Choosing How to Install Fedora
Fedora can be installed in a variety of ways using different techniques and hardware.
Most users will install Fedora by booting to the installation directly from a CD-ROM. Other options include
Booting to an installation using a floppy diskette.
Booting to an installation using Fedora's mini CD-ROM.
Using a hard drive partition to hold the installation software.
Booting from a DOS command line.
Booting via a virtual network session. (See the file Release Notes included on the first Fedora Core CD-ROM for details.)
Booting to an installation and installing software over a network using FTP or HTTP protocols.
Booting to an installation and installing software from an NFS-mounted hard drive.
How you choose to install (and use) Fedora depends on your system's hardware, networking capabilities, corporate information service policy, or personal preference. The following sections describe the issues surrounding each of these types of installation.
Installing from CD-ROM
Most PCs' BIOS support booting directly from a CD-ROM drive, and offer the capability to set a specific order of devices (such as floppy, hard drive, CD-ROM, or USB) to search for bootable software. Turn on your PC, set your PC's BIOS if required (usually accessed by pressing an F or Del key after powering on); then insert Fedora Core's first CD-ROM, and boot to install Fedora.
To use this installation method, your computer must support booting from CD-ROM, and the CD-ROM drive must be recognizable by the Linux kernel. You can verify this by checking your BIOS and then booting your PC.
Older PCs with some CD-ROM drives might prove problematic when you desire to boot to an install using optical media. The good news is that this should no longer be a problem with most post-1995 personal computers. However, you can consult Table 3.1, which lists a driver disk image that can be used to support older drives.
The file boot.iso listed in Table 3.1 is a 4.7MB CD-ROM image found under the images directory on the first Fedora Core CD-ROM. The image can be burned onto a CDR, mini CDR, or business-card sized CDR and supports booting to a network install. This is a convenient way to boot to a network install on a PC with a bootable CD-ROM drive, but no installed floppy drive, or when you don't want to use multiple floppies during an install requiring driver diskettes.
You burn the image onto optical media using the cdrecord command. For example, copy the file to your hard drive, insert a blank CDR into your CDRW drive, and then use a command line like so:
# cdrecord -v speed=4 dev=0,0,0 -data -eject boot.iso
This example will create a bootable CD-ROM, and then eject the new CD-ROM after writing the image. The speed (4 in this example) depends on the capabilities of your CD writing device. The device numbers are those returned by running cdrecord with its scanbus option, like so:
# cdrecord -scanbus
Booting to an Install from DOS
As previously mentioned, you can use a DOS utility such as LOADLIN (or BOOTLIN) to either directly boot to an install from the CD-ROM or to load the install kernel.
See the dosutils directory on the first Fedora Core CD-ROM included with this book, and read the README file under the dosutils directory for an overview of the DOS utilities. The directory contains a one-line DOS batch file (.bat file) that can help boot to an install:
loadlin autoboot\vmlinuz initrd=autoboot\initrd.img
In this example, the LOADLIN command will boot the install kernel residing under the dosutils/autoboot directory, and then load the installation software to launch an install.
A half-dozen DOS and Windows-based utilities will be found under the dosutils directory. The software also includes programs such as the fips.exe FAT partition editor, and a Windows equivalent command of RAWRITE (to create boot diskettes) named RAWRITEWIN. Documentation is also included to help you get started.
Making an Installation Boot Diskette
Your installation can also be started using a boot floppy. A boot floppy can be used to support booting the Install CD-ROM on older CD-ROM drives. A boot floppy can also be used to start an install over a network when the PC does not have a CD-ROM drive but has a network interface card (NIC). Boot floppies are created from floppy images (.img files). These images are contained in the images directory on the first Fedora Core CD-ROM.
Fedora Core provides a number of images, as listed in Table 3.1.
Table 3.1 Fedora Boot and Driver Disk Images
CD-ROM boot image supporting single-media network installs
Enables booting to an install using all install methods
Utility image containing various hardware drivers, such as SCSI
Auxiliary network interface card drivers to support network installs
PCMCIA driver disk with support for SCSI, network, and other adapters
Directory containing a PXE-enabled Linux boot kernel and RAM-disk image use to support booting various installs
Most of the images listed in Table 3.1 support booting to an install. The pxeboot directory contains a kernel that supports a remote booting protocol named PXE that enables installation, or upgrades of Fedora Core for network-only PCs. Use of this software requires a properly configured DHCP server and a supported BIOS. Although it might be possible to use a floppy boot disk, this method is not supported by Fedora. Browse to http://www.compaq.com/products/servers/linux/redhat-whitepapers.html and read the Compaq white paper titled "Configuring a Preboot eXecution Environment (PXE) using Red Hat Linux 7.1 on Compaq ProLiant Servers" to see one way of booting to an install using PXE. Browse to http://www.europe.redhat.com/documentation/mini-HOWTO/Remote-Boot.php3 to read how to set up remote boot environments using older versions of Red Hat.
If your PC's BIOS supports booting from a USB device, you can use a USB floppy drive to boot to an install. It is also possible to boot from an IEEE-1394 (FireWire) CD-ROM if supported by your PC's BIOS.
