how to install a linux Server

1 Introduction to Linux

chap-introIntroduction to Linux     Linux is quite possibly the most important free software achievement since the original Space War, or, more recently, Emacs. It has developed into an operating system for business, education, and personal productivity. Linux is no longer only for UNIX wizards who sit for hours in front of a glowing console (although we assure you that many users fall into this category). This book will help you get the most from Linux.
Linux (pronounced with a short i, as in LIH-nucks) is a UNIX operating system clone which runs on a variety of platforms, especially personal computers with Intel 80386 or better processors. It supports a wide range of software, from TeX, to the X Window System, to the GNU C/C++ compiler, to TCP/IP. It's a versatile, bona fide implementation of UNIX, freely distributed under the terms of the GNU General Public License (see Appendix C).
Linux can turn any 80386 or better personal computer into a workstation that puts the full power of UNIX at your fingertips. Businesses install Linux on entire networks of machines, and use the operating system to manage financial and hospital records, distributed computing environments, and telecommunications. Universities worldwide use Linux to teach courses on operating system programming and design. Computing enthusiasts everywhere use Linux at home for programming, productivity, and all-around hacking.
What makes Linux so different is that it is a free implementation of UNIX. It was and still is developed cooperatively by a group of volunteers, primarily on the Internet, who exchange code, report bugs, and fix problems in an open-ended environment. Anyone is welcome to join the Linux development effort. All it takes is interest in hacking a free UNIX clone, and some programming know-how. The book in your hands is your tour guide.

1.1 About this book.

This book is an installation and entry-level guide to Linux. The purpose is to get new users up and running by consolidating as much important material as possible into one book. Instead of covering volatile technical details which tend to change with rapid development, we give you the straight background to find out more on your own.
Linux is not difficult to install and use. However, as with any implementation of UNIX, there is often black magic involved to get everything working correctly. We hope that this book will get you on the Linux tour bus and show you how great an operating system can be.
In this book, we cover the following topics:

• What is Linux? The design and philosophy of this unique operating system, and what it can do for you.
• Details of running Linux, including suggestions on recommended hardware configuration.
• Specific instructions to install various Linux distributions, including Debian, Red Hat Software, and Slackware.
• A brief, introductory UNIX tutorial for users with no previous UNIX experience. This tutorial should provide enough material for novices to find their way around the system.
• An introduction to system administration under Linux. This covers the most important tasks that Linux administrators need to perform, like creating user accounts and managing file systems.
• Information on configuring more advanced features of Linux, like the X Window System, TCP/IP networking, and electronic mail and news.

This book is for the personal computer user who wishes to get started with Linux. We don't assume previous UNIX experience but do expect novices to refer to other material along the way. For those unfamiliar with UNIX, a list of useful references is given in Appendix A. In general, this book is meant to be read in addition to another book on basic UNIX concepts.

1.2 A brief history of Linux.

UNIX is one of the most popular operating systems worldwide because of its large support base and distribution. It was originally developed at AT&T as a multitasking system for minicomputers and mainframes in the 1970's, but has since grown to become one of the most widely-used operating systems anywhere, despite its sometimes confusing interface and lack of central standardization.
Many hackers feel that UNIX is the Right Thing--the One True Operating System. Hence, the development of Linux by an expanding group of UNIX hackers who want to get their hands dirty with their own system.
Versions of UNIX exist for many systems, from personal computers to supercomputers like the Cray Y-MP. Most versions of UNIX for personal computers are expensive and cumbersome. At the time of this writing, a one-machine version of UNIX System V for the 386 runs about US$1500.
Linux is a free version of UNIX developed primarily by Linus Torvalds at the University of Helsinki in Finland, with the help of many UNIX programmers and wizards across the Internet. Anyone with enough know-how and gumption can develop and change the system. The Linux kernel uses no code from AT&T or any other proprietary source, and much of the software available for Linux was developed by the GNU project of the Free Software Foundation in Cambridge, Massachusetts, U.S.A. However, programmers from all over the world have contributed to the growing pool of Linux software.
Linux was originally developed as a hobby project by Linus Torvalds. It was inspired by Minix, a small UNIX system developed by Andy Tanenbaum. The first discussions about Linux were on the Usenet newsgroup, comp.os.minix. These discussions were concerned mostly with the development of a small, academic UNIX system for Minix users who wanted more.
The very early development of Linux mostly dealt with the task-switching features of the 80386 protected-mode interface, all written in assembly code. Linus writes,

``After that it was plain sailing: hairy coding still, but I had some devices, and debugging was easier. I started using C at this stage, and it certainly speeds up development. This is also when I started to get serious about my megalomaniac ideas to make `a better Minix than Minix.' I was hoping I'd be able to recompile gcc under Linux someday... ``Two months for basic setup, but then only slightly longer until I had a disk driver (seriously buggy, but it happened to work on my machine) and a small file system. That was about when I made 0.01 available (around late August of 1991): it wasn't pretty, it had no floppy driver, and it couldn't do much of anything. I don't think anybody ever compiled that version. But by then I was hooked, and didn't want to stop until I could chuck out Minix.''

No announcement was ever made for Linux version 0.01. The 0.01 sources weren't even executable. They contained only the bare rudiments of the kernel source and assumed that you had access to a Minix machine to compile and experiment with them.
On October 5, 1991, Linus announced the first ``official'' version of Linux, which was version 0.02. At that point, Linus was able to run bash (the GNU Bourne Again Shell) and gcc (the GNU C compiler), but not much else. Again, this was intended as a hacker's system. The primary focus was kernel development--user support, documentation, and distribution had not yet been addressed. Today, the Linux community still seems to treat these issues as secondary to ``real programming''--kernel development.
As Linus wrote in comp.os.minix,

``Do you pine for the nice days of Minix-1.1, when men were men and wrote their own device drivers? Are you without a nice project and just dying to cut your teeth on an OS you can try to modify for your needs? Are you finding it frustrating when everything works on Minix? No more all-nighters to get a nifty program working? Then this post might be just for you. ``As I mentioned a month ago, I'm working on a free version of a Minix-look-alike for AT-386 computers. It has finally reached the stage where it's even usable (though may not be, depending on what you want), and I am willing to put out the sources for wider distribution. It is just version 0.02...but I've successfully run bash, gcc, gnu-make, gnu-sed, compress, etc. under it.''

After version 0.03, Linus bumped up the version number to 0.10, as more people started to work on the system. After several further revisions, Linus increased the version number to 0.95 in March, 1992, to reflect his expectation that the system was ready for an ``official'' release soon. (Generally, software is not assigned the version number 1.0 until it is theoretically complete or bug-free.). Almost a year and a half later, in late December of 1993, the Linux kernel was still at version 0.99.pl14--asymptotically approaching 1.0. At the time of this writing, the current stable kernel version is 2.0 patchlevel 33, and version 2.1 is under development.
Most of the major, free UNIX software packages have been ported to Linux, and commercial software is also available. More hardware is supported than in the original kernel versions. Many people have executed benchmarks on 80486 Linux systems and found them comparable with mid-range workstations from Sun Microsystems and Digital Equipment Corporation. Who would have ever guessed that this ``little'' UNIX clone would have grown up to take on the entire world of personal computing?

1.3 System features.

Linux supports features found in other implementations of UNIX, and many which aren't found elsewhere. In this section, we'll take a nickel tour of the features of the Linux kernel.
Linux is a complete multitasking, multiuser operating system, as are all other versions of UNIX. This means that many users can log into and run programs on the same machine simultaneously.
The Linux system is mostly compatible with several UNIX standards (inasmuch as UNIX has standards) at the source level, including IEEE POSIX.1, UNIX System V, and Berkely System Distribution UNIX. Linux was developed with source code portability in mind, and it's easy to find commonly used features that are shared by more than one platform. Much of the free UNIX software available on the Internet and elsewhere compiles under Linux ``right out of the box.'' In addition, all of the source code for the Linux system, including the kernel, device drivers, libraries, user programs, and development tools, is freely distributable.
Other specific internal features of Linux include POSIX job control (used by shells like csh and bash), pseudoterminals ( pty devices), and support for dynamically loadable national or customized keyboard drivers. Linux supports virtual consoles that let you switch between login sessions on the same system console. Users of the screen program will find the Linux virtual console implementation familiar.
The kernel can emulate 387-FPU instructions, and systems without a math coprocessor can run programs that require floating-point math capability.
Linux supports various file systems for storing data, like the ext2 file system, which was developed specifically for Linux. The Xenix and UNIX System V file systems are also supported, as well as the Microsoft MS-DOS and Windows 95 VFAT file systems on a hard drive or floppy. The ISO 9660 CD-ROM file system is also supported. We'll talk more about file systems in chapters 2 and 4.
Linux provides a complete implementation of TCP/IP networking software. This includes device drivers for many popular Ethernet cards, SLIP (Serial Line Internet Protocol) and PPP (Point-to-Point Protocol), which provide access to a TCP/IP network via a serial connection, PLIP (Parallel Line Internet Protocol), and NFS (Network File System). The complete range of TCP/IP clients and services is also supported, which includes FTP, telnet, NNTP, and SMTP. We'll talk more about networking in Chapter .
The Linux kernel is developed to use protected-mode features of Intel 80386 and better processors. In particular, Linux uses the protected-mode, descriptor based, memory-management paradigm, and other advanced features. Anyone familiar with 80386 protected-mode programming knows that this chip was designed for multitasking systems like UNIX. Linux exploits this functionality.
The kernel supports demand-paged, loaded executables. Only those segments of a program which are actually in use are read into memory from disk. Also, copy-on-write pages are shared among executables. If several instances of a program are running at once, they share physical memory, which reduces overall usage.
In order to increase the amount of available memory, Linux also implements disk paging. Up to one gigabyte of swap space may be allocated on disk (upt to 8 partitions of 128 megabytes each). When the system requires more physical memory, it swaps inactive pages to disk, letting you run larger applications and support more users. However, swapping data to disk is no substitute for physical RAM, which is much faster.
The Linux kernel also implements a unified memory pool for user programs and disk cache. All free memory is used by the cache, which is reduced when running large programs.
Executables use dynamically linked, shared libraries: code from a single library on disk. This is not unlike the SunOS shared library mechanism. Executable files occupy less disk space, especially those which use many library functions. There are also statically linked libraries for object debugging and maintaining ``complete'' binary files when shared libraries are not installed. The libraries are dynamically linked at run time, and the programmer can use his or her own routines in place of the standard library routines.
To facilitate debugging, the kernel generates core dumps for post-mortem analysis. A core dump and an executable linked with debugging support allows a developer to determine what caused a program to crash.

1.4 Software features.

Virtually every utility one would expect of a standard UNIX implementation has been ported to Linux, including basic commands like ls, awk, tr, sed, bc, and more. The familiar working environment of other UNIX systems is duplicated on Linux. All standard commands and utilities are included. (Novice UNIX or Linux users should see Chapter 3 for an introduction to basic UNIX commands.)
Many text editors are available, including vi, ex, pico, jove, and GNU emacs, and variants like Lucid emacs, which incorporates extensions of the X Window System, and joe. The text editor you're accustomed to using has more than likely been ported to Linux.
The choice of a text editor is an interesting one. Many UNIX users prefer ``simple'' editors like vi. (The original author wrote this book with vi.) But vi has many limitations due to its age, and modern editors like emacs have gained popularity. emacs supports a complete, Lisp based macro language and interpreter, powerful command syntax, and other extensions. There are emacs macro packages which let you read electronic mail and news, edit directory contents, and even engage in artificially intelligent psychotherapy sessions (indispensible for stressed-out Linux hackers).
Most of the basic Linux utilities are GNU software. GNU utilities support advanced features that are not found in the standard versions of BSD and UNIX System Vprograms. For example, the GNU vi clone, elvis, includes a structured macro language that differs from the original implementation. However, GNU utilities are intended to remain compatible with their BSD and System V counterparts. Many people consider the GNU versions to be superior to the originals.
A shell is a program which reads and executes commands from the user. In addition, many shells provide features like job control, managing several processes at once, input and output redirection, and a command language for writing shell scripts. A shell script is a program in the shell's command language and is analogous to a MS-DOS batch file.
Many types of shells are available for Linux. The most important difference between shells is the command language. For example, the C SHell (csh) uses a command language similar to the C programming language. The classic Bourne SHell sh uses another command language. The choice of a shell is often based on the command language it provides, and determines, to a large extent, the qualities of your working environment under Linux.
The GNU Bourne Again Shell (bash) is a variation of the Bourne Shell which includes many advanced features like job control, command history, command and filename completion, an emacs-like interface for editing command lines, and other powerful extensions to the standard Bourne Shell language. Another popular shell is tcsh, a version of the C Shell with advanced functionality similar to that found in bash. Other shells include zsh, a small Bourne-like shell; the Korn Shell (ksh); BSD's ash; and rc, the Plan 9 shell.
If you're the only person using the system and refer to use vi and bash exclusively as your editor and shell, there's no reason to install other editors or shells. This ``do it yourself'' attitude is prevalent among Linux hackers and users.

1.4.1 Text processing and word processing.

Almost every computer user needs a method of preparing documents. In the world of personal computers, word processing is the norm: editing and manipulating text in a ``What-You-See-Is-What-You-Get'' (WYSIWYG) environment and producing printed copies of the text, complete with graphics, tables, and ornamentation.
Commercial word processors from Corel, Applix, and Star Division are available in the UNIX world, but text processing, which is quite different conceptually, is more common. In text processing systems, text is entered in a page-description language, which describes how the text should be formatted. Rather than enter text within a special word processing environment, you can modify text with any editor, like vi or emacs. Once you finish entering the source text (in the typesetting language), a separate program converts the source to a format suitable for printing. This is somewhat analogous to programming in a language like C, and ``compiling'' the document into printable form.
Many text processing systems are available for Linux. One is groff, the GNU version of the classic troff text formatter originally developed by Bell Labs and still used on many UNIX systems worldwide. Another modern text processing system is TeX, developed by Donald Knuth of computer science fame. Dialects of TeX, like LaTeX, are also available.
Text processors like TeX and groff differ mostly in the syntax of their formatting languages. The choice of one formatting system over another is based upon what utilities are available to satisfy your needs, as well as personal taste.
Many people consider groff's formatting language to be a bit obscure and use find TeX more readable. However, groff produces ASCII output which can be viewed on a terminal more easily, while TeX is intended primarily for output to a printing device. Various add-on programs are required to produce ASCII output from TeX formatted documents, or convert TeX input to groff format.
Another program is texinfo, an extension to TeX which is used for software documentation developed by the Free Software Foundation. texinfo can produce printed output, or an online-browsable hypertext ``Info'' document from a single source file. Info files are the main format of documentation used in GNU software like emacs.
Text processors are used widely in the computing community for producing papers, theses, magazine articles, and books. (This book is produced using LaTeX.) The ability to process source language as a text file opens the door to many extensions of the text processor itself. Because a source document is not stored in an obscure format that only one word processor can read, programmers can write parsers and translators for the formatting language, and thus extend the system.
What does a formatting language look like? In general, a formatted source file consists mostly of the text itself, with control codes to produce effects like font and margin changes, and list formatting.
Consider the following text:

Mr. Torvalds: We are very upset with your current plans to implement post-hypnotic suggestions in the Linux terminal driver code. We feel this way for three reasons:

1. Planting subliminal messages in the terminal driver is not only immoral, it is a waste of time;
2. It has been proven that ``post-hypnotic suggestions'' are ineffective when used upon unsuspecting UNIX hackers;
3. We have already implemented high-voltage electric shocks, as a security measure, in the code for login.

We hope you will reconsider.

This text might appear in the LaTeX formatting language as the following:

\begin{quote}
Mr. Torvalds: 

We are very upset with your current plans to implement 
{\em post-hypnotic suggestions\/} in the {\bf Linux} terminal
driver code. We feel this way for three reasons:
\begin{enumerate}
\item Planting subliminal messages in the kernel driver is not only
      immoral, it is a waste of time;
\item It has been proven that ``post-hypnotic suggestions''
      are ineffective when used upon unsuspecting UNIX hackers;
\item We have already implemented high-voltage electric shocks, as
      a security measure, in the code for {\tt login}.
\end{enumerate}
We hope you will reconsider. 
\end{quote}

The author enters the text using any text editor and generates formatted output by processing the source with LaTeX. At first glance, the typesetting language may appear to be obscure, but it's actually quite easy to understand. Using a text processing system enforces typographical standards when writing. All the enumerated lists within a document will look the same, unless the author modifies the definition of an enumerated list. The goal is to allow the author to concentrate on the text, not typesetting conventions.
When writing with a text editor, one generally does not think about how the printed text will appear. The writer learns to visualize the finished text's appearance from the formatting commands in the source.
WYSIWYG word processors are attractive for many reasons. They provide an easy-to-use visual interface for editing documents. But this interface is limited to aspects of text layout which are accessible to the user. For example, many word processors still provide a special format language for producing complicated expressions like mathematical formulae. This is text processing, albeit on a much smaller scale.
A not-so-subtle benefit of text processing is that you specify exactly which format you need. In many cases, the text processing system requires a format specification. Text processing systems also allow source text to be edited with any text editor, instead of relying on format codes which are hidden beneath a word processor's opaque user interface. Further, the source text is easily converted to other formats. The tradeoff for this flexibility and power is the lack of WYSIWYG formatting.
Some programs let you preview the formatted document on a graphics display device before printing. The xdvi program displays a ``device independent'' file generated by the TeX system under X. Applications like xfig and gimp provide WYSIWYG graphics interfaces for drawing figures and diagrams, which are subsequently converted to text processing language for inclusion in your document.
Text processors like troff were around long before WYSIWYG word processing was available. Many people still prefer their versatility and independence from a graphics environment.
Many text-processing-related utilities are available. The powerful METAFONT system, which is used to design fonts for TeX, is included in the Linux port of TeX. Other programs include ispell, an interactive spelling checker and corrector; makeindex, which generates indices in LaTeX documents; and many other groff and TeXbased macro packages which format many types of technical and mathematical texts. Conversion programs that translate between TeX or groff source to a myriad of other formats are also available.
A newcomer to text formatting is YODL, written by Karel Kubat. YODL is an easy-to-learn language with filters to produce various output formats, like LaTeX, SGML, and HTML.

1.4.2 Programming languages and utilities.

Linux provides a complete UNIX programming environment which includes all of the standard libraries, programming tools, compilers, and debuggers which you would expect of other UNIX systems.
Standards like POSIX.1 are supported, which allows software written for Linux to be easily ported to other systems. Professional UNIX programmers and system administrators use Linux to develop software at home, then transfer the software to UNIX systems at work. This not only saves a great deal of time and money, but also lets you work in the comfort of your own home. (One of the authors uses his system to develop and test X Window System applications at home, which can be directly compiled on workstations elsewhere.) Computer Science students learn UNIX programming and explore other aspects of the system, like kernel architecture.
With Linux, you have access to the complete set of libraries and programming utilities and the complete kernel and library source code.
Within the UNIX software world, systems and applications are often programmed in C or C++. The standard C and C++ compiler for Linux is GNU gcc, which is an advanced, modern compiler that supports C++, including AT&T 3.0 features, as well as Objective-C, another object-oriented dialect of C.
Besides C and C++, other compiled and interpreted programming languages have been ported to Linux, like Smalltalk, FORTRAN, Java, Pascal, LISP, Scheme, and Ada (if you're masochistic enough to program in Ada, we aren't going to stop you). In addition, various assemblers for writing protected-mode 80386 code are available, as are UNIX hacking favorites like Perl (the script language to end all script languages) and Tcl/Tk (a shell-like command processing system which has support for developing simple X Window System applications).
The advanced gdb debugger can step through a program one line of source code at a time, or examine a core dump to find the cause of a crash. The gprof profiling utility provides performance statistics for your program, telling you where your program spends most of its execution time. As mentioned above, the emacs text editor provides interactive editing and compilation environments for various programming languages. Other tools include GNU make and imake, which manage compilation of large applications, and RCS, a system for source code locking and revision control.
Finally, Linux supports dynamically linked, shared libraries (DLLs), which result in much smaller binaries. The common subroutine code is linked at run-time. These DLLs let you override function definitions with your own code. For example, if you wish to write your own version of the malloc() library routine, the linker will use your new routine instead of the one in the libraries.

Introduction to the X Window System.

The X Window System, or simply X, is a standard graphical user interface (GUI) for UNIX machines and is a powerful environment which supports many applications. Using the X Window System, you can have multiple terminal windows on the screen at once, each having a different login session. A pointing device like a mouse is often used with X, although it isn't required.
Many X-specific applications have been written, including games, graphics and programming utilities, and documentation tools. Linux and X make your system a bona fide workstation. With TCP/IP networking, your Linux machine can display X applications running on other machines.
The X Window System was originally developed at the Massachusetts Institute of Technology and is freely distributable. Many commercial vendors have distributed proprietary enhancements to the original X Window System as well. The version of X for Linux is XFree86, a port of X11R6 which is freely distributable. XFree86 supports a wide range of video hardware, including VGA, Super VGA, and accelerated video adaptors. XFree86 is a complete distribution of the X Windows System software, and contains the X server itself, many applications and utilities, programming libraries, and documents.
Standard X applications include xterm, a terminal emulator used for most text-based applications within a window, xdm, which handles logins, xclock, a simple clock display, xman, a X-based manual page reader, and xmore. The many X applications available for Linux are too numerous to mention here, but their number includes spreadsheets, word processors, graphics programs, and web browsers like the Netscape Navigator. Many other applications are available separately. Theoretically, any application written for X should compile cleanly under Linux.
The interface of the X Window System is controlled largely by the window manager. This user-friendly program is in charge of the placement of windows, the user interface for resizing and moving them, changing windows to icons, and the appearance of window frames, among other tasks. XFree86 includes twm, the classic MIT window manager, and advanced window managers like the Open Look Virtual Window Manager (olvwm) are available. Popular among Linux users is fvwm--a small window manager that requires less than half the memory of twm. It provides a 3-dimensional appearance for windows and a virtual desktop. The user moves the mouse to the edge of the screen, and the desktop shifts as though the display were much larger than it really is. fvwm is greatly customizable and allows access to functions from the keyboard as well as mouse. Many Linux distributions use fvwm as the standard window manager. A version of fvwm called fvwm95-2 offers Microsoft Windows 95-like look and feel.
The XFree86 distribution includes programming libraries for wily programmers who wish to develop X applications. Widget sets like Athena, Open Look, and Xaw3D are supported. All of the standard fonts, bitmaps, manual pages, and documentation are included. PEX (a programming interface for 3-dimensional graphics) is also supported.
Many X application programmers use the proprietary Motif widget set for development. Several vendors sell single and multiple user licenses for binary versions of Motif. Because Motif itself is relatively expensive, not many Linux users own it. However, binaries statically linked with Motif routines can be freely distributed. If you write a program using Motif, you may provide a binary so users without the Motif libraries can use the program.
A major caveat to using the X Window System is its hardware requirements. A 80386-based CPU with 4 megabytes of RAM is capable of running X, but 16 megabytes or more of physical RAM is needed for comfortable use. A faster processor is nice to have as well, but having enough physical RAM is much more important. In addition, to achieve really slick video performance, we recommend getting an accelerated video card, like a VESA Local Bus (VLB) S3 chipset card. Performance ratings in excess of 300,000 xstones have been achieved with Linux and XFree86. Using adequate hardware, you'll find that running X and Linux is as fast, or faster, than running X on other UNIX workstations.
In Chapter  we discuss how to install and use X on your system.

1.4.4 Introduction to Networking.

Would you like to communicate with the world? Linux supports two primary UNIX networking protocols: TCP/IP and UUCP. TCP/IP (Transmission Control Protocol/Internet Protocol) is the networking paradigm which allows systems all over the world to communicate on a single network, the Internet. With Linux, TCP/IP, and a connection to the Internet, you can communicate with users and machines via electronic mail, Usenet news, and FTP file transfer.
Most TCP/IP networks use Ethernet as the physical network transport. Linux supports many popular Ethernet cards and interfaces for personal computers, including pocket and PCMCIA Ethernet adaptors.
However, because not everyone has an Ethernet connection at home, Linux also supports SLIP (Serial Line Internet Protocol) and PPP (Point-to-Point Protocol), which provide Internet access via modem. Many businesses and universities provide SLIP and PPP servers. In fact, if your Linux system has an Ethernet connection to the Internet and a modem, your system can become a SLIP or PPP server for other hosts.
NFS (Network File System) lets your system seamlessly share file systems with other machines on the network. FTP (File Transfer Protocol) lets you transfer files with other machines. sendmail sends and receives electronic mail via the SMTP protocol; C-News and INN are NNTP based new systems; and telnet, rlogin, and rsh let you log in and execute commands on other machines on the network. finger lets you get information about other Internet users.
Linux also supports Microsoft Windows connectivity via Samba, and Macintosh connectivity with AppleTalk and LocalTalk. Support for Novell's IPX protocol is also included.
The full range of mail and news readers is available for Linux, including elm, pine, rn, nn, and tin. Whatever your preference, you can configure a Linux system to send and receive electronic mail and news from all over the world.
The system provides a standard UNIX socket programming interface. Virtually any program that uses TCP/IP can be ported to Linux. The Linux X server also supports TCP/IP, and applications running on other systems may use the display of your local system.
In Chapter , we discuss the installation of TCP/IP software, including SLIP and PPP.
UUCP (UNIX-to-UNIX Copy) is an older mechanism to transfer files, electronic mail, and electronic news between UNIX machines. Historically, UUCP machines are connected over telephone lines via modem, but UUCP is able to transfer data over a TCP/IP network as well. If you do not have access to a TCP/IP network or a SLIP or PPP server, you can configure your system to send and receive files and electronic mail using UUCP. See Chapter  for more information.

1.4.5 Telecommunications and BBS software.

If you have a modem, you'll be able to communicate with other machines via telecommunications packages available for Linux. Many people use telecommunications software to access bulletin board systems (BBS's) as well as commercial, online services like Prodigy, CompuServe, and America Online. People use modems to connect to UNIX systems at work or school. Modems can send and receive faxes.
A popular communications package for Linux is seyon, which provides a customizable, ergonomic interface undex X and has built-in support for the Kermit and ZModem file transfer protocols. Other telecommunications programs include C-Kermit, pcomm, and minicom. These are similar to communications programs found on other operating systems, and are quite easy to use.
If you do not have access to a SLIP or PPP server (see the previous section), you can use term to multiplex your serial line. The term program allows you to open more than one login session over a modem connection. It lets you redirect X client connections to your local X server via a serial line. Another software package, KA9Q, implements a similar, SLIP-like interface.
Operating a Bulletin Board System (BBS) is a favorite hobby and means of income for many people. Linux supports a wide range of BBS software, most of which is more powerful than that available for other operating systems. With a phone line, modem, and Linux, you can turn your system into a BBS and provide dial-in access for users worldwide. BBS software for Linux includes XBBS and UniBoard BBS packages.
Most BBS software locks the user into a menu based system where only certain functions and applications are available. An alternative to BBS access is full UNIX access, which lets users dial into your system and log in normally. This requires a fair amount of maintenance by the system administrator, but providing public UNIX access is not difficult. In addition to TCP/IP networking, you can make electronic mail and news access available on your system.
If you do not have access to a TCP/IP network or UUCP feed, Linux lets you communicate with BBS networks like FidoNet, which let you exchange electronic news and mail over a telephone line. You can find more information on telecommunications and BBS software under Linux in Chapter .

1.4.6 World Wide Web.

It is worth noting that Linux includes web server software as well as web browsers. The most common server is Apache. Thousands of Linux systems run Apache on the Internet today, including the Linux Resources site, www.linuxresources.com.
Linux distributions include different web browsers, and other browsers can be downloaded from the Internet. Available browsers include Lynx, Mosaic, Netscape, Arena, and Amaya.
Linux provides complete support for Java and CGI applets, and Perl is a standard tool in the Linux programming environment.

Interfacing and MS-DOS.

Various utilities exist to interface with MS-DOS. The most well-known application is the Linux MS-DOS Emulator, which lets you run MS-DOS applications directly from Linux. Although Linux and MS-DOS are completely different operating systems, the 80386 protected-mode environment allows MS-DOS applications to behave as if they were running in their native 8086 environment.
The MS-DOS emulator is still under development, but many popular applications run under it. Understandably, MS-DOS applications that use bizarre or esoteric features of the system may never be supported, because of the limitations inherent in any emulator. For example, you shouldn't expect to run programs that use 80386 protected-mode features, like Microsoft Windows (in 386 enhanced mode, that is).
Standard MS-DOS commands and utilities like PKZIP.EXE work under the emulators, as do 4DOS, a COMMAND.COM replacement, FoxPro 2.0, Harvard Graphics, MathCad, Stacker 3.1, Turbo Assembler, Turbo C/C++, Turbo Pascal, Microsoft Windows 3.0 (in real mode), and WordPerfect 5.1.
The MS-DOS Emulator is meant mostly as an ad-hoc solution for those who need MS-DOS for only a few applications and use Linux for everything else. It's not meant to be a complete implementation of MS-DOS. Of course, if the Emulator doesn't satisfy your needs, you can always run MS-DOS as well as Linux on the same system. Using the LILO boot loader, you can specify at boot time which operating system to start. Linux can also coexist with other operating systems, like OS/2.
Linux provides a seamless interface to transfer files between Linux and MS-DOS. You can mount a MS-DOS partition or floppy under Linux, and directly access MS-DOS files as you would any file.
Currently under development is WINE--a Microsoft Windows emulator for the X Window System under Linux. Once WINE is complete, users will be able to run MS-Windows applications directly from Linux. This is similar to the commercial WABI Windows emulator from Sun Microsystems, which is also available for Linux.
In Chapter , we talk about the MS-DOS tools available for Linux.

1.4.8 Other applications.

A host of miscellaneous programs and utilities exist for Linux, as one would expect of such a hodgepodge operating system. Linux's primary focus is UNIX personal computing, but this is not the only field where it excels. The selection of business and scientific software is expanding, and commercial software vendors have begun to contribute to the growing pool Linux applications.
Several relational databases are available for Linux, including Postgres, Ingres, and Mbase. These are full-featured, professional, client/server database applications, similar to those found on other UNIX platforms. Many commercial database systems are available as well.
Scientific computing applications include FELT (finite element analysis); gnuplot (data plotting and analysis); Octave (a symbolic mathematics package similar to MATLAB); xspread (a spreadsheet calculator); xfractint (an X-based port of the popular Fractint fractal generator); and xlispstat (statistics). Other applications include SPICE (circuit design and analysis) and Khoros (image and digital signal processing and visualization). Commercial packages like Maple and MathLab are available.
Many more applications have been ported to Linux. If you absolutely cannot find what you need, you can attempt to port the application from another platform to Linux yourself. Whatever your field, porting standard UNIX applications to Linux is straightforward. Linux's complete UNIX programming environment is sufficient to serve as the base for any scientific application.
Linux also has its share of games. These include classic text based dungeon games like Nethack and Moria; MUDs (multi-user dungeons, which allow many users to interact in a text-based adventure) like DikuMUD and TinyMUD; and a slew of X games like xtetris, netrek, and xboard, the X11 version of gnuchess. The popular shoot-em-up, arcade-style game, Doom, has also been ported to Linux.
For audiophiles, Linux supports various sound cards and related software, like CDplayer, which makes a CD-ROM drive into an audio CD player, MIDI sequencers and editors, which let you compose music for playback through a synthesizer or other MIDI controlled instrument, and sound editors for digitized sounds.
Can't find the application you're looking for? The Linux Software Map, described in Appendix A, lists software packages which have been written or ported to Linux. Another way to find Linux applications is to look at the INDEX files found on Linux FTP sites, if you have Internet access.
Most freely-distributable, UNIX based software will compile on Linux with little difficulty. If all else fails, you can write the application yourself. If you're looking for a commercial application, there may be a free ``clone'' available. Or, you can encourage the software company to consider releasing a binary version for Linux. Several individuals have contacted software companies and asked them to port their applications to Linux, with various degrees of success.

