LTSP is an abbreviation for "Linux Terminal Server Project". Installed on a server it can supply many workstations (so called thin-clients) with identical environments. All applications run on the server and therefore you can use old PCs and convert them into XTerminals. This reduces costs and maintenance especially in an environment where you need to have an uniform workspace on each computer you login, e.g. in schools or firms.

All of the examples in this document presume that your server's IP is 192.168.0.254, your domain is named yourdomain.com and your network is 192.168.0.0/24.

First of all, you must have a working Gentoo system. Please read the Gentoo Installation Manual for your architecture on Gentoo's Documentation Website.

Then lets begin with the easiest step, installing the ltsp core utils:

# emerge ltsp-core

This will install the following packages as dependencies:

• XFree: They are called XTerminals, guess why :)
• DHCP: DHCP is a protocol for automating the configuration of computers that use TCP/IP, used by ltsp to distribute IPs to the workstations.
• NFS: NFS is a protocol to allow access to harddisks through the network, used by ltsp to mount a base system for the workstations.
• TFTP: TFTP is a simple file transfer protocol, used by ltsp to transfer the kernel to the workstations.
• XINETD: Xinetd is a powerful replacement for inetd, with advanced features, used by ltsp to start tftp.

To analyze problems easier, the system logger must be configured to accept remote connections. Please read the documentation of your system logger on how to achieve this.

Next step is to edit your /etc/exports file, in order to allow the workstations to mount the root filesystem. There should be at least two lines in it:

/opt/ltsp/i386             192.168.0.0/255.255.255.0(ro,no_root_squash,async)
/var/opt/ltsp/swapfiles    192.168.0.0/255.255.255.0(rw,no_root_squash,async)

Now start NFS.

# rc-update add nfs default
# /etc/init.d/nfs start

By default TFTP won't be started, to change this edit /etc/xinetd.d/tftp and replace disable=yes with disable=no. Afterwards, start xinetd.

# rc-update add xinetd default
# /etc/init.d/xinetd start

In order for the workstation to reach all resources, a correct name resolving must be available. There are several ways to achieve this. One is to configure a DNS server for the local network, the other (and more simple) is to have almost identical /etc/hosts files on all systems. We are going to use the latter.

All workstations must be listed in /etc/hosts. Take a look at the example:

127.0.0.1        localhost
192.168.0.254    server      server.yourdomain.com
192.168.0.1      ws-1        ws-1.yourdomain.com

This is the most complicated step in my opinion, you have to create a valid DHCP Config (/etc/dhcp/dhcpd.conf). Here is an example:

Some general options
default-lease-time            21600;
max-lease-time                21600;
use-host-decl-names           on;
ddns-update-style             ad-hoc;

Bootp options
allow booting;
allow bootp;

Network Options
option subnet-mask            255.255.255.0;
option broadcast-address      192.168.0.255;
option routers                192.168.0.254;
option domain-name-servers    192.168.0.254;
option log-servers            192.168.0.254;
option domain-name            "yourdomain.com";

LTSP Path Options
option root-path              "192.168.0.254:/opt/ltsp/i386";
filename                      "/lts/vmlinuz-2.4.19-ltsp-1";

If your workstations have ISA NICs uncomment the following
lines and alter the driver and IO
#option option-128 code 128 = string;
#option option-129 code 129 = text;
#option option-128 e4:45:74:68:00:00;
#option option-129 "NIC=ne IO=0x300";

shared-network WORKSTATIONS {
  subnet 192.168.0.0 netmask 255.255.255.0 {
    Distribute dynamic IPs to the workstations
    range dynamic-bootp 192.168.1.1 192.168.1.16;
    Workstation specific configuration for PXE booting
    #host ws001 {
    #  hardware ethernet     00:E0:06:E8:00:84;
    #  fixed-address         192.168.0.1;
    #}
  }
}

If your workstations support PXE, you should list each one of them as we have done with host ws001 (don't forget to uncomment it). Don't give them an adress in the dynamic range, otherwise it would be possible that more workstations have the same IP (which is troublesome).

For more documentation on this item read the official dhcp handbook: http://www.dhcp-handbook.com/

Now start DHCP as you did with NFS and xinetd:

# rc-update add dhcp default
# /etc/init.d/dhcp start

There are many options to configure your workstations, visit http://www.ltsp.org/documentation/ltsp-3.0-4-en.html#AEN903 for a full description of /opt/ltsp/i386/etc/lts.conf.

Now you have to change your displaymanager's configuration to also accept remote connections.

XDM: In /etc/X11/xdm/xdm-config comment out DisplayManager.requestPort: 0

KDM: In /usr/kde/3.1/share/config/kdm/kdmrc look for the [Xdmcp] section and change Enable = false to Enable = true.

GDM: In /etc/X11/gdm/gdm.conf look for the [xdmcp] section and change Enable = false to Enable = True.

Then start the displaymanager:

# rc-update add xdm default
# /etc/init.d/xdm start

If you workstations don't support PXE booting, you must create a boot-floppy, which is needed to start your workstations. Go to http://www.rom-o-matic.net/5.0.9/, select your NIC, press Get ROM and write the image to a floppy:

# cat nicfile.lzdsk > /dev/fd0