You can create the boot floppies using the DOS RAWRITE command or the Linux dd command. You'll need one or more blank floppies. Use the RAWRITE command after starting DOS like this:
Follow the prompts to create the images, entering a source filename and a target drive (such as A or B). To get a copy of the latest version of RAWRITE, browse to http://www.tux.org/pub/dos/rawrite/.
To create a boot diskette while running Linux, use the dd command. The first step is to insert the first Fedora Core CD-ROM and to make sure that it is mounted:
$ df Filesystem 1k-blocks Used Available Use% Mounted on /dev/hda2 18714368 10410280 7353444 59% / /dev/hda1 46636 13247 30981 30% /boot none 120016 0 120016 0% /dev/shm /dev/cdrom 655808 655808 0 100% /mnt/cdrom
This example uses the disk filesystem or the df command to show free hard drive space and currently mounted devices and partitions. As you can see, the contents of the first Fedora Core CD-ROM is found under the /mnt/cdrom directory. If the CD-ROM is not mounted, use the mount command to manually mount the CD-ROM. You will need to be the superuser or Linux root operator to do this (see Chapter 5, "First Steps with Fedora," for details about how to become root):
# mount -t iso9660 /dev/cdrom /mnt/cdrom
You can then use the convert and copy (dd) command like so:
# dd if=/mnt/cdrom/images/nameofimage.img of=/dev/fd0
This will take the contents of the specified image file (from Table 3.1) and create a diskette in the DOS drive A (represented by /dev/fd0). Use /dev/fd1 if you want to use an installed secondary floppy drive.
PC notebook users installing via a network or external CD-ROM drive using a PCMCIA adapter should also create the pcmciadd.img diskette.
Hard Drive Partition Installation
Another way to install Fedora Core is to use a Linux or DOS hard partition to either boot to the install or hold the software required for an install. The partition must be large enough to hold .iso images (binary images of a CD-ROM). Copy the images of the first and second Fedora Core CD-ROMs into a directory on the local hard drive. If you use this type of install and don't need the required hard drive space later on, keep the images installed on the hard drive because Fedora can then be quickly reinstalled from the partition.
The .iso images can be downloaded from the Fedora Project or a mirror FTP site (see http://fedora.redhat.com for links). To perform this installation, you will need to know the hard drive's device name (such as /dev/hdb), along with the partition number and the name of directory containing the images (such as /dev/hdb1 and /redhat/images; if you simply copy the images to the formatted DOS or Linux partition, you don't need the directory information). See Chapter 10, "Managing the Filesystem," for more information about using hard drive partitions.
Installing Using a Network
Fedora can be installed using a local network (or even over the Internet if you have broadband access). You will need access to a Web, FTP, or NFS server hosting the installation packages. To boot to a network install, use a network boot floppy, a bootable CD-ROM created using the boot.iso boot image, or the first Fedora Core CD-ROM included with this book. Boot your PC with the boot floppy or, if you use CD-ROM, type linux askmethod at the boot prompt. Follow the prompts, and you'll then be asked to choose the type of network installation.
Just press Enter at the boot prompt if you boot to a network install using a CDR created with the boot.iso image. You'll boot a graphical network install.
To install using FTP, select the network IP address assignment for your target PC, such as DHCP, or manually enter an IP address along with optional gateway IP address and nameserver addresses. You'll then be asked for the FTP site name. You can enter the name or IP address of a remote FTP server hosting the Fedora Core release. The name of the remote directory will depend on where the Fedora install files are located on the remote server.
Installing Fedora using the File Transfer Protocol (FTP) will require access to an FTP server (see Chapter 18, "Secure File Transfer Protocol (FTP) Service," to see how to set up a server and use FTP). You'll need to know the hostname or IP address of the server, along with the path (directory) holding the Fedora Core software. One way to prepare a server to host installs is to
Create a directory named Fedora under the FTP server's pub directory. The directory will usually be /var/ftp/pub on a Linux server.
Create a directory named base and a directory named RPMS underneath the Fedora directory.
Copy or download all RPM packages included with Fedora Core into the pub/Fedora/RPMS directory.
Copy all original base files (comps.rpm, comps.xml, hdlist, hdlist2, hdstg2.img, netstg2.img, stage2.img, TRANS.TBL) from the first CD-ROM's base directory into the pub/Fedora/base directory.
Using this approach, enter pub when asked for the name of the remote directory holding the Fedora Core install software.
Installing Fedora Core using a remotely mounted Network File System (NFS) is similar to a hard drive installation, but requires access to an NFS server. You'll need access permission, a permitted IP address or hostname for your computer, the hostname or IP address of the NFS server, and the path to the Fedora Core software. See Chapter 13, "Network Connectivity," for more information about NFS and network addressing.
To install Fedora using HTTP, you will need the hostname or IP address of the remote Web server, along with the directory containing Fedora's software. See Chapter 16, "Apache Web Server Management," to see how to set up a Web server.
See Chapter 18 for details on how to configure the vsftpd FTP server. Chapter 16 provides information on how to set up and configure Apache for Web service. See Chapter 13 for Samba settings. Note that you can have your server perform all three duties.