Troubleshooting the Installation

One common installation problem is that the Red Hat installer may not detect some key hardware such as the SCSI controller and network interface card on your system. The installer takes you through a sequence of installation steps that depends on what hardware the installer detects. For example, if the installer cannot detect the network card, it skips the network configuration step.

Another common installation problem crops up when you restart the PC and, instead of a graphical login screen, you get a text terminal or worse, the system seems to be hung. This means that there is something wrong with the X Window System (or X) configuration.

There is an alternate way to install Red Hat Linux so that you can force it to configure the network card and the SCSI controller (if you have one). You can also troubleshoot any problems with X by reconfiguring X.

Cross Ref

If you have problems with X Window System, printer, sound, or network, consult Chapters 3 through 6 for more information. In particular, Chapter 3 shows you how to configure X, and Chapter 4 shows you how to set up printers.

Using Text Mode Installation

The Red Hat installation program attempts to use a minimal X server to display the GUI mode installation screens. If the program fails to detect a video card, X does not start. If—for this reason or any other reason—it fails to start X, you can always fall back on the text mode installation program.

To use text mode installation, type linux text at the boot: prompt after you start the PC from the Red Hat Linux boot floppy. From then on, the basic sequence is similar to that of the graphical installation described previously in this chapter. However, many small details are different. You should be able to respond to the prompts and perform the installation.

In text mode, when the installation program fails to detect the video card, it displays a list of video cards from which you can select one. By selecting the video card, X may work when you install in text mode. If it does not, you can configure X using the information in Chapter 3.

Using the linux noprobe Command

If the Red Hat installation program does not detect your SCSI controller or network card, you can specify these devices manually by typing the linux noprobe command at the boot prompt.

Look for any indication of SCSI or network devices in the messages the Linux kernel displays as it boots. To view these boot messages during installation, press Ctrl-Alt-F4. This switches to a text-mode virtual console on which the messages appear. (A virtual console is a screen of text or graphical information stored in memory that you can view on the physical screen by pressing the appropriate key sequence.)

Another sign of undetected hardware is when the installation program skips a step. For example, if the Linux kernel does not detect the network card, the installation program skips the network configuration step.

To manually install devices, type linux noprobe at the boot: prompt in the initial text screen. The installation program then displays a dialog box that gives you the opportunity to add devices. Press Tab to highlight the Add Device button, then press Enter. The installation program then displays a dialog box that prompts you to select a driver from a list. You can then select the driver and press Enter. Repeat the process for as many drivers as you want to load.

After you finish adding any SCSI controllers and network cards, the installation program switches to graphics mode and guides you through the rest of the installation, as described previously in this chapter.

Troubleshooting X at the First Reboot

Sometimes you may run into a curious installation problem. During installation, the X configuration step works fine. But when you reboot the PC for the first time after installation, the graphical login screen does not appear. Instead, the boot process seems to hang just as it starts something called firstboot. If this happens to you, here’s how you can troubleshoot the problem.

When the firstboot process runs during the first system startup, it can either be uneventful or an aggravation. If all goes well, firstboot runs the Red Hat Setup agent that enables you to perform one-time setups such as date and time configuration as well as install any other CDs. Unfortunately firstboot may get stuck if the X Window System is not working on your system. You have to get around the firstboot process to configure X again and continue with the normal course of events that Red Hat planned for you. Here’s how you can stop firstboot and configure X to tide you over this problem:

Press Ctrl-Alt-F1 to get back to the text-mode boot screen. You see the text display with the boot messages that stop at a line displaying information about firstboot.
Press Ctrl-Alt-Del to reboot the PC. The PC starts to boot, and you get to a screen where the GRUB boot loader prompts you to press Enter to boot the Red Hat Linux (if you have GRUB password, you have to enter that password before you get to this point).
Press a to add an option for use by the Linux kernel. The GRUB boot loader then displays a command line for the Linux kernel and prompts you to add what you want.
Type a space followed by the word single and then press Enter. The Linux kernel boots in a single-user mode and displays a prompt that looks like the following:
```
sh-2.05b# 
```
Now you can configure X.

X uses a configuration file, called XF86Config, to figure out the type of display card, monitor, and the kind of screen resolution you want. The Red Hat installer prepares the configuration file, but sometimes the configuration is not correct.

To quickly create a working XF86Config file, follow these steps:

Type the following command:
```
/usr/X11R6/bin/XFree86 -configure
```
The screen goes blank and then XFree86 exists after displaying some messages. The last line of the message says the following:
```
To test the server, run 'XFree86 -xf86config //XF86Config.new'
```
Try the new configuration file by typing the command:
```
/usr/X11R6/bin/XFree86 -xf86config //XF86Config.new
```
If you see a blank screen with a X-shaped cursor, the configuration file is probably working fine.
Press Ctrl-Alt-Backspace to kill the X server.
Copy the new XF86Config file to the /etc/X11 directory with the following command:
```
cp //XF86Xonfig.new /etc/X11/XF86Config
```
That should provide a working X configuration file. Now, you can reboot the PC by pressing Ctrl-Alt-Del or typing reboot.

Cross Ref

The XF86Config file created by using the -configure option of the X server is not displayed at the best resolution possible. To fine-tune the configuration file, you should turn to Chapter 3, where you can learn more about configuring the X Window System and editing the /etc/X11/XF86Config file further.

Resolving Other Installation Problems

I am sure I have not covered all the installation problems that at least someone out there may encounter. There are so many different combinations of components in Intel x86 PCs that there is bound to be some combination of hardware that the installation program cannot handle. This section lists a few known problems. For others, I would advise you to go to Google Groups (groups.google.com) and type in some of the symptoms of the trouble. Assuming that others are running into similar problems, you should get some indication of how to troubleshoot out of your particular predicament.

Not Being Able to Boot from Boot Disk

Sometimes the PC does not seem to boot with the Red Hat installation boot disk. If this happens to you, try creating another boot disk using a fresh floppy disk and see if that takes care of the problem. Otherwise, you may need updated boot images. New boot images, if any, should be available at Red Hat’s support website (http://www.redhat.com/apps/support/errata/). After you download the image file, save it on your Windows system, give it a short name such as new.img. Then, create a new boot disk by following the steps explained in the “Creating the Red Hat Boot Disk” section. Remember to specify the new boot image file in response to the question that asks for the image source filename.

Understanding the Fatal Signal 11 Error

Some people get a fatal signal 11 error during installation. This usually happens past the initial boot screen as the anaconda installer is starting its GUI or text interface. The most likely cause of a signal 11 error during installation is a hardware error related to memory or the cache associated with the CPU (microprocessor).

Signal 11, also known as SIGSEGV (short for Segment Violation Signal), can occur in other Linux applications. A segment violation occurs when a process tries to access a memory location that it is not supposed to access. The operating system catches the problem before it happens and stops the offending process by sending it a signal 11. When it happens during installation, it means that anaconda had error accessing memory, and the most likely reason is some hardware problem.

A commonly suggested cure for the signal 11 problem is to turn off CPU cache in the BIOS. To do this, you have to enter setup as the PC boots (by pressing a function key such as F2) and then turn off CPU cache from the BIOS setup menus.

If the problem is due to hardware error in memory (in other words, the result of bad memory chips), you could try swapping the memory modules around in their slots. You may also consider replacing an existing memory module with another memory module, if you have it handy.

You can read more about the signal 11 problem at the following website: