does not occur automatically. The goal of disaster recovery is to restore full
functionality as quickly as possible. Depending on the degree of fault toler-
ance your systems have, disaster recovery may not be necessary at all.
Planning for fault tolerance and disaster recovery requires that you assess
the needs of your systems. The following two questions are the most impor-
tant ones you should ask:
■■ How critical are the systems to daily operation?
■■ Could the systems be down and not affect operations?
Obviously, if your systems are used for business, they could not be down for
long, if at all. You must determine how vital a given system is to your opera-
tion. Vital systems require greater fault tolerance than nonvital systems do. Be
sure to keep in mind that greater fault tolerance costs more than less fault tol-
erance does. Another important consideration is the amount of money avail-
able for building fault tolerance into your system. Balance your need for fault
tolerance with the amount of money you can spend on it.
Clustering Solutions
If your systems must be up 24/7, you have to rely on a clustering solution for
your fault tolerance. You basically have two choices for clustering: failover
clustering and true clustering. Both of these solutions are costly to implement
and require a great deal of configuration. Configuring these types of systems
is well beyond the scope of this book, but a brief description of the two types
of clustering is useful.
Failover clustering typically requires two systems. The first system is the
active system that responds to service requests. The second, failover system is
an exact copy of the first system that is connected to the first system by a dedi-
cated link. The second system uses the dedicated link to listen for a signal —
called a heartbeat — from the first system at a specified interval. The second
system does nothing but listen to the heartbeat signal from the first system. If
the second system does not receive the signal, it assumes that the first system
has gone offline and immediately begins accepting requests for services. When
the first system comes back online, the second system returns to monitoring the

possibly happen and figuring out a way around it. However, that which is ideal
often does not square with what’s practical, especially when it involves spend-
ing money to guard against an infinitesimal likelihood. Fortunately, the things
that save you from likely disasters save you from the most unlikely ones, too.
Just as security planning requires attention to threats from outside and
inside the network, there are two parts to disaster planning. The first is doing
everything you can to prevent a catastrophe from taking place.
Only you, or other administrators at your organization, know how impor-
tant your system is and how much money is budgeted to keep it running.
Chances are good that an uninterruptible power supply (UPS) that keeps the
network up long enough to save and close files and shut down the system in
an orderly fashion fits within the available budget. Agood UPS system is espe-
cially useful if your enterprise has a generator backup that kicks on in the
event of power failure because generators do not always start instantly and,
when they do, the electricity provided is not always clean enough for com-
puter use. A battery backup can protect you from both of these potential prob-
lems. If your enterprise is important enough to have an emergency generator,
it’s probably important enough to keep the network running.
Renegade electricity is one of the worst enemies of system reliability. Small
power strips with surge suppression are better than nothing, but more robust
power conditioning is needed if really important equipment and data are to be
protected. In fact, be sure to protect all lines from the outside world that attach
Planning the Network 25
07_599496 ch02.qxd 8/30/05 6:21 PM Page 25
to your computer or its peripherals, be they phone lines or cable or DSL con-
nections. Likewise, put the peripherals themselves on protected circuits.
Second, formulate a regular (daily or better) backup scheme, with one set of
backups stored in a safe place off-site as protection against loss of data in the
event of fire, flood, tornado, or other physical disaster. One way of making this
process relatively painless, albeit an expensive one, is to rent storage from a

A very important part of network planning is to put it all down on paper and
to save that piece of paper. Working out your network’s design is best done by
actually diagramming the network, making multiple diagrams to explore dif-
ferent strategies. Once you settle on a design, draw a more formal diagram.
26 Chapter 2
07_599496 ch02.qxd 8/30/05 6:21 PM Page 26
Sometimes it’s a good idea to save your discarded designs as well, with a note
on each version explaining why it wasn’t chosen. Formalizing the network
design and saving the discarded ideas is useful for several reasons. It bolsters
your decisions in case you’re second-guessed, it demonstrates that you con-
sidered all the possibilities, and the formal diagram is a valuable tool should
someone need to administer the system in your absence.
A written security policy is essential in the enterprise and not a bad idea
even for a home network. An additional security file you should always keep
is a full security log. Such a record might begin by detailing what security mea-
sures you have designed into the system. It should include copies of any secu-
rity notices you have received, as well as an initialed notation of when the
recommended security patch was applied. If log files show an attempted crack
of your network, hard copies of the relevant portions should be kept there, too.
When users or management complain about how you have the system so
tight that it seems inconvenient even for them to log in, there’s nothing like
proving that the system is regularly under attack — and it will be, by port
scanners and others — to demonstrate the wisdom of tight security. One very
big company has made huge amounts of money by putting user convenience
over security, and many companies have paid a high price for adopting their
products. Your Red Hat system costs a very small amount in user inconve-
nience in exchange for greatly enhanced system security. It’s useful to be able
to prove that the threat is real.
A security log is also the place to keep copies of any security-related email
messages from within the company, from log listings of employees who have

you begin to construct it, discovered some of the options available to you, and
found out some of the reasons why you might choose one over another. You
learned that network security is a never-ending task made easier by careful
planning and that threats can come both from outside the network and from
among its users. Working to prevent catastrophic failures and having a plan to
recover from them is something you’ve learned to do. You now know the
importance of putting it all on paper as you proceed, too.
28 Chapter 2
07_599496 ch02.qxd 8/30/05 6:21 PM Page 28
29
Standard
Installation
IN THIS CHAPTER
■■ Exploring Your PC’s Components
■■ Checking for Supported Hardware
■■ Creating the Red Hat Boot Disk
■■ Starting the Installation
■■ Partitioning the Hard Disk
■■ Using Disk Druid to Partition Your Disks
■■ Configuring the Installation
■■ Selecting the Package Groups to Install
■■ Running Firstboot
This chapter explains the steps necessary to install Red Hat Enterprise Linux and
Fedora Core on a single system. You begin by making a list of your PC’s hard-
ware. You use this hardware inventory later when you begin the installation.
NOTE When you purchase Red Hat Enterprise Linux, you are eligible for
installation support from Red Hat. Also, an online installation manual is
available on the Red Hat Web site at www.redhat.com/docs. There is no
official support for Fedora Core from Red Hat.
NOTE The installation processes for Red Hat Enterprise Linux and Fedora

before you begin to install Red Hat Linux on your PC.
Processor
The central processing unit (CPU) — or just the processor — is an integrated cir-
cuit chip that performs nearly all control and processing functions in the PC.
Both Red Hat Enterprise Linux and Fedora Core run on an Intel 80386 proces-
sor or newer, as well as compatibles made by AMD or Cyrix. However, you
probably don’t want to use any processor older than a Pentium-class proces-
sor. Red Hat Linux also supports motherboards with multiple processors that
use the symmetric multiprocessing (SMP) Linux kernel.
Bus
The bus provides the electrical connection between the processor and its periph-
erals. Several types of PC buses exist on the motherboard with slots to accept
peripheral components. Each of the slots is colored to help in its identification.
The most recent is the Peripheral Component Interconnect (PCI) bus, and it is found
on all current production motherboards. The PCI slot is white and is available
in 32- and 64-bit form as well as 33 and 64 MHz. The new PCI-X standard will
30 Chapter 3
08_599496 ch03.qxd 8/30/05 6:20 PM Page 30
support speeds up to 533 MHz. Another type of slot is also based on the PCI bus
specifications, but offers significant advantages over the PCI bus. The Accelerated
Graphics Port (AGP) is a special slot on the motherboard designed to accept an
AGP graphics card. The AGP slot is brown. Another is the Industry Standard
Architecture (ISA) bus, formerly called the AT bus because IBM introduced it in
the IBM PC-AT computer in 1984. The ISA bus is black. Other, less frequently
encountered, buses because of their aging status include Extended Industry Stan-
dard Architecture (EISA); VESA local (VL-bus); and Micro Channel Architecture
(MCA). Red Hat Enterprise Linux supports all of these buses.
Memory
Referred to as random access memory, or RAM, is not a consideration in deter-
mining compatibility. This means that Linux does not care what kind of mem-

