Using the Installer

We designed the installer to be fairly easy to use. Most of the installation process consists of waiting while the client downloads and installs packages. However, once the installer has finished you need to carry out some manual configuration.

The installation procedure takes about 45 minutes on a Pentium 90, and about an hour and a half on a 486-66. Fortunately, most of the installation can be left unattended.

This section describes the steps you need to follow in order to complete a typical installation.

Booting with a floppy

Each FAI boot floppy should be marked with an IP address. This IP address will be assigned to the client. Make sure that no other computer is using this computer (that is, that no other currently-installing client used this boot floppy).

In the BIOS of the computer, check that the boot sequence will boot from floppy first. Then put in the boot floppy and (re) start the machine.

Startup

The computer should boot from the floppy. First the computer loads a Linux kernel from the floppy. At this point you should see a message similar to:

LILO 22.1: Booting FAI....

Once the kernel has loaded off the floppy, the client boots into Linux and tries to connect to the server. At this point you will see a message like the following flash by:

---------------------------
FAI 2.3.1, 16 apr 2002
Fully Automatic Installation for Debian GNU/Linux

Copyright (c) 1999-2002, Thomas Lange
lange@informatik.uni-koeln.de
---------------------------

At this point, the installer takes control of the machine.

Customization

You should now be presented with a few configuration screens. In each of these screens, use the space bar to toggle selections, and press $\langle\textnormal{enter}\rangle$ to move to the next screen.

(Screenshots? -P.)

Additional classes

The first screen allows you to select additional FAI classes that should be installed on the system. Depending on the size of the system's hard drive, some classes available on this screen will be defined regardless of your choice. For more information about this process, see Section 6.3.1

These selections do not take into account hard drive size, so you should make sure that the hard drive is sufficiently big to hold any extra packages you install.

Unless the user specifically requested certain packages, it is best to stick with the default applications. However, you may need to select additional classes in the following situations:

• If you are installing a sound card, you will want to select classes to help you configure and use the card. Read Section 5.8 for more information.

By default, this screen defaults to selecting the following classes:

AUTOLOGIN:

This class configures GDM to log in a user account automatically. See Section 5.3.2 for more information.

X-server selections

In order for the X-Window graphical environment to work, a video card driver (called an “X-server”) needs to be selected and installed. There are several versions of X-Windows available for Linux. The version we use is called XFree86.

Selecting the right X-server can be tricky. One complication is that Debian currently supports two versions of XFree86 - version 3.3.6 and the newer version 4.1.0 . Many older video cards that work with XFree86 3.3.6 do not work with 4.1.0 . Thus, you have to do some detective work to find out which package to install:

1. First, figure out the make and model of your video card. Ideally, you have done this before you started the installation. If not, switch to the second virtual terminal by pressing $\langle\textnormal{alt}\rangle$- $\langle\textnormal{F2}\rangle$ and type

   discover -format "Card %V %M, Xserver: %S" video

   If you are lucky, this will show you the make and model of the video card. If you are unlucky your display will go blank, and you will have to reboot the system to see anything.

   Another detection program is called detect. To use it, type

   detect

   If all goes well, a file called report.txt will be generated in the current directory.

2. Once you know the make and model of the video card, go to the server, open a terminal window, and type

   zcat /usr/share/doc/xserver-xfree86/Status.gz | less

   This will display a file containing the status of all video cards supported by XFree86. Look up the card in this list. You will discover one of three things:

   1. The card is supported by XFree86 4.1.0 . In this case, you want to note the driver for the card, and select the XSERVER-XFREE86 package (which contains all drivers for XFree86 4.1.0).

   2. The card is not supported by XFree86 4.1.0, but is supported by 3.3.6 . In this case, you have to look up the xserver package you need to install.

   3. The card is not supported by either XFree86 versions 3 or 4. In this case you are out of luck, and you should swap the video card for something that is supported, or swap the computer altogether.

3. Once you know the X-server you need, go back to the installer (by pressing $\langle\textnormal{alt}\rangle$- $\langle\textnormal{F1}\rangle$) and see whether the package you need is available. If it is, select it. If it isn't, then you will need to install the X-server later. See Section 8.6 to learn about installing extra packages.

It is possible to install multiple X-servers. However, this will just waste disk space, since you can only use one X-server at a time.

Keep track of which X-server you installed. You will need this information when configuring the system (see Section 5.2).

Hostname

In this screen, you are asked to provide a hostname for the machine. A hostname is the computer's name. If your user named the new system, type in this name. Otherwise stick with the defaults.

Conventionally, hostnames are made up of a combination of lower case letters, numbers, and dashes (`_' and `-'). You may be able to use other characters as well, but definitely should not use periods.

(Restrictions? -P.)

Automated install

After configuration, the installer takes over. First, it partitions and formats the hard disks:

Calling task_partition
Partitioning local harddisks

You will probably see a few screenfuls of a line similar to the following:

modprobe: modprobe: Can't locate module block-major-105

These messages are harmless.

Next, the installer starts installing software. You will see the installer print out a large list of software that needs to be installed. The installer will first download each of these packages from the server. Then it will install these packages. First, packages are unpacked:

Selecting previously deselected package libgtk1.2-common.
Unpacking libgtk1.2-common (from .../libgtk1.2-common_1.2.10-11_all.deb) ...

When all packages have been unpacked, they are configured:

Setting up libgtk1.2-common (1.2.10-11) ...

Finally, the installer runs some customization scripts and makes the system bootable. When the installer has completed you will hear a beep and see the message:

FAI finished.
Calling task_chboot
Calling task_savelog

At this point you can reboot the system.

Checking the installation process

Restoring a blank screen

When the keyboard has not been touched for a few minutes the screen goes blank. This is just Linux saving energy. To restore the screen press an arrow key (pressing $\langle\textnormal{enter}\rangle$ is not such a good idea because you will reboot the system if the installer has finished).

Using the second virtual terminal

During the install, you can check what is going on by pressing $\langle\textnormal{Alt}\rangle$- $\langle\textnormal{F2}\rangle$ . This activates the second virtual terminal. You should see a command line prompt, which you can use to check the system.

Some useful things you might want to do include:

• Running the top program to check on memory and CPU usage. You can start top by typing top , and get out of it by pressing q .

• Looking at the messages the installer has printed. To do this type

  less /tmp/fai/rcS.log

  To get out of less, type q .

  There are lots of other log files and configuration files in /tmp/fai as well. See Section 6.1.3 for more information about these logfiles.

• Looking at disk usage. Type df to see this. The local hard disk(s) will be mounted to directory /tmp/target .

To get back to the installer's main screen, press $\langle\textnormal{Alt}\rangle$- $\langle\textnormal{F1}\rangle$.

Reboot the system

Remove the boot floppy from the computer's drive, then press $\langle\textnormal{enter}\rangle$. The system should reboot.

Before rebooting, you may see the following message:

(write down the message -P.)

It is safe to ignore this.

The system should boot into Linux, and you should get a login prompt. If the system does not boot, the installer failed.

If the system does boot properly into Linux, you can log in as the root user and continue configuring the system.