp is not already on your host, you must build it from the source included on the installation media. Refer to the Making the Utilities section in Chapter 1, Installing the Software for more information.
    1. Verify that the bootpd and bootptab files are in the /etc or /usr/etc directory:
      # ls bootp*
    2. If they are not there, make the utility, then:
      # cp /tekxp/src/bootp/bootpd /etc
      # cp /tekxp/examples/bootptab /etc
    3. Edit the /etc/hosts file to add the network computer internet addresses and names:
      128.07.60.30 portland
      128.07.60.31 corvallis
    4. Some hosts may require an explicit update to the arp table to add the new entries:
      # arp -s ether portland 08:00:11:01:00:45
      # arp -s ether corvallis 08:00:11:01:00:46
      The ether switch indicates you are providing an ethernet address. The arp command can also be run with a -f filename option, where filename is a file of all the network computer entries to set.
    5. Set up the bootptab file.
    6. Start or restart the inetd or bootpd daemon.

Setting Up the bootptab File

The bootptab file contains the data used by bootp. A sample bootptab file is located in /tekxp/examples. Refer to the bootp man page in /tekxp/man for more information. The file contains three sections:

Note: T134 is a user-defined variable to the bootptab file. It is intended for use with the bootp and bootptab file included on the media.

Example 8-1 Sample bootptab File

# Global entries for all hosts
global.d:\
:sm=255.255.255.0:\
:ds=128.07.5.200 128.07.61.150:\
:hn:\
:hd=/tekxp:
# Master entries for each subnet template
subnet60: :tc=global.d:gw=128.07.60.100:
subnet61: :tc=global.d:gw=128.07.61.150:
# Individual network computer entries
# Sample entry for model type XP100, XP200, XP350, or XP400
portland: tc=subnet60:ht=ethernet:ha=080011010045:\
ip=128.07.60.30: bf=/boot/os.350:\
T134=config/admin.cnf:
# Sample entry for model type NC200 or NC400
corvallis: tc=subnet60: ht=ethernet: ha=080011010046:\
ip=128.07.60.31: bf=/boot/os.500:
# Sample entry for model type NC900
astoria: tc=subnet60: ht=ethernet: ha=080011010047:\
ip=128.07.60.32: bf=/boot/os.900:

Running BOOTP

There are two methods for using bootp:

Select the method that best fits your environment.

Only Running BOOTP on Request
    1. If the bootpd daemon is started within the inetd.conf file, it is only run when bootp requests are sent to the host. If the bootps command line does not exist in your inetd.conf file, you need to add it. To verify that the bootps startup command is in the file:
      # grep bootp /etc/inetd.conf
      bootps dgram udp wait root /etc/bootpd bootpd
      Your startup could be different, check the man page. Another example:
      bootps dgram udp wait nobody /usr/etc/bootpd bootpd
    2. Verify that bootpd resides where it is specified in inetd.conf. For the preceding examples, bootpd would be located in /usr/etc. If there is a mismatch between the actual location and the specified location, either move the bootpd binary or change the inetd.conf file.
    3. The port numbers for bootp listening and replying must be set in the /etc/services file. If they are not set, edit the file and add them. To verify that the ports are set in the file:
      # grep bootp /etc/services
      bootps 67/udp
      bootpc 68/udp
    4. If you edited inetd.conf, restart the inetd daemon. Use the appropriate ps command to find the process ID number (PID) for inetd. In this example the PID is 112.
      # ps -aux
      # ps -eaf
      root 112 1 0 Jul 27 ? 0:01 /usr/etc/inet/inetd
      1. If inetd is running, use the kill command with the PID to restart the inetd daemon.
        # kill -HUP PID
      2. If inetd is not running, either reboot the host or type:
        # inetd
Always Running BOOTP
    1. If the bootpd daemon is started within the /etc/rc.local file or a similar startup file, it is always running and waiting for bootp requests. If the bootp command line does not exist in your /etc/rc.local file, you need to add it. To verify that the bootp startup command is in the file:
      # grep bootp /etc/rc.local
      /etc/bootpd -s ; echo -n ' bootpd'
      The -s option specifies continuous execution.
    2. Verify that bootpd resides where it is specified in /etc/rc.local. For the preceding examples, bootpd would be located in /etc. If there is a mismatch between the actual location and the specified location, either move the bootpd binary or change the /etc/rc.local file.
    3. The port numbers for bootp listening and replying must be set in the /etc/services file. If they are not set, edit the file and add them. To verify that the ports are set in the file:
      # grep bootp /etc/services
      bootps 67/udp
      bootpc 68/udp
    4. To start the bootpd daemon without rebooting the system:
      # /etc/bootpd -s &