1.5 Copyright issues.

Linux is covered by what is known as the GNU General Public License, or GPL. The GPL was developed for the GNU project by the Free Software Foundation and specifies several provisions for the distribution and modification of free software. Free, in this sense, refers to distribution, not cost. The GPL has always been subject to misinterpretation. We hope that this summary will help you understand the extent and goals of the GPL and its effect on Linux. A complete copy of the GPL is printed in Appendix C.
Originally, Linus Torvalds released Linux under a license more restrictive than the GPL, which allowed the software to be freely distributed and modified, but prevented any money from changing hands for its distribution and use. On the other hand, the GPL allows people to sell and profit from free software, but does not allow them to restrict another's right to distribute the software in any way.
First, it should be explained that free software that is covered by the GPL is not in the public domain. Public domain software by definition is not copyrighted and is literally owned by the public. Software covered by the GPL, on the other hand, is copyrighted by the author. The software is protected by standard international copyright laws, and the author is legally defined. The GPL provides for software which may be freely distributed but is not in the public domain.
GPL-licensed software is also not shareware. Generally, shareware is owned and copyrighted by an author who requires users to send in money for its use. Software covered by the GPL may be distributed and used free of charge.
The GPL also lets people take, modify, and distribute their own versions of the software. However, any derived works of GPL software must also be covered by the GPL. In other words, a company may not take Linux, modify it, and sell it under a restrictive license. If the software is derived from Linux, that software must be covered under the GPL also.
The GPL allows free software to be distributed and used free of charge. It also lets a person or organization distribute GPL software for a fee, and even make a profit from its sale and distribution. However, a distributor of GPL software cannot take those rights away from a purchaser. If you purchase GPL software from a third-party source, you may distribute the software for free, and sell it yourself as well.
This may sound like a contradiction. Why sell software when the GPL allows you to get it for free? Let's say that a company decided to bundle a large amount of free software on a CD-ROM and distribute it. That company would need to charge for the overhead of producing and distributing the CD-ROM, and may even decide to profit from the sales of the software. This is allowed by the GPL.
Organizations that sell free software must follow certain restrictions set forth in the GPL. They cannot restrict the rights of users who purchase the software. If you buy a CD-ROM that contains GPL software, you can copy and distribute the CD-ROM free of charge, or resell it yourself. Distributors must make obvious to users that the software is covered by the GPL. Distributors must also provide, free of charge, the complete source code to the software distributed. This permits anyone who purchases GPL software to make modifications to that software.
Allowing a company to distribute and sell free software is a good thing. Not everyone has access to the Internet and the ability to download software for free. Many organizations sell Linux on diskette, tape, or CD-ROM via mail order, and profit from the sales. Linux developers may never see any of this profit; that is the understanding reached between the developer and the distributor when software is licensed by the GPL. In other words, Linus Torvalds knew that companies may wish to sell Linux, and that he might not see a penny of the profits.
In the free software world, the important issue is not money. The goal of free software is always to develop and distribute fantastic software and allow anyone to obtain and use it. In the next section, we'll discuss how this applies to the development of Linux.

1.6 The design and philosophy of Linux.

New users often have a few misconceptions and false expectations about Linux. It is important to understand the philosophy and design of Linux in order to use it effectively. We'll start by describing how Linux is not designed.
In commercial UNIX development houses, the entire system is developed under a rigorous quality assurance policy that utilizes source and revision control systems, documentation, and procedures to report and resolve bugs. Developers may not add features or change key sections of code on a whim. They must validate the change as a response to a bug report and subsequently ``check in'' all changes to the source control system, so that the changes may be reversed if necessary. Each developer is assigned one or more parts of the system code, and only that developer can alter those sections of the code while it is ``checked out'' (that is, while the code is under his or her control).
Organizationally, a quality assurance department runs rigorous tests on each new version of the operating system and reports any bugs. The developers fix these bugs as reported. A complex system of statistical analysis is used to ensure that a certain percentage of bugs are fixed before the next release, and that the operating system as a whole passes certain release criteria.
The software company, quite reasonably, must have quantitative proof that the next revision of the operating system is ready to be shipped; hence, the gathering and analysis of statistics about the performance of the operating system. It is a big job to develop a commercial UNIX system, often large enough to employ hundreds, if not thousands, of programmers, testers, documenters, and administrative personnel. Of course, no two commercial UNIX vendors are alike, but that is the general picture.
The Linux model of software development discards the entire concept of organized development, source code control systems, structured bug reporting, and statistical quality control. Linux is, and likely always will be, a hacker's operating system. (By hacker, I mean a feverishly dedicated programmer who enjoys exploiting computers and does interesting things with them. This is the original definition of the term, in contrast to the connotation of hacker as a computer wrongdoer, or outlaw.)
There is no single organization responsible for developing Linux. Anyone with enough know-how has the opportunity to help develop and debug the kernel, port new software, write documentation, and help new users. For the most part, the Linux community communicates via mailing lists and Usenet newsgroups. Several conventions have sprung up around the development effort. Anyone who wishes to have their code included in the ``official'' kernel, mails it to Linus Torvalds. He will test and include the code in the kernel as long as it doesn't break things or go against the overall design of the system.
The system itself is designed using an open-ended, feature-minded approach. The number of new features and critical changes to the system has recently diminished, and the general rule is that a new version of the kernel will be released every few weeks. Of course, this is a rough figure. New release criteria include the number of bugs to be fixed, feedback from users testing pre-release versions of the code, and the amount of sleep Linus Torvalds has had this week.
Suffice it to say that not every bug is fixed, nor is every problem ironed out between releases. As long as the revision appears to be free of critical or recurring bugs, it is considered to be stable, and the new version is released. The thrust behind Linux development is not to release perfect, bug-free code: it is to develop a free UNIX implementation. Linux is for the developers, more than anyone else.
Anyone who has a new feature or software application generally makes it available in an alpha version--that is, a test version, for those brave users who want to hash out problems in the initial code. Because the Linux community is largely based on the Internet, alpha software is usually uploaded to one or more Linux FTP sites (see Appendix B), and a message is posted to one of the Linux Usenet newsgroups about how to obtain and test the code. Users who download and test alpha software can then mail results, bug fixes, and questions to the author.
After the initial bugs have been fixed, the code enters a beta test stage, in which it is usually considered stable but not complete. It works, but not all of the features may be present. The software may also go directly to a final stage, in which the software is considered complete and usable.
Keep in mind that these are only conventions--not rules. Some developers may feel so confident of their software that they decide it isn't necessary to release alpha or test versions. It is always up to the developer to make these decisions.
You might be amazed at how such an unstructured system of volunteers who program and debug a complete UNIX system gets anything done at all. As it turns out, this is one of the most efficient and motivated development efforts ever employed. The entire Linux kernel is written from scratch, without code from proprietary sources. It takes a huge amount of work to port all the free software under the sun to Linux. Libraries are written and ported, file systems are developed, and hardware drivers are written for many popular devices--all due to the work of volunteers.
Linux software is generally released as a distribution, a set of prepackaged software which comprises an entire system. It would be difficult for most users to build a complete system from the ground up, starting with the kernel, adding utilities, and installing all of the necessary software by hand. Instead, many software distributions are available which include everything necessary to install and run a complete system. There is no single, standard distribution--there are many, and each has its own advantages and disadvantages. We describe installation of the various Linux distributions starting on page .

1.7 Differences between Linux and other operating systems.

 
It is important to understand the differences between Linux and other operating systems, like MS-DOS, OS/2, and the other implementations of UNIX for personal computers. First of all, Linux coexists happily with other operating systems on the same machine: you can run MS-DOS and OS/2 along with Linux on the same system without problems. There are even ways to interact between various operating systems, as we'll see.

Why use Linux?

Why use Linux, instead of a well known, well tested, and well documented commercial operating system? We could give you a thousand reasons. One of the most important, however, is that Linux is an excellent choice for personal UNIX computing. If you're a UNIX software developer, why use MS-DOS at home? Linux allows you to develop and test UNIX software on your PC, including database and X Window System applications. If you're a student, chances are that your university computing systems run UNIX. You can run your own UNIX system and tailor it to your needs. Installing and running Linux is also an excellent way to learn UNIX if you don't have access to other UNIX machines.
But let's not lose sight. Linux isn't only for personal UNIX users. It is robust and complete enough to handle large tasks, as well as distributed computing needs. Many businesses--especially small ones--have moved their systems to Linux in lieu of other UNIX based, workstation environments. Universities have found that Linux is perfect for teaching courses in operating systems design. Large, commercial software vendors have started to realize the opportunities which a free operating system can provide.

Linux vs. MS-DOS.

It's not uncommon to run both Linux and MS-DOS on the same system. Many Linux users rely on MS-DOS for applications like word processing. Linux provides its own analogs for these applications, but you might have a good reason to run MS-DOS as well as Linux. If your dissertation is written using WordPerfect for MS-DOS, you may not be able to convert it easily to TeX or some other format. Many commercial applications for MS-DOS aren't available for Linux yet, but there's no reason that you can't use both.
MS-DOS does not fully utilize the functionality of 80386 and 80486 processors. On the other hand, Linux runs completely in the processor's protected mode, and utilizes all of its features. You can directly access all of your available memory (and beyond, with virtual RAM). Linux provides a complete UNIX interface which is not available under MS-DOS. You can easily develop and port UNIX applications to Linux, but under MS-DOS you are limited to a subset of UNIX functionality.
Linux and MS-DOS are different entities. MS-DOS is inexpensive compared to other commercial operating systems and has a strong foothold in the personal computer world. No other operating system for the personal computer has reached the level of popularity of MS-DOS, because justifying spending $1,000 for other operating systems alone is unrealistic for many users. Linux, however, is free, and you may finally have the chance to decide for yourself.
You can judge Linux vs. MS-DOS based on your expectations and needs. Linux is not for everybody. If you always wanted to run a complete UNIX system at home, without the high cost of other UNIX implementations for personal computers, Linux may be what you're looking for.

Linux vs. The Other Guys.

A number of other advanced operating systems have become popular in the PC world. Specifically, IBM's OS/2 and Microsoft Windows have become popular for users upgrading from MS-DOS.
Both OS/2 and Windows NT are full featured multitasking operating systems, like Linux. OS/2, Windows NT, and Linux support roughly the same user interface, networking, and security features. However, the real difference between Linux and The Other Guys is the fact that Linux is a version of UNIX, and benefits from contributions of the UNIX community at large.
What makes UNIX so important? Not only is it the most popular operating system for multiuser machines, it is a foundation of the free software world. Much of the free software available on the Internet is written specifically for UNIX systems.
There are many implementations of UNIX from many vendors. No single organization is responsible for its distribution. There is a large push in the UNIX community for standardization in the form of open systems, but no single group controls this design. Any vendor (or, as it turns out, any hacker) may develop a standard implementation of UNIX.
OS/2 and Microsoft operating systems, on the other hand, are proprietary. The interface and design are controlled by a single corporation, which develops the operating system code. In one sense, this kind of organization is beneficial because it sets strict standards for programming and user interface design, unlike those found even in the open systems community.
Several organizations have attempted the difficult task of standardizing the UNIX programming interface. Linux, in particular, is mostly compliant with the POSIX.1 standard. As time goes by, it is expected that the Linux system will adhere to other standards, but standardization is not the primary goal of Linux development.

Linux vs. other implementations of UNIX.

Several other implementations of UNIX exist for 80386 or better personal computers. The 80386 architecture lends itself to UNIX, and vendors have taken advantage of this.
Oher implementations of UNIX for the personal computer are similar to Linux. Almost all commercial versions of UNIX support roughly the same software, programming environment, and networking features. However, there are differences between Linux and commercial versions of UNIX.
Linux supports a different range of hardware than commercial implementations. In general, Linux supports most well-known hardware devices, but support is still limited to hardware which the developers own. Commercial UNIX vendors tend to support more hardware at the outset, but the list of hardware devices which Linux supports is expanding continuously. We'll cover the hardware requirements for Linux in Section 1.8.
Many users report that Linux is at least as stable as commercial UNIX systems. Linux is still under development, but the two-pronged release philosophy has made stable versions available without impeding development.
The most important factor for many users is price. Linux software is free if you can download it from the Internet or another computer network. If you do not have Internet access, you can still purchase Linux inexpensively via mail order on diskette, tape, or CD-ROM.
Of course, you may copy Linux from a friend who already has the software, or share the purchase cost with someone else. If you plan to install Linux on a large number of machines, you need only purchase a single copy of the software--Linux is not distributed with a ``single machine'' license.
The value of commercial UNIX implementations should not be demeaned. In addition to the price of the software itself, one often pays for documentation, support, and quality assurance. These are very important factors for large institutions, but personal computer users may not require these benefits. In any case, many businesses and universities have found that running Linux in a lab of inexpensive personal computers is preferable to running a commercial version of UNIX in a lab of workstations. Linux can provide workstation functionality on a personal computer at a fraction of the cost.
Linux systems have travelled the high seas of the North Pacific, and manage telecommunications and data analysis for an oceanographic research vessel. Linux systems are used at research stations in Antarctica. Several hospitals maintain patient records on Linux systems.
Other free or inexpensive implementations of UNIX are available for the 80386 and 80486. One of the best known is 386BSD, an implementation of BSD UNIX for the 80386. The 386BSD package is comparable to Linux in many ways, but which one is better depends on your needs and expectations. The only strong distinction we can make is that Linux is developed openly, and any volunteer can aid in the development process, while 386BSD is developed by a closed team of programmers. Because of this, serious philosophical and design differences exist between the two projects. The goal of Linux is to develop a complete UNIX system from scratch (and have a lot of fun in the process), and the goal of 386BSD is in part to modify the existing BSD code for use on the 80386.
NetBSD is another port of the BSD NET/2 distribution to several machines, including the 80386. NetBSD has a slightly more open development structure, and is comparable to 386BSD in many respects.
Another project of note is HURD, an effort by the Free Software Foundation to develop and distribute a free version of UNIX for many platforms. Contact the Free Software Foundation (the address is given in Appendix C) for more information about this project. At the time of this writing, HURD is still under development.
Other inexpensive versions of UNIX exist as well, like Minix, an academic but useful UNIX clone upon which early development of Linux was based. Some of these implementations are mostly of academic interest, while others are full fledged systems.

1.8 Hardware requirements.

You must be convinced by now of how wonderful Linux is, and of all the great things it can do for you. However, before you rush out and install Linux, you need to be aware of its hardware requirements and limitations.
Keep in mind that Linux is developed by users. This means, for the most part, that the hardware supported by Linux is that which the users and developers have access to. As it turns out, most popular hardware and peripherals for personal computers are supported. Linux supports more hardware than some commercial implementations of UNIX. However, some obscure devices aren't supported yet.
Another drawback of hardware support under Linux is that many companies keep their hardware interfaces proprietary. Volunteer Linux developers can't write drivers for the devices because the manufacturer does not make the technical specifications public. Even if Linux developers could develop drivers for proprietary devices, they would be owned by the company which owns the device interface, which violates the GPL. Manufacturers that maintain proprietary interfaces write their own drivers for operating systems like MS-DOS and Microsoft Windows. Users and third-party developers never need to know the details of the interface.
In some cases, Linux programmers have attempted to write hackish device drivers based on assumptions about the interface. In other cases, developers work with the manufacturer and try to obtain information about the device interface, with varying degrees of success.
In the following sections, we attempt to summarize the hardware requirements for Linux. The Linux Hardware HOWTO (see Section 1.9) contains a more complete listing of hardware supported by Linux.
Disclaimer: Much hardware support for Linux is in the development stage. Some distributions may or may not support experimental features. This section lists hardware which has been supported for some time and is known to be stable. When in doubt, consult the documentation of your Linux distribution. See Section 2.2 for more information about Linux distributions.
Linux is available for many platforms in addition to Intel 80x86 systems. These include Macintosh, Amiga, Sun SparcStation, and Digital Equipment Corporation Alpha based systems. In this book, however, we focus on garden-variety Intel 80386, 80486, and Pentium processors, and clones by manufacturers like AMD, Cyrix, and IBM.

Motherboard and CPU requirements.

Linux currently supports systems with the Intel 80386, 80486, or Pentium CPU, including all variations like the 80386SX, 80486SX, 80486DX, and 80486DX2. Non-Intel clones work with Linux as well. Linux has also been ported to the DEC Alpha and the Apple PowerMac.
If you have an 80386 or 80486SX, you may also wish to use a math coprocessor, although one isn't required. The Linux kernel can perform FPU emulation if the machine doesn't have a coprocessor. All standard FPU couplings are supported, including IIT, Cyrix FasMath, and Intel.
Most common PC motherboards are based on the PCI bus but also offer ISA slots. This configuration is supported by Linux, as are EISA and VESA-bus systems. IBM's MicroChannel (MCA) bus, found on most IBM PS/2 systems, is significantly different, and support has been recently added.

Memory requirements.

Linux requires very little memory, compared to other advanced operating systems. You should have 4 megabytes of RAM at the very least, and 16 megabytes is strongly recommended. The more memory you have, the faster the system will run. Some distributions require more RAM for installation.
Linux supports the full 32-bit address range of the processor. In other words, it uses all of your RAM automatically.
Linux will run with only 4 megabytes of RAM, including bells and whistles like the X Window System and emacs. However, having more memory is almost as important as having a faster processor. For general use, 16 megabytes is enough, and 32 megabytes, or more, may be needed for systems with a heavy user load.
Most Linux users allocate a portion of their hard drive as swap space, which is used as virtual RAM. Even if your machine has more than 16 megabytes of physical RAM, you may wish to use swap space. It is no replacement for physical RAM, but it can let your system run larger applications by swapping inactive portions of code to disk. The amount of swap space that you should allocate depends on several factors; we'll come back to this question in Chapter 2.

Hard drive controller requirements.

It is possible to run Linux from a floppy diskette, or, for some distributions, a live file system on CD-ROM, but for good performance you need hard disk space. Linux can co-exist with other operating systems--it only needs one or more disk partitions.
Linux supports all IDE and EIDE controllers as well as older MFM and RLL controllers. Most, but not all, ESDI controllers are supported. The general rule for non-SCSI hard drive and floppy controllers is that if you can access the drive from MS-DOS or another operating system, you should be able to access it from Linux.
Linux also supports a number of popular SCSI drive controllers. This includes most Adaptec and Buslogic cards as well as cards based on the NCR chip sets.

Hard drive space requirements.

Of course, to install Linux, you need to have some amount of free space on your hard drive. Linux will support more than one hard drive on the same machine; you can allocate space for Linux across multiple drives if necessary.
How much hard drive space depends on your needs and the software you're installing. Linux is relatively small, as UNIX implementations go. You could run a system in 20 megabytes of disk space. However, for expansion and larger packages like X, you need more space. If you plan to let more than one person use the machine, you need to allocate storage for their files. Realistic space requirements range from 200 megabytes to one gigabyte or more.
Also, you will likely want to allocate disk space as virtual RAM. We will discuss installing and using swap space in Chapter 2.
Each Linux distribution comes with literature to help you gauge the precise amount of storage required for your software configuration. Look at the information which comes with your distribution or the appropriate installation section in Chapter 2.

Monitor and video adaptor requirements.

Linux supports standard Hercules, CGA, EGA, VGA, IBM monochrome, Super VGA, and many accelerated video cards, and monitors for the default, text-based interface. In general, if the video card and monitor work under an operating system like MS-DOS, the combination should work fine under Linux. However, original IBM CGA cards suffer from ``snow'' under Linux, which is not pleasant to view.
Graphical environments like X have video hardware requirements of their own. Rather than list them here, we relegate that discussion to Section 5.1. Popular video cards are supported and new card support is added regularly.

Miscellaneous hardware.

You may also have devices like a CD-ROM drive, mouse, or sound card, and may be interested in whether or not this hardware is supported by Linux.

Mice and other pointing devices.

Typically, a mouse is used only in graphical environments like X. However, several Linux applications that are not associated with a graphical environment also use mice.
Linux supports standard serial mice like Logitech, MM series, Mouseman, Microsoft (2-button), and Mouse Systems (3-button). Linux also supports Microsoft, Logitech, and ATIXL bus mice, and the PS/2 mouse interface.
Pointing devices that emulate mice, like trackballs and touchpads, should work also.

CD-ROM drives.

Many common CD-ROM drives attach to standard IDE controllers. Another common interface for CD-ROM is SCSI. SCSI support includes multiple logical units per device so you can use CD-ROM ``jukeboxes.'' Additionally, a few proprietary interfaces, like the NEC CDR-74, Sony CDU-541 and CDU-31a, Texel DM-3024, and Mitsumi are supported.
Linux supports the standard ISO 9660 file system for CD-ROMs, and the High Sierra file system extensions.

Tape drives.

Any SCSI tape drive, including quarter inch, DAT, and 8MM are supported, if the SCSI controller is supported. Devices that connect to the floppy controller like floppy tape drives are supported as well, as are some other interfaces, like QIC-02.

Printers.

Linux supports the complete range of parallel printers. If MS-DOS or some other operating system can access your printer from the parallel port, Linux should be able to access it, too. Linux printer software includes the UNIX standard lp and lpr software. This software allows you to print remotely via a network, if you have one. Linux also includes software that allows most printers to handle PostScript files.

Modems.

As with printer support, Linux supports the full range of serial modems, both internal and external. A great deal of telecommunications software is available for Linux, including Kermit, pcomm, minicom, and seyon. If your modem is accessible from another operating system on the same machine, you should be able to access it from Linux with no difficulty.

Ethernet cards.

Many popular Ethernet cards and LAN adaptors are supported by Linux. Linux also supports some FDDI, frame relay, and token ring cards, and all Arcnet cards. A list of supported network cards is included in the kernel source of your distribution.

1.9 Sources of Linux information.

Many other sources of information about Linux are available. In particular, a number of books about UNIX in general will be of use, especially for readers unfamiliar with UNIX. We suggest that you peruse one of these books before attempting to brave the jungles of Linux.
Information is also available online in electronic form. You must have access to an online network like the Internet, Usenet, or Fidonet to access the information. A good place to start is www.linuxresources.com (see Appendix A). If you do not, you might be able to find someone who is kind enough to give you hard copies of the documents.

1.9.1 Online documents.

Many Linux documents are available via anonymous FTP from Internet archive sites around the world and networks like Fidonet and CompuServe. Linux CD-ROM distributions also contain the documents mentioned here. If you are can send mail to Internet sites, you may be able to retrieve these files using one of the FTP e-mail servers that mail you the documents or files from the FTP sites. See Appendix B for more information on using FTP e-mail servers.
A list of well-known Linux archive sites is given in Appendix B. To reduce network traffic, you should use a FTP site that is geographically close to you.
Appendix A contains a partial list of the Linux documents available via anonymous FTP. The filenames vary depending on the site. Most sites keep Linux-related documents in the docs subdirectory of their Linux archive. For example, the FTP site sunsite.unc.edu, keeps Linux files in /pub/Linux, with Linux-related documentation in /pub/Linux/docs.
Examples of available online documents are Linux Frequently Asked Questions with Answers, a collection of frequently asked questions about Linux; Linux HOWTO documents, which describe specific aspects of the system, like the Installation HOWTO, Printing HOWTO, and Ethernet HOWTO; and the Linux META-FAQ, which is a list of information sources on the Internet.
Many of these documents are also regularly posted to one or more Linux-related Usenet newsgroups; see Section 1.9.4 below.

1.9.2 Linux on the World Wide Web.

  The Linux Documentation Project Home Page is on the World Wide Web at http://sunsite.unc.edu/LDP This web page lists many HOWTOs and other documents in HTML format, as well as pointers to other sites of interest to Linux users, like ssc.com, home of the Linux Journal, a monthly magazine. You can find their home page at http://www.ssc.com/.

1.9.3 Books and other published works.

The books of the Linux Documentation Project are the result of an effort carried out over the Internet to write and distribute a bona fide set of manuals for Linux, analogs of the documentation which comes with commercial UNIX versions and covers installation, operation, programming, networking, and kernel development.
Linux Documentation Project manuals are available via anonymous FTP and by mail order. Appendix A lists the manuals available and describes how to obtain them.
Many large publishers, including MIS:Press, Digital Press, O'Reilly & Associates, and SAMS have jumped onto the Linux bandwagon. Check with computer bookstores or SSC's web page at http://www.ssc.com/, or the book reviews in Linux Journal, sometimes made available on their site, http://www.linuxjournal.com
A large number of books about UNIX in general are applicable to Linux. In its use and programming interface, Linux does not differ greatly from other implementations of UNIX. Almost everything you would like to know about using and programming Linux can be found in general UNIX texts. In fact, this book is meant to supplement the library of UNIX books currently available. Here, we present the most important Linux-specific details and hope that you will look to other sources for in-depth information.
Armed with good books about UNIX as well as this book, you should be able to tackle just about anything. Appendix A lists several UNIX books which are recommended highly for UNIX newcomers and wizards.
The Linux Journal magazine is distributed worldwide, and is an excellent way to keep in touch with the goings-on of the Linux community, especially if you do not have access to Usenet news (see below). See Appendix A for information on subscribing to the Linux Journal.

1.9.4 Usenet newsgroups.

 
Usenet is a worldwide electronic news and discussion forum with a diverse selection of newsgroups, which are discussion areas devoted to specific topics. Much discussion about Linux development occurrs over the Internet and Usenet. Not surprisingly, a number of Usenet newsgroups are dedicated to Linux.
The original Linux newsgroup, alt.os.linux, was created to move some of the discussion about Linux from comp.os.minix and various mailing lists. Soon, the traffic on alt.os.linux grew large enough that a newsgroup in the comp hierarchy was warranted. A vote was taken in February, 1992, and comp.os.linux was created.
comp.os.linux quickly became one of the most popular (and loudest) of the Usenet groups, more popular than any other group in the comp.os hierarchy. In December, 1992, a vote was taken to split the newsgroup to reduce traffic; only comp.os.linux.announce passed this vote. In July, 1993, the group was finally split into a new hierarchy. Almost 2,000 people voted in the comp.os.linux reorganization, making it one of the largest Usenet Calls For Votes ever.
If you do not have Usenet, there are mail-to-news gateways available for many (if not all) of the newsgroups below.



This list is by no means complete. New groups are created when a need for a subdivision of discussion is advisable, and there are linux groups in other hierarchies as well.

1.9.5 Internet mailing lists.

  If you have access to Internet electronic mail, you can participate in several mailing lists, even if you do not have Usenet access. If you are not directly on the Internet, you can join one of these mailing lists if you can exchange electronic mail with the Internet (for example, through UUCP, Fidonet, CompuServe, or other networks which exchange Internet mail).
For more information about the Linux mailing lists, send e-mail to

majordomo@vger.rutgers.edu

Include a line with the word help in the body of the message, and a message will be returned to you which describes how to subscribe and unsubscribe to various mailing lists. The word lists on a line by itself will retrieve the names of mailing lists which are accessible through the majordomo.vger.rutgers.edu server.
There are several special-purpose mailing lists for Linux as well. The best way to find out about these is to watch the Linux Usenet newsgroups for announcements, as well as to read the list of publicly-available mailing lists, which is posted to the Usenet news.answers group.

1.10 Getting Help with Linux.

You will undoubtedly need assistance during your adventures in the Linux world. Even UNIX wizards are occasionally stumped by some quirk or feature of Linux. It's important to know how, where, and when to find help.
The primary means of obtaining help is through Internet mailing lists and newsgroups as discussed in Section 1.9. If you don't have access to these sources, you may be able to find comparable Linux discussion forums on online services, like BBS's and CompuServe. Also available online are Linux Journal's Best of Technical Support columns, at http://www.linuxjournal.com/techsup.html.
Several businesses provide commercial support for Linux. These services allow you to pay a subscription fee that lets you call consultants for help with your Linux problems.
Keeping the following suggestions in mind will greatly improve your experience with Linux and guarantee more success in finding help.
Consult all available documentation...first! You should do this when you first encounter a problem. Various sources of information are listed in Section 1.9 and Appendix A. These documents are laboriously written for people who need help with the Linux system, like you. As mentioned above, books written for UNIX are applicable to Linux, and you should use them, too.
If you have access to Usenet news, or any of the Linux-related mailing lists, be sure to read the information there before posting. Often, solutions to common problems that are not easy to find in the documentation are well-covered in newsgroups and mailing lists. If you only post to these groups but don't read them, you are asking for trouble.
Learn to appreciate self-reliance. You asked for it by running Linux in the first place. Remember, Linux is all about hacking and fixing problems. It is not a commercial operating system, nor does it try to be one. Hacking won't kill you. In fact, it will be enlightening to investigate and solve problems yourself--you may even one day call yourself a Linux guru. Learn to appreciate the full value of hacking the system and fixing problems yourself. You shouldn't expect to run a complete, homebrew Linux system without some handiwork.
Remain calm. Nothing is earned by taking an axe--or worse, a powerful electromagnet--to your Linux box. A large punching bag or a long walk is a good way to relieve occasional stress attacks. As Linux matures and distributions become more reliable, we hope this problem will disappear. However, even commercial UNIX implementations can be tricky. When all else fails, sit back, take a few deep breaths, and return to the problem when you feel relaxed. Your mind and conscience will be clearer.
Refrain from posting spuriously. Many people make the mistake of posting or mailing messages pleading for help prematurely. When encountering a problem, do not rush immediately to the nearest terminal and post a message to one of the Linux Usenet groups. First try to resolve the problem yourself, and be absolutely certain what the problem is. Does your system not respond when switched on? Perhaps it is unplugged.
When you post for help, make it worthwhile. Remember that people who read your post are not necessarily there to help you. Therefore, it is important to remain as polite, terse, and informative as possible.
How does one accomplish this? First, you should include as much relevant information about your system and your problem as possible. Posting the simple request, ``I cannot seem to get e-mail to work'' will probably get you nowhere unless you include information about your system, what software you're using, what you have attempted to do so far, and what the results were. When you include technical information, it is also a good idea to include general information about the version of your software (the Linux kernel version, for example), as well as a brief summary of your hardware configuration. But don't overdo it--your monitor type and brand is probably irrelevant if you're trying to configure network software.

2 Obtaining and Installing Linux

chap-installObtaining and Installing Linux  
   
David Bandel rewrote and revised the first section on installing Linux. Parts of the work by the following authors of the different sections on Linux distributions were also added to this first section.
Boris Beletsky wrote the Debian section. Sean Dreilinger wrote the section on Slackware. Henry Pierce wrote the section on Red Hat Linux. Evan Leibovitch wrote the section on Caldera OpenLinux. Larry Ayers wrote the section on S.u.S.E. Linux.

2.1 Generic installation.

Generic-installation
  Unlike most other operating systems, Linux can be obtained free of charge. Due to the GNU General Public License under which Linux is distributed (see Appendix C), no one can sell you a license for the software. You can use Linux at no charge and are encouraged to make it available to others.
But that doesn't mean companies aren't entitled to reimbursement for copying costs plus a profit. They may also add software that is not free that runs on the system.
This gives you the freedom to choose. If purchasing a CD-ROM is not within your budget, you may simply borrow a friend's copy or download the source from the Internet. Whether purchased from a major Linux distributor or downloaded from their FTP site (see Appendix B), you get the same operating system and the software packages that they offer. In fact, you can get more free software from one of the FTP sites than the companies can distribute on CD, due to restrictions some authors place on the distribution of their software.

2.1.1 Major Linux distributions.

  An in-depth look at some of the Linux distributions begins on page . These distributions are: Debian, Red Hat, Caldera, Slackware, and S.u.S.E. Each section has more information on where to obtain that distribution. But remember, Linux is the kernel. The software is part of the distribution, not Linux. Most of the software is freely available and can be ported between various UNIX platforms. After taking into account what the kernel itself will support, the biggest difference comes in what the libraries (software called from within the applications) support.
Each distribution has its own installation and maintenance utilities that ease installation and system administration. Each is apparently aimed at a different audience. Any distribution will get you started and keep you running. So I recommend that you read about each distribution and talk to any knowledgeable friends. Most large cities have a Linux User Group, most with experienced users, who argue at length over which distribution is the best, and why. I suggest that you listen to some of their arguments and then decide. You can also join mailing lists (I recommend joining only one at a time) and reading user posts and answers from the list gurus. As different as each distribution is, so too are the mailing lists that provide assistance. Making the right choice for yourself is important, because changing distributions generally means reinstalling from scratch.

2.1.2 Common concerns.

This section makes the assumptions that the average newcomer to Linux:

• has a computer with MS-DOS and Windows or OS/2;
• has a basic understanding of MS-DOS but not UNIX;
• knows or can find out what kind of hardware the computer has installed;
• has a desire to "try out" Linux for whatever reason, though probably not switch to it exclusively (yet); and
• has neither a spare machine nor second disk drive available, but several hundred megabytes on an existing drive free for use.

These assumptions are not extreme, and may even be a bit conservative. Some say that if your VCR still blinks 12:00, Linux isn't for you, but then that would leave me out as well. My VCR still blinks 12:00.
Before we begin, we must know where we are going. While it is certainly possible to get from New York to California (eventually) by striking out in almost any random direction, most of us would opt to go in a more or less direct route. So it is with installing Linux.

2.1.3 Hardware.

This section explains all of the installation steps necessary short of the actual install. Each distribution handles this preparation slightly differently. While the installs look different, they accomplish the same things and have more in common than not. All require:

• planning;
• gathering system hardware information;
• backing up your old system (optional, but strongly recommended);
• preparing Linux partitions;
• deciding on a boot loader (for dual boot systems);
• booting a Linux kernel;
• installing the kernel;
• choosing and installing software packages;
• loading the software;
• making final configuration adjustments; and
• rebooting into a running system.

Now that I've sufficiently oversimplified the process, let's go down the list. Hang on, it's not that bad when you learn from others' mistakes.

2.1.4 Planning.

  I can't overemphasize this step. Any pilot will tell you that the landing is only as good as the approach. The same goes for Linux installation.
First, determine what kind of hardware you have. A checklist has been included to assist you. Be as precise as possible, but don't get carried away. For example, if you have an Ethernet card, you need to know what kind (e.g., SMC-Ultra, 3Com 3C509, etc.), base I/O (e.g., io=0x300), interrupt (IRQ 10), but not the hardware address (00 00 a6 27 bf 3c). Not all information will be needed for your hardware. If you have Windows 95 or Windows NT running, you can copy the values from the system hardware device information screen. Otherwise, consult the hardware manuals or the hardware company's Web site. Since it is important, we'll review this worksheet here.

2.1.5 System planning worksheet.

 
General

Processor:		 Type:		 386 486 Pentium PPro



Speed (optional):



Mfg:



Intel AMD Cyrix



Motherboard:    		 Make:         		 Chip Set:



Example: 		 Make:  unknown  		 Chip Set: triton II 



Mouse: 		 Mfg:         		 Type: bus  PS/2  serial port



If serial: 		  COM1 (ttyS0) 		  COM2 (ttyS1)



Hard disk drive(s): 		 Type: 		 IDE/MFM/RLL/ESDI    SCSI



Size (list each drive):



If SCSI Controller: 		 Make: 		 Model:



Example: 		 Make: BusLogic  		 Model: 948 



Boot: 		 Linux   		 DOS/Windows 		 OS/2 		 Other



Disk: 		 Partition:     		 Size:          		 Boot:



Disk: 		 Partition:     		 Size:          		 Boot:



Disk: 		 Partition:     		 Size:          		 Boot:



Disk: 		 Partition:     		 Size:          		 Boot:



CD-ROM:		 IDE/ATAPI  SCSI  Proprietary



Mfg: 		 Model:



(Proprietary only):



X-Windows:



Video Card: 		  Mfg: 		 Model:



RAM: 		 1Mb     2Mb     4Mb     8Mb     16Mb



Monitor: 		  Mfg: 		  Model: 		  Max scan rate:



Networking:



Modem: 		 Mfg:  		  Model:



Serial port:  		  COM1 		 COM2 		 COM3 		 COM4



		(ttyS0)  		  (ttyS1)   		  (ttyS2) 		   (ttyS3)



Computer hostname:  		 		 Example: rainier

The following answers are only needed if using network interface card (NIC): (do not configure networking if you do not have a NIC installed)

NIC Type: 		 ethernet  		 token ring  		 FDDI 		  other



NIC Mfg: 		  Model:



Network domain name: 		 		  (Example: mountains.net)



IP Address: 		 		 (Ex: 192.168.1.2)



Network address: 		 		 (Ex: 192.168.1.0)



Netmask: 		 		 (Ex: 255.255.255.0)



Broadcast address:		 		 (Ex: 192.168.1.255)



Gateway(s): 		 		 (Ex: none or 192.168.1.1)



DNS(s): 		 		  (Ex: 192.168.1.2)

  Some of the General Section is there for future reference. Specifically, we don't need to know right now our CPU processor type. We can also do without ever knowing what chip set we have on the motherboard. But if the information is available, it is good to have.

2.1.6 Mice.

  Other information, beginning with the mouse, we do need, if we expect to use the mouse. We need to know the mouse manufacturer, because different brands implement internal signal functions differently. Here, attention to detail is everything. If you have a mouse with a Microsoft brand on it, it may have a serial or PS/2 interface. Looking at the connector for the computer won't help, either. A number of computers come with mice that look like serial mice and have a serial-type connector, but are connected to the motherboard internally as a PS/2 mouse.
Read the print on the bottom of the mouse carefully before deciding. Also, if you have a mouse with three buttons, but it has a switch on the bottom which you can change between, say, Microsoft and PC systems, choose PC system. The Microsoft setting doesn't implement the middle button, which is useful in UNIX. For manufacturer, choose the switch setting, since that is the signaling protocol used. No drivers exist for a "Cutie" mouse, but do exist for the switch settings of Microsoft and Mouse System found on the bottom of the mouse.
While not specifically asked for, the only additional information you may want to add is the device through which the system accesses the mouse. Linux must know how the device is referred to. If you have a PS/2 mouse, you will normally use either /dev/psaux, the auxiliary port for a PS/2 pointing device, or /dev/psmouse, a synonym sometimes available for use. Bus mice are accessed through a file specifically created for that proprietary mouse, like /dev/atibm ATI bus mice, /dev/logibm for Logitech bus mice, /dev/inportbm for InPort bus mice, or their respective synonyms of atimouse, logimouse, and so on. For serial mice, if you know the MS-DOS COM: port, substitute /dev/ttyS0 for COM1: and /dev/ttyS1 for COM2:. I'll refrain from explaining the origins of the tty name of ttyS0 since that will take up several paragraphs and is already explained in many UNIX references

Considering Hard drives and CD-ROMs.

Before you begin installation, you need to take stock of how much hard disk real estate you will dedicate to Linux versus how much you have. Deciding during installation how you want to divide up your hard disk is only asking for problems and will probably end up in lost time, lost data, and a reinstall.
Your hard drive will be one of several kinds. For our purposes, IDE, MFM, RLL, and ESDI are equivalent, and I will use the term IDE. This also encompasses EIDE, the most common interface currently in home computer systems on the market and favored for its low price.
If the hard drive has a SCSI interface, this will be evident during booting. You will need to know the make and model of the SCSI controller. The most common are the Adaptec and BusLogic controllers, but by no means are they the only ones. These also have specific models, like the AHA-1572 or BTC-958. This information is often displayed during system initialization.
To allocate space, we need to assess the hard drive size. Under OS/2, you can use the entire hard disk for OS/2, then install Microsoft Windows on the partition with OS/2 and run Microsoft Windows under OS/2. If you have MS-DOS and Microsoft Windows, or OS/2 on your computer, Linux should have its own partition. It can be loaded on a MS-DOS partition with UMSDOS, which is not covered here. While Linux has DOS emulators and can read and even run some DOS programs, DOS cannot usually ``see'' what is on a Linux partition.
If you have and want to keep MS-DOS (assumed), you must determine how much space to reserve for it. Subtract this number from the hard disk total and that is what you have to work with. For now, annotate the total size of the drive(s) you have and the second number with how much to dedicate to Linux.
For your CD-ROM drive(s) you need similar information. A CD-ROM drive is either IDE/ATAPI, the most common in home systems marketed today; SCSI; or an older, proprietary drive, like those connected to sound cards. If you have an IDE or SCSI drive, so much the better. If you have a proprietary drive, you must know the make and model because Linux identifies proprietary CD-ROM drives by manufacturer and the specific drive.

2.1.8 Disk drives under Linux.

For newcomers to Linux who are familiar only with MS-DOS, and for those who come from other UNIX platforms, devices under Linux have have peculiar references. These references are used almost from the start, and some understanding of them is necessary.
Under Linux, as under any UNIX, devices are special files. Hard drives are treated as files and are referred to by name, as are modems, your monitor screen, and other hardware devices. UNIX treats them as files to be read from and written to. Since Linux sees them as files, they will all be located in a directory dedicated to devices. After installation you will be able to see them under the directory /dev, for devices.
Even though these devices are seen by Linux as files, they are special. They come in two ``flavors,'' block and character, which refer to the way the device communicates, in blocks of data or individual characters. They are created automatically during installation.
These naming conventions are discussed on page .

2.1.9 Installing The X Window System

  I've included on your worksheet information regarding your video card and monitor. While this is not absolutely necessary, most of those who come from the Microsoft Windows or OS/2 world want to install and configure a graphical user interface (GUI) for use. Some distributions walk you through this set up, others point you to post-installation programs. This information will be important then.
You must know the manufacturer and specific model of your video card. Some cards can be probed for RAM or chip sets, others can't. In either case, knowing how much RAM is on the card and the chips used, like the S3 or S3-Virge, is important. This information saves much time and grief. The most difficult and frustrating part of any Linux installation and setup is the X Window System.
The data for your monitor is often more difficult to obtain. If you have one of the more obscure brands of monitors, you may need to supply vertical and horizontal scan rates yourself.
If in doubt, always err on the conservative side. Overdriving your system can result in damage to the monitor or video card.
We already have most of the information that we need for the mouse, the only other subsystem the X server needs. The information that Linux needs to know about your mouse is described on page .

2.1.10 Networking hardware.

This section is not as significant yet as the worksheet suggests. Networking is explained in detail in Chapter . But if you have a network interface card (NIC), be it Ethernet, token ring, or other system, you must read up on the card before proceeding. This information is needed during installation to use the NIC.
During initial Linux installation, if you do not have a NIC, you can skip over some of the networking part. However, all computers must have a name under Linux. The example on the worksheet assumes you've chosen a theme, like mountains, and will name your computers after names of mountains, but any scheme your concoct is fine.
If you have a modem, you must know where it is connected. This is a serial port, /dev/ttyS0 through /dev/ttyS3, corresponding to MS-DOS COM: ports 1-4. ISDN is treated similarly, but is normally set up post-installation with special, multiple device designations.
That finishes our worksheet and about half of the planning that we need to do. One notation that is not on our worksheet is the amount of random access memory (RAM) the system has. Linux runs happily on a system with less than 4 MB of RAM, but this has a significant impact on installation and subsequent system usage. If you have 4 MB of RAM or less, then you must follow special procedures for low memory machines where applicable. With the current low price of RAM and with few machines being sold today with less than 16MB of RAM, this generally isn't an issue. If it is, be sure to check your distribution for special instructions.

2.1.11 Planning, Part 2.

Portions of the following section, particularly the disk partitioning strategies, are highly contentious among seasoned installers, but I will give you my thoughts on it. You are welcome to deviate as you see fit. Most differences in opinion, though, come from a difference in the ultimate purpose for the system; i.e., as a workstation, Web server, News server, or other function.

2.1.12 Partitioning strategies.

Few seasoned Linux users will tell you to make one Linux native partition and one swap partition and start installation. There are several reasons for this, and I subscribe to most of them, so I have several Linux native partitions. But to me, the most compelling reason of all, is that one day you will want to upgrade, and that will require reformatting the file system(s). In fact, the Slackware distribution has, at last look, no means to even attempt an ``in-place'' upgrade, or any indication that it will in the future. Upgrading from kernel 0.99 to 1.2.13 required me to reformat, as did the upgrade from 1.2.13 to 2.0.0, and I suspect that another will be required for 2.2.0 (or whatever the next stable kernel is). What I don't ever want to do is lose the files that I have accumulated in my home directory. Yes, I have a backup. But keeping my /home directory intact is easier, especially since I moved all my special files to a subdirectory there.
Another reason is that any bootable partition must be within the first 1024 cylinders of the hard drive. When any PC boots up, a sequence of events occurs which ends in the loading of the operating system. Due to limitations in the BIOS (Basic Input/Output System), until the operating system is loaded, only the first 1024 cylinders of the first or second hard disk can be accessed.
To get a feel for exactly what we're talking about, I'm going to describe a standard Linux file system and how Linux handles partitions.
Under MS-DOS, every partition is a different drive, and little distinction is made between whether that is a physical drive or a logical drive (partition). Under Linux, physical and logical drives are much less rigidly designated.
During installation, you must choose a partition as your root partition. The root partition is designated as ``/''. When we refer to ``/dev'', this is really two directories, ``/'' and ``dev''. Your Linux kernel will be located on the root partition, but can be in a subdirectory as long as that subdirectory resides on the root partition. For example, some distributions use /boot to hold the kernel, system map, and boot up files.
The following structure (as a minimum) will be set up on your root partition during install:

You may see others like /boot, /mnt, /cdrom, /floppy, /opt, and so on, but the above are essential.
What about other partitions? Linux can use a directory name (say /usr) as a mount point. That is, the other partition on the disk (or on another disk) is mounted under it (in this case /usr).
If you unmount the other partition and look in the subdirectory Linux uses as a mount point, you will (or should) see nothing--no files or directories. When the other partition is mounted, you will see files and directories which are on that partition under the mount point. So if you have two drives, one with 120 MB and another with 840 MB, you can make one partition on the 120MB drive (let's say it's the root partition) and mount any partitions you have created on the 840MB drive (this could be one big partition, or several smaller partitions) under their respective mount points, one partition per mount point, creating, in effect, one, 960-MB file system.
The one restriction is that you cannot use certain directories on the root drive as mount points, because they contain files that are needed to either boot the system or mount other systems. Obviously if the command used to mount other partitions is located on another partition and you can't access that partition until you've mounted it, you'll be like the dog chasing its tail.
The directories you cannot use as mount points are: /bin, /dev, /etc, /lib, /lost+found, /proc, /root, and /sbin.
A detailed description of what files are contained in these standard system directories is given on page .
Let's look at a small example. You are an aspiring Internet Service Provider (ISP). You have four machines, and each has a 1-gigabyte drive. So, you decide to allocate space as follows:


You probably noticed that I arbitrarily assigned the root partition 120MB, and allocated the rest to whatever ( /usr, /home, /var/spool/mail and so forth). I also didn't allocate any space to a swap partition. So, let's look at what we will likely need, understanding that ``it depends,'' is key. I will discuss this from the perspective of a home situation with only a few users, lots of programs, and no other remarkable needs.
The best place to start is to tell you what my primary home computer looks like. I have two drives, /dev/hda (1.2 GB) and /dev/hdb (540 MB). df (disk free) displays


You can see that I have a half-used 150-MB root (/) partition, a nearly full /usr partition, a largely used /usr/X11R6 partition, and a large, but cramped, 500-MB /home partition. The remainder of the drive /dev/hdb is a swap partition.
At a realistic minimum, I would suggest reserving 80-100 MB for your root partition, about 10 MB per user on your /home partition, as much space as you can reserve for swap, within reason (see the next section), and the rest to /usr. I have a five-user system at home, but I personally have over 400 MB of the /home directory tied up, much of that in graphics--a photo album of family and friends. Your /usr partition should probably be at least 250 MB, but the minimum will depend on what you decide to install. As you can see, it can rapidly fill with over 800 MB of programs, libraries, and data. Also remember that partitions give you flexibility that you lose with one, giant partition.

2.1.13 The swap partition.

You must give thought to a swap partition. Unlike Microsoft Windows, Linux uses a dedicated swap partition for speed. Although it is possible to create a swap file, it is not recommended. Linux can use up to 128 MB of swap space. I recommend a practical minimum of 16 MB. The optimum is probably as much as you can spare between 32 and 64MB--the more, the better.
One last consideration before you decide to how best to carve up the disk. Remember that I said the BIOS cannot ``see'' past sector 1023 on the hard drive (about 512MB). So, the Linux kernel (a file probably called vmlinuz on your boot disk), or any OS kernel for that matter, must reside entirely on one of the first two disk drives ( /dev/hda or /dev/hdb) and within the first 1024 sectors, or the BIOS will be unable to load it. To insure that it can, plan to make your root partition (as well as any other boot partition) fall entirely within this limitation on either the first or second hard drive.

2.1.14 Repartitioning.

At the beginning of this chapter I said I'd make a few assumptions. One was that you would want to keep your comfortable MS-DOS and Microsoft Windows operating system around. And since the computer you bought only has MS-DOS on it, it doesn't make sense to have multiple partitions, so the one drive you have is probably entirely dedicated to MS-DOS.
One way or another, then, we will have two operating systems on this computer. If you currently have nothing on your disk (lucky you), that is great, but you're not quite ready to skip ahead. Linux is comfortable wherever you put it. Your BIOS may not be capable of booting it, but once running, it will not complain if it's relegated to the fourth partition of the fourth hard drive. But MS-DOS and Microsoft Windows aren't so forgiving. They want the first drive and the first partition and may refuse to boot from any other position. I have seen MS-DOS boot from the first partition on the second hard drive, but the first hard drive did not have any MS-DOS partitions, so MS-DOS didn't recognize the drive. The best strategy is often the path of least resistance. If at all possible, put MS-DOS on the first drive and the first partition.
A second consideration in a multiple OS situation is which operating system to load first. If you're tempted to partition the hard disk and install Linux first (reserving /dev/hda1 for MS-DOS, then installing MS-DOS second, don't. Windows 95 is the worst offender, but Microsoft products in general will delete any previous boot loader you had installed on the master boot record (what the BIOS uses to point to bootable kernels). In fact, you may even hear this referred to as the ``Microsoft virus''. This is not a virus in the true sense of the word, just arrogance on the part of Microsoft, that one would only want a Microsoft operating system to boot. Linux does not cause such problems, and in fact provides a way to choose the default boot image. It also allows you to intervene during the boot process to specify which operating system to boot. This is a standard part of Linux installation procedures.

2.1.15 Backing up your old system.

Before we actually get to work on the partition table, I will walk through procedures to protect the data that you have on the hard disk. These procedures assume that you have a DOS partition. Other operating systems may or may not have a way to accomplish the same thing.
The first thing that you should do is perform a complete backup. The tools that you will use work as they should. But these procedures are inherently dangerous. Any time you work with a hard disk partition table, you can easily lose all the data contained on the drive. Back up your hard disk before you proceed.
Once you have your disk backed up, create a boot floppy disk for the system. You can either use the MS-DOS command

which formats the floppy and puts the required system files on it, or, using a formatted disk, issue the command


Once you have created a boot floppy and tested it to insure that it works, copy the following files from your MS-DOS system to the boot floppy: FDISK.EXE, SCANDISK.EXE, and SYS.COM. Also copy the file RESTORRB.EXE from a Linux distribution CD or Linux FTP archive. (See Appendix B).
Run a defragmentation program on your DOS drive to defragment and group the files together at the front of the disk. If defragmeter encounters any errors, you need to run SCANDISK.EXE to fix the problems. Once you have defragmented the disk and ensured that the files are compressed toward the front of the drive (as indicated in the graphical portrayal of your disk), you're ready to run FIPS.EXE to shrink the MS-DOS partition.

2.1.16 FIPS.EXE

On your Linux distribution CD (or an Internet distribution site), you'll find a copy of FIPS.EXE, which can shrink the MS-DOS partition. Note that FIPS.EXE only works for MS-DOS partitions. If you have other partitions that you need to shrink, the program Partition Magic may help, but is not free. Copy FIPS.EXE to your boot floppy and reboot using this floppy. This accomplishes two things: it insures that the boot floppy works, and insures that you are booted into MS-DOS Real Mode and are not running Microsoft Windows.
At the A:> prompt, type FIPS (upper or lower case). You will be greeted and asked which drive you want to operate on (if you have more than one). Select the drive to shrink. Once you confirm your choice, let FIPS.EXE make a copy of your boot and root sectors to the floppy in case something untoward happens.
You will then be asked if all of the free space on your partition should be used to create a second partition. If you say, ``yes,'' you will not have any free space on the MS-DOS partition to save data to, so say, ``no.'' You will then be able to alter the amount of space allocated between the first and second partitions. Note that if you didn't properly defragment your drive, you won't have much to work with on the second partition. Also, if you use MS-DOS mirroring software, a file is created at the very end of the partition, and FIPS.EXE tells you that you have no space to create a second partition. Exit and correct the problem by deleting the MIRROR.FIL file, then restart FIPS.EXE.
You can edit and re-edit the table until you are satisfied. Once you are happy with the distribution of space between the partitions, confirm your changes and write out the table.
Once FIPS.EXE has finished, remove the boot floppy and reboot your computer. In this example, we'll destroy and recreate the second partition during installation to create at least two partitions for Linux: a swap partition and a Linux native partition. But you can create as many as you like.

2.1.17 Preparing to boot Linux.

In order to install Linux, we must begin by booting the Linux kernel. This is accomplished in exactly the same manner as if you wanted to reload MS-DOS: we need a boot disk. But most distributions come only with a CD-ROM, and even if we had a running Linux system, the command to create boot disks for Linux is different than for MS-DOS. If you bought a new computer with a bootable CD-ROM, some distributions allow you to boot in this manner. But we'll go through the process of creating a boot disk for the rest of us.

2.1.18 Creating a Linux boot disk under DOS.

   
Each distribution CD contains a MS-DOS program that allows you to write a raw disk image to a formatted floppy disk. You must have a high density floppy, and some distributions require this to be a 3.5-inch, 1.44 Mb floppy. Insert the floppy in the drive. On the CD (or on your disk drive if you downloaded it) find RAWRITE2.EXE (you may have the older RAWRITE.EXE).
Then cd to the directory that has the disk image(s) that you need to boot with. There may be only one, or many which are configured for different hardware. You will have to consult the distribution documentation. Running RAWRITE2.EXE with no arguments results in your having to answer two questions: the path name of the disk image file to write. and the destination disk drive, either A: or B:. To shortcut the prompts with RAWRITE.EXE or RAWRITE2.EXE, issue the arguments on the MS-DOS command line

Repeat this step for any additional disk images your system needs.
If you can check the floppy disks with SCANDISK.EXE and do a surface scan before writing the images to the floppy, you may save yourself some time later. Most initial install failures come from boot disks that are bad, and RAWRITE2.EXE doesn't verify the disks.
This is also true if you create boot disks under Linux. The badblocks(1) manual page describes how to check disks for errors.
Label the disks that you create for future use.

2.1.19 Creating a Linux boot disk under Linux.

If you have an operational Linux system; for example, if you upgrade and want to create the disk images with Linux, you change to the directory with the disk images and issue the command

Substitute the disk image name for diskimage and the correct floppy device (almost always /dev/fd0), and repeat for each disk that you need. The dd arguments are: if for input file; of for output file, and here we want to use the floppy device; bs for block size, in this case 512 bytes; conv=sync ensures that the output file is exactly the same size as the input file. The trailing ``sync'' insures that we flush the buffers to disk immediately.
An alternate method that works, though will often be shunned by ``real'' Linux administrators, is the cp (copy) command

Again, substitute the disk image file name for diskimage, the correct boot floppy device, and repeat the step for each disk that you need. You may receive a message asking if you want to replace the boot floppy device with diskimage. Obviously this won't happen, since the floppy diskette is not a true file but a device, but cp doesn't pay attention to that detail. Just say, ``yes,'' if you are asked.
With the Linux installation boot disks in hand, you're ready to install our system. Most distributions invoke fdisk, the Linux version, so you can create a native Linux partition and a swap partition. The install programs continue by creating the file system (the equivalent of formatting a MS-DOS disk) for both the Linux and swap partitions, and initialize the swap partition and mount the Linux partition.
One question that you will be asked is whether you want to check your hard disk for bad blocks. If you are using a SCSI drive, answer ``no.'' SCSI drives have built-in error checking and correcting. IDE and similar drives don't have this and need to map out bad blocks. If you have an older drive you want to do this. If you say, ``yes,'' the installation program will invoke the badblocks program to maps out all of the bad blocks it finds. This takes time. If in doubt, say, ``yes.''

2.1.20 Partitioning the hard disk: fdisk and cfdisk.

   Every operating system, be it MS-DOS and Microsoft Windows, or Linux, has its own version of fdisk. If you want to create a partition for use by MS-DOS, use the MS-DOS version, FDISK.EXE, to create the partition and write the table. If you are going to create a partition for Linux, you must create it with the Linux version, fdisk.
Under Linux, two disk partitioning programs are available: the original fdisk, and a friendlier cfisk. The difference between the two is that in fdisk you issue all commands via the keyboard with letters and numbers. With cfdisk, you use the arrow keys to highlight the options you want, and press Enter to execute the command. The only time you use anything but the arrow and Enter keys is when you specify a number for the size of the partition.
For starters, all Linux boot disks are created essentially equal. Reboot the computer with the boot floppy in the boot drive. You will be greeted with a screen with some instructions and a prompt

and a flashing cursor. If you use the Tab key, you should see a list of names. The names differ depending on the distribution, but look for one that says ``rescue'' or ``expert.'' The ``install'' label starts the installation program after loading the kernel, so if you want to let the installation program walk you through the partitioning and filesystem initialization process, you can use the ``install'' label; otherwise, choose a different label. You may also need to provide Linux some boot parameters. For our purposes, this should not be necessary, but you'll soon find out if this is the case.
Enter a label name and press /keyReturn. When the Linux kernel finishes the bootup process, you may be presented with any of a number of prompts, depending on the distribution. If you have a shell prompt, like the pound sign (#) or a dollar sign ($), you're where you need to be. If not, try presssing Alt-F2 or Alt-Shift-F2. You should be able to activate one of the system's virtual consoles.
Once you have a prompt (you should not need to log in), you will be working as ``root'' (more on this in Chapter 4). Enter the command

If an error is returned, try cfdisk. This is the disk partition utility. It defaults to /dev/hda, so if you need to work on the second hard drive, use the command


In fdisk, press m to see a menu. The commands you will use are: n to create a new partition; d to destroy a partition; t to change the partition type (83 is Linux Native, 82 is Linux Swap); p prints to the screen the partition information currently in memory (not what's on the disk); w writes the partition table to disk; and q quits.
Until you issue the w command, you are not committed and can make changes or quit without making any changes.
Pay attention to prefixes and suffixes of the partition size. With the partition size you need to specify ``+'' if the size will be other than the ending partition number, and a suffix of ``k'' or ``M'' (case does not matter) to specify KB or MB.
One final note on partitions: you can create up to four primary partitions. If you need more than four partitions, you will create three primary partitions and then extended partitions. The extended partition numbers begin with 5, so you may have /dev/hda1, /dev/hda2, /dev/hda3, /dev/hda5, and /dev/hda6 if you need five partitions.
As a final check before you write the partition table, ensure that your partitions do not overlap. As long as the start and end segments don't overlap with any other start and end segments, you can be sure the partition boundaries are okay. A beginning number may be listed as 1024 for partitions with numbers starting higher than that. For now, just consider that a reminder that the BIOS will not be able to read (or boot from) that partition.
 cfdisk does exactly the same thing as fdisk, but displays on the screen the state of the partition table in memory (but not on the disk) at all times. Use the Up and Down arrow keys to select a partition to work on, and the Right and Left arrow keys to select the action to be performed. Then press Enter to perform the action. You will have to input numbers for the size you want to make the partition, but all information is given on the screen, just follow the instructions. cfdisk defaults to /dev/hda, so you must to give it the argument /dev/hdb if you want to change the partition table on a second disk drive. Remember to write the table before you quit. This is the hardest part of cfdisk. It doesn't ask for confirmation before exiting. So select Write and press Enter before you select Quit and press Enter.

2.2 Linux distributions.

 
You are now faced with the task of deciding which particular distribution of Linux suits your needs. Not all distributions are alike. Many of them come with just about all of the software you'd need to run a complete system--and then some. Other Linux distributions are ``small'' distributions intended for users without copious amounts of disk space. Many distributions contain only the core Linux software, and you are expected to install larger software packages, such as the X Window System, yourself. (In Chapter  we'll show you how.)
The Linux Distribution HOWTO (see Appendix A) contains a list of Linux distributions available on the Internet as well as by mail order.
If you have access to USENET news, or another computer conferencing system, you might want to ask there for personal opinions from people who have installed Linux. Also, Linux Journal maintains a table of features comparing Linux Distributions and periodically publishes software reviews of distributions (check http://www.linuxjournal.com/selected.html for on-line versions of the table and articles). Even better, if you know someone who installed Linux, ask them for help and advice. There are many factors to consider when choosing a distribution; however, everyone's needs and opinions are different. In actuality, most of the popular Linux distributions contain roughly the same set of software, so the distribution you select is more or less arbitrary.
 

2.3 Debian GNU/Linux.

   
This section on Debian GNU/Linux was written by Boris Beletsky.

2.3.1 Debian GNU/Linux installation features.

2.3.2 Getting floppy images.

If you have fast, cheap Internet access, the best way to get Debian is via anonymous FTP (see Appendix B). The home ftp site of Debian is located at ftp.debian.org in the /pub/debian directory. The structure of Debian archive is described in the table on page .
  
Table 2.1: Debian GNU/Linux archive structure.

For a base installation of Debian you need about 12 megabytes of disk space and some floppies. First, you need boot and driver floppy images. Debian provides two sets of boot floppy images, for 1.2 and 1.44 Mb floppy disks, and one set of the base images which work with either type of floppy. Check what floppy drive your system boots from, and download the appropriate disk set.
  
Table 2.2: Debian GNU/Linux installation floppies.

Choose the appropriate floppy set for your hardware from the table on page  and write the images to floppy as described on page .

2.3.3 Downloading the packages.

To install and use Debian, you need more than the base system. To decide what packages you want, download the file Packages from:



This file is a current list of Debian packages available in the stable Debian distribution. The file comes in a special format; every package has its own entry separated by a blank line. Each package's information is broken up into fields. The table on page  describes the fields and their possible values. It should give you an idea of how to build your personal download list. When you have the list of the packages you want, you need to decide how to download them. If you are an experienced user, you may want to download the netbase package--and SLIP and PPP, if necessary--so you can download packages later, via Linux. Otherwise, you can download all of the packages with your current operating system and install them later from a mounted partition.

2.3.4 Booting from floppies and installing Debian GNU/Linux.

The Rescue floppy.

Place the Rescue floppy in the boot drive and reboot. In a minute or two, you should see a screen introduce the Rescue floppy and the boot prompt.
It's called the Rescue floppy because you can use it to boot your system and perform repairs if there is a problem that makes your hard disk unbootable. Save this floppy after you install the system.
You can do two things at the boot: prompt: press the function keys F1 through F10 to view a few pages of helpful information, or boot the system. If you have any hardware devices that Linux doesn't access correctly at boot time, you may find a parameter to add to the boot command line in the screens you see by pressing F3, F4, and F5.
If you add parameters to the boot command line, be sure to type the word ``linux'' and a space before the first parameter. If you simply press Enter, that's the same as typing ``linux'' without any special parameters.
If this is the first time you're booting the system, press Enter and see if it works correctly. It probably will. If not, you can reboot later and look for any special parameters that inform the system about your hardware.
Once you press Enter, you should see the messages

then there is a page or so of cryptic information about the hardware in the system. There may be many messages in the form of, ``can't find something,'' ``something not present,'' ``can't initialize something,'' or even ``this driver release depends on something,'' Most of these are harmless. The installation boot disk is built to run on computers with many different peripheral devices. Obviously, no computer will have every possible peripheral device, and the operating system may emit a few complaints while it looks for peripherals you don't own. You may also see the system pause for a while. This happens if it is waiting for a device to respond that is not present on your system. If you find that the time it takes to boot the system unacceptably long, you can create a custom kernel after you install the system which doesn't have the drivers for non-existent devices.

Low memory systems.

If your system has 4MB of RAM, you may see a paragraph about low memory and a text menu with three choices. If your system has enough RAM, you won't see this at all, and you'll go directly to the color or monochrome dialog box. If you get the low-memory menu, you should go through its selections in order. Partition your disk, activate the swap partition, and start the graphical installation system. The program that is used to partition your disk is called cfdisk, and you should see the manual page for cfdisk and the instructions on page  for assistance.
cfdisk is used to create a Linux Swap partition (type 82) on the hard drive. You need the swap partition for virtual memory during installation, because the procedure likely uses more memory than you have physical RAM for. Select the amount of virtual memory that you intend to use once your system is installed. It is exactly equal to the amount of disk space required. Sixteen megabytes is probably the smallest practical amount, but use 32 megabytes if you can spare the disk space, and 64 megabytes if the disk is large enough and you won't miss the space.

The color or monochrome dialog box.

Once the system finishes booting, you should see the color or monochrome dialog box. If your monitor displays black and white (monochrome), press Enter and continue with the installation. Otherwise, use the arrow key to move the cursor to the Color menu item and then press Enter. The display should change from black and white to color. Press Enter again to continue with the installation.

The Main Menu

You may see a dialog box that says,

On some systems, this message flashes by too quickly to read. It is displayed between steps in the installation process. The installation program checks the state of the system after each step. This allows you to restart the installation without losing the work that you have already done, if you halt the system in the middle of the installation. If you need to restart an installation, you will be prompted to select color or monochrome again, configure the keyboard, reactivate the swap partition, and remount any disks that have been initialized. Any other installation on the system will be saved.
During the entire process, you are presented with the main menu. The choices at the top of the menu change to indicate your progress in installing the system. Phil Hughes wrote in Linux Journal that you could teach a chicken to install Debian. He meant that the installation process was mostly just pecking at the Enter key. The first choice on the installation menu is the next action you should perform according to what the system detects you have already done. It should say Next, and, at this point, the next item should be

Configuring the keyboard.

Make sure that the highlight is on the Next item, and press Enter for the keyboard configuration menu. Select a keyboard that conforms to the layout used for your national language, or select something close to it if the keyboard layout you want isn't shown. After installation you can select a keyboard layout from a wider range of choices. Move the highlight to the keyboard selection and press Enter. Use the arrow keys to move the highlight--they are in the same place on all national language keyboard layouts and are independent of the keyboard configuration.