grated circuit that controls the monitor and causes the monitor to display out-
put. You can find out the name of the video chipset used in a video card from
the card’s documentation.
Your video card’s name may not be in the list at the Red Hat site. The impor-
tant thing to note is the name of the video chipset. Many popular video cards
made by different manufacturers use the same video chipsets. Look for the
name of the video chipsets listed at the Red Hat site. In nearly all cases, the Red
Hat installation program automatically detects the video chipset as it sets up
the X Window System.
Hard Drive
Red Hat Linux supports any IDE hard drive that your PC’s basic input/output
system (BIOS) supports, as long as the system BIOS supports the hard drive
without any additional drivers. This would include EIDE- and ATA-compatible
drives as well.
For hard drives connected to your PC through a Small Computer System
Interface (SCSI) controller card, Red Hat Linux must have a driver that enables
the SCSI controller to access and use the hard drive. If you have a recent SCSI
controller card, there is most likely a driver for it already included with the dis-
tribution.
Also supported are Serial Advanced Technology Attachment (SATA) drives,
which use serial technology instead of the parallel ATA technology currently
used by IDE drives. SATA provides a significant speed increase over IDE.
As for the size (storage capacity) of the drive, most new systems seem to
have drives 20 GB or larger. You should buy the highest capacity drive you can
afford.
Floppy Disk Drive
Linux drivers use the PC BIOS to access the floppy disk drive, so any floppy
disk drive is compatible with Red Hat Linux. The Red Hat installation program
can be started from the CD-ROM if your PC has one and is able to boot from it.
If not, you have to boot Red Hat Linux from a floppy disk drive during the

ROM discs are capable of storing up to 4.7 GB and are most frequently used to
record digital video, but can be used to hold any data.
CD-RW and DVD-R/RW and DVD+R/RW drives are used to create CDs
and DVDs, respectively. Either of these types of drives can be used in your Red
Hat system. Any IDE/ATAPI-compatible drive, as well as SCSI drives, will
work with Red Hat Enterprise Linux.
Sound Card
If you are configuring a server, you probably aren’t too interested in playing
sounds. But with Red Hat Linux you can play sound on a sound card to enjoy
multimedia programs and games. If you have a sound card, you can also play
audio CDs. Nearly all sound cards available today, whether built into the
motherboard or a separate card that plugs into a bus socket, are supported.
Standard Installation 33
08_599496 ch03.qxd 8/30/05 6:20 PM Page 33
Network Card
A network interface card (NIC) is necessary if you connect your Red Hat Linux
PC to a local area network (LAN), which is usually an Ethernet network. If you
are configuring a server, you certainly want to configure at least one network
card. Red Hat Enterprise Linux supports a variety of Ethernet network cards.
ARCnet and IBM’s Token Ring network are also supported. Check the hard-
ware list on the Red Hat site to see if your NIC is supported. Nearly all NICs
currently in use are supported.
For any Red Hat Linux PC connected to a network, you need the following
information:
■■ Hostname of the PC
■■ Domain name of the network
■■ Internet Protocol (IP) address of the PC
■■ Address of the gateway
■■ IP address of name servers
NOTE If you plan to use DHCP to obtain your IP information, you do not need