Enabling RARP

In addition to bootp, rarp (TCP/IP Reverse Address Resolution Protocol) can be used to establish communications. The network computer broadcasts its ethernet address across the network, and receives its internet address from the first host to respond. The /tekxp directory must be on the host.

Note: rarp is not supported for Token-Ring networks.
    1. If the rarpd daemon is started within the /etc/rc.local file or a similar startup file, it is always running and waiting for rarp broadcasts. If the rarp command line does not exist in your /etc/rc.local file, you need to add it. To verify that the rarp startup command is in the file:
      # grep rarp /etc/rc.local
      /usr/etc/rarpd -a ; echo -n ' rarpd'
    2. Verify that rarpd resides where it is specified in /etc/rc.local. For the preceding examples, rarpd would be located in /usr/etc. If there is a mismatch between the actual location and the specified location, either move the rarpd binary or change the /etc/rc.local file.
    3. Edit the /etc/hosts file to add the network computer internet addresses and names:
      128.07.60.30 portland
      128.07.60.31 corvallis
    4. Edit the /etc/ethers file to add the network computer ethernet addresses and names:
      08.00.11.01.00.45 portland
      08.00.11.01.00.46 corvallis
    5. Some hosts require an explicit update to the arp table to add new entries:
      # arp -s ether portland 08:00:11:01:00:45
      # arp -s ether corvallis 08:00:11:01:00:46
      The ether switch indicates an ethernet address. arp can be run with a
      -f filename option, where filename is a file of the network computer entries to set.
    6. Use the appropriate ps command to see if the rarpd daemon is running:
      # ps -aux
      # ps -eaf
      root 69 1 0 Jul 27 ? 0:01 /etc/rarpd
      1. If rarp is running, use kill with the PID to restart the daemon:
        # kill -HUP PID
      2. If rarp is not running, type:
        # /usr/etc/rarpd -a &

Adding DNS Entries

This section assumes that your site is currently running Domain Name Service (DNS). The instructions in this section deal strictly with adding network computer names and addresses to your existing zone and local host files on the name server. The following structure is used for the examples:

DOC 128.07.50.1 is the primary name server for the DOC company, oregon 128.07.60.1 is the secondary name server, serving the site's network computers portland 128.07.60.30 and corvallis 128.07.60.31 are network computers.

    1. Edit the local host file for the name server. The typical file name is named.local. Depending on how you set up your system, the full path name for this example could be /var/named/oregon.DOC.COM/named.local. Add the network computer entries to the file. For example:
      ; named.local for server oregon
      IN NS oregon.DOC.COM
      30 IN PTR portland.oregon.DOC.COM
      31 IN PTR corvallis.oregon.DOC.COM
      IN = Internet Protocol, NS = name server, PTR = set reverse pointers
    2. Edit the zone file for the name server. The typical file name is the name of your server. Depending on how you set up your system, the full path name for this example could be /var/named/hosts/oregon.DOC.COM. Add the network computer entries to the file. For example:
      ; zone hosts file for server oregon
      oregon A 128.07.60.1
      HINFO Sparc2
      portland A 128.07.60.30
      HINFO network computer
      corvallis A 128.07.60.31
      HINFO network computer
      A = IP address record, HINFO = information entry.
    3. Use the appropriate ps command to see if named is running:
      # ps -aux
      # ps -eaf
      root 88 1 0 Jul 27 ? 0:01 named
      1. If named is not running, type:
        # named
      2. If named is running, use kill and the PID to restart the daemon:
        # kill -HUP PID

TOC PREV NEXT INDEX