The shell.

If you are an experienced UNIX or Linux user, press LeftAlt and F2 in unison for the second virtual console. That's the Alt key on the left-hand side of the Space bar and the F2 function key. You'll see a separate window running a Bourne shell clone called ash. At this point, the root file system is on the RAM disk, and there is a limited set of UNIX utilities available for your use. You can see what programs are available with the command


The shell and commands are there only in case something goes wrong. In particular, you should always use the menus, not the shell, to activate your swap partition, because the menu software can't detect whether you've done this from the shell. Press LeftAlt-F1 to get back to menus. Linux provides up to 64 virtual consoles, but the Rescue floppy only uses a few of them.
Last chance! Have you backed up your disks? Here's your first chance to wipe out all of the data on your disks, and your last chance to save your old system. If you haven't backed up all of your disks, remove the floppy from the drive, reset the system, and create a backup.

Partition your hard disks.

If you have not already partitioned your disks for Linux Native and Linux Swap file systems, the menu item Next will be

If you have already created at least one Linux Native and one Linux Swap disk partition, the Next menu selection will be

or you may even skip that step if your system has little RAM and the installation software asked you to activate the swap partition as soon as the system started. Whatever the Next menu selection is, you can use the down-arrow key to select


The Partition a Hard Disk menu item presents you with a list of disk drives you can partition and runs the cfdisk program (see page ), which allows you to create and edit disk partitions. You must create at least one Linux (type 83) disk partition.
Your swap partition will be used to provide virtual memory for the system and should be between 16 and 128 megabytes in size, depending on how much disk space you have and how many large programs you want to run. Linux will not use more than 128 megabytes of swap, so there's no reason to make your swap partition larger than that. A swap partition is strongly recommended, but you can do without one if you insist and system has more than 16 Mb of RAM.

Initialize and Activate the Swap Disk Partition.

This is the Next menu item after you create one disk partition. You have the choice of initializing and activating a new swap partition, activating a previously initialized partition, and doing without a swap partition. It's always permissible to re-initialize a swap partition, so select Initialize and Activate the Swap Disk Partition unless you are sure that you know what you are doing. This menu choice will give you the option to scan the entire partition for unreadable disk blocks caused by defects on the surface of the hard disk platters. This is useful if you have MFM, RLL, or older IDE disks, and checking the disk never hurts. Properly working SCSI disks don't need to be scanned. They have their own internal mechanism for mapping out bad disk blocks.
The swap partition provides virtual memory to supplement the RAM in your system, and it's even used while the system is being installed. That's why we initialize it first.

Initialize a Linux disk partition.

At this point, the Next menu item should be

If it isn't, you haven't completed the disk partitioning process, or you haven't made one of the menu choices dealing with your swap partition.
You can initialize a Linux disk partition, or alternately you can mount a previously initialized partition.
The boot floppies will not upgrade an old system without removing the files--Debian provides a different procedure than using the boot floppies for upgrading existing Debian systems. Thus, if you are using old disk partitions that are not empty, you should initialize them, which erases all of the files. You must initialize any partition that you created in the disk partitioning step. About the only reason to mount a partition without initializing it at this point would be to mount a partition upon which you have user files, like /home, that you don't want deleted.
Select the Next menu item, to initialize and mount the root (the ``/'' directory) disk partition. The first partition you mount or initialize, after the swap partition, if you're using it, is the partition mounted as root. You will be offered the choice to scan the disk partition for bad blocks, as when you initialized the swap partition. It never hurts to scan for bad blocks. Keep in mind that this step can take 10 minutes or more if you have a large disk.

Install the base system.

Once you've mounted the root partition, the Next menu item will be

unless you already performed some of the installation steps. You can use the arrow keys to select the menu items to initialize or mount disk partitions if you have additional partitions to set up. If you have created separate partitions for /var, /usr, or other file systems, you should initialize and mount them now.
There will be a pause while the system looks for a local copy of the base system. This search is for CD-ROM installations and will not succeed. You are then offered a menu of drives from which to read the base floppies. Select the appropriate drive. Feed in the Base 1, Base 2, Base 3, and Base 4 floppies--and Base 5 if you are using 1.2MB floppies--as requested by the program. If one of the base floppies is unreadable, you need to create a replacement floppy and feed all five floppies into the system again. After the floppies have been read, the system installs the files. This can take ten minutes or more on a slow system.

Install the operating system kernel.

At this point, the Next menu item should be

Select it, and you will be prompted to select a floppy drive and insert the Rescue floppy. This copies the kernel onto the hard disk. This kernel is used later to create a custom boot floppy for your system and make the hard disk bootable without a floppy.

Install the device drivers.

Select the menu item to install the device drivers. You will be prompted to insert the Device Drivers floppy, and the drivers will be copied onto your hard disk. Select the

menu item and look for devices which are on your system. Configure those device drivers, so they will be loaded whenever your system boots.
There is a menu selection for PCMCIA device drivers, but you do not need to use it. After installation you can install the pcmcia-cs package. This detects PCMCIA cards automatically and configures those it finds. It also recognizes cards that are hot swapped when the system is running--they will all be configured as they are plugged in, and de-configured when unplugged.

Configure the base system.

At this point the system read in all of the files that make up a minimal Debian system, but you must perform some configuration before the system will run. Select

This asks you to select your time zone. Look for your time zone or region of the world in the menu, and type it at the prompt. This may lead to another menu where you can select more specific information.
Next, you are asked if your system clock should be set to Greenwich Mean Time (GMT) or local time. Select GMT if are running only Linux or another UNIX on your system. Select local time if you use another operating system like MS-DOS or Microsoft Windows. UNIX systems keep GMT time on the system clock and use software which converts it to the local time. This allows them to keep track of daylight savings time and leap years, and even allows users who are logged in from other time zones to individually set the time zone on their terminal. If you run the system clock on GMT and your locality uses daylight savings time, the system adjusts for daylight savings time properly on the days it starts and ends.

Configure the network.

You must configure the network even if you don't have one, but you only have to answer the first two questions:

If you are connected to a network, check with your system administrator or ISP vendor if you don't know the following information:

• your computer's host name;
• your computer's or ISP's domain name;
• your computer's IP address;
• the netmask to use with your network;
• the IP address of your network;
• the broadcast address to use on your network;
• if your network has a gateway, the IP address of the default gateway system to which you should route packets;
• the system on your network to use for Domain Name Service (DNS); and
• whether you connect to the network using Ethernet.

The program will guess that the network IP address is the bitwise AND of your system's IP address and netmask. It will guess that the broadcast address is the bitwise OR of your system's IP address with the bitwise negation of the netmask. It will guess that your gateway system is also your DNS server. If you can't find any of these answers, use the system's guesses--if necessary, you can alter them after installation by editing the /etc/init.d/network file.

Make the hard disk bootable.

If you choose to make the hard disk boot directly to Linux, you are asked to install a master boot record. If you aren't using a boot manager (this is probably the case if you don't know what a boot manager is), answer ``yes'' to this question. The next question is whether you want to boot Linux automatically from the hard disk when you turn on the system. This sets Linux to be the bootable partition--the one that will be loaded from the hard disk. If you answer ``no'' to this question, you can set the bootable partition later using the MS-DOS FDISK.EXE program, or the Linux fdisk or activate programs.

Make a boot floppy.

You should make a boot floppy even if you intend to boot the system from the hard disk. The reason for this is that it's possible for the hard disk bootstrap to be installed incorrectly. A boot floppy will almost always work. Select

from the menu and feed the system a blank floppy as directed. Make sure that the floppy isn't write protected. The software attempts to format and write it. Mark this diskette the ``Custom Boot'' floppy and write-protect it once it has been written.

The moment of truth.

This is what electrical engineers call the ``smoke test''--what happens when you power up a new system for the first time. Remove the floppy disk from the floppy drive and select

from the menu. If the Linux system doesn't start up, insert the Custom Boot floppy you created in the previous step and reset the system. Linux should boot. You should see the same messages as when you first booted the installation boot floppy, followed by some new messages.

Add a user account and password.

After you've added logins, (Chapter 4 discusses this in some detail), you are dropped into dselect, the Debian package management program.
You should read the tutorial before attempting to install packages with dselect.
dselect allows you to select the packages that you want installed on your system. The Debian package management software is described in detail starting on page . If you have a CD-ROM or hard disk with the additional Debian packages or are connected to the Internet, you may want to read that section now. Otherwise, exit dselect. You can use the package management software after you have transferred the Debian package files to your system.
You must be the superuser (root) to use dselect.
If you install the X Window System and do not use a US keyboard, read the X11 Release note for non-US keyboards.

Log in.

After you exit dselect, you are at the login: prompt. Log in using the personal login and password you selected. Your system is ready to use.

2.3.5 Running Debian GNU/Linux.

  This section describes the Debian packaging system and Debian-specific utilities. The Debian/GNU Linux Packages file format is shown in the table on page .

  
Table 2.3: Fields in a Debian/GNU Linux Packages file record.


Debian distributions come in archives called packages. Every package is a collection of files (programs, usually) that can be installed using dpkg or dselect. In addition, the package contains some information about itself that is read by the installation utilities.

Package classifications.

The packages that are included with Debian GNU/Linux are classified according to how essential they are (priority) and their functionality (section).
The priority of a package indicates how essential or necessary it is. Debian GNU/Linux classifies all packages into four different priority levels:

Required.

These packages must be installed for the system to operate correctly and have been installed as part of the base system.
Never remove a required package from the system unless you are absolutely sure of what you are doing. This bears repeating: Never, never, never remove a required package from the system unless you are absolutely sure of what you are doing. It is likely that doing so will render your system completely unusable.
Required packages are abbreviated in dselect as Req.

Important.

Important packages are found on almost all UNIX-like operating systems. These packages include cron, man, and vi.
Important packages are abbreviated in dselect as Imp.

Standard.

Standard packages are packages that, more or less, comprise the ``standard,'' character based, Debian GNU/Linux system. The Standard system includes a fairly complete software development environment and GNU Emacs.
Standard packages are abbreviated in dselect as Std.

Optional.

Optional packages comprise a fairly complete system. The Optional system includes TeX and the X Window System.
Optional packages are abbreviated in dselect as Opt.

Extra

Extra packages are only useful to a small or select group of people, or are installed for a specific purpose. Extra packages might include such programs as electronics and ham radio applications.
Extra packages are abbreviated in dselect as Xtr.
By default, dselect automatically selects the Standard system if the user doesn't want to individually select the packages to be installed.
The section of a package indicates its functionality or use. Packages on the CD-ROM and in FTP archives are arranged in subdirectories according to function. The directory names are fairly self-explanatory: for example, the directory admin contains packages for system administration and the directory devel contains packages for software development and programming. Unlike priority levels, there are many sections, and more may be added in the future, so we do not individually describe them in this guide.

Package relationships.

Each package includes information about how it relates to the other packages included with the system. There are four package relationships in Debian GNU/Linux: conflicts, dependencies, recommendations, and suggestions.
A conflict occurs when two or more packages cannot be installed on the same system at the same time. A good example of conflicting packages are mail transfer agents (MTAs). A MTA is a program that delivers electronic mail to users on the system and other machines on the network. Debian GNU/Linux has two mail transfer agents: sendmail and smail.
Only one mail transfer agent may be installed at a time. They both do the same job and are not designed to coexist. Therefore, the sendmail and smail packages conflict. If you try to install sendmail when smail is already installed, the Debian GNU/Linux package maintenance system will refuse to install it. Likewise, if you try to install smail when sendmail is already installed, dselect (or dpkg; see below) will refuse to install it.
A dependency occurs when one package requires another package to function properly. Using our electronic mail example, users read mail with programs called mail user agents (MUAs). Popular MUAs include elm, pine, and emacs RMAIL mode. It is normal to install several MUAs at once because they do not conflict. But MUAs do not deliver mail--that is the job of the MTA. So all mail user agent packages depend on a mail transfer agent.
A package can also recommend or suggest other related packages.

2.3.6 dselect.

This section is a brief tutorial on Debian dselect. For more detailed information, refer to the dselect manual at


dselect is simple, menu-driven interface which helps install packages. It takes you through the package installation process in the order of the on-screen menu:



There are two ways to select an option from the menu: choose it with arrows, or press the key of the corresponding letter in brackets.

Access.

In this menu you choose the method to obtain and install the packages.

Update.

dselect reads the Packages database (described above) and creates a database of the packages available on your system.

Select.

This section of the program selects the packages. Choose your the package you want and press Enter. If you have a slow machine, the screen may clear and remain blank for 15 seconds. The first thing that appears is Page 1 of the Help file. You can view this screen by pressing "?" at any point in the Select screens, and you can page through the help screens by pressing the . (period) key.
To exit the Select screen after all of the selections are complete, press Enter. This returns you to the main screen if there are no problems with your selection. You must resolve those problems first. When you are satisfied with any given screen, press Enter.
Dependency conflicts are quite normal and to be expected. If you select package A and that package requires the unselected package B in order to run, dselect warns you of the problem and will most likely suggest a solution. If package A conflicts with package B, you must decide between them.

Install

dselect runs through the entire 800 packages and installs the ones that are selected. You will need to make decisions during this process. It is often useful to switch to a different shell to compare, for example, an old configuration file with a new one. If the old file is called conf.modules, for example the new file will be called conf.modules.dpkg-new.
The screen scrolls by fairly quickly on faster machines. You can halt the display is by pressing Control-S and restart it with Control-Q. At the end of the run, there will be a list of any uninstalled packages.

Configure.

Most packages are configured in Step 3, but anything remaining can be configured here.

Remove.

Remove packages that are no longer needed.

Quit.

Au revoir.

2.3.7 dpkg.

This is a command line tool that installs and manipulates Debian packages. It has several options that allow you to install, configure, update, remove, and perform other operations on Debian packages. You can even build your own packages. dpkg also allows you to list the available packages, files "owned" by packages, which package owns a file, and so on.

Installing or updating new or existing packages.

Type the following command:


where filename is the name of the file containing a Debian package, like tcsh_6.06-11_i386.deb. dpkg is partly interactive; during the installation it may ask additional questions, like whether to install the new version of a configuration file or keep the old version.
You may also unpack a package without configuring it by entering:



If a package depends on an uninstalled package or a newer version of a package you already have, or if any other dependency problem occurs during the installation, dpkg will exit without configuring it.

Configuring installed packages.

If dpkg aborts during installation and leaves a package installed, the package is left unconfigured. The Debian packaging system requires the package to be configured to avoid dependency problems. Some packages also require configuration to work properly.
To configure a package, type:

where package is the name of the package, like tcsh. (Notice that this is not the original name of the file from which tcsh was installed, which was longer, included a version number, and ended in .deb.)

Removing installed packages.

In the Debian package system, there are two ways to eliminate packages: remove and purge. The remove option removes the specified package; the purge option removes both the specified package and its configuration files. The usage is:


If there are any installed packages that depend on the one you wish to remove, the package will not be removed, and dpkg will abort with an error message.

Reporting package status.

To report the status of the package (e.g., installed, not installed, or unconfigured), enter:

Listing available packages.

To list the installed packages that match some pattern, type:


where package-name-pattern is an optional argument specifying a pattern for the package names to match, like *sh. Normal shell wildcards are allowed. If you don't specify the pattern, all of the installed packages are listed.

Listing files owned by packages.

To list all the files owned by a particular package, simply type:

However, this does not list files created by package-specific installation scripts.

Finding the package that owns a file.

To find the package which owns a particular file, type the following command:


where filename-pattern is the pattern with which to search the package names for a match. Again, normal shell wildcards are allowed.

Summary.

dpkg is simple to use and is preferred over dselect when all that you need to do is install, upgrade, or remove a small number of packages. It also has some functionality that dselect (an interface to dpkg) doesn't have, like finding which package owns a file. For the full list of options, refer to the dpkg(8) manual page.

2.3.8 About Debian GNU/Linux.

The Debian GNU/Linux Project was created by Ian Murdock in 1993, initially under the sponsorship of the Free Software Foundation's GNU project. Later, Debian separated from FSF. Debian is the result of a volunteer effort to create a free, high-quality UNIX-compatible operating system based on the Linux kernel, complete with a suite of applications.
The Debian community is a group of more than 150 unpaid volunteers from all over the world who collaborate via the Internet. The founders of the project have formed the organization Software in the Public Interest (SPI) to sponsor Debian GNU/Linux development.
Software in the Public Interest is a non-profit organization formed when the FSF withdrew their sponsorship of Debian. The purpose of the organization is to develop and distribute free software. Its goals are very much like those of FSF, and it encourages programmers to use the GNU General Public License on their programs. However, SPI has a slightly different focus in that it is building and distributing a Linux system that diverges in many technical details from the GNU system planned by FSF. SPI still communicates with FSF and cooperates in sending them changes to GNU software and in asking its users to donate to FSF and the GNU project.
SPI can be reached by post at:

Software in the Public Interest
P.O. Box 70152
Pt. Richmond, CA 94807-0152

2.3.9 Mailing lists.

 
There are several Debian-related mailing lists:



There are also several mailing lists for Debian developers.
You can subscribe to those mailing list by mail or the World Wide Web. For more information, please visit http://www.debian.org/.

2.3.10 Bug tracking system.

The Debian project has a bug tracking system that handles bug reports by users. As soon as the bug is reported, it is given a number and all of the information provided on the particular bug is stored in a file and mailed to the maintainer of the package. When the bug is fixed, it must be marked as done (``closed'') by the maintainer. If it was closed by mistake, it may be reopened.
To receive more information on the bug tracing system, send e-mail to request@bugs.debian.org with help in the body of the message.

2.3.11 Debian Acknowledgments.

Many thanks to Bruce Perens and the other authors of Debian related materials that were used to write this chapter. Thanks also to Vadik Vygonets, my beloved cousin, who helped me quite a bit. Lastly, thanks are also due to the members of Debian community for their hard work. Let's hope that Debian GNU/Linux becomes even better.

2.3.12 Last note.

Debian GNU/Linux changes very quickly, and many facts can change more quickly than this book. The source text of this section is updated regularly. You can find it at



   

2.4 Red Hat Linux.

red-hat-installation
   
This section on Red Hat Linux was written by Henry Pierce.

2.4.1 Red Hat Linux installation features.

2.4.2 The RPM package management system.

Red Hat Linux's RPM package management system manages software by defining how a software package is built for installation and collects information about its components and installation methods during the build process. A RPM package has an organized packet of data in the header of package.rpm which can be added to a database that describes where the package belongs, what supporting packages are required, whether the required packages are installed, and provides a means to determine software dependencies.
RPM gives system administrators the ability to: upgrade individual components or entire systems while preserving the configuration of the system or package; query the database for the location of files, packages, or related information; perform package verification so packages can be installed properly, or at all; keep source packages ``pristine'' (provide the package author's original source and second-party patches separately) so that porting issues can be tracked. Because RPM does this, you can install, upgrade, or remove packages with a single command line in text mode or a few clicks of the mouse in X Package Management Tool. Examples of using RPM from the command line are:

Package naming conventions.

A properly built package.rpm has the following characteristics: its name identifies the package, the version, the build revision, the architecture, and the extension .rpm, which identifies it as a RPM package.
Take, for example, bash-1.14.7-1.i386.rpm. The name itself contains useful information: the package is bash (the Bourne Again SHell), it is version 1.14.7, and it is build 1 of the current version for Red Hat Linux. It was built for Intel or compatible 80386 or higher CPUs, and it is in RPM format. So, if you see a package named bash-1.14.7-2.i386.rpm, you know that it is the second build of bash version 1.14.7, and probably contains fixes for problems of the previous build and is more current. While the internal organization of a *.rpm file is beyond the scope of this section, a properly built package contains an executable file, any configuration files, the documentation (at least manual pages), any miscellaneous files directly related to the package, a record of where the packages files are to be installed, and a record of any required packages. After successful installation, information about the package is registered in the system's RPM database. A more thorough discussion of RPM package management system may be found in the RPM HOWTO (see Appendix A). It is also available at

2.4.3 A note about upgrading Red Hat Linux.

Only upgrades from version 2.0 of Red Hat Linux and onward are supported due to major changes in Linux's binary format. Otherwise, upgrades can be performed from the same installation methods of CD-ROM, NFS, FTP, and hard disk. As of Red Hat Linux version 4.0, the upgrade option is incorporated into the Boot diskette instead of a separate program. If you upgrade from v2.1 to v3.0.3 and want to upgrade again to version 4.0, you need to create the Boot diskette instead of looking for an upgrade script, the same as Red Hat 4.x installation from scratch. This method does not reformat your partitions nor delete your configuration files.

2.4.4 Creating the installation floppies.

To create an Installation Floppy Kit, you need the following:

1. The Red Hat Boot diskette image, boot.img, which is available at:
   
   or in the images directory of a Red Hat CD-ROM.
2. The Red Hat Supplemental diskette image, supp.img, which is available at
   
   or in the images directory of a Red Hat CD-ROM. This diskette is required if your method of installation is not CD-ROM based, or you need PCMCIA support for any device, like a CD-ROM on a laptop, to install properly. This diskette can also be used with the Boot diskette as an emergency start disk for an installed system.
3. The program RAWRITE.EXE, which is available at:
   
   or in the DOS directory of a Red Hat CD-ROM.
4. MS-DOS and Windows 95 users installing Red Hat Linux for the first time on a machine that will have Linux installed as a second operating system should also obtain:
   
   and unzip the files into: C:FIPS if you need to free space on your hard drive.
5. An Emergency Boot diskette for an existing operating system on the target machine on which Linux will be installed as a second operating system must be created.

2.4.5 Installation media.

After you create the Installation floppies using RAWRITE.EXE or dd as described on page , insure that your installation method is properly set up for the Red Hat installation diskettes. For CD-ROM, NFS, FTP, and hard drive installation methods, the source must have the directory /RedHat on the ``top level'' with the directories /base and /RPMS underneath:

NFS installation.

For NFS installation, you will either need a Red Hat CD-ROM on a machine (such as an existing Linux box) that can support and export an ISO-9660 file system with Rockridge Extensions, or you need to mirror one of the Red Hat distributions with the directory tree organized as described above--and of course, the proper files in each directory. The directory /RedHat needs to be exported to the machines on the network that are to have Red Hat Linux installed or upgraded. This machine must be on an Ethernet; you can not do an NFS install via dialup link.

Hard drive installation.

Hard drive installations must have the /RedHat directory created relative to the root directory of the partition (it doesn't matter which partition) that will contain the Red Hat distribution obtained either from CD-ROM or an FTP site. For example, on the primary DOS partition, the path to \RedHat should be C:\RedHat. On a MS-DOS file system, it does not matter that the package.rpm names are truncated. All you need to do is make sure the \RedHat\base directory contains the base files from a CD-ROM or FTP site and the \RedHat\RPMS directory contains all of the package.rpm files from the CD-ROM or FTP site. Then you can install or upgrade from that partition. If you have an existing Linux partition that is not needed for an installation or upgrade, you can set it up as outlined here and use it.

FTP installation.

To install via FTP over the Internet, all you need is the IP address of the FTP server and the root directory path for the Red Hat Linux system you wish to install. See Appendix B for a list of Linux FTP sites and mirrors. If you intend to do an FTP installation over a low-bandwidth connection (anything slower than a 128K ISDN link), it is highly recommended you copy the files to an existing MS-DOS hard drive partition and then install from the hard drive. The total size of the packages in the /RedHat/RPMS directory is approximately 170MB and will take many hours to install. If something goes wrong with the installation, such as the link going down, you need to start from the beginning. If you get the files first and set up your hard drive to install Linux, it is then less work and less confusing to recover from a failed installation. You don't even need to download all of the files in /RedHat/RPMS to successfully install a minimal system which can grow with your needs. See the next section for details.

2.4.6 Customizing your NFS or hard drive installation.

You can customize the installation process. This is not for the faint of heart--only those already familiar with Linux should attempt it. As of Red Hat Linux version 4.x, the /RedHat/RPMS directory contains approximately 170MB of .rpm files. RPM compresses these packages and assumes that the packages need an average of 2 to 3MB of hard drive space for every 1MB of RPM package volume. If package.rpm is 6MB in size, you need between 12 and 18MB of free space to install the package.
Customizing which packages are available for installation is an option when installing the system via FTP, NFS, and hard drive. CD-ROMs (typically) cannot be written to, but you can copy the files to the hard drive and install from there with the customized package list. FTP and NFS installations can only be designed if you have root access to the server(s) on your network or your system administrator is willing to work with you. The following installation situations make custom installation desirable: when obtaining Red Hat Linux via FTP over a low-bandwidth connection or when designing a suite of software to be used by all Red Hat Linux workstations on a network.
To customize the installation, you must obtain the /base/comps file which provides you with the list of packages that a full installation would normally include. Then, the packages you actually want to install from /base/comps need to be downloaded. The /base/comps file must be edited to reflect the packages that you obtained and are going to install.
If you have local RPM packages, you can add them to the comps file as well.

The comps file.

The Red Hat installation program uses the file /RedHat/base/comps (the file here is an example from Red Hat Linux version 4.0) to determine what packages are available in the /RedHat/RPMS directory for each category to be installed. The file is organized by category, and each category contains a list of packages Red Hat believes are the minimum required for that section. NOTE: only the package part of a package's name (package-version-build.rpm) is listed in the file. This means the comps file is generally usable from one version of Red Hat to the next. A section in this file has the structure:



That is a tag to identify the category number, the category, a list of the package names in the category, and the tag ``end'' to mark the end of the category.
Without exception, everyone needs all of the software packages listed in the Base section of the file. The other sections, though, can generally be customized or eliminated to suit a particular need. For example, there are three types of Networked Stations: ``plain'', management, and dial-up. An examination of these sections shows that many of the software packages are listed in all three categories, but some software packages are specific to the category. If you are creating a Dial-up Networked Station, then you can safely eliminate the ``Plain'' and ``Management'' sections and any software unique to those categories. Conversely, if you only need basic networking capability for networked work stations, the other sections can be eliminated from the file as well as the software unique to those sections. All you need to do is make sure that you have all of the software packages listed in that category. If you have local custom packages (those not provided by Red Hat Software), you should add them to an existing category that is appropriate rather than creating a new category.
Because the list of packages in each category only contains the name of the package (i.e., not the entire package-name-version-build.rpm), you can substitute any updates Red Hat has made available in the updates directory on:


or one of Red Hat's mirror sites for the original package found in the distribution's original /RedHat/RPMS directory. The installation program is relatively version-insensitive. The only warning here is to insure that package dependencies are met. When an RPM package is built, RPM itself tries to determine what packages must be installed for the package to work (the RPM developer also has direct control of this as well--he or she can add dependencies that RPM might not ordinarily detect). This is where a little experimentation or research may be needed. For example, one way to determine package dependencies (if you have user access to your NFS server on an existing Red Hat Linux box) is to telnet or login into it (or if you have the CD-ROM, mount it and go to the RedHat/RPMS directory) and query the package for its dependencies:



The ``-q'' puts rpm in query mode, the ``-p'' tells rpm to query an uninstalled package, and the ``-R'' tells rpm to list the target package's dependencies. In this example, we see libc.so.5 and libtermcap.so.2 are required. Since libc and termcap are part of the base of required software (as is bash), you must insure that the libc and libtermcap packages (the dependency packages) are present to be able to install bash (the target). As long as you get the entire base system installed, you can boot the system when the installation program completes. You can add additional packages to Red Hat Linux as required even if the installation program reports that a package failed to install because its dependencies were not met.
The table on page  describes the categories of software found in /base/comps in Red Hat v4.0:

  
Table 2.4: Important Red Hat Linux packages.

2.4.7 Recommended minimal installation.

It is difficult to determine how much space an installation will require. However, someone installing via FTP should get the Base system and the Dialup Networked Station and install these. Then, additional software can be obtained and added as the need arises. Of course, if you want to do C programming, you should get the relevant packages and edit the comps file appropriately.
If you encounter a package during the installation which requires another package that you don't have available, or you make a mistake in the comps file, you can generally finish the installation and have a bootable, working system. You can correct the problem by manually adding the failed packages and their dependencies later. Overall, get the entire Base system and one of the Networked Station packages installed, and you can add anything you need or want later.

2.4.8 How much space do you really need?

The table on page  gives approximate disk space requirements Red Hat Linux and various subsystems.
  
Table 2.5: Typical Red Hat Linux disk space requirements.

2.4.9 Installation.

By now, you should have created the Installation Floppy Kit, prepared your hard drive, and have your installation media ready. The details of the installation follow. You first begin by booting your system and configuring the installation program to install from your selected medium. After this the installation proceeds with the same steps for everyone. You need to begin by booting your computer with the diskette labeled ``Boot diskette.''

2.4.10 Installation media revisited.

As the boot diskette starts up, the kernel attempts to detect any hardware for which the boot kernel has drivers compiled directly into it. Once booting is complete, a message appears which asks if you have a color screen (if you do, select ``OK''). Next comes the Red Hat Welcome screen. Choose ``OK'' to continue. The next question asks if you need PCMCIA support. You must answer ``yes'' if you are installing to a laptop, inserting the Supplemental diskette when prompted. Once PCMCIA support is enabled if necessary, you are presented with a screen that asks what type of installation method to use. Follow the instructions in the following sections for the method you chose.

CD-ROM installation.

To install from CD-ROM, highlight ``Local CD-ROM'' from the list of installation types. Then click ``OK''. You will be asked if you have a SCSI, IDE/ATAPI, or proprietary CD-ROM. This is where some of the hardware research pays off: if you have 4X or faster CD-ROM drive that was made recently and bundled with a Sound Blaster or other sound card, you most likely have an IDE/ATAPI type drive. This is one of the most confusing issues facing you.
If you choose SCSI, you must know what kind of SCSI card you have and will be presented a list. Scroll down the list until you find your SCSI card. After you select it, you will be asked if you wish to AUTOPROBE for it or SPECIFY OPTIONS. Most people should choose AUTOPROBE, which causes the program to scan for your SCSI card and enable the SCSI support for your card when found.
After the Installation Program has successfully located the Red Hat CD-ROM, you should read the next section.

Hard drive installation.

To install from a hard drive, highlight this option and choose `` OK''. If you have not already chosen PCMCIA support, you will be prompted to insert the Supplemental diskette.

NFS installation.

To install via NFS, highlight this option and choose ``OK''. You must choose the Ethernet card installed on the target machine so the Installation Program can load the correct driver. Highlight the appropriate card from the list, and then select ``OK'', allowing the Installation Program to AUTOPROBE for your card.
If your machine locks up, you must press Ctrl-Alt-Delete to reboot the system. Most of the time, when this happens, it is because the probing interferes with a non-Ethernet card. If this happens, try again and choose SPECIFY OPTIONS, and give the data about your card in this form:


This instructs the probe to look at the location specified by the values IRQ and IO_PORT for the Ethernet card. If your Ethernet card is configured for IRQ 11 and IO_PORT 0x300, specify:


After the card has been successfully found, you will be prompted for TCP/IP information about your machine and the NFS server with the Linux installation packages. First, you will be asked to provide the target machine's IP Address, Netmask, Default Gateway, and Primary Name Server. For example:


After you select OK, you are prompted for the target machine's Domain name and Host name. For example, if your domain name is infomagic.com and host name is vador, enter:



The last screen prompts you for the NFS server and the exported directory containing the Red Hat distribution. For example, if your NFS server is redhat.infomagic.com, enter:



If you do not know these values, ask your system administrator. After you enter the values, select OK to continue. If the installation program reports an error locating the Red Hat distribution, make sure that you have the correct values filled in above and that your network administrator has given you export permission for the target machine.

FTP installation.

FTP installation is similar to the NFS installation described above. You are prompted for the Ethernet card and your machine's TCP/IP information. However, you will be asked for the FTP site name and Red Hat directory on the Red Hat mirror site, instead of NFS server information. One warning about performing an FTP installation: find the closest and least busy FTP site to your location. See Appendix B for a list of Linux FTP sites.
If your hardware isn't detected, you may need to provide an override for the hardware to be enabled properly. You may also want to check:



to see if Red Hat has updated boot diskettes for your hardware.