However, you may not be able to access the CD-ROM in MS-DOS mode because
the startup files (AUTOEXEC.BAT and CONFIG.SYS) may not be configured
correctly. To access the CD-ROM from DOS, you typically must add a CD-ROM
driver in CONFIG.SYS and add a line in AUTOEXEC.BAT that runs the MSCDEX
program. Try restarting your PC in MS-DOS mode and see whether the CD-ROM
can be accessed.
The Red Hat boot disk starts your PC and the Red Hat Linux installation
program. After you install Red Hat Linux, you no longer need the Red Hat
boot disk (except when you want to reinstall Red Hat Linux from the
CD-ROMs).
The Red Hat boot disk contains an initial version of the Red Hat Linux
installation program that you use to start Red Hat Enterprise Linux, prepare
the hard disk, and load the rest of the installation program. Creating the Red
Hat boot disk involves using a utility program called RAWRITE.EXE to copy a
special file called the Red Hat Linux boot image to a disk.
To create the Red Hat boot disk under Windows, follow these steps:
1. In Windows 95/98/ME open an MS-DOS window (select Start ➪ Pro-
grams ➪ MS-DOS Prompt). In Windows 2000 or XP, select Start ➪ Run
and enter cmd in the dialog box.
2. In the MS-DOS window, enter the following commands at the MS-DOS
prompt. (Our comments are in parentheses and your input is in
boldface.)
Standard Installation 35
08_599496 ch03.qxd 8/30/05 6:20 PM Page 35
a. d: (use the drive letter for the CD-ROM drive)
b. cd \dosutils
c. rawrite
d. Enter disk image source filename: \images\boot.img
e. Enter target disk drive: a
f. Insert a formatted disk into drive A and press ENTER.

using the Tab key to highlight Skip and then pressing Enter. If you purchased
Red Hat Enterprise Linux, you are not prompted to check the disks.
After the Linux kernel loads, it automatically starts the Red Hat Linux
installation program. This, in turn, starts the X Window System, which pro-
vides a graphical user interface for the installation.
You should compile all the configuration information explained earlier in
this chapter before you begin. If the installation program detects your hard-
ware, installing Red Hat Linux from the CD-ROM on a 200-MHz or better Pen-
tium PC should take 30 to 40 minutes.
NOTE During installation, the Red Hat installation program tries to determine
the hardware in your PC and alters the installation steps as required. For
example, if the installation program detects a network card, the program
displays the appropriate network configuration screens. If a network card is not
detected, the network configuration screens are not displayed. So, depending
on your specific hardware, the screens you see during installation may differ
from those shown in this section.
CROSS-REFERENCE If you run into any problems during the installation,
refer to Chapter 35 to learn how to troubleshoot common installation
problems.
You go through the following steps before moving on to disk setup and
installation:
1. The installation program starts the X Window System and displays a
welcome message that provides some explanatory information on the
left side of the screen. Take a look at this information and be sure to
click the Release Notes button. When you are finished reading, click
the Next button to proceed.
Standard Installation 37
08_599496 ch03.qxd 8/30/05 6:20 PM Page 37
2. After clicking Next, a list of languages to use during the installation is
displayed, as shown in Figure 3-1. Use your mouse to select the lan-

plan to use the system. After making your selection, click Next to go on
to the Disk Partitioning Setup screen shown in Figure 3-5.
The next major phase of installation involves partitioning the hard disk.
Figure 3-5 Choosing to use automatic partitioning or manual partitioning.
Partitioning the Hard Disk
Red Hat Linux requires you to partition and prepare a hard disk before you
can begin installation. With a new PC that you purchase from a vendor, you
usually do not perform this step because the vendor normally takes care of
preparing the hard disk and installing the operating system and other appli-
cations on the hard disk. Because you are installing Red Hat Linux from
scratch, however, you have to perform this crucial step yourself. As you see in
the following sections, this task is just a matter of following instructions.
42 Chapter 3
08_599496 ch03.qxd 8/30/05 6:20 PM Page 42
The Red Hat Linux installation program offers you two choices for parti-
tioning your hard drive. You can choose to have the installation program auto-
matically partition your disk or you can choose to use Disk Druid to manually
partition your drives. If you select automatic partitioning, the installation pro-
gram does all the hard work for you and creates the partitions and allocates
space for your file system. If you want to manually partition your disks, go to
the section “Using Disk Druid to Partition Your Disks.”
1. To use automatic partitioning be sure the radio button in front of this
choice is checked, and click Next on the Disk Partitioning Setup screen.
2. The Automatic Partitioning screen, shown in Figure 3-6, appears. Here
you decide how automatic partitioning should handle existing parti-
tions. You can choose to remove Linux partitions, remove all partitions,
or keep all partitions and use free space. If you are installing a new sys-
tem, you should choose to remove all partitions. A warning screen will
appear asking if you are sure you want to remove all partitions. Click
Yes to continue.

Because Linux treats a device as a file in the /dev directory, the hard disk
names start with /dev. Table 3-1 lists the hard disk and floppy drive names
that you may have to use.
Table 3-1 Hard Disk and Floppy Drive Names
NAME DESCRIPTION
/dev/hda First Integrated Drive Electronics (IDE) hard drive (the C drive in
DOS and Windows) connected to the first IDE controller as the
master drive
/dev/hdb Second (IDE) hard drive connected to the first IDE controller as the
slave drive
/dev/hdc First (IDE) hard drive connected to the second IDE controller as the
master drive
/dev/hdd Second (IDE) hard drive connected to the second IDE controller as
the slave drive
/dev/sda First Small Computer System Interface (SCSI) drive or first SATA drive
/dev/sdb Second SCSI drive or second SATA drive
/dev/fd0 First floppy drive (the A drive in DOS)
/dev/fd1 Second floppy drive (the B drive in DOS)
Standard Installation 45
08_599496 ch03.qxd 8/30/05 6:20 PM Page 45
TIP When Disk Druid displays the list of partitions, the partition names take
the form hda1, hda2, and so on. Linux constructs each partition name by
appending the partition number (1 through 4 for the four primary partitions on
a hard disk) to the disk’s name. Therefore, if your PC’s single IDE hard drive has
two partitions, notice that the installation program uses hda1 and hda2 as the
names of these partitions.
Mounting a File System
In Red Hat Linux, you use a physical disk partition by associating it with a
specific part of the file system. This arrangement is a hierarchical directory —
a directory tree. If you have more than one disk partition (you may have other

Red Hat Enterprise Linux supports virtual memory and can make use of a
swap partition. When you create the Linux partitions, you also create a swap
partition. With the Disk Druid utility program, described in the next section,
creating a swap partition is easy. Simply mark a partition type as a swap
device, choose the size, and let Disk Druid perform the necessary tasks.
Preparing Disk Partitions
After you select Disk Druid to manually partition your disks, the Disk Setup
screen, shown in Figure 3-7, reappears.
Before beginning to partition the drive, consider exactly how you want to
create the partitions. You must create one partition on the drive to be used as
the root (/) partition. This works well in most cases, but it can cause some prob-
lems. If the root partition should become full, the system could crash. Many
times the partition fills up because of system logging, email, and print queue
files. These files are all written to the /var directory by default, so it may be a
good idea to create a separate partition for /var to prevent the root partition
from filling up with system logs, email, and print files. You might also want to
create a separate partition for your user’s home directories (/home) if you have
a large number of users.
You also need to create a swap partition. A swap partition is used for virtual
memory to hold data that is too large to fit into system RAM. Your swap par-
tition should be at least 32 MB or two times your system’s RAM, whichever is
larger.
Disk Druid gathers information about the hard drives on your system and
displays a list of disk drives in the lower part of the screen and the current par-
tition information for one of the drives in the Partitions area in the upper part.
For each partition, Disk Druid shows seven fields:
Standard Installation 47
08_599496 ch03.qxd 8/30/05 6:20 PM Page 47
■■ Device refers to the partition’s device name. For example, hda1 is the
first partition on the first IDE drive.

port a RAID device.
NOTE The reference to RAID in this section is for a software RAID
configuration.
48 Chapter 3
08_599496 ch03.qxd 8/30/05 6:20 PM Page 48