2.4.11 Walking through the rest of the installation.

 

1. Next, you are asked if you are installing to a New System or Upgrading Red Hat Linux 2.0 or greater. If upgrading, you will not be offered the chance to partition your hard drive or configure anything with your system except LILO. Select either INSTALL or UPGRADE to continue.
2. If you are upgrading, you will be asked for the root partition of your existing Red Hat system. Highlight the appropriate partition and press OK. If you are installing for the first time, you need to partition your hard disk with the free space determined above.
3. After you create the necessary Linux Native and Linux Swap partitions, you must initialize and enable the swap partition. You will then be asked to which partition(s) you intend to install Linux If upgrading, select the root partition. You must configure and choose at least one partition, which will be the root partition. Highlight the root partition. Then, unless you are upgrading, you are presented with a table of other available partitions. Choose the appropriate partitions and EDIT to indicate which partitions will be used for which directories. If you have more than one partition for the Linux installation, now is the time to designate those as well.
4. Next, a list of software categories to install is presented, followed by a chance to customize which software packages from each category are to be installed. If you have not installed Red Hat or other distributions of Linux before, simply choose the category of software to install and let the setup program install the defaults for each category. If you need a package that wasn't installed originally, you can always install it later. While the software is installing, you will see a progress indicator and you should get a cup or two of coffee. Installation can take thirty minutes to an hour or more, depending on software choices and hardware configuration.
5. After the software installation is done, you will be asked to configure your mouse. A discussion mouse protocols and devices starts on page .
6. Next is the X Window System configuration. It is recommend you wait until after you boot your system for the first time to configure X. If something goes wrong with the X configuration, you may need to start the installation procedure from the beginning if the Installation Program isn't able to recover.
7. If you do not have an Ethernet card, do not configure your network at this time. If you have a network card and didn't configure it earlier, you should configure it now. Configuration for a dialup network should be done after the installation is complete.
8. Next, you need to configure the system clock. UTC is a good choice if you are on a network and want daylight savings time handled properly. Local Time is okay if the computer is a stand-alone machine.
9. If you do not have a US keyboard, you will need specify the configuration for your keyboard.
10. You are prompted for the root system password. Don't forget it. Recovering the password is not a trivial matter. You will need the password to access the system when you first reboot.
11. Finally, you will be asked to configure LILO.

If you have not installed a root partition that begins and ends between cylinder 0-1023, Do not install LILO. When you reboot the system for the first time, if LILO does not allow you to boot your system correctly, use the Emergency MS-DOS and Windows 95 boot diskette and, at A:\> enter FDISK /mbr. This allows your system to boot into an existing MS-DOS or Windows 95 system as it did before LILO was installed. You can then use the Red Hat Boot diskette with the following parameters at the boot: prompt to boot your system on the hard drive:



Where xxxx is the root partition.
After the installation procedure is complet, you are ready to reboot your system and use Linux.

2.4.12 After installation.

Now that you have installed Linux and booted your system for the first time, there are some useful things to know about using your system

Understanding the LILO prompt.

When you power up or reboot the system, you may see the LILO prompt, which you hopefully configured for a 30-second or so delay before it boots the system. When LILO appears on the screen, if you do nothing, the default operating system will boot at the prescribed timeout period. However, from LILO, you can control several aspects of how Linux boots, or tell LILO to boot an alternative operating system. If you wish to override the default behavior of LILO, pressing the Shift key at the appearance of LILO will cause a ``boot:'' prompt to appear. Pressing Tab at this prompt will produce a list of available operating systems:



This tells us that ``dos'' is the default operating system, which will boot if nothing is typed; to boot Linux, type `` linux''. However, LILO lets you pass parameters to the Linux kernel which override the default behavior. For example, you may have been experimenting with start-up configuration files and did something to prevent the system from coming up properly. If so, you want to boot the system up to the point where it reads the configuration files and no further . The override for this is ``single'':


boots the system in single user mode so you can take corrective action. This is also useful if your system doesn't boot all the way to the login: prompt for some reason.

Logging in the first time.

Now that you are faced with the login: prompt for the first time, you may be wondering how to get into the system. At this point on a newly installed system, there is only one account to log in to--the administrative account, ``root''. This account is used to manage your system and do things like configure the system, add and remove users, software, and so on. To login into the account, enter ``root'' at the login: prompt and press Enter. You are aked for the password you entered during installation. Enter that password at the password: prompt. The system prompt [root@locahost] # appears after you have successfully negotiated the login. The system prompt tells you two things: you are logged in as root, and in this case, your machine is called localhost. If you named your machine during the installation process, your host name will appear instead of localhost.
 

2.5 Caldera OpenLinux

caldera-installation
This section on Caldera OpenLinux was written by Evan Leibovitch.
This section is intended to be a complement to the Getting Started Guides that Caldera ships with all of its Linux-based products. References to the Getting Started Guide for Caldera Open Linux Base is indicated throughout this section as ``the Guide''.

2.5.1 Obtaining Caldera OpenLinux.

Unlike most other Linux distributions, Caldera OpenLinux is not available for downloading from the Internet, nor can it be distributed freely, nor passed around. This is because of the commercial packages which are part of COL; while most of the components of COL are under the GNU Public License, the commercial components, such as Looking Glass and Metro-X, are not. In the list of packages included on the COL media starting on page 196 of the Guide, the commercial packages are noted by an asterisk.
COL is available directly from Caldera, or through a network of Partners around the world who have committed to supporting Caldera products. These Partners can usually provide professional assistance, configuration and training for Caldera users. For a current list of Partners, check the Caldera web site.

2.5.2 Preparing to install Caldera OpenLinux.

Caldera supports the same hardware as any other release based on Linux 2.0 kernels. Appendix A of the Guide lists mosts of the SCSI hosts supported and configuration parameters necessary for many hardware combinations.
Caldera's Guide provides an installation worksheet that assists you in having at hand all the details of your system that you'll need for installation. It is highly recommended you complete this before starting installation; while some parameters, such as setting up your network, are not required for installation, doing it all at one time is usually far easier than having to come back to it. Sometimes this can't be avoided, but do as much at installation time as possible.

2.5.3 Creating boot/modules floppies.

The COL distribution does not come with the floppy disks required for installation. There are two floppies involved; one is used for booting, and the other is a ``modules'' disk which contains many hardware drivers.
While the Guide recommends that you create the floppies by copying them from the CD-ROM, it is better to get newer versions of the disks from the Caldera web site. The floppy images on older CD-ROMs have errors that cause problems, especially with installations using SCSI disks and large partitions.
To get newer versions of the floppy images, download them from Caldera's FTP site. In directory pub/col-1.0/updates/Helsinki, you'll find a bunch of numbered directories. Check out the directories in descending order--that will make sure you get the latest versions.
If you find one of these directories has a subdirectory called bootdisk, the contents of that directory are what you want.
You should find two files:


The XXX is replaced by the version number of the disk images. At the time of writing, the current images are 034 and located in the 001 directory.
After you have these images, transfer them onto two floppies as described for generic installations on page , using the MS-DOS program RAWRITE.EXE from the Caldera CD-ROM or dd from a Linux system.
Caldera's CD-ROM is bootable if your system's BIOS allows it, but use the downloaded floppies if possible. They are newer and will contain bug-fixes that won't be in the CD versions.

2.5.4 Preparing the hard disks.

This procedure is no different than other Linux distributions. You must use fdisk on your booted hard disk to allocate at least two Linux partitions, one for the swap area and one for the root file system. If you are planning to make your system dual-boot COL with another operating system, like Microsoft Windows, MS-DOS, or OS/2, it's usually preferable to install COL last. The Linux fdisk programs recognizes ``foreign'' OS types better than the disk partitioning tools of most other operating systems.
To run the Linux fdisk, you must start your system with the boot (and maybe the modules) floppy described above. You must tell COL what kind of disk and disk controller you have. You can't even get as far as entering fdisk if Linux doesn't recognize your hard disk!
To do this, follow the bootup instructions in the Guide, from step 2 on pages 33-36. Don't bother going through the installation or detection of CDROMs or network cards at this time; all that matters at this point is that Linux ``sees'' the boot hard disk so you can partition it with fdisk. A brief description of the use of the Linux fdisk is provided on page 28 of the Guide.
Remember that when running fdisk, you need to set up both your root file system as Linux Native (type 83) and your Swap space (type 82) as new partitions. A brief discussion of how much swap space to allocate is offered on page 10 of the Guide.
As soon as you have allocated the partitions and written the partition table information to make it permanent, you must reboot.

2.6 Slackware

 
       
This section on Linux Slackware was written by Sean Dreilinger.

2.6.1 Slackware is not for you. (Or maybe it is.)

Welcome to the Slackware distribution of Linux! This section aims to help the new Linux user or administrator evaluate Slackware, plan a Slackware system, and install Slackware Linux.
Whether or not to choose Slackware as the flavor of Linux you will use is a serious consideration. It may seem like a trivial decision now, but Linux boxes have a way of taking on more and more responsibility in organizational computing environments. Plenty of Linux experiments have evolved in their first year to become mission-critical machines serving many more users and purposes than originally intended. Slackware is one of the most widely used distributions of Linux. When it comes to finding the newest, easiest, or most carefully planned distribution of Linux, Slackware may be ``none of the above''. Some background on the life and times of Slackware put things into perspective.

2.6.2 A quick history.

In 1993, Soft Landing System created one of the first organized distributions of Linux. Although it was a great start, the SLS distribution had many shortcomings (it didn't exactly work, for starters). Slackware, a godsend from Patrick Volkerding, solved most of these issues, was mirrored via FTP and pressed onto CD-ROMs worldwide, and quickly became the most widely used flavor of Linux. For a while, Slackware was the only full featured Linux ``solution.'' Other Linux distribution maintainers, both commercial and nonprofit, have gradually developed distributions that are also well worth your consideration.
By January 1994, Slackware had achieved such widespread use that it earned a popular notoriety normally reserved for rock stars and cult leaders. Gossip spread through the Usenet suggesting that the entire Slackware project was the work of witches and devil-worshippers!
``Linux, the free OS....except for your SOUL! MOUHAHAHAHA!''

2.6.3 Why, then?

If you are a system administrator, you may already be dealing with one or more key servers running Slackware. Unless you have time to experiment at work, sticking to the tried-and-true distribution may be the easiest way to go. If you expect to get help from UNIX literate friends and colleagues, you had better make sure they're running something compatible--odds are they're running Slackware. Its shortcomings are widely acknowledged, for the most part discovered, documented, and patched whenever possible. You can put together a Slackware box, close the known security holes, and install some complementary tools from the other Linux distributions to create an excellent UNIX server or desktop workstation, all in about half a day.
Have a look also at the Buyer's Guide published in the Linux Journal, which gives a thorough comparison and evaluation of each major distribution. For a straightforward listing of Linux flavors, have a look at the Linux Distribution HOWTO (see Appendix A).

2.6.4 Upgrade? Think twice!

One thing we don't hear too often with Slackware is the U-word. Slackware's setup program is designed to put a fresh operating system onto empty hard disks or empty disk partitions. Installing on top of a previous Slackware installation can erase your custom applications and cause compatibility problems between updated applications and older files on the same system. When Slackware was first put together, everyone was a first-time Linux user, and the system was always experimental--reinstalling the entire operating system and applications was the norm in a developmental system. Today, many institutions and businesses run mission-critical applications on Slackware Linux. In such environment, a simple reboot is a planned activity and taking down the system and overwriting all the user files or custom applications is absolutely unacceptable.
Teaching you how to finagle a Slackware upgrade is beyond the scope of this chapter, but it is workable if you are an experienced UNIX administrator and you've taken precautions to preserve your local modifications and user files. There is an Internet resource that claims to analyze your distribution and bring it up to date across the Internet. you might want to have a look at this URL if you're facing an upgrade situation:



Or read, weep, and learn from the upgrade expertise of Greg Louis in his mini HOWTO document: Upgrading Your Linux Distribution available where finer LDP publications are mirrored:

2.6.5 Select an installation method.

Slackware can be installed from a variety of media and network sources to fit your needs and budget. Every installation method requires you to have at least three floppy diskettes available to get started.

CD-ROM.

Installation from CD-ROM is fast, popular, and convenient. Although someone has to break down and pay for the initial purchase of a CD-ROM, sharing CD's is encouraged. Because Linux and the Slackware distribution are copylefted, you may make as many copies as you like. CD-ROM installation is also a bit better practice in terms of netiquette, since you're not hogging bandwidth for an all-day FTP transfer. Finally, you may be grateful for the extra utilities and documentation that accompany the CD-ROM, especially if you run into installation hassles or need to add components in the future.

Party!

If you're a hobbyist (or want to watch a few dozen Slackware installs before taking on the task at work), see if there is a LUG (Linux User Group) in your area that sponsors install parties. Imagine a roomful of generous and knowledgeable hackers uniting to share CD-ROMs and expertise with other enthusiasts.

FTP.

Once you transfer Slackware from the closest possible FTP mirror, you'll still need to put the Slackware 'disk sets' onto installation media such as a hard drive partition or laboriously copy them onto 50-odd floppy diskettes.

NFS.

In a networked environment, it is possible to install Slackware on a shared file system and allow everyone on the Local net to attach to this shared location and install. If you have the technical know-how or a geeked out system administrator who is Linux-literate, this is a great way to go. The initial distribution of Slackware can be added to the network via CD-ROM, FTP, Loading floppies, tape, or even via a remote NFS share across the Internet! For details on such a remote share, see these URLs:

Floppy.

It's time consuming, but it works--you can create the pile of floppies needed to install Slackware and then feed them into your box one-by-one when prompted. Slackware ``disk sets'' are actually designed and arranged to fit floppy diskettes. If you happen to have a huge stack of recycled, high-density floppy diskettes at your disposal, this can be the most economical way to go.

Hard disk.

This is the way to do it if you've transferred the Slackware distribution across the Internet via FTP--you'll escape the floppy trap by merely creating boot, root, and rescue diskettes. It requires you to have an extra disk or disk partition with extra space to hold the Slackware files during installation (you can erase them afterwards). Installation from the hard drive is also a workaround if you bought the CD but your CD-ROM drive is not supported by any of the Linux kernels that come with the Slackware CD. You can use your present operating system to transfer the Slackware files onto spare hard disk space, then boot into the Slackware installation.

Tape.

Still experimental as of this writing, tape offers a great compromise of speed and economy when installing Slackware--worth considering if a friend with compatible tape drive can dupe a CD or FTP archive for you. Get the latest details from the Tape section of the INSTALL.TXT file that accompanies your Slackware distribution.

2.6.6 Boot disks: always a good thing.

Even if you're gifted with a direct T-3 Internet connection that allows you to suck up a new distribution of Slackware right off the 'Net, you'll be wise to start by building the two Slackware setup disks (boot and root) before proceeding. In the event of an unfortunate accident (power outage, feline friends traversing the keyboard, or even human error), these two little disks may be able to revive your system or at least rescue your personal files.

2.6.7 Slackware setup worksheet.

After the files are all copied, Slackware can go on to do most of the system and network configuration, if you're ready. To help you plan your decisions, this section consists of a worksheet derived from the text-based Slackware setup program. You can use this worksheet to record answers in advance (while your computer is still working!), so you'll be ready with the necessary details-partitions, IP addresses, modem and mouse IRQs, host and domain names, and others that you're required to provide during setup.

1. Keyboard: Slackware setup will want to know if you need to remap your keyboard to something other than a standard USA 101 key layout? yes or no
2. Swap Configuration: Do you have one or more partitions prepared as type 82 (Linux Swap)? yes or no
   Do you want setup to use mkswap on your swap partitions? Most likely ``yes'', unless you have less than 4MB of RAM and have already done this to help setup work better.
   yes or no
3. Prepare Main Linux Partition: setup will list any partitions marked as type 83 (Linux Native) and ask which one to use for the root (/) of the Linux file system. Use a format like /dev/hda3 or whatever the device name is. partition name
   Last chance to back out! When using the install from scratch option, you must install to a blank partition. If you have not already formatted it manually, then you must format it when prompted. Enter ``I'' to install from scratch, or ``a'' to add software to your existing system.
   [i]nstall or [a]dd
   (Re)format the main Linux partition. Would you like to format this partition?
   [y]es, [n]o, or [c]heck sectors, too
   ext2fs defaults to one inode per 4096 bytes of drive space. If you're going to have many small files on your drive, you may need more inodes (one is used for each file entry). You can change the density to one inode per 2048 bytes, or even per 1024 bytes. Enter 2048 or 1024, or just hit Enter to accept the default of 4096.
   4096 (default). 2048, or 1024
4. Prepare Additional Linux Partitions: You can mount some other partitions for /usr or /usr/X11 or whatever (/tmp--you name it). Would you like to use some of the other Linux partitions to mount some of your directories? [y]es or [n]o
   These are your Linux partitions (partition list displayed). These partitions are already in use (partition list displayed). Enter the partition you would like to use, or type q to quit adding new partitions. Use a format such as: /dev/hda3 or whatever the device name is.
   Partition name or [q]uit
   Would you like to format this partition?
   [y]es, [n]o, or [c]heck sections, too
   Now this new partition must be mounted somewhere in your new directory tree. For example, if you want to put it under /usr/X11R6, then respond: /usr/X11R6 Where would you like to mount this new partition?
   Mount point
   Would you like to mount some more additional partitions?
   [y]es or [n]o
5. DOS and OS/2 Partition Setup: The following DOS FAT or OS/2 HPFS partitions were found: (partition list displayed). Would you like to set up some of these partitions to be visible from Linux? [y]es or [n]o
   Please enter the partition you would like to access from Linux, or type q to quit adding new partitions. Use a format such as: /dev/hda3 or whatever the device name is.
   Partition name or [q]uit
   Now this new partition must be mounted somewhere in your directory tree. Please enter the directory under which you would like to put it. for instance, you might want to reply /dosc, /dosd, or something like that. Where would you like to mount this partition?
   Mount point
6. Source Media Selection:
   
   1, 2, 3, 4, or 5
7. Install from a hard drive partition: To install directly from the hard disk, you must have a partition with a directory containing the Slackware distribution such that each disk other than the boot disk is contained in a subdirectory. For example, if the distribution is in /stuff/slack, then you need to have directories named /stuff/slack/a1, /stuff/slack/a2, and so on, each containing the files that would be on that disk. You may install from DOS, HPFS, or Linux partitions. Enter the partition where the Slackware sources can be found, or p to see a partition list. Partition name or [p]artition list
   What directory on this partition can the Slackware sources be found. In the example above, this would be: /stuff/slack. What directory are the Slackware sources in?
   Directory name
   What type of file system does your Slackware source partition contain?
   
   
   1, 2, 3, 4, or 5
8. Install from a pre-mounted directory: Okay, we will install from a directory that is currently mounted. This can be mounted normally or through NFS. You need to specify the name of the directory that contains the subdirectories for each source disk. Which directory would you like to install from? Directory name
9. Install from floppy disks: The base Slackware series (A) can be installed from 1.2M or 1.44M media. Most of the other disks will not fit on 1.2M media, but can be downloaded to your hard drive and installed from there later. Which drive would you like to install from (1/2/3/4)?
   
   
   1, 2, 3, or 4
10. Install via NFS: You're running off the hard drive file system. Is this machine currently running on the network you plan to install from? If so, we won't try to reconfigure your ethernet card. Are you up and running on the network? [y]es or [n]o
    You will need to enter the IP address you wish to assign to this machine. Example: 111.112.113.114. What is your IP address?
    IP address
    Now we need to know your netmask. Typically this will be 255.255.255.0. What is your netmask?
    IP address
    Do you have a gateway (y/n)?
    [y]es or [n]o
    What is your gateway address?
    IP address
    Good! We're all set on the local end, but now we need to know where to find the software packages to install. First, we need the IP address of the machine where the Slackware sources are stored. Since you're already running on the network, you should be able to use the hostname instead of an IP address if you wish. What is the IP address of your NFS server?
    IP address
    There must be a directory on the server with the Slackware sources for each disk in subdirectories beneath it. setup needs to know the name of the directory on your server that contains the disk subdirectories. For example, if your A3 disk is found at /slackware/a3, then you would respond: /slackware. What is the Slackware source directory?
    Directory name
11. Install from CD-ROM: What type of CD-ROM drive do you have?
    
    1, 2, 3, 4, 5, 6, 7 8, 9, 10, 11, 12, or 13
    IDE CD-ROM: Enter the device name that represents your IDE CD-ROM drive. This will probably be one of these (in the order of most to least likely): /dev/hdb /dev/hdc /dev/hdd /dev/hde /dev/hdf /dev/hdg /dev/hdh /dev/hda
    Device name
    SCSI CD-ROM: Which SCSI CD-ROM are you using? If you're not sure, select /dev/scd0.
    1. /dev/scd0
    2. /dev/scd1
    installation method: With the Slackware CD, you can run most of the system from the CD if you're short of drive space or if you just want to test Linux without going through a complete installation. Which type of installation do you want (slakware or slaktest)?
    

    slakware

    Normal installation to hard drive

    slaktest

    Link /usr->/cdrom/live/usr
    
    to run mostly from CD-ROM

    slakware or slaktext
12. Series Selection: Identify which Packages you plan to install. You may specify any combination of disk sets at the prompt which follows. For example, to install the base system, the base X Window System, and the Tcl toolkit, you would enter: a x tcl. Which disk sets do you want to install?
    
    Any combination of a ap d e f k n q t tcl x xap xd xv y and other disk sets offered, separated by spaces
13. Software Installation: Next, software packages are going to be transferred on to your hard drive. If this is your first time installing Linux, you should probably use PROMPT mode. This will follow a defaults file on the first disk of each series you install that will ensure that required packages are installed automatically. You will be prompted for the installation of other packages. If you don't use PROMPT mode, the install program will just go ahead and install everything from the disk sets you have selected. Do you want to use PROMPT mode (y/n)? [y]es or [n]o
    These defaults are user definable--you may set any package to be added or skipped automatically by editing your choices into a file called TAGFILE that will be found on the first disk of each series. There will also be a copy of the original tagfile called TAGFILE.ORG available in case you want to restore the default settings. The tagfile contains all the instructions needed to completely automate your installation. Would you like to use a special tagfile extension? You can specify an extension consisting of a ``.'' followed by any combination of 3 characters other than tgz. For instance, I specify ``.pat'', and then whenever any tagfiles called ``tagfile.pat'' are found during the installation they are used instead of the default ``tagfile'' files. If the install program does not find tagfiles with the custom extension, it will use the default tagfiles. Enter your custom tagfile extension (including the leading ``.''), or just press Enter to continue without a custom extension.
    Tagfile extension Enter
14. Extra Configuration: If you wish, you may now go through the options to reconfigure your hardware, make a bootdisk, and install LILO. If you've installed a new kernel image, you should go through these steps again. Otherwise, it's up to you. [y]es or [n]o
15. Boot Disk Creation: It is recommended that you make a boot disk. Would you like to do this? [y]es or [n]o
    Now put a formatted floppy in your boot drive. This will be made into your Linux boot disk. Use this to boot Linux until LILO has been configured to boot from the hard drive. Any data on the target disk will be destroyed. Insert the disk and press Return, or s if you want to skip this step.
    Enter or [s]kip
16. Modem Setup: A link in /dev will be created from your callout device (cua0, cua1, cua2 , cua3) to /dev/modem. You can change this link later if you put your modem on a different port. Would you like to set up your modem? [y]es or [n]o
    These are the standard serial I/O devices, Which device is your modem attached to (0, 1, 2, 3)?
    
    
    0, 1, 2, or 3
17. Mouse Setup: A link will be created in /dev from your mouse device to /dev/mouse. You can change this link later if you switch to a different type of mouse. Would you like to set up your mouse? [y]es or [n]o
    These types are supported. Which type of mouse do you have (1, 2, 3, 4, 5, 6, 7)?
    
    
    1, 2, 3, 4, 5, 6, or 7
    These are the standard serial I/O devices. Which device is your mouse attached to (0, 1, 2, 3)?
    
    
    0, 1, 2, or 3
18. Network Configuration: Now we will attempt to configure your mail and TCP/IP. This process probably won't work on all possible network configurations, but should give you a good start. You will be able to reconfigure your system at any time by typing netconfig. First, we'll need the name you'd like to give your host. Only the base host name is needed right now (not the domain). Enter the host name. Host name
    Now, we need the domain name. Do not supply a leading ``.'' Enter the domain name.
    Domain name
    If you only plan to use TCP/IP through loopback, then your IP address will be 127.0.0.1, and we can skip a lot of the following questions. Do you plan to only use loopback?
    [y]es or [n]o
    Enter your IP address for the local machine. Example: 111.112.113.114. Enter the IP address for this machine (aaa.bbb.ccc.ddd).
    IP address
    Enter your gateway address, such as 111.112.113.1. If you don't have a gateway, you can edit /etc/rc.d/rc.inet1 later,or you can probably get away with entering your own IP address here. Enter the gateway address (aaa.bbb.ccc.ddd).
    IP address
    Enter your netmask. This will generally look something like this: 255.255.255.0. Enter the netmask (aaa.bbb.ccc.ddd).
    IP address
    Will you be accessing a name server?
    [y]es or [n]o
    Please give the IP address of the name server to use. You can add more Domain Name Servers by editing /etc/resolv.conf. Name server for your domain (aaa.bbb.ccc.ddd)?
    IP address

You may now reboot your computer by pressing Ctrl-Alt-Delete. If you installed LILO, remove the boot disk from your computer before rebooting. Don't forget to create your /etc/fstab if you don't have one (see page )!

2.6.8 Making Slackware happen.

If you've taken the time to plot and plan as recommended in the preceding sections, then the actual installation is a piece of cake. There isn't much writing needed to explain the actual process of loading Slackware on your computer(s). Follow the steps to build boot and root diskettes, then answer a long series of questions asked by the menu driven Slackware installation program. If you've completed the Slackware Installation Worksheet, these questions will be familiar and everything will run smoothly.

2.6.9 Build some boot disks.

Choose your kernel!

When installing Slackware Linux, you must create a boot diskette with a Linux kernel that is specially prepared to recognize your system hardware. For example, to install Slackware from an IDE CD-ROM drive onto a SCSI hard drive, the kernel that you put onto the boot diskette will need to have drivers for your SCSI card and your IDE CD-ROM drive.
The kernels are stored as compressed binary image files that you can access from most any operating system to create a Slackware Boot diskette. On the Slackware FTP site, CD-ROM, or NFS mount, you'll find a subdirectory called bootdsks.144, containing 1.44 MB kernel images for creating boot disks on 1.44MB high density 3.5'' floppy diskettes. If you're working from a 5.25'' floppy diskette drive, look in a directory called bootdsks.12 for kernel images that will fit the smaller diskette format.
The table on page  provides a quick reference of the kernel images available as we went to press. Information and up-to-date boot disk image information is available from this URL:



  
Table 2.6: Slackware IDE boot disk images.

  
Table 2.7: Slackware SCSI/IDE boot disk images.

2.6.10 Boot into action.

Here's the big anticlimax. After all of this planning, preparation, and partitioning, you're in the home stretch. Make sure that the boot floppy is in the diskette drive, and restart your computer. Now is a good time to go get some coffee (or whatever you like to keep you company) and return to the machine ready to play the part of a button-pushing drone, answering yes-no questions for an hour or so.
Log in as root (no password) and type setup or setup.tty.

2.6.11 The Slackware setup program.

Slackware comes with two versions of an excellent setup program. One is a colorful, dialog-based, menu-driven version. An alternative, setup.tty, is a text-only version that you may actually prefer, because detailed diagnostics and error messages stay on the screen and are not covered up by the next dialog box. If you're attempting a Slackware installation on sketchy hardware, I strongly recommend the less colorful setup.tty routine. If you don't know much about UNIX and would feel more comfortable with an attractive, ``clean'' interface to the same process, then by all means go for the beautiful setup.



Transferring Slackware onto your system from here should involve little more than selecting what you want from the menus. By filling out Section 3 of the worksheet in advance, you should be able progress quickly through each menu in order, until you reach the INSTALL option, at which point things may s l o w down: you are advised to select the PROMPT feature and read about each software package, deciding whether or not you'd like it to end up on your Slackware system. The last part of a regular setup is the CONFIGURE section on the setup menu, and the questions you must answer bear a striking resemblance to the second half of the Section 3 worksheet.

2.6.12 Is that all?

Definitely not! At this point, you either have some annoying obstacle that is preventing the setup from completing, or more likely, you're looking at the root prompt


and wondering ``What Next?''
Well, if you're plagued by problems, you'll want to proceed directly to the next section on troubleshooting. If things appear to be in working order, you've still got some details to attend to. It's sort of like purchasing a new automobile--after you select and pay for a car, there are still some things that you need before you can drive it with confidence--insurance, a steering wheel club, and perhaps some luxuries that make the driving experience closer to Fahrvergnügen than FAQ!

2.6.13 Troubleshooting difficult deliveries.

Not every Slackware installation is born on cue to expecting system administrators. I've pulled a few all-nighters, sitting down after work one evening to upgrade a Slackware box and still there struggling to get the damn thing back online at dawn, before people start bitching about their missing mail and news. This section will look at a few common Slackware setup problems, solutions, and where to look for additional assistance.

Slackware installation FAQs.

Patrick Volkerding, the father of Slackware, has dealt with many questions of new users by listening, answering, and anticipating repeat queries. To catch the new Slackware users before they ask the same question for the 5,000th time, Patrick has kindly created documentation and included it with the Slackware distribution. Three files that you may find very helpful in answering your initial questions are FAQ.TXT, INSTALL.TXT, and BOOTING.TXT.

Web Support For Slackware.

The easiest way to access finding Linux documents in general is the Linux Documentation Project Home Page. See page  for a description of the LDP Home Page.
At this time, the Slackware-specific help you'll find on the Internet tends to be highly customized--like how to NFS-mount the distribution on computers within a certain university or how to wire your dorm room into a particular residential WAN using Slackware.

Usenet Groups For Slackware.

The comp.os.linux.* hierarchy of Usenet is a treasure trove of Linux information, not necessarily Slackware-specific. At present, 11 separate Linux forums handle a high volume of discussion in this hierarchy, which is described on page .

Mailing lists for Slackware.

At this time, there are no electronic mail discussions devoted to Slackware per se. You can participate in some excellent Linux-related talk via e-mail, try http://www.linux.org, and ask in the Usenet newsgroups for a few good subscription lists.
There is a general Linux mailing list server, majordomo@vger.rutgers.edu. See page  for a description of how to subscribe to mailing lists via this server.

You get what you pay for (commercial support).

Commercial support for Linux is available from some of the CD-ROM vendors and a long list of Linux Consultants, who can be contacted through the Linux Commercial and Consultants HOWTO documents:

2.6.14 Basking in the afterglow.

Don't rest on your laurels quite yet, especially if your Slackware machine is a shared computer or lives in a networked environment. Grooming a computer for community and network use is a bit more demanding than just running the setup program and then forgetting about it. We'll leave you with a few pointers to securing and sharing your new Slackware system.

2.6.15 Consider reinstalling!

I know you just sat through what may have been a long and perplexing installation session. But before you move into the house you just built, consider tearing it down and starting over again. Friedrich Nietzsche had a quote:

A man learns what he needs to know about building his house only after he's finished.

If, in the process of installing the system, you had some thoughts about how you might do it differently, now is the time. If your Slackware Linux box will be a multi-user machine or a network server, there may never be such a convenient opportunity to re-install or reconfigure the system in radical ways.

2.6.16 Secure the system.

Get off the LAN at once.

Out of the box, Slackware is an insecure system. Although Patrick Volkerding does his best to create a secure distribution, a few inevitable holes become known, and patches or workarounds are made available in the system administration (and cracker) communities. If you installed Slackware from a network source like a NFS-mounted drive, you should temporarily disconnect your box from the LAN after a successful installation, while you plug a few holes.

Give root a password.

By default, a new Slackware box will not require a password for the root user. When you're comfortable that your new Slackware system is stable (after a few hours, not days or weeks), add a password to protect the root account. Login as root and type:

Give yourself an account.

On large, shared systems, the super-user root account is not used as a working login account by any individual. If you're interested in system administration or are running a networked machine, this is a good precedent to follow. Use the /sbin/adduser program and make yourself a login account, rather than working out of the root login. I always smile when I see students and hobbyists posting proudly to the Usenet as root@mymachine.mydomain. Be humble and safe: create another login account for your daily work and use su (rather than login) to enter the root account sparingly.
Read Chapter 4 for a discussion of what you should do with the root account (or shouldn't).

Deny root logins.

Not only is it uncommon to work as the root user, it is not considered secure to login as root across the network. Administrative users usually connect to a UNIX box as their regular, user-name login, then su to root as needed. To prevent crackers, hackers, and ignorant users from logging in directly as root, edit the file /etc/securetty and comment out (prepend a pound (#) sign before) all but the local terminals:



After this fix, users who attempt to login in as root across the network will be denied:

Apply the simple fixes.

Slackware installs itself with some very real security problems. Rather than master UNIX security and sleuth out these vulnerabilities yourself, you can jump start the hole-patching process by visiting a Web resource maintained for just this purpose, called Slackware SimpleFixes:

Check for patches on ftp.cdrom.com

As an actively maintained Linux distribution, Slackware updates and patches are available from:

Stay current.

You might like to subscribe to one or more electronic mailing lists that alert users to issues in Linux administration, such as:

2.6.16.1 Back up.

Like how things are running? Save it for a rainy day by backing up. Amanda (the Advanced Maryland Automatic Network Disk Archiver) is one of several backup options for Linux installations. You can learn more about Amanda from:

2.7 S.u.S.E.

 
   
This section on S.u.S.E. Linux was written by Larry Ayers.
The SuSE distribution began a few years ago as an adaptation of Slackware. Patrick Volkerding of Slackware helped the SuSE developers at first, but before too long, the distribution began to assume an identity of its own. Several new features intended to aid the first-time user increase the probability an installation won't need to be immediately redone. Given the cross-pollination endemic in the free software world, I wouldn't be surprised to learn some of these features have shown up in newer Slackware releases.

2.7.1 Beginning the installation.

When booting your machine from the single installation disk, you are really booting a miniature Linux system designed for this purpose. A colored screen appears, ready to ask a series of questions which with any luck will guide you through the process. YAST (Yet Another Set-up Tool) shows its Slackware ancestry inasmuch as it uses the dialog program. This tool enables shell scripts to present dialog boxes, radio buttons, and check lists which allow a user to make choices and direct the course of an installation.
While no distribution can guarantee a painless installation, the developers at S.u.S.E. GmbH have managed to anticipate several problems that new Linux users are liable to have. One of the more frustrating problems is finding that your CD-ROM drive isn't recognized. Copying the packages needed to get started to a hard drive and installing them from there is a solution, but it's awkward and time-consuming. Rather than provide a selection of several disk images, one of which probably has the CD-ROM drive support you need, the single S.u.S.E. boot disk contains a small, basic kernel with all drivers available-if needed-in the form of modules. The kernel daemon is a background process which ensures the relevant module will be loaded if a modular function is needed. This helps to eliminate one stumbling block. Another common trap is underestimating the disk space which you need. This forces the installation to abort itself due to lack of room. When this happens, the crucial final steps (like LILO installation) haven't yet been reached, and starting over is usually necessary. Script-based installations are necessarily sequential in nature; you may know that skipping one step won't hurt anything, but it's hard to anticipate every eventuality in a shell script, and if things go awry the script usually aborts.
During the S.u.S.E. installation, a running tally of partition space remaining is displayed on the YAST screen; while selecting packages, you can try various combinations while keeping in mind how much free disk space you would prefer to remain free. Partitioning and formatting disks, as well as creation and activation of a swap partition, are processes that aren't much different than in other distributions. They all use the same underlying tools; the procedure has become more or less standardized.

Dependencies.

The use of dependencies, which consist of information included in a software package concerning what other packages are necessary for it to run, has spread rapidly among Linux distributions. Unfortunately no universal format for dependencies has arisen. Each distribution uses a different format. Redhat's RPM format, used in several distributions, is powerful and effective, but it has a few drawbacks. It works best on an all-RPM system, as the dependency checking done by the RPM program only knows about RPM packages. S.u.S.E. 5.1 uses srpm-format. The dependencies are only checked if a package is installed from within the YAST program, allowing the option (for a skilled user) of unarchiving a package in another location, then checking out the files and configuration before final installation. Dependencies are most useful during the initial setup and while becoming familiar with a new installation. Once you've used the system for a while, you'll have an idea of what libraries and programs are available. Most software packages for Linux also contain information as to what needs to be present on a system in order for the package to function. It is wise to read through the entire rc.config file before running SuSEconfig and committing any changes you may have made. Some of the default actions the script will take you may prefer to handle yourself, but they are easily disabled by editing the file.
Users familiar with the Slackware layout of initialization files will need to make some adjustments; the files usually found in /etc/rc.d are instead in /sbin/init.d.

S.u.S.E Post-installation.

YAST is also intended to be used after installation for routine system maintenance. The multiplicity of resource files necessary for Linux to boot and run can be bewildering to beginners. YAST offers a menu-driven interface to these files, including the sendmail configuration file, the cron (scheduling) files, initialization scripts, and various networking files. The changes made within the YAST session are written to a single file in the /etc directory, rc.config, which can also be edited directly. These changes are then written to the various ``real'' configuration files by a script called SuSEconfig. This script is automatically run by YAST at the end of a YAST session; if /etc/rc.config is edited directly, SuSEconfig must be started manually. This sounds like a complicated procedure, but it's much easier than tracking down the individual files, learning the correct syntax needed to edit them, and making them do what you want.
After you have S.u.S.E. Linux up and running, it's a good idea to install the kernel source (available on the CD-ROM, it's an optional package which can be installed during initial set-up). S.u.S.E. installs a generic kernel, and you probably need only a few of of the accompanying modules. This is an excellent opportunity to familiarize yourself with the mechanics of source code compilation, and you'll end up with a smaller customized kernel with only the capabilities you need. The gcc compiler and accompanying tools must be installed in order to compile a kernel; these tools are a near-necessity on a Linux system even if you're not a programmer. The YAST dependency checking will help insure that all of the required compilation tools are installed.
Kernel compilation can seem daunting to a beginner, but it is a fairly intuitive process. Three interfaces are available for the initial configuration step. The first (and oldest) is a console-mode script invoked via the command make config. This script asks a series of questions and uses the results to write a file which guides the compiler in its work. You need to know some basic facts about your hardware such as what type of hard disk and CD-ROM drive you have. If you want sound support you'll need to know the IRQ your card uses, as well as a few other parameters that can be gathered from the card's manual or the output of the MS-DOS msd utility.
The other two interfaces are menuconfig and xconfig. The first uses a modified version of the dialog program mentioned above, which runs on a virtual console or a xterm and resembles the YAST setup tool. xconfig is a Tk-based version, designed to run in a X window. All three accomplish the same task. The latter two let you make choices without typing much. The kernel sources are well-documented. The README file in the top-level directory contains enough information to nearly guarantee a successful build.

2.7.3 Getting X up and running.

Successfully configuring the X Window System (specifically XFree86, which is included with S.u.S.E. and most other distributions) can be a stumbling block. There is such a multiplicity of monitors and video cards that each installation of X must be individually configured. The difficulty has been eased somewhat with the release of XFree86 3.2, which is included with the most recent S.u.S.E. release. A dialog based configuration tool can now be used in place of the previous xf86config. Both are based on shell scripts similar to the one that is used to configure the Linux kernel. Nonetheless, you will still need to know your monitor's horizontal and vertical refresh rates as well as the chip set installed on your video card. It helps to initially set your sites low, i.e., get X functioning at a low resolution first before attempting to make full use of your video card's capabilities.
The S.u.S.E. developers have taken some pains in configuring the various window managers, for example, fvwm95. The first time you start X, many of the applications you elected for installation will be available from the mouse activated root window menu. Another entry on the menu allows you to change the window background.
Many well-designed icons are supplied with the S.u.S.E. distribution. This gives new users something of a reprieve. After getting Linux and X running finally, there is enough to do just learning the system without feeling compelled to customize the environment, in order to make it tolerable to view!

2.7.4 Later upgrades.

The minute you finish installing even the most up-to-date distribution, it begins to incrementally become outdated. This is a slow process, but eventually you will feel the need to upgrade some part of the system. with S.u.S.E. 5.1, YaST can now update via ftp.

Post-installation procedures.

At this point it's a good idea to explain how to reboot and shutdown the system as you're using it. You should never reboot or shutdown your Linux system by pressing the reset switch. You shouldn't simply switch off the power, either. As with most UNIX systems, Linux caches disk writes in memory. Therefore, if you suddenly reboot the system without shutting down ``cleanly'', you can corrupt the data on your drives, causing untold damage.
The easiest way to shut down the system is with the shutdown command. As an example, to shutdown and reboot the system immediately, use the following command as root:

  This cleanly reboots your system. The manual page for shutdown describes the other command-line arguments that are available. Use the command man shutdown to see the manual page for shutdown.
Note, however, that many Linux distributions do not provide the shutdown command on the installation media. This means that the first time you reboot your system after installation, you may need to use the Ctrl-Alt-Del combination.
After you have a chance to explore and use the system, there are several configuration chores that you should undertake. The first is to create a user account for yourself (and, optionally, any other users that might have access to the system). Creating user accounts is described in Chapter 4. Usually, all that you have to do is login as root, and run the adduser (sometimes useradd) program. This leads you through several prompts to create new user accounts.
If you create more than one filesystem for Linux, or if you're using a swap partition, you may need to edit the file /etc/fstab in order for those filesystems to be available automatically after rebooting. If you're using a separate filesystem for /usr, and none of the files that should be in /usr appear to be present, you may simply need to mount that filesystem. See page  for a description of the /etc/fstab file.

2.9 Running into trouble.

 
Almost everyone runs into some kind of snag or hangup when attempting to install Linux the first time. Most of the time, the problem is caused by a simple misunderstanding. Sometimes, however, it can be something more serious, like an oversight by one of the developers, or a bug.
If your installation appears to be successful, but you received unexpected error messages, these are described here as well.

2.9.1 Problems with booting the installation media

   
When attempting to boot the installation media for the first time, you may encounter a number of problems. These are listed below. Note that the following problems are not related to booting your newly installed Linux system. See page  for information on these kinds of pitfalls.

• Floppy or media error when attempting to boot. The most popular cause for this kind of problem is a corrupt boot floppy. Either the floppy is physically damaged, in which case you should re-create the disk with a brand new floppy, or the data on the floppy is bad, in which case you should verify that you downloaded and transferred the data to the floppy correctly. In many cases, simply re-creating the boot floppy will solve your problems. Retrace your steps and try again.
  If you received your boot floppy from a mail order vendor or some other distributor, instead of downloading and creating it yourself, contact the distributor and ask for a new boot floppy--but only after verifying that this is indeed the problem.
• System hangs during boot or after booting. After the installation media boots, you will see a number of messages from the kernel itself, indicating which devices were detected and configured. After this, you will usually be presented with a login prompt, allowing you to proceed with installation (some distributions instead drop you right into an installation program of some kind). The system may appear to hang during several of these steps. Be patient: loading software from floppy is comparatively slow. In many cases, the system has not hung at all but is merely taking a long time. Verify that there is no drive or system activity for at least several minutes before assuming that the system is hung.
  
   
  1. After booting from the LILO prompt, the system must load the kernel image from floppy. This may take several seconds; you will know that things are going well if the floppy drive light is still on.
  2. While the kernel boots, SCSI devices must be probed for. If you do not have any SCSI devices installed, the system will hang for up to 15 seconds while the SCSI probe continues; this usually occurs after the line
     
     appears on your screen.
  3. After the kernel is finished booting, control is transferred to the system boot-up files on the floppy. Finally, you will be presented with a login prompt, or be dropped into an installation program. If you are presented with a login prompt such as
     
     you should then login (usually as root or install--this varies with each distribution). After entering the user name, the system may pause for 20 seconds or more while the installation program or shell is being loaded from floppy. Again, the floppy drive light should be on. Don't assume that the system is hung.
  Any of the above items may be the source of your problem. However, it is possible that the system actually may hang while booting, which can be due to several causes. First of all, you may not have enough available RAM to boot the installation media. (See the following item for information on disabling the ramdisk to free up memory.)
  The cause of many system hangs is hardware incompatibility. The last chapter presented an overview of supported hardware under Linux. Even if your hardware is supported, you may run into problems with incompatible hardware configurations which are causing the system to hang. See page , below, for a discussion of hardware incompatibilities.
• System reports out-of-memory errors while attempting to boot or install the software. This item deals with the amount of RAM that you have available. On systems with 4 megabytes of RAM or less, you may run into trouble booting the installation media or installing the software itself. This is because many distributions use a RAM disk, a file system loaded directly into RAM, for operations while using the installation media. The entire image of the installation boot floppy, for example, may be loaded into a RAM disk, which may require more than a megabyte of RAM.
  You may not see an ``out of memory'' error when attempting to boot or install the software; instead, the system may unexpectedly hang, or fail to boot. If your system hangs, and none of the explanations in the previous section seem to be the cause, try disabling the ramdisk. See your distribution's documentation for details.
  Keep in mind that Linux itself requires at least 2 megabytes of RAM to run at all; most modern distributions of Linux require 4 megabytes or more.
• The system reports an error like ``permission denied'' or ``file not found'' while booting. This is an indication that your installation bootup media is corrupt. If you try to boot from the installation media (and you're sure that you're doing everything correctly), you should not see any errors like this. Contact the distributor of your Linux software and find out about the problem, and perhaps obtain another copy of the boot media if necessary. If you downloaded the boot disk yourself, try re-creating it and see if this solves the problem.
• The system reports the error ``VFS: Unable to mount root'' when booting. This error message means that the root file system (found on the boot media itself), could not be found. This means that either your boot media is corrupt in some way, or that you are not booting the system correctly.
  For example, many CD-ROM distributions require that you have the CD-ROM in the drive when booting. Be sure that the CD-ROM drive is on and check for any activity. It's also possible that the system is not locating your CD-ROM drive at boot time; see page  for more information.
  

   

2.9.2 Hardware problems.

 
   
The most common form of problem when attempting to install or use Linux is an incompatibility with hardware. Even if all of your hardware is supported by Linux, a misconfiguration or hardware conflict can sometimes cause strange results--your devices may not be detected at boot time, or the system may hang.
It is important to isolate these hardware problems if you suspect that they may be the source of your trouble.

Isolating hardware problems

 
If you experience a problem that you believe to be hardware-related, the first thing that you should to do is attempt to isolate the problem. This means eliminating all possible variables and (usually) taking the system apart, piece-by-piece, until the offending piece of hardware is isolated.
This is not as frightening as it may sound. Basically, you should remove all nonessential hardware from your system, and then determine which device is causing the trouble--possibly by reinserting each device, one at a time. This means that you should remove all hardware other than the floppy and video controllers, and of course the keyboard. Even innocent-looking devices such as mouse controllers can wreak unknown havoc on your peace of mind unless you consider them nonessential.
For example, let's say that the system hangs during the Ethernet board detection sequence at boot time. You might hypothesize that there is a conflict or problem with the Ethernet board in your machine. The quick and easy way to find out is to pull the Ethernet board, and try booting again. If everything goes well, then you know that either (a) the Ethernet board is not supported by Linux (see page ), or (b) there is an address or IRQ conflict with the board.
  ``Address or IRQ conflict?'' What on earth does that mean? All devices in your machine use an IRQ, or interrupt request line, to tell the system that they need something done on their behalf. You can think of the IRQ as a cord that the device tugs when it needs the system to take care of some pending request. If more than one device is tugging on the same cord, the kernel won't be able to detemine which device it needs to service. Instant mayhem.
Therefore, be sure that all of your installed devices use unique IRQ lines. In general, the IRQ for a device can be set by jumpers on the card; see the documentation for the particular device for details. Some devices do not require the use of an IRQ at all, but it is suggested that you configure them to use one if possible. (The Seagate ST01 and ST02 SCSI controllers are good examples).
In some cases, the kernel provided on your installation media is configured to use certain IRQs for certain devices. For example, on some distributions of Linux, the kernel is preconfigured to use IRQ 5 for the TMC-950 SCSI controller, the Mitsumi CD-ROM controller, and the bus mouse driver. If you want to use two or more of these devices, you'll need to first install Linux with only one of these devices enabled, then recompile the kernel in order to change the default IRQ for one of them. (See Chapter 4 for information on recompiling the kernel.)
Another area where hardware conflicts can arise is with DMA (direct memory access) channels, I/O addresses, and shared memory addresses. All of these terms describe mechanisms through which the system interfaces with hardware devices. Some Ethernet boards, for example, use a shared memory address as well as an IRQ to interface with the system. If any of these are in conflict with other devices, then the system may behave unexpectedly. You should be able to change the DMA channel, I/O or shared memory addresses for your various devices with jumper settings. (Unfortunately, some devices don't allow you to change these settings.)
The documentation for various hardware devices should specify the IRQ, DMA channel, I/O address, or shared memory address that the devices use, and how to configure them. Again, the simple way to get around these problems is to temporarily disable the conflicting devices until you have time to determine the cause of the problem.
The table below is a list of IRQ and DMA channels used by various ``standard'' devices on most systems. Almost all systems have some of these devices, so you should avoid setting the IRQ or DMA of other devices in conflict with these values.
  
Table 2.8: Common device settings.

 

Problems recognizing hard drive or controller.

 
When Linux boots, you should see a series of messages on your screen such as:

Here, the kernel is detecting the various hardware devices present on your system. At some point, you should see the line

followed by a list of recognized partitions, for example:

If, for some reason, your drives or partitions are not recognized, then you will not be able to access them in any way.
There are several things that can cause this to happen:

• Hard drive or controller not supported. If you have a hard drive controller (IDE, SCSI, or otherwise) that is not supported by Linux, the kernel will not recognize your partitions at boot time.  
• Drive or controller improperly configured. Even if your controller is supported by Linux, it may not be configured correctly. (This is particularly a problem for SCSI controllers. Most non-SCSI controllers should work fine without any additional configuration). Refer to the documentation for your hard drive and/or controller. In particular, many hard drives need to have a jumper set to be used as a slave drive (the second device on either the primary or secondary IDE bus). The acid test of this kind of condition is to boot MS-DOS or some other operating system that is known to work with your drive and controller. If you can access the drive and controller from another operating system, then it is not a problem with your hardware configuration.
  See page , above, for information on resolving possible device conflicts, and page , below, for information on configuring SCSI devices.
• Controller properly configured, but not detected. Some BIOS-less SCSI controllers require the user to specify information about the controller at boot time. A description of how to force hardware detection for these controllers begins on page .
• Hard drive geometry not recognized. Some systems, like the IBM PS/ValuePoint, do not store hard drive geometry information in the CMOS memory, where Linux expects to find it. Also, certain SCSI controllers need to be told where to find drive geometry in order for Linux to recognize the layout of your drive. Most distributions provide a bootup option to specify the drive geometry. In general, when booting the installation media, you can specify the drive geometry at the LILO boot prompt with a command such as:
  
  where cylinders, heads, and sectors correspond to the number of cylinders, heads, and sectors per track for your hard drive.
  After installing Linux, you will be able to install LILO, allowing you to boot from the hard drive. At that time, you can specify the drive geometry to LILO, making it unnecessary to enter the drive geometry each time you boot. See Chapter 4 for more information about LILO.
  

 

Problems with SCSI controllers and devices.

    Presented here are some of the most common problems with SCSI controllers and devices like CD-ROMs, hard drives, and tape drives. If you have problems getting Linux to recognize your drive or controller, read on.
The Linux SCSI HOWTO (see Appendix A) contains much useful information on SCSI devices in addition to that listed here. SCSI can be particularly tricky to configure at times.

• A SCSI device is detected at all possible IDs. This is caused by strapping the device to the same address as the controller. You need to change the jumper settings so that the drive uses a different address than the controller.
• Linux reports sense errors, even if the devices are known to be error-free. This can be caused by bad cables or bad termination. If your SCSI bus is not terminated at both ends, you may have errors accessing SCSI devices. When in doubt, always check your cables.
• SCSI devices report timeout errors. This is usually caused by a conflict with IRQ, DMA, or device addresses. Also check that interrupts are enabled correctly on your controller.
• SCSI controllers that use BIOS are not detected. Detection of controllers that use BIOS will fail if the BIOS is disabled, or if your controller's signature is not recognized by the kernel. See the Linux SCSI HOWTO, available from the sources in Appendix A, for more information about this.
• Controllers using memory mapped I/O do not work. This is caused when the memory-mapped I/O ports are incorrectly cached. Either mark the board's address space as uncacheable in the XCMOS settings, or disable cache altogether.
• When partitioning, you get a warning that ``cylinders > 1024'', or you are unable to boot from a partition using cylinders numbered above 1023. BIOS limits the number of cylinders to 1024, and any partition using cylinders numbered above this won't be accessible from the BIOS. As far as Linux is concerned, this affects only booting; once the system has booted you should be able to access the partition. Your options are to either boot Linux from a boot floppy, or boot from a partition using cylinders numbered below 1024.
• CD-ROM drive or other removeable media devices are not recognized at boot time. Try booting with a CD-ROM (or disk) in the drive. This is necessary for some devices.

If your SCSI controller is not recognized, you may need to force hardware detection at boot time. This is particularly important for BIOS-less SCSI controllers. Most distributions allow you to specify the controller IRQ and shared memory address when booting the installation media. For example, if you are using a TMC-8xx controller, you may be able to enter

at the LILO boot prompt, where interrupt is the IRQ of controller, and memory-address is the shared memory address. Whether or not this is possible depends on the distribution of Linux; consult your documentation for details.
   
   

2.9.3 Problems installing the software.

Actually installing the Linux software should be quite trouble-free, if you're lucky. The only problems that you might experience would be related to corrupt installation media or lack of space on your Linux filesystems. Here is a list of these common problems.

 
• System reports ``Read error'', ``file not found'', or other errors while attempting to install the software. This indicates a problem with the installation media. If you install from floppy, keep in mind that floppies are quite succeptible to media errors of this type. Be sure to use brand-new, newly formatted floppies. If you have an MS-DOS partition on your drive, many Linux distributions allow you to install the software from the hard drive. This may be faster and more reliable than using floppies. If you use a CD-ROM, be sure to check the disc for scratches, dust, or other problems that may cause media errors.
  The cause of the problem may be that the media is in the incorrect format. Many Linux distributions require that the floppies be formatted in high-density MS-DOS format. (The boot floppy is the exception; it is not in MS-DOS format in most cases.) If all else fails, either obtain a new set of floppies, or recreate the floppies (using new diskettes) if you downloaded the software yourself.
• System reports errors such as ``tar: read error'' or ``gzip: not in gzip format''. This problem is usually caused by corrupt files on the installation media. In other words, your floppy may be error-free, but the data on the floppy is in some way corrupted. If you downloaded the Linux software using text mode, rather than binary mode, then your files will be corrupt, and unreadable by the installation software.
• System reports errors like ``device full'' while installing. This is a clear-cut sign that you have run out of space when installing the software. Not all Linux distributions can pick up the mess cleanly; you shouldn't be able to abort the installation and expect the system to work. The solution is usually to re-create your file systems (with mke2fs) which deletes the partially installed software. You can attempt to re-install the software, this time selecting a smaller amount of software to install. In other cases, you may need to start completely from scratch, and rethink your partition and filesystem sizes.
• System reports errors such as ``read_intr: 0x10'' while accessing the hard drive. This usually indicates bad blocks on your drive. However, if you receive these errors while using mkswap or mke2fs, the system may be having trouble accessing your drive. This can either be a hardware problem (see page ), or it might be a case of poorly specified geometry. If you used the
  
  option at boot time to force detection of your drive geometry, and incorrectly specified the geometry, you could be prone to this problem. This can also happen if your drive geometry is incorrectly specified in the system CMOS.
• System reports errors like ``file not found'' or ``permission denied''. This problem can occur if not all of the necessary files are present on the installation media (see the next paragraph) or if there is a permissions problem with the installation software. For example, some distributions of Linux have been known to have bugs in the installation software itself. These are usually fixed very rapidly, and are quite infrequent. If you suspect that the distribution software contains bugs, and you're sure that you have not done anything wrong, contact the maintainer of the distribution to report the bug.

If you have other strange errors when installing Linux (especially if you downloaded the software yourself), be sure that you actually obtained all of the necessary files when downloading. For example, some people use the FTP command

when downloading the Linux software via FTP. This will download only those files that contain a ``.'' in their filenames; if there are any files without the ``.'', you will miss them. The correct command to use in this case is


The best advice is to retrace your steps when something goes wrong. You may think that you have done everything correctly, when in fact you forgot a small but important step somewhere along the way. In many cases, re-downloading and re-installing the software can solve the problem. Don't beat your head against the wall any longer than you have to!
Also, if Linux unexpectedly hangs during installation, there may be a hardware problem of some kind. See page  for hints.

2.9.4 Problems after installing Linux.

 
 
You've spent an entire afternoon installing Linux. In order to make space for it, you wiped your MS-DOS and OS/2 partitions, and tearfully deleted your copies of SimCity and Wing Commander. You reboot the system, and nothing happens. Or, even worse, something happens, but it's not what should happen. What do you do?
On page , we cover some of the most common problems that can occur when booting the Linux installation media--many of those problems may apply here. In addition, you may be victim to one of the following maladies.

Problems booting Linux from floppy.

    If you use a floppy to boot Linux, you may need to specify the location of your Linux root partition at boot time. This is especially true if you are using the original installation floppy itself, and not a custom boot floppy that was created during installation.
While booting the floppy, hold down Shift or Ctrl. This should present you with a boot menu. Press Tab to see a list of available options. For example, many distributions allow you to type

at the boot menu, where partition is the name of the Linux root partition, like /dev/hda2. Consult the documentation for your distribution for details.

Problems booting Linux from the hard drive.

    If you opted to install LILO instead of creating a boot floppy, you should be able to boot Linux from the hard drive. However, the automated LILO installation procedure used by many distributions is not always perfect. It may make incorrect assumptions about your partition layout, and you will need to re-install LILO to get everything correct. LILO installation is covered in Chapter 4.

• System reports ``Drive not bootable--Please insert system disk.'' The hard drive's master boot record is corrupt in some way. In most cases, it's harmless, and everything else on your drive is still intact. There are several ways around this:
  1. While partitioning your drive using fdisk, you may have deleted the partition that was marked as ``active''. MS-DOS and other operating systems attempt to boot the ``active'' partition at boot time (Linux pays no attention to whether the partition is ``active'' or not). You may be able to boot MS-DOS from floppy and run FDISK.EXE to set the active flag on your MS-DOS paritition, and all will be well. Another command to try (with MS-DOS 5.0 and higher) is
     
     This command attempts to rebuild the hard drive master boot record for booting MS-DOS, by overwriting LILO. If you no longer have MS-DOS on your hard drive, you need to boot Linux from floppy and attempt to install LILO later.
  2. If you created a MS-DOS partition using Linux's version of fdisk, or vice versa, you may get this error. You should create MS-DOS partitions only using MS-DOS's version, FDISK.EXE. (This applies to operating systems other than MS-DOS.) The best solution is either to start from scratch and repartition the drive correctly, or to merely delete and re-create the offending partitions with the correct version of fdisk.    
  3. The LILO installation procedure may have failed. In this case, you should either boot from your Linux boot floppy (if you have one), or from the original installation media. Either of these should provide options for specifying the Linux root partition to use when booting. Hold down Shift or Ctrl at boot time, and press Tab from the boot menu for a list of options.
• When booting the system from the hard drive, MS-DOS (or another operating system) starts instead of Linux. First of all, be sure that you actually installed LILO when installing the Linux software. If not, then the system still boots MS-DOS (or whatever other operating system you may have) when you attempt to boot from the hard drive. In order to boot Linux from the hard drive, you need to install LILO (see Chapter 4). On the other hand, if you did install LILO, and another operating system boots instead of Linux, then you have LILO configured to boot that other operating system by default. While the system is booting, hold down Shift or Ctrl, and press Tab at the boot prompt. This should present you with a list of possible operating systems to boot; select the appropriate option (usually ``linux'') to boot Linux.
  If you wish to select Linux as the default operating system, you must re-install LILO. See Chapter 4.
  It may also be possible that you attempted to install LILO, but the installation procedure failed in some way. See the previous item.
  

Problems logging in

    After booting Linux, you should be presented with a login prompt, like

At this point, either the distribution's documentation or the system itself will tell you what to do. For many distributions, you simply log in as root, with no password. Other possible user names to try are guest or test.
Most newly installed Linux systems should not require a password for the initial log in. However, if you are asked to enter a password, there may be a problem. First, try using a password equivalent to the username; that is, if you are logging in as root, use ``root'' as the password.
If you simply can't log in, there may be a problem. First, consult your distribution's documentation; the user name and password to use may be buried in there somewhere. The user name and password may have been given to you during the installation procedure, or they may be printed on the login banner.
One cause may be a problem with installing the Linux login program and initialization files. You may need to reinstall (at least parts of) the Linux software, or boot your installation media and attempt to fix the problem by hand--see Chapter 4 for hints.

Problems using the system.

If logging in is successful, you should be presented with a shell prompt (like ``#'' or ``$'') and can happily roam around your system. However, there are some initial problems with using the system that sometimes creep up.
The most common initial configuration problem is incorrect file or directory permissions. This can cause the error message

to be printed after logging in (in fact, any time you see the message ``permission denied'' you can be fairly certain that it is a problem with file permissions).
    In many cases, it's a simple matter of using chmod to fix the permissions of the appropriate files or directories. For example, some distributions of Linux once used the (incorrect) file mode 0644 for the root directory (/). The fix was to issue the command

as root. However, in order to issue this command, you needed to boot from the installation media and mount your Linux root filesystem by hand--a hairy task for most newcomers.
As you use the system, you may run into places where file and directory permissions are incorrect, or software does not work as configured. Welcome to the world of Linux! While most distributions are quite trouble-free, very few of them are perfect. We don't want to cover all of those problems here. Instead, throughout the book we help you to solve many of these configuration problems by teaching you how to find them and fix them yourself. In Chapter 1 we discussed this philosophy in some detail. In Chapter 4, we give hints for fixing many of these common configuration problems.

3.1 Introduction.

If you're new to UNIX and Linux, you may be a bit intimidated by the size and apparent complexity of the system before you. This chapter does not go into great detail or cover advanced topics. Instead, we want you to hit the ground running.
We assume very little here about your background, except perhaps that you have some familiarity with personal computer systems, and MS-DOS. However, even if you're not an MS-DOS user, you should be able to understand everything here. At first glance, Linux looks a lot like MS-DOS--after all, parts of MS-DOS were modeled on the CP/M operating system, which in turn was modeled on UNIX. However, only the most superficial features of Linux resemble MS-DOS. Even if you're completely new to the PC world, this tutorial should help.
And, before we begin: Don't be afraid to experiment. The system won't bite you. You can't destroy anything by working on the system. Linux has built-in security features to prevent ``normal'' users from damaging files that are essential to the system. Even so, the worst thing that can happen is that you may delete some or all of your files and you'll have to re-install the system. So, at this point, you have nothing to lose.

3.2 Basic Linux concepts.

            Linux is a multitasking, multiuser operating system, which means that many people can run many different applications on one computer at the same time. This differs from MS-DOS, where only one person can use the system at any one time. Under Linux, to identify yourself to the system, you must log in, which entails entering your login name (the name the system uses to identify you), and entering your password, which is your personal key for logging in to your account. Because only you know your password, no one else can log in to the system under your user name.
On traditional UNIX systems, the system administrator assigns you a user name and an initial password when you are given an account on the system. However, because in Linux tt you are the system administrator, you must set up your own account before you can log in. For the following discussions, we'll use the imaginary user name, ``larry.''
  In addition, each system has a host name assigned to it. It is this host name that gives your machine a name, gives it character and charm. The host name is used to identify individual machines on a network, but even if your machine isn't networked, it should have a host name. For our examples below, the system's host name is ``mousehouse''.

3.2.1 Creating an account.

      Before you can use a newly installed Linux system, you must set up a user account for yourself. It's usually not a good idea to use the root account for normal use; you should reserve the root account for running privileged commands and for maintaining the system as discussed below.
In order to create an account for yourself, log in as root and use the useradd or adduser command. See Section 4.6 for information on this procedure.

3.2.2 Logging in.

    At login time, you'll see a prompt resembling the following:



Enter your user name and press the Enter key. Our hero, larry, would type:



Next, enter your password. The characters you enter won't be echoed to the screen, so type carefully. If you mistype your password, you'll see the message

and you'll have to try again.
Once you have correctly entered the user name and password, you are officially logged in to the system, and are free to roam.

3.2.3 Virtual consoles.

    The system's console is the monitor and keyboard connected directly to the system. (Because Linux is a multiuser operating system, you may have other terminals connected to serial ports on your system, but these would not be the console.) Linux, like some other versions of UNIX, provides access to virtual consoles (or VCs), that let you have more than one login session on the console at one time.
To demonstrate this, log in to your system. Next, press Alt-F2. You should see the login: prompt again. You're looking at the second virtual console. To switch back to the first VC, press Alt-F1. Voila! You're back to your first login session.
A newly-installed Linux system probably lets you to access only the first half-dozen or so VCs, by pressing Alt-F1 through Alt-F4, or however many VCs are configured on your system. It is possible to enable up to 12 VCs--one for each function key on your keyboard. As you can see, use of VCs can be very powerful because you can work in several different sessions at the same time.
While the use of VCs is somewhat limiting (after all, you can look at only one VC at a time), it should give you a feel for the multiuser capabilities of Linux. While you're working on the first VC, you can switch over to the second VC and work on something else.

3.2.4 Shells and commands.

    For most of your explorations in the world of Linux, you'll be talking to the system through a shell, a program that takes the commands you type and translates them into instructions to the operating system. This can be compared to the COMMAND.COM program under MS-DOS, which does essentially the same thing. A shell is just one interface to Linux. There are many possible interfaces--like the X Window System, which lets you run commands by using the mouse and keyboard.
As soon as you log in, the system starts the shell, and you can begin entering commands. Here's a quick example. Larry logs in and is waiting at the shell prompt.

  The last line of this text is the shell's prompt, indicating that it's ready to take commands. (More on what the prompt itself means later.) Let's try telling the system to do something interesting:



Well, as it turns out, make is the name of an actual program on the system, and the shell executed this program when given the command. (Unfortunately, the system was being unfriendly.)
      This brings us to the burning question: What is a command? What happens when you type ``make love''? The first word on the command line, ``make'', is the name of the command to be executed. Everything else on the command line is taken as arguments to this command. Example:

The name of this command is ``cp'', and the arguments are ``foo'' and ``bar''.
When you enter a command, the shell does several things. First, it checks the command to see if it is internal to the shell. (That is, a command which the shell knows how to execute itself. There are a number of these commands, and we'll go into them later.) The shell also checks to see if the command is an alias, or substitute name, for another command. If neither of these conditions apply, the shell looks for a program, on disk, having the specified name. If successful, the shell runs the program, sending the arguments specified on the command line.
In our example, the shell looks for a program called make, and runs it with the argument love. Make is a program often used to compile large programs, and takes as arguments the name of a ``target'' to compile. In the case of ``make love'', we instructed make to compile the target love. Because make can't find a target by this name, it fails with a humorous error message, and returns us to the shell prompt.
    What happens if we type a command to a shell and the shell can't find a program having the specified name? Well, we can try the following:

Quite simply, if the shell can't find a program having the name given on the command line (here, ``eat''), it prints an error message. You'll often see this error message if you mistype a command (for example, if you had typed ``mkae love'' instead of ``make love'').

3.2.5 Logging out.

    Before we delve much further, we should tell you how to log out of the system. At the shell prompt, use the command

to log out. There are other ways of logging out, but this is the most foolproof one.

3.2.6 Changing your password.

    You should also know how to change your password. The command passwd prompts you for your old password, and a new password. It also asks you to reenter the new password for validation. Be careful not to forget your password--if you do, you will have to ask the system administrator to reset it for you. (If you are the system administrator, see page .)

3.2.7 Files and directories.

  Under most operating systems (including Linux), there is the concept of a file, which is just a bundle of information given a name (called a filename). Examples of files might be your history term paper, an e-mail message, or an actual program that can be executed. Essentially, anything saved on disk is saved in an individual file.
  Files are identified by their file names. For example, the file containing your history paper might be saved with the file name history-paper. These names usually identify the file and its contents in some form that is meaningful to you. There is no standard format for file names as there is under MS-DOS and some other operating systems; in general, a file name can contain any character (except the / character--see the discussion of path names, below) and is limited to 256 characters in length.
  With the concept of files comes the concept of directories. A directory is a collection of files. It can be thought of as a ``folder'' that contains many different files. Directories are given names, with which you can identify them. Furthermore, directories are maintained in a tree-like structure; that is, directories may contain other directories.
  Consequently, you can refer to a file by its path name, which is made up of the filename, preceded by the name of the directory containing the file. For example, let's say that Larry has a directory called papers, which contains three files: history-final, english-lit, and masters-thesis. Each of these three files contains information for three of Larry's ongoing projects. To refer to the english-lit file, Larry can specify the file's pathname, as in:


  As you can see, the directory and filename are separated by a single slash (/). For this reason, filenames themselves cannot contain the / character. MS-DOS users will find this convention familiar, although in the MS-DOS world the backslash (\) is used instead.
  As mentioned, directories can be nested within each other as well. For example, let's say that there is another directory within papers, called notes. The notes directory contains the files math-notes and cheat-sheet. The pathname of the file cheat-sheet would be


    Therefore, a path name is really like a path to the file. The directory that contains a given subdirectory is known as the parent directory. Here, the directory papers is the parent of the notes directory.

3.2.8 The directory tree.

            Most Linux systems use a standard layout for files so that system resources and programs can be easily located. This layout forms a directory tree, which starts at the ``/'' directory, also known as the ``root directory''. Directly underneath / are important subdirectories: /bin, /etc, /dev, and /usr, among others. These directories in turn contain other directories which contain system configuration files, programs, and so on.
    In particular, each user has a home directory, which is the directory set aside for that user to store his or her files. In the examples above, all of Larry's files (like cheat-sheet and history-final) are contained in Larry's home directory. Usually, user home directories are contained under /home, and are named for the user owning that directory. Larry's home directory is /home/larry.
The diagram on page  shows a sample directory tree, which should give you an idea of how the directory tree on your system is organized.
=1.0pt
  
Figure 3.1: A typical (abridged) Linux directory tree.

3.2.9 The current working directory.

        At any moment, commands that you enter are assumed to be relative to your current working directory. You can think of your working directory as the directory in which you are currently ``located''. When you first log in, your working directory is set to your home directory--/home/larry, in our case. Whenever you refer to a file, you may refer to it in relationship to your current working directory, rather than specifying the full pathname of the file.
Here's an example. Larry has the directory papers, and papers contains the file history-final. If Larry wants to look at this file, he can use the command

The more command simply displays a file, one screen at a time. However, because Larry's current working directory is /home/larry, he can instead refer to the file relative to his current location by using the command

  If you begin a filename (like papers/final) with a character other than /, you're referring to the file in terms relative to your current working directory. This is known as a relative path name.
    On the other hand, if you begin a file name with a /, the system interprets this as a full path name--that is, a path name that includes the entire path to the file, starting from the root directory, /. This is known as an absolute path name.

3.2.10 Referring to home directories.

      Under both tcsh and bash you can specify your home directory with the tilde character (~). For example, the command

is equivalent to

The shell replaces the ~ character with the name of your home directory.
You can also specify other user's home directories with the tilde character. The pathname ~karl/letters translates to /home/karl/letters by the shell (if /home/karl is karl's home directory). Using a tilde is simply a shortcut; there is no directory named ~--it's just syntactic sugar provided by the shell.
 

3.3 First steps into Linux.

Before we begin, it is important to know that all file and command names on a Linux system are case-sensitive (unlike operating systems such as MS-DOS). For example, the command make is very different from Make or MAKE. The same is true for file and directory names.

3.3.1 Moving around.

Now that you can log in, and you know how to refer to files using pathnames, how can you change your current working directory, to make life easier?
    The command for moving around in the directory structure is cd, which is short for ``change directory''. Many often-used Linux commands are two or three letters. The usage of the cd command is

where directory is the name of the directory which you wish to become the current working directory.
As mentioned earlier, when you log in, you begin in your home directory. If Larry wanted to switch to the papers subdirectory, he'd use the command

As you can see, Larry's prompt changes to reflect his current working directory (so he knows where he is). Now that he's in the papers directory, he can look at his history final with the command


      Now, Larry is stuck in the papers subdirectory. To move back up to the next higher (or parent) directory, use the command

(Note the space between the ``cd'' and the ``..''.) Every directory has an entry named ``..'' which refers to the parent directory. Similarly, every directory has an entry named ``.'' which refers to itself. Therefore, the command

gets us nowhere.
You can also use absolute pathnames with the cd command. To cd into Karl's home directory, we can use the command


Also, using cd with no argument will return you to your own home directory.

 

3.3.2 Looking at the contents of directories.

          Now that you know how to move around directories, you might think, ``So what?'' Moving around directories is fairly useless by itself, so let's introduce a new command, ls. The ls command displays a listing of files and directories, by default from your current directory. For example:



Here we can see that Larry has three entries in his current directory: Mail, letters, and papers. This doesn't tell us much--are these directories or files? We can use the -F option of the ls command to get more detailed information.


From the / appended to each filename, we know that these three entries are in fact subdirectories.
    Using ls -F may also append ``*'' to the end of a filename in the resulting list which would indicate that the file is an executable, or a program which can be run. If nothing is appended to the filename using ls -F, the file is a ``plain old file'', that is, it's neither a directory nor an executable.
In general, each UNIX command may take a number of options in addition to other arguments. These options usually begin with a ``-'', as demonstrated above with the -F option. The -F option tells ls to give more information about the type of the files involved--in this case, printing a / after each directory name.
If you give ls a directory name, the system will print the contents of that directory.


Or, for a more interesting listing, let's see what's in the system's /etc directory.



If you're a MS-DOS user, you may notice that the filenames can be longer than 8 characters, and can contain periods in any position. You can even use more than one period in a filename.
Let's move to the top of the directory tree, and then down to another directory with the commands


You can also move into directories in one step, as in cd /usr/bin.
Try moving around various directories, using ls and cd. In some cases, you may run into the foreboding ``Permission denied'' error message. This is simply UNIX security kicking in: in order to use the ls or cd commands, you must have permission to do so. We talk more about this starting on         page .

3.3.3 Creating new directories.

    It's time to learn how to create directories. This involves the use of the mkdir command. Try the following:



Congratulations! You made a new directory and moved into it. Since there aren't any files in this new directory, let's learn how to copy files from one place to another.

3.3.4 Copying files.

      To copy files, use the command cp, as shown here:



The cp command copies the files listed on the command line to the file or directory given as the last argument. Notice that we use ``.'' to refer to the current directory.

3.3.5 Moving files.

      The mv command moves files, rather than copying them. The syntax is very straightforward:



Notice that the termcap file has been renamed sells. You can also use the mv command to move a file to a completely new directory.
Note: mv and cp will overwrite a destination file having the same name without asking you. Be careful when you move a file into another directory. There may already be a file having the same name in that directory, which you'll overwrite!

3.3.6 Deleting files and directories.

      You now have an ugly rhyme developing with the use of the ls command. To delete a file, use the rm command, which stands for ``remove'', as shown here:



We're left with nothing but shells, but we won't complain. Note that rm by default won't prompt you before deleting a file--so be careful.
      A related command to rm is rmdir. This command deletes a directory, but only if the directory is empty. If the directory contains any files or subdirectories, rmdir will complain.

3.3.7 Looking at files.

      The commands more and cat are used for viewing the contents of files. more displays a file, one screenful at a time, while cat displays the whole file at once.
To look at the file shells, use the command


In case you're interested what shells contains, it's a list of valid shell programs on your system. On most systems, this includes /bin/sh, /bin/bash, and /bin/csh. We'll talk about these different types of shells later.
While using more, press Space to display the next page of text, and b to display the previous page. There are other commands available in more as well, these are just the basics. Pressing q will quit more.
Quit more and try cat /etc/termcap. The text will probably fly by too quickly for you to read it all. The name ``cat'' actually stands for ``concatenate'', which is the real use of the program. The cat command can be used to concatenate the contents of several files and save the result to another file. This will be again in section  3.14.1.

3.3.8 Getting online help.

      Almost every UNIX system, including Linux, provides a facility known as manual pages. These manual pages contain online documentation for system commands, resources, configuration files and so on.
  The command used to access manual pages is man. If you're interested in learning about other options of the ls command, you can type

and the manual page for ls will be displayed.
Unfortunately, most manual pages are written for those who already have some idea of what the command or resource does. For this reason, manual pages usually contain only the technical details of the command, without much explanation. However, manual pages can be an invaluable resource for jogging your memory if you forget the syntax of a command. Manual pages will also tell you about commands that we don't cover in this book.
I suggest that you try man for the commands that we've already gone over and whenever I introduce a new command. Some of these commands won't have manual pages, for several reasons. First, the manual pages may not have been written yet. (The Linux Documentation Project is responsible for manual pages under Linux as well. We are gradually accumulating most of the manual pages available for the system.) Second, the the command might be an internal shell command, or an alias (discussed on page ), which would not have a manual page of its own. One example is cd, which is an internal shell command. The shell itself actually processes the cd--there is no separate program that implements this command.

Accessing MS-DOS files.

      If, for some twisted and bizarre reason, you want to access files from MS-DOS, it's easily done under Linux.
    The usual way to access MS-DOS files is to mount an MS-DOS partition or floppy under Linux, allowing you to access the files directly through the file system. For example, if you have an MS-DOS floppy in /dev/fd0, the command

will mount it under /mnt. See Section 4.8.4 for more information on mounting floppies.
You can also mount an MS-DOS partition of your hard drive for access under Linux. If you have an MS-DOS partition on /dev/hda1, the command

mounts it. Be sure to umount the partition when you're done using it. You can have a MS-DOS partition automatically mounted at boot time if you include the entry in /etc/fstab. See Section 4.4 for details. The following line in /etc/fstab will mount an MS-DOS partition on /dev/hda1 on the directory /dos.


You can also mount the VFAT file systems that are used by Windows 95:

This allows access to the long filenames of Windows 95. This only applies to partitions that actually have the long filenames stored. You can't mount a normal FAT16 file system and use this to get long filenames.
  The Mtools software may also be used to access MS-DOS files. The commands mcd, mdir, and mcopy all behave like their MS-DOS counterparts. If you install Mtools, there should be manual pages available for these commands.
      Accessing MS-DOS files is one thing; running MS-DOS programs is another. There is an MS-DOS Emulator under development for Linux; it is widely available, and included in most distributions. It can also be retrieved from a number of locations, including the various Linux FTP sites listed in Appendix B. The MS-DOS Emulator is reportedly powerful enough to run a number of applications, including WordPerfect, from Linux. However, Linux and MS-DOS are vastly different operating systems. The power of any MS-DOS emulator under UNIX is limited. In addition, a Microsoft Windows emulator that runs under X Windows is under development.

3.5 Summary of basic UNIX commands.

    This section introduces some of the most useful basic commands of a UNIX system, including those that are covered in the previous section.
    Note that options usually begin with ``-'', and in most cases you can specify more than one option with a single ``-''. For example, rather than use the command ls -l -F, you can use ls -lF.
Rather than listing every option for each command, we only present useful or important commands at this time. In fact, most of these commands have many options that you'll never use. You can use man to see the manual pages for each command, which list all of the available options.
Also note that many of these commands take as arguments a list of files or directories, denoted in this table by ``file1 ...fileN''. For example, the cp command takes as arguments a list of files to copy, followed by the destination file or directory. When copying more than one file, the destination must be a directory.



 

3.6 Exploring the file system.

    A file system is the collection of files and the hierarchy of directories on a system. The time has now come to escort you around the file system.
You now have the skills and the knowledge to understand the Linux file system, and you have a roadmap. (Refer to diagram on page ).
First, change to the root directory (cd /), and then enter ls -F to display a listing of its contents. You'll probably see the following directories: bin, dev, etc, home, install, lib, mnt, proc, root, tmp, user, usr, and var.
Now, let's take a look at each of these directories.



The various directories described above are essential for the system to operate, but most of the items found in /usr are optional. However, it is these optional items that make the system useful and interesting. Without /usr, you'd have a boring system that supports only programs like cp and ls. /usr contains most of the larger software packages and the configuration files that accompany them.


 

3.7 Types of shells.

  As mentioned before, Linux is a multitasking, multiuser operating system. Multitasking is very useful, and once you understand it, you'll use it all of the time. Before long, you'll run programs in the background, switch between tasks, and pipeline programs together to achieve complicated results with a single command.
Many of the features we'll cover in this section are features provided by the shell itself. Be careful not to confuse Linux (the actual operating system) with a shell--a shell is just an interface to the underlying system. The shell provides functionality inaddition to Linux itself.
  A shell is not only an interpreter for the interactive commands you type at the prompt, but also a powerful programming language. It lets you to write shell scripts, to ``batch'' several shell commands together in a file. If you know MS-DOS you'll recognize the similarity to ``batch files''. Shell scripts are a very powerful tool, that will let you automate and expand your use of Linux. See page  for more information.
            There are several types of shells in the Linux world. The two major types are the ``Bourne shell'' and the ``C shell''. The Bourne shell uses a command syntax like the original shell on early UNIX systems, like System III. The name of the Bourne shell on most Linux systems is /bin/sh (where sh stands for ``shell''). The C shell (not to be confused with sea shell) uses a different syntax, somewhat like the programming language C, and on most Linux systems is named /bin/csh.
              Under Linux, several variations of these shells are available. The two most commonly used are the Bourne Again Shell, or ``Bash'' ( /bin/bash), and ``Tcsh'' (/bin/tcsh). bash is a form of the Bourne shell that includes many of the advanced features found in the C shell. Because bash supports a superset of the Bourne shell syntax, shell scripts written in the standard Bourne shell should work with bash. If you prefer to use the C shell syntax, Linux supports tcsh, which is an expanded version of the original C shell.
The type of shell you decide to use is mostly a religious issue. Some folks prefer the Bourne shell syntax with the advanced features of bash, and some prefer the more structured C shell syntax. As far as normal commands such as cp and ls are concerned, the shell you use doesn't matter--the syntax is the same. Only when you start to write shell scripts or use advanced features of a shell do the differences between shell types begin to matter.
As we discuss the features of the various shells, we'll note differences between Bourne and C shells. However, for the purposes of this manual most of those differences are minimal. (If you're really curious at this point, read the man pages for bash and tcsh).
 

3.8 Wildcards.

        A key feature of most Linux shells is the ability to refer to more than one file by name using special characters. These wildcards let you refer to, say, all file names that contain the character `` n''.
    The wildcard ``*'' specifies any character or string of characters in a file name. When you use the character ``*'' in a file name, the shell replaces it with all possible substitutions from file names in the directory you're referencing.
Here's a quick example. Suppose that Larry has the files frog, joe, and stuff in his current directory.


To specify all files containing the letter ``o'' in the filename, use the command

As you can see, each instance of ``*'' is replaced with all substitutions that match the wildcard from filenames in the current directory.
The use of ``*'' by itself simply matches all filenames, because all characters match the wildcard.


Here are a few more examples:


    The process of changing a ``*'' into a series of filenames is called wildcard expansion and is done by the shell. This is important: an individual command, such as ls, never sees the ``*'' in its list of parameters. The shell expands the wildcard to include all filenames that match. So, the command

is expanded by the shell to


  One important note about the ``*'' wildcard: it does not match file names that begin with a single period (``.''). These files are treated as hidden files--while they are not really hidden, they don't show up on normal ls listings and aren't touched by the use of the ``*'' wildcard.
Here's an example. We mentioned earlier that each directory contains two special entries: ``.'' refers to the current directory, and ``..'' refers to the parent directory. However, when you use ls, these two entries don't show up.

If you use the -a switch with ls, however, you can display filenames that begin with ``.''. Observe:

The listing contains the two special entries, ``.'' and `` ..'', as well as two other ``hidden'' files--.bash_profile and .bashrc. These two files are startup files used by bash when larry logs in. They are described starting on page .
Note that when you use the ``*'' wildcard, none of the filenames beginning with ``.'' are displayed.

This is a safety feature: if the ``*'' wildcard matched filenames beginning with ``.'', it would also match the directory names ``.'' and ``..''. This can be dangerous when using certain commands.
    Another wildcard is ``?''. The ``?'' wildcard expands to only a single character. Thus, ``ls ?'' displays all one-character filenames. And ``ls termca?'' would display ``termcap'' but not ``termcap.backup''. Here's another example:


As you can see, wildcards lets you specify many files at one time. In the command summary that starts on page , we said that the cp and mv commands actually can copy or move more than one file at a time. For example,

copies all filenames in /etc beginning with ``s'' to the directory /home/larry. The format of the cp command is really

where files lists the filenames to copy, and destination is the destination file or directory. mv has an identical syntax.
If you are copying or moving more than one file, the destination must be a directory. You can only copy or move a single file to another file.
     

3.9 Linux plumbing.

 

3.9.1 Standard input and standard output.

        Many Linux commands get input from what is called standard input and send their output to standard output (often abbreviated as stdin and stdout). Your shell sets things up so that standard input is your keyboard, and standard output is the screen.
Here's an example using the cat command. Normally, cat reads data from all of the files specified by the command line, and sends this data directly to stdout. Therefore, using the command

displays the contents of the file history-final followed by masters-thesis.
However, if you don't specify a filename, cat reads data from stdin and sends it back to stdout. Here's an example:

    Each line that you type is immediately echoed back by cat. When reading from standard input, you indicate the input is ``finished'' by sending an EOT (end-of-text) signal, in general, generated by pressing Ctrl-D.
Here's another example. The sort command reads lines of text (again, from stdin, unless you specify one or more filenames) and sends the sorted output to stdout. Try the following.

Now we can alphabetize our shopping list... isn't Linux useful?

3.9.2 Redirecting input and output.

          Now, let's say that you want to send the output of sort to a file, to save our shopping list on disk. The shell lets you redirect standard output to a filename, by using the ``>'' symbol. Here's how it works:

As you can see, the result of the sort command isn't displayed, but is saved to the file named shopping-list. Let's look at this file:

Now you can sort your shopping list, and save it, too! But let's suppose that you are storing the unsorted, original shopping list in the file items. One way of sorting the information and saving it to a file would be to give sort the name of the file to read, in lieu of standard input, and redirect standard output as we did above, as follows:

      However, there's another way to do this. Not only can you redirect standard output, you can redirect standard input as well, using the ``<'' symbol.

Technically, sort < items is equivalent to sort items, but lets you demonstrate the following point: sort < items behaves as if the data in the file items was typed to standard input. The shell handles the redirection. sort wasn't given the name of the file (items) to read; as far as sort is concerned, it still reads from standard input as if you had typed the data from your keyboard.
  This introduces the concept of a filter. A filter is a program that reads data from standard input, processes it in some way, and sends the processed data to standard output. Using redirection, standard input and standard output can be referenced from files. As mentioned above, stdin and stdout default to the keyboard and screen respectively. sort is a simple filter. It sorts the incoming data and sends the result to standard output. cat is even simpler. It doesn't do anything with the incoming data, it simply outputs whatever is given to it.

3.9.3 Using pipes.

  We already demonstrated how to use sort as a filter. However, these examples assume that you have data stored in a file somewhere or are willing to type the data from the standard input yourself. What if the data that you wanted to sort came from the output of another command, like ls?
The -r option to sort sorts the data in reverse-alphabetical order. If you want to list the files in your current directory in reverse order, one way to do it is follows:

Now redirect the output of the ls command into a file called file-list:

Here, you save the output of ls in a file, and then run sort -r on that file. But this is unwieldy and uses a temporary file to save the data from ls.
    The solution is pipelining. This is a shell feature that connects a string of commands via a ``pipe.'' The stdout of the first command is sent to the stdin of the second command. In this case, we want to send the stdout of ls to the stdin of sort. Use the ``|'' symbol to create a pipe, as follows:

This command is shorter and easier to type.
Here's another useful example, the command

displays a long list of files, most of which fly past the screen too quickly for you to read. So, let's use more to display the list of files in /usr/bin.

Now you can page down the list of files at your leisure.
But the fun doesn't stop here! You can pipe more than two commands together. The command head is a filter that displays the first lines from an input stream (in this case, input from a pipe). If you want to display the last filename in alphabetical order in the current directory, use commands like the following:

where head -1 displays the first line of input that it receives (in this case, the stream of reverse-sorted data from ls).  

Non-destructive redirection of output.

    Using ``>'' to redirect output to a file is destructive: in other words, the command

overwrites the contents of the file file-list. If instead, you redirect with the symbol ``>>'', the output is appended to (added to the end of) the named file instead of overwriting it. For example,

appends the output of the ls command to file-list.
Keep in mind that redirection and pipes are features of the shell--which supports the use of ``>'', ``>>'' and `` |''. It has nothing to do with the commands themselves.
   

3.10 File permissions.

  
   

3.10.1 Concepts of file permissions.

      Because there is typically more than one user on a Linux system, Linux provides a mechanism known as file permissions, which protect user files from tampering by other users. This mechanism lets files and directories be ``owned'' by a particular user. For example, because Larry created the files in his home directory, Larry owns those files and has access to them.
Linux also lets files be shared between users and groups of users. If Larry desired, he could cut off access to his files so that no other user could access them. However, on most systems the default is to allow other users to read your files but not modify or delete them in any way.
  Every file is owned by a particular user. However, files are also owned by a particular group, which is a defined group of users of the system. Every user is placed into at least one group when that user's account is created. However, the system administrator may grant the user access to more than one group.
    Groups are usually defined by the type of users who access the machine. For example, on a university Linux system users may be placed into the groups student, staff, faculty or guest. There are also a few system-defined groups (like bin and admin) which are used by the system itself to control access to resources--very rarely do actual users belong to these system groups.
Permissions fall into three main divisions: read, write, and execute. These permissions may be granted to three classes of users: the owner of the file, the group to which the file belongs, and to all users, regardless of group.
                  Read permission lets a user read the contents of the file, or in the case of directories, list the contents of the directory (using ls). Write permission lets the user write to and modify the file. For directories, write permission lets the user create new files or delete files within that directory. Finally, execute permission lets the user run the file as a program or shell script (if the file is a program or shell script). For directories, having execute permission lets the user cd into the directory in question.

3.10.2 Interpreting file permissions.

        Let's look at an example that demonstrates file permissions. Using the ls command with the -l option displays a ``long'' listing of the file, including file permissions.


The first field in the listing represents the file permissions. The third field is the owner of the file (larry) and the fourth field is the group to which the file belongs (users). Obviously, the last field is the name of the file (stuff). We'll cover the other fields later.
This file is owned by larry, and belongs to the group users. The string -rw-r-r- lists, in order, the permissions granted to the file's owner, the file's group, and everybody else.
The first character of the permissions string (``-'') represents the type of file. A ``-'' means that this is a regular file (as opposed to a directory or device driver). The next three characters (``rw-'') represent the permissions granted to the file's owner, larry. The ``r'' stands for ``read'' and the ``w'' stands for ``write''. Thus, larry has read and write permission to the file stuff.
As mentioned, besides read and write permission, there is also ``execute'' permission--represented by an ``x''. However, a ``-'' is listed here in place of an ``x'', so Larry doesn't have execute permission on this file. This is fine, as the file stuff isn't a program of any kind. Of course, because Larry owns the file, he may grant himself execute permission for the file if he so desires. (This will be covered shortly.)
The next three characters, (``r-''), represent the group's permissions on the file. The group that owns this file is users. Because only an ``r'' appears here, any user who belongs to the group users may read this file.
The last three characters, also (``r-''), represent the permissions granted to every other user on the system (other than the owner of the file and those in the group users). Again, because only an ``r'' is present, other users may read the file, but not write to it or execute it.
Here are some other examples of permissions:

3.10.3 Permissions Dependencies.

      The permissions granted to a file also depend on the permissions of the directory in which the file is located. For example, even if a file is set to -rwxrwxrwx, other users cannot access the file unless they have read and execute access to the directory in which the file is located. For example, if Larry wanted to restrict access to all of his files, he could set the permissions to his home directory /home/larry to -rwx---. In this way, no other user has access to his directory, and all files and directories within it. Larry doesn't need to worry about the individual permissions on each of his files.
In other words, to access a file at all, you must have execute access to all directories along the file's pathname, and read (or execute) access to the file itself.
Typically, users on a Linux system are very open with their files. The usual set of permissions given to files is -rw-r-r-, which lets other users read the file but not change it in any way. The usual set of permissions given to directories is -rwxr-xr-x, which lets other users look through your directories, but not create or delete files within them.
However, many users wish to keep other users out of their files. Setting the permissions of a file to -rw---- will prevent any other user from accessing the file. Likewise, setting the permissions of a directory to -rwx--- keeps other users out of the directory in question.

3.10.4 Changing permissions.

        The command chmod is used to set the permissions on a file. Only the owner of a file may change the permissions on that file. The syntax of chmod is


Briefly, you supply one or more of all, user, group, or other. Then you specify whether you are adding rights (+) or taking them away (-). Finally, you specify one or more of read, write, and execute. Some examples of legal commands are:



   

3.11 Managing file links.

          Links let you give a single file more than one name. Files are actually identified by the system by their inode number, which is just the unique file system identifier for the file. A directory is actually a listing of inode numbers with their corresponding filenames. Each filename in a directory is a link to a particular inode.

3.11.1 Hard links.

  The ln command is used to create multiple links for one file. For example, let's say that you have a file called foo in a directory. Using ls -i, you can look at the inode number for this file.

Here, foo has an inode number of 22192 in the file system. You can create another link to foo, named bar, as follows:

With ls -i, you see that the two files have the same inode.

Now, specifying either foo or bar will access the same file. If you make changes to foo, those changes appear in bar as well. For all purposes, foo and bar are the same file.
These links are known as hard links because they create a direct link to an inode. Note that you can hard-link files only when they're on the same file system; symbolic links (see below) don't have this restriction.
When you delete a file with rm, you are actually only deleting one link to a file. If you use the command

then only the link named foo is deleted, bar will still exist. A file is only truly deleted on the system when it has no links to it. Usually, files have only one link, so using the rm command deletes the file. However, if a file has multiple links to it, using rm will delete only a single link; in order to delete the file, you must delete all links to the file.
  The command ls -l displays the number of links to a file (among other information).

The second column in the listing, ``2'', specifies the number of links to the file.
As it turns out, a directory is actually just a file containing information about link-to-inode associations. Also, every directory contains at least two hard links: ``.'' (a link pointing to itself), and ``..'' (a link pointing to the parent directory). The root directory (/) ``..'' link just points back to /. (In other words, the parent of the root directory is the root directory itself.)

3.11.2 Symbolic links.

  Symbolic links, or symlinks, are another type of link, which are different from hard links. A symbolic link lets you give a file another name, but doesn't link the file by inode.
The command ln -s creates a symbolic link to a file. For example, if you use the command

you will create a symbolic link named bar that points to the file foo. If you use ls -i, you'll see that the two files have different inodes, indeed.

However, using ls -l, we see that the file bar is a symlink pointing to foo.


The file permissions on a symbolic link are not used (they always appear as rwxrwxrwx). Instead, the permissions on the symbolic link are determined by the permissions on the target of the symbolic link (in our example, the file foo).
Functionally, hard links and symbolic links are similar, but there are differences. For one thing, you can create a symbolic link to a file that doesn't exist; the same is not true for hard links. Symbolic links are processed by the kernel differently than are hard links, which is just a technical difference but sometimes an important one. Symbolic links are helpful because they identify the file they point to; with hard links, there is no easy way to determine which files are linked to the same inode.
Links are used in many places on the Linux system. Symbolic links are especially important to the shared library images in /lib. See page  for more information.    

3.12 Job control.

   

3.12.1 Jobs and processes.

  
  Job control is a feature provided by many shells (including bash and tcsh) that let you control multiple running commands, or jobs, at once. Before we can delve much further, we need to talk about processes.
      Every time you run a program, you start what is called a process. The command ps displays a list of currently running processes, as shown here:

  The PID listed in the first column is the process ID, a unique number given to every running process. The last column, COMMAND, is the name of the running command. Here, we're looking only at the processes which Larry himself is currently running. (There are many other processes running on the system as well--``ps -aux'' lists them all.) These are bash (Larry's shell), and the ps command itself. As you can see, bash is running concurrently with the ps command. bash executed ps when Larry typed the command. After ps has finished running (after the table of processes is displayed), control is returned to the bash process, which displays the prompt, ready for another command.
  A running process is also called a job. The terms process and job are interchangeable. However, a process is usually referred to as a ``job'' when used in conjunction with job control--a feature of the shell that lets you switch between several independent jobs.
In most cases users run only a single job at a time--whatever command they last typed to the shell. However, using job control, you can run several jobs at once, and switch between them as needed.
How might this be useful? Let's say you are editing a text file and want to interrupt your editing and do something else. With job control, you can temporarily suspend the editor, go back to the shell prompt and start to work on something else. When you're done, you can switch back to the editor and be back where you started, as if you didn't leave the editor. There are many other practical uses of job control.

3.12.2 Foreground and background.

            Jobs can either be in the foreground or in the background. There can only be one job in the foreground at a time. The foreground job is the job with which you interact--it receives input from the keyboard and sends output to your screen, unless, of course, you have redirected input or output, as described starting on page ). On the other hand, jobs in the background do not receive input from the terminal--in general, they run along quietly without the need for interaction.
Some jobs take a long time to finish and don't do anything interesting while they are running. Compiling programs is one such job, as is compressing a large file. There's no reason why you should sit around being bored while these jobs complete their tasks; just run them in the background. While jobs run in the background, you are free to run other programs.
  Jobs may also be suspended. A suspended job is a job that is temporarily stopped. After you suspend a job, you can tell the job to continue in the foreground or the background as needed. Resuming a suspended job does not change the state of the job in any way--the job continues to run where it left off.
    Suspending a job is not equal to interrupting a job. When you interrupt a running process (by pressing the interrupt key, which is usually Ctrl-C), the process is killed, for good. Once the job is killed, there's no hope of resuming it. You'll must run the command again. Also, some programs trap the interrupt, so that pressing Ctrl-C won't immediately kill the job. This is to let the program perform any necessary cleanup operations before exiting. In fact, some programs don't let you kill them with an interrupt at all.

3.12.3 Backgrounding and killing jobs.

Let's begin with a simple example. The command yes is a seemingly useless command that sends an endless stream of y's to standard output. (This is actually useful. If you piped the output of yes to another command which asked a series of yes and no questions, the stream of y's would confirm all of the questions.)
Try it out:

        The y's will continue ad infinitum. You can kill the process by pressing the interrupt key, which is usually Ctrl-C. So that we don't have to put up with the annoying stream of y's, let's redirect the standard output of yes to /dev/null. As you may remember, /dev/null acts as a ``black hole'' for data. Any data sent to it disappears. This is a very effective method of quieting an otherwise verbose program.

Ah, much better. Nothing is printed, but the shell prompt doesn't come back. This is because yes is still running, and is sending those inane y's to /dev/null. Again, to kill the job, press the interrupt key.
Let's suppose that you want the yes command to continue to run but wanted to get the shell prompt back so that you can work on other things. You can put yes into the background, allowing it to run, without need for interaction.
    One way to put a process in the background is to append an ``&'' character to the end of the command.

As you can see, the shell prompt has returned. But what is this ``[1] 164''? And is the yes command really running?
The ``[1]'' represents the job number for the yes process. The shell assigns a job number to every running job. Because yes is the one and only job we're running, it is assigned job number 1. The ``164'' is the process ID, or PID, number given by the system to the job. You can use either number to refer to the job, as you'll see later.
  You now have the yes process running in the background, continuously sending a stream of y's to /dev/null. To check on the status of this process, use the internal shell command jobs.

Sure enough, there it is. You could also use the ps command as demonstrated above to check on the status of the job.
      To terminate the job, use the kill command. This command takes either a job number or a process ID number as an argument. This was job number 1, so using the command

kills the job. When identifying the job with the job number, you must prefix the number with a percent (``%'') character.
Now that you've killed the job, use jobs again to check on it:

The job is in fact dead, and if you use the jobs command again nothing should be printed.
You can also kill the job using the process ID (PID) number, displayed along with the job ID when you start the job. In our example, the process ID is 164, so the command

is equivalent to

You don't need to use the ``%'' when referring to a job by its process ID.

3.12.4 Stopping and restarting jobs.

  There is another way to put a job into the background. You can start the job normally (in the foreground), stop the job, and then restart it in the background.
First, start the yes process in the foreground, as you did before:

Again, because yes is running in the foreground, you shouldn't get the shell prompt back.
Now, rather than interrupt the job with Ctrl-C, suspend the job. Suspending a job doesn't kill it: it only temporarily stops the job until you restart it. To do this, press the suspend key, which is usually Ctrl-Z.

While the job is suspended, it's simply not running. No CPU time is used for the job. However, you can restart the job, which causes the job to run again as if nothing ever happened. It will continue to run where it left off.
    To restart the job in the foreground, use the fg command (for ``foreground'').

    The shell displays the name of the command again so you're aware of which job you just put into the foreground. Stop the job again with Ctrl-Z. This time, use the bg command to put the job into the background. This causes the command to run just as if you started the command with ``&'' as in the last section.

And you have your prompt back. Jobs should report that yes is indeed running, and you can kill the job with kill as we did before.
How can you stop the job again? Using Ctrl-Z won't work, because the job is in the background. The answer is to put the job in the foreground with fg, and then stop it. As it turns out, you can use fg on either stopped jobs or jobs in the background.
There is a big difference between a job in the background and a job that is stopped. A stopped job is not running--it's not using any CPU time, and it's not doing any work (the job still occupies system memory, although it may have been swapped out to disk). A job in the background is running and using memory, as well as completing some task while you do other work.
However, a job in the background may try to display text on your terminal, which can be annoying if you're trying to work on something else. For example, if you used the command

without redirecting stdout to /dev/null, a stream of y's would be displayed on your screen, without any way for you to interrupt it. (You can't use Ctrl-C to interrupt jobs in the background.) In order to stop the endless y's, use the fg command to bring the job to the foreground, and then use Ctrl-C to kill it.
Another note. The fg and bg commands normally affect the job that was last stopped (indicated by a ``+'' next to the job number when you use the jobs command). If you are running multiple jobs at once, you can put jobs in the foreground or background by giving the job ID as an argument to fg or bg, as in

(to put job number 2 into the foreground), or

(to put job number 3 into the background). You can't use process ID numbers with fg or bg.
Furthermore, using the job number alone, as in

is equivalent to


Just remember that using job control is a feature of the shell. The fg, bg and jobs commands are internal to the shell. If for some reason you use a shell that doesn't support job control, don't expect to find these commands available.
In addition, there are some aspects of job control that differ between bash and tcsh. In fact, some shells don't provide job control at all--however, most shells available for Linux do.
 

3.13 Using the vi editor.

        A text editor is a program used to edit files that are composed of text: a letter, C program, or a system configuration file. While there are many such editors available for Linux, the only editor that you are guaranteed to find on any UNIX or Linux system is vi-- the ``visual editor.'' vi is not the easiest editor to use, nor is it very self-explanatory. However, because vi is so common in the UNIX/Linux world, and sometimes necessary, it deserves discussion here.
    Your choice of an editor is mostly a question of personal taste and style. Many users prefer the baroque, self-explanatory and powerful emacs--an editor with more features than any other single program in the UNIX world. For example, Emacs has its own built-in dialect of the LISP programming language, and has many extensions (one of which is an Eliza-like artificial intelligence program). However, because Emacs and its support files are relatively large, it may not be installed on some systems. vi, on the other hand, is small and powerful but more difficult to use. However, once you know your way around vi, it's actually very easy.
This section presents an introduction to vi--we won't discuss all of its features, only the ones you need to know to get started. You can refer to the man page for vi if you're interested in learning more about this editor's features. Alternatively, you can read the book Learning the vi Editor from O'Reilly and Associates, or the VI Tutorial from Specialized Systems Consultants (SSC) Inc. See Appendix A for information.

3.13.1 Concepts.

While using vi, at any one time you are in one of three modes of operation. These modes are called command mode, insert mode, and last line mode.
  When you start up vi, you are in command mode. This mode lets you use commands to edit files or change to other modes. For example, typing ``x'' while in command mode deletes the character underneath the cursor. The arrow keys move the cursor around the file you're editing. Generally, the commands used in command mode are one or two characters long.
  You actually insert or edit text within insert mode. When using vi, you'll probably spend most of your time in this mode. You start insert mode by using a command such as ``i'' (for ``insert'') from command mode. While in insert mode, you can insert text into the document at the current cursor location. To end insert mode and return to command mode, press Esc.
  Last line mode is a special mode used to give certain extended commands to vi. While typing these commands, they appear on the last line of the screen (hence the name). For example, when you type ``:'' in command mode, you jump into last line mode and can use commands like ``wq'' (to write the file and quit vi), or ``q!'' (to quit vi without saving changes). Last line mode is generally used for vi commands that are longer than one character. In last line mode, you enter a single-line command and press Enter to execute it.

3.13.2 Starting vi.

  The best way to understand these concepts is to fire up vi and edit a file. The example ``screens'' below show only a few lines of text, as if the screen were only six lines high instead of twenty-four.
The syntax for vi is

where filename is the name of the file to edit.
Start up vi by typing

to edit the file test. You should see something like


The column of ``~'' characters indicates you are at the end of the file. The represents the cursor.

3.13.3 Inserting text.

  The vi program is now in command mode. Insert text into the file by pressing i, which places the editor into insert mode, and begin typing.


Type as many lines as you want (pressing Enter after each). You may correct mistakes with the Backspace key.
To end insert mode and return to command mode, press Esc.
In command mode you can use the arrow keys to move around in the file. (If you have only one line of text, trying to use the up- or down-arrow keys will probably cause vi to beep at you.)
There are several ways to insert text other than the i command. The a command inserts text beginning after the current cursor position, instead of at the current cursor position. For example, use the left arrow key to move the cursor between the words ``good'' and ``men.''

Press a to start insert mode, type ``wo'', and then press Esc to return to command mode.


To begin inserting text at the next line, use the o command. Press o and enter another line or two:


 

3.13.4 Deleting text.

  From command mode, the x command deletes the character under the cursor. If you press x five times, you'll end up with:

Now press a and insert some text, followed by esc:


You can delete entire lines using the command dd (that is, press d twice in a row). If the cursor is on the second line and you type dd, you'll see:


To delete the word that the cursor is on, use the dw command. Place the cursor on the word ``good'', and type dw.

 

3.13.5 Changing text.

  You can replace sections of text using the R command. Place the cursor on the first letter in ``party'', press R, and type the word ``hungry''.

Using R to edit text is like the i and a commands, but R overwrites, rather than inserts, text.
The r command replaces the single character under the cursor. For example, move the cursor to the beginning of the word ``Now'', and press r followed by C, you'll see:


The ``~'' command changes the case of the letter under the cursor from upper- to lower-case, and back. For example, if you place the cursor on the ``o'' in ``Cow'' above and repeatedly press ~, you'll end up with:

 

3.13.6 Commands for moving the cursor.

  You already know how to use the arrow keys to move around the document. In addition, you can use the h, j, k, and l commands to move the cursor left, down, up, and right, respectively. This comes in handy when (for some reason) your arrow keys aren't working correctly.
The w command moves the cursor to the beginning of the next word; the b command moves it to the beginning of the previous word.
The 0 command (that's the zero key) moves the cursor to the beginning of the current line, and the $ command moves it to the end of the line.
When editing large files, you'll want to move forwards or backwards through the file a screenful at a time. Pressing Ctrl-F moves the cursor one screenful forward, and Ctrl-B moves it a screenful back.
To move the cursor to the end of the file, press G. You can also move to an arbitrary line; for example, typing the command 10G would move the cursor to line 10 in the file. To move to the beginning of the file, use 1G.
You can couple moving commands with other commands, such as those for deleting text. For example, the d$ command deletes everything from the cursor to the end of the line; dG deletes everything from the cursor to the end of the file, and so on.

3.13.7 Saving files and quitting vi.

      To quit vi without making changes to the file, use the command :q!. When you press the ``:'', the cursor moves to the last line on the screen and you'll be in last line mode.

In last line mode, certain extended commands are available. One of them is q!, which quits vi without saving. The command :wq saves the file and then exits vi. The command ZZ (from command mode, without the ``:'') is equivalent to :wq. If the file has not been changed since the last save, it merely exits, preserving the modification time of the last change. Remember that you must press Enter after a command entered in last line mode.
To save the file without quitting vi, use :w.

3.13.8 Editing another file.

  To edit another file, use the :e command. For example, to stop editing test and edit the file foo instead, use the command

If you use :e without saving the file first, you'll get the error message

which means that vi doesn't want to edit another file until you save the first one. At this point, you can use :w to save the original file, and then use :e, or you can use the command

The ``!'' tells vi that you really mean it--edit the new file without saving changes to the first.

3.13.9 Including other files.

  If you use the :r command, you can include the contents of another file in the current file. For example, the command

inserts the contents of the file foo.txt in the text at the location of the cursor.

3.13.10 Running shell commands.

  You can also run shell commands within vi. The :r! command works like :r, but rather than read a file, it inserts the output of the given command into the buffer at the current cursor location. For example, if you use the command

you'll end up with


You can also ``shell out'' of vi, in other words, run a command from within vi, and return to the editor when you're done. For example, if you use the command

the ls -F command will be executed and the results displayed on the screen, but not inserted into the file you're editing. If you use the command

vi starts an instance of the shell, letting you temporarily put vi ``on hold'' while you execute other commands. Just log out of the shell (using the exit command) to return to vi.

3.13.11 Getting vi help.

vi doesn't provide much in the way of interactive help (most Linux programs don't), but you can always read the man page for vi. vi is a visual front-end to the ex editor; which handles many of the last-line mode commands in vi. So, in addition to reading the man page for vi, see ex as well.  

3.14 Customizing your environment.

  A shell provides many mechanisms to customize your work environment. As mentioned above, a shell is more than a command interpreter--it is also a powerful programming language. Although writing shell scripts is an extensive subject, we'd like to introduce you to some of the ways that you can simplify your work on a Linux system by using these advanced features of the shell.
As mentioned before, different shells use different syntaxes when executing shell scripts. For example, Tcsh uses a C-like syntax, while Bourne shells use another type of syntax. In this section, we won't be encountering many differences between the two, but we will assume that shell scripts are executed using the Bourne shell syntax.

3.14.1 Shell scripts.

      Let's say that you use a series of commands often and would like to save time by grouping all of them together into a single ``command''. For example, the three commands

concatenates the files chapter1, chapter2, and chapter3 and places the result in the file book. The second command displays a count of the number of lines in book, and the third command lp book prints book. Rather than type all these commands, you can group them into a shell script. The shell script used to run all these commands might look like this:

Shell scripts are just plain text files; you can create them with an editor such as emacs or vi, which is described starting on page .
Let's look at this shell script. The first line, ``#!/bin/sh'', identifies the file as a shell script and tells the shell how to execute the script. It instructs the shell to pass the script to /bin/sh for execution, where /bin/sh is the shell program itself. Why is this important? On most Linux systems, /bin/sh is a Bourne-type shell, like bash. By forcing the shell script to run using /bin/sh, you ensure that the script will run under a Bourne-syntax shell (rather than a C shell). This will cause your script to run using the Bourne syntax even if you use tcsh (or another C shell) as your login shell.
  The second line is a comment. Comments begin with the character ``#'' and continue to the end of the line. Comments are ignored by the shell--they are commonly used to identify the shell script to the programmer and make the script easier to understand.
The rest of the lines in the script are just commands, as you would type them to the shell directly. In effect, the shell reads each line of the script and runs that line as if you had typed it at the shell prompt.
    Permissions are important for shell scripts. If you create a shell script, make sure that you have execute permission on the script in order to run it. When you create text files, the default permissions usually don't include execute permission, and you must set them explicitly. See the discussion of file permissions on page  for details. Briefly, if this script were saved in the file called makebook, you could use the command

to give yourself execute permission for the shell script makebook.
You can use the command

to run all the commands in the script.

3.14.2 Shell variables and the environment.

        A shell lets you define variables, as do most programming languages. A variable is just a piece of data that is given a name.
tcsh, as well as other C-type shells, use a different mechanism for setting variables than is described here. This discussion assumes the use of a Bourne shell like bash. See the tcsh manual page for details.
When you assign a value to a variable (using the ``='' operator), you can access the variable by prepending a ``$'' to the variable name, as demonstrated below.

The variable foo is given the value hello there. You can then refer to this value by the variable name prefixed with a ``$'' character. For example, the command

produces the same results as


These variables are internal to the shell, which means that only the shell can access them. This can be useful in shell scripts; if you need to keep track of a filename, for example, you can store it in a variable, as above. Using the set command displays a list of all defined shell variables.
    However, the shell lets you export variables to the environment. The environment is the set of variables that are accessible by all commands that you execute. Once you define a variable inside the shell, exporting it makes the variable part of the environment as well. Use the export command to export a variable to the environment.
  Again, here we differ between bash and tcsh. If you use tcsh, another syntax is used for setting environment variables (the setenv command is used). See the tcsh manual page for more information.
  The environment is very important to the UNIX system. It lets you configure certain commands just by setting variables which the commands know about.
Here's a quick example. The environment variable PAGER is used by the man command and it specifies the command to use to display manual pages one screenful at a time. If you set PAGER to the name of a command, it uses that command to display the man pages, instead of more (which is the default).
Set PAGER to ``cat''. This causes output from man to be displayed to the screen all at once, without pausing between pages.

Now, export PAGER to the environment.

Try the command man ls. The man page should fly past your screen without pausing for you.
Now, if we set PAGER to ``more'', the more command is used to display the man page.

Note that we don't have to use the export command after we change the value of PAGER. We only need to export a variable once; any changes made to it thereafter will automatically be propagated to the environment.
It is often necessary to quote strings in order to prevent the shell from treating various characters as special. For example, you need to quote a string in order to prevent the shell from interpreting the special meaning of characters such as ``*'', ``?'' or a space. There are many other characters that may need to be protected from interpretation. A detailed explanation and desription of quoting is described in SSC's Bourne Shell Tutorial.
The manual pages for a particular command tell you if the command uses any environment variables. For example, the man man page explains that PAGER is used to specify the pager command.
Some commands share environment variables. For example, many commands use the EDITOR environment variable to specify the default editor to use when one is needed.
The environment is also used to keep track of important information about your login session. An example is the HOME environment variable, which contains the name of your home directory.


Another interesting environment variable is PS1, which defines the main shell prompt. For example,

To set the prompt back (which contains the current working directory followed by a ``#'' symbol),

The bash manual page describes the syntax used for setting the prompt.

The PATH environment variable.

  When you use the ls command, how does the shell find the ls executable itself? In fact, ls is in /bin on most systems. The shell uses the environment variable PATH to locate executable files for commands you type.
For example, your PATH variable may be set to

This is a list of directories for the shell to search, each directory separated by a ``:''. When you use the command ls, the shell first looks for /bin/ls, then /usr/bin/ls, and so on.
Note that the PATH has nothing to do with finding regular files. For example, if you use the command

the shell does not use PATH to locate the files foo and bar--those filenames are assumed to be complete. The shell only uses PATH to locate the cp executable.
This saves you time, and means that you don't have to remember where all the command executables are stored. On many systems, executables are scattered about in many places, such as /usr/bin, /bin, or /usr/local/bin. Rather than give the command's full pathname (such as /usr/bin/cp), you can set PATH to the list of directories that you want the shell to automatically search.
Notice that PATH contains ``.'', which is the current working directory. This lets you create a shell script or program and run it as a command from your current directory without having to specify it directly (as in ./makebook). If a directory isn't in your PATH, then the shell will not search it for commands to run; this also includes the current directory.

3.14.3 Shell initialization scripts.

      In addition to the shell scripts that you create, there are a number of scripts that the shell itself uses for certain purposes. The most important of these are initialization scripts, which are scripts executed by the shell when you log in. The initialization scripts themselves are simply shell scripts. However, they initialize your environment by executing commands automatically when you log in. If you always use the mail command to check your mail when you log in, you place the command in the initialization script so it will execute automatically.
  Both bash and tcsh distinguish between a login shell and other invocations of the shell. A login shell is a shell invoked when you log in. Usually, it's the only shell you'll use. However, if you ``shell out'' of another program like vi, you start another instance of the shell, which isn't your login shell. In addition, whenever you run a shell script, you automatically start another instance of the shell to execute the script.
            The initialization files used by bash are: /etc/profile (set up by the system administrator and executed by all bash users at login time), $HOME/.bash_profile (executed by a login bash session), and $HOME/.bashrc (executed by all non-login instances of bash). If .bash_profile is not present, .profile is used instead.
    tcsh uses the following initialization scripts: /etc/csh.login (executed by all tcsh users at login time), $HOME/.tcshrc (executed at login time and by all new instances of tcsh), and $HOME/.login (executed at login time, following .tcshrc). If .tcshrc is not present, .cshrc is used instead.
A complete guide to shell programming would be beyond the scope of this book. See the manual pages for bash or tcsh to learn more about customizing the Linux environment.
 

3.15 So you want to strike out on your own?

This chapter should give you enough information for basic Linux use. The manual pages are indispensable tools for learning about Linux. They may appear confusing at first, but if you dig beneath the surface, there is a wealth of information.

4 System Administration

   
This chapter covers the most important things that you need to know about system administration under Linux in sufficient detail to start using the system comfortably. In order to keep the chapter manageable, it covers just the basics and omits many important details. The Linux System Administrator's Guide, by Lars Wirzenius (see Appendix A) provides considerably more detail on system administration topics. It will help you understand better how things work and hang together. At least, skim through the SAG so that you know what it contains and what kind of help you can expect from it.

4.1 The root account.

Linux differentiates between different users. What they can do to each other and the system is regulated. File permissions are arranged so that normal users can't delete or modify files in directories like /bin and /usr/bin. Most users protect their own files with the appropriate permissions so that other users can't access or modify them. (One wouldn't want anybody to be able to read one's love letters.) Each user is given an account that includes a user name and home directory. In addition, there are special, system defined accounts which have special privileges. The most important of these is the root account, which is used by the system administrator. By convention, the system administrator is the user, root.
There are no restrictions on root. He or she can read, modify, or delete any file on the system, change permissions and ownerships on any file, and run special programs like those which partition a hard drive or create file systems. The basic idea is that a person who cares for the system logs in as root to perform tasks that cannot be executed as a normal user. Because root can do anything, it is easy to make mistakes that have catastrophic consequences.
If a normal user tries inadvertently to delete all of the files in /etc, the system will not permit him or her to do so. However, if root tries to do the same thing, the system doesn't complain at all. It is very easy to trash a Linux system when using root. The best way to prevent accidents is:

• Sit on your hands before you press Enter for any command that is non-reversible. If you're about to clean out a directory, re-read the entire command to make sure that it is correct.
• Use a different prompt for the root account. root's .bashrc or .login file should set the shell prompt to something different than the standard user prompt. Many people reserve the character ``#'' in prompts for root and use the prompt character ``$'' for everyone else.
• Log in as root only when absolutely necessary. When you have finished your work as root, log out. The less you use the root account, the less likely you are to damage the system. You are less likely to confuse the privileges of root with those of a normal user.

Picture the root account as a special, magic hat that gives you lots of power, with which you can, by waving your hands, destroy entire cities. It is a good idea to be a bit careful about what you do with your hands. Because it is easy to wave your hands in a destructive manner, it is not a good idea to wear the magic hat when it is not needed, despite the wonderful feeling.
We'll talk in greater detail about the system administrator's responsibilities starting on page .

4.2 Booting the system.

  Some people boot Linux with a floppy diskette that contains a copy of the Linux kernel. This kernel has the Linux root partition coded into it, so it knows where to look for the root file system. This is the type of floppy created by Slackware during installation, for example.
To create your own boot floppy, locate the kernel image on your hard disk. It should be in the file /vmlinuz, or /vmlinux. In some installations, /vmlinuz is a soft link to the actual kernel, so you may need to track down the kernel by following the links.
Once you know where the kernel is, set the root device of the kernel image to the name of your Linux root partition with the rdev command. The format of the command is


where kernel-name is the name of the kernel image, and root-device is the name of the Linux root partition. For example, to set the root device in the kernel /vmlinuz to /dev/hda2, use the command

rdev can set other options in the kernel, like the default SVGA mode to use at boot time. The command

prints a help message on the screen. After setting the root device, simply copy the kernel image to the floppy. Before copying data to any floppy, however, it's a good idea to use the MS-DOS FORMAT.COM or the Linux fdformat program to format the diskette. This lays down the sector and track information that is appropriate to the floppy's capacity.
Floppy diskette formats and their device driver files are discussed further starting on page .
Device driver files, as mentioned earlier, reside in the /dev directory. To copy the kernel in the file /etc/Image to the floppy in /dev/fd0, use the command

This floppy should now boot Linux.

4.2.1 Using LILO.

  LILO is a separate boot loader which resides on your hard disk. It is executed when the system boots from the hard drive and can automatically boot Linux from a kernel image stored there.
LILO can also be used as a first-stage boot loader for several operating systems, which allows you to select the operating system you to boot, like Linux or MS-DOS. With LILO, the default operating system is booted unless you press Shift during the boot-up sequence, or if the prompt directive is given in the lilo.conf file. In either case, you will be provided with a boot prompt, where you type the name of the operating system to boot (such as ``linux'' or ``msdos''). If you press Tab at the boot prompt, a list of operating systems that the system knows about will be provided.
The easy way to install LILO is to edit the configuration file, /etc/lilo.conf. The command

rewrites the modified lilo.conf configuration to the boot sector of the hard disk, and must be run every time you modify lilo.conf.
The LILO configuration file contains a ``stanza'' for each operating system that you want to boot. The best way to demonstrate this is with an example. The lilo.conf file below is for a system which has a Linux root partition on /dev/hda1 and a MS-DOS partition on /dev/hda2.



The first operating system stanza is the default operating system for LILO to boot. Also note that if you use the ``root ='' line, above, there's no reason to use rdev to set the root partition in the kernel image. LILO sets it at boot time.
The Microsoft Windows '95 installer will overwrite the LILO boot manager. If you are going to install Windows '95 on your system after installing LILO, make sure to create a boot disk first (see Section 4.2). With the boot disk, you can boot Linux and re-install LILO after the Windows '95 installation is completed. This is done simply by typing, as root, the command /sbin/lilo, as in the step above. Partitions with Windows '95 can be configured to boot with LILO using the same lilo.conf entries that are used to boot the MS-DOS partition.
The Linux FAQ (see Appendix A) provides more information on LILO, including how to use LILO to boot with the OS/2 Boot Manager.

4.3 Shutting down.

 
Shutting down a Linux system can be tricky. You should never simply turn off the power or press the reset switch. The kernel keeps track of the disk read/write data in memory buffers. If you reboot the system without giving the kernel a chance to write its buffers to disk, you can corrupt the file systems.
Other precautions are taken during shutdown as well. All processes are sent a signal that allows them to die gracefully (by first writing and closing all files, for example). File systems are unmounted for safety. If you wish, the system can also alert users that the system is going down and give them a chance to log off.
The easiest way to shut down is with the shutdown command. The format of the command is

The time argument is the time to shut down the system (in the format hh:mm:ss), and warning-message is a message displayed on all user's terminals before shutdown. Alternately, you can specify the time as ``now'', to shut down immediately. The -r option may be given to shutdown to reboot the system after shutting down.
For example, to shut down and reboot the system at 8:00 pm, use the command


The command halt may be used to force an immediate shutdown without any warning messages or grace period. halt is useful if you're the only one using the system and want to shut down and turn off the machine.
Don't turn off the power or reboot the system until you see the message:

It is very important that you shut down the system, ``cleanly,'' using the shutdown or halt command. On some systems, pressing Ctrl-Alt-Del will be trapped and cause a shutdown. On other systems, using the ``Vulcan nerve pinch'' will reboot the system immediately and cause disaster.

4.3.1 The /etc/inittab file.

  Immediately after Linux boots and the kernel mounts the root file system, the first program that the system executes is init. This program is responsible for starting the system startup scripts, and modifies the system operatiing from its initial boot-up state to its standard, multiuser state. init also spawns the login: shells for all of the tty devices on the system, and specifies other startup and shutdown procedures.
After startup, init remains quietly in the background, monitoring and if necessary altering the running state of the system. There are many details that the init program must see to. These tasks are defined in the /etc/inittab file. A sample /etc/inittab file is shown below.
Modifying the /etc/inittab file incorrectly can prevent you from logging in to your system. At the very least, when changing the /etc/inittab file, keep on hand a copy of the original, correct file, and a boot/root emergency floppy in case you make a mistake.



At startup, this /etc/inittab starts six virtual consoles, a login: prompt on the modem attached to /dev/ttyS0, and a login: prompt on a character terminal connected via a RS-232 serial line to /dev/ttyS1.
Briefly, init steps through a series of run levels, which correspond to various operationing states of the system. Run level 1 is entered immediately after the system boots, run levels 2 and 3 are the normal, multiuser operation modes of the system, run level 4 starts the X Window System via the X display manager xdm, and run level 6 reboots the system. The run level(s) associated with each command are the second item in each line of the /etc/inittab file.
For example, the line

will maintain a login prompt on a serial terminal for runlevels 1-5. The ``s2'' before the first colon is a symbolic identifier used internally by init. respawn is an init keyword that is often used in conjunction with serial terminals. If, after a certain period of time, the agetty program, which spawns the terminal's login: prompt, does not receive input at the terminal, the program times out and terminates execution. `` respawn'' tells init to re-execute agetty, ensuring that there is always a login: prompt at the terminal, regardless of whether someone has logged in. The remaining parameters are passed directly to agetty and instruct it to spawn the login shell, the data rate of the serial line, the serial device, and the terminal type, as defined in /etc/termcap or /etc/terminfo.
The /sbin/agetty program handles many details related to terminal I/O on the system. There are several different versions that are commonly in use on Linux systems. They include mgetty, psgetty, or simply, getty.
In the case of the /etc/inittab line

which allows users to log in via a modem connected to serial line /dev/ttyS0, the /sbin/agetty parameters ``-mt60'' allow the system to step through all of the modem speeds that a caller dialing into the system might use, and to shut down /sbin/agetty if there is no connection after 60 seconds. This is called negotiating a connection. The supported modem speeds are enumerated on the command line also, as well as the serial line to use, and the terminal type. Of course, both of the modems must support the data rate which is finally negotiated by both machines.
Many important details have been glossed over in this section. The tasks that /etc/inittab maintains would comprise a book of their own. For further information, the manual pages of the init and agetty programs, and the Linux Documentation Project's Serial HOWTO, available from the sources listed in Appendix A, are starting points.

4.4 Managing file systems.

  Another task of the system administrator is caring for file systems. Most of this job entails periodically checking the file systems for damage or corrupted files. Many Linux systems also automatically check the file systems at boot time.

4.4.1 Mounting file systems.

Before a file system is accessible to the system, it must be mounted on a directory. For example, if you have a file system on a floppy, you must mount it under a directory like /mnt in order to access the files on the floppy (see page ). After mounting the file system, all of the files in the file system appear in that directory. After unmounting the file system, the directory (in this case, /mnt) will be empty.
The same is true of file systems on the hard drive. The system automatically mounts file systems on your hard drive at bootup time. The so-called ``root file system'' is mounted on the directory /. If you have a separate file system for /usr, it is mounted on /usr. If you only have a root file system, all files (including those in /usr) exist on that file system.
mount and umount (not unmount) are used to mount and unmount file systems. The command


is executed automatically by the file /etc/rc at boot time, or by the file /etc/rc.d/boot (see page ) on some Linux systems. The file /etc/fstab provides information on file systems and mount points. An example /etc/fstab file is
 


The first field, device, is the name of the partition to mount. The second field is the mount point. The third field is the file system type, like ext2 (for ext2fs) or minix (for Minix file systems). Table 4.1 lists the various file system types that are mountable under Linux. Not all of these file system types may be available on your system, because the kernel must have support for them compiled in. See page  for information on building the kernel.
  
Table 4.1: Linux File system Types

The last field of the fstab file are the mount options. This is normally set to defaults.
Swap partitions are included in the /etc/fstab file. They have a mount directory of none, and type swap. The swapon -a command, which is executed from /etc/rc or /etc/init.d/boot, is used to enable swapping on all of the swap devices that are listed in /etc/fstab.
The /etc/fstab file contains one special entry for the /proc file system. As described on page , the /proc file system is used to store information about system processes, available memory, and so on. If /proc is not mounted, commands like ps will not work.
The mount command may be used only by root. This ensures security on the system. You wouldn't want regular users mounting and unmounting file systems on a whim. Several software packages are available which allow non-root users to mount and unmount file systems, especially floppies, without compromising system security.
The mount -av command actually mounts all of the file systems other than the root file system (in the table above, /dev/hda2). The root file system is automatically mounted at boot time by the kernel.
Instead of using mount -av, you can mount a file system by hand. The command

is equivalent to mounting the file system with the entry for /dev/hda3 in the example /etc/fstab file, above.

4.4.2 Device driver names.

  In addition to the partition names listed in the /etc/fstab file, Linux recognizes a number of fixed and removable media devices. They are classified by type, interface, and the order they are installed. For example, the first hard drive on your system, if it is an IDE or older MFM hard drive, is controlled by the device driver pointed to by /dev/hda. The first partition on the hard drive is /dev/hda1, the second partition is /dev/hda2, the third partition is /dev/hda3, and so on. The first partition of the second IDE drive is often /dev/hdb1, the second partition /dev/hdb2, and so on. The naming scheme for the most commonly installed IDE drives for Intel-architecture, ISA and PCI bus machines, is given in Table 4.2.
  
Table 4.2: IDE device driver names.

CD-ROM and tape drives which use the extended IDE/ATAPI drive interface also use these device names.
Many machines, however, including high-end personal computer workstations, and machines based on Digital Equipment Corporation's Alpha processor, use the Small Computer System Interface (SCSI). The naming conventions for SCSI devices are somewhat different than that given above, due the greater flexibility of SCSI addressing. The first SCSI hard drive on a system is /dev/sda, the second SCSI drive is /dev/sdb, and so on. A list of common SCSI devices is given in Table 4.3.
  
Table 4.3: SCSI device drivers

Note that SCSI CD-ROM and tape drives are named differently than SCSI hard drives. Removable SCSI media, like the Iomega Zip drive, follow naming conventions for non-removable SCSI drives. The use of a Zip drive for making backups is described starting on page 
Streaming tape drives, like those which read and write QIC-02, QIC-40, and QIC-80 format magnetic tapes, have their own set of device names, which are described on page .
Floppy disk drives use still another naming scheme, which is described on page .

4.4.3 Checking file systems.

  It is usually a good idea to check your file systems for damaged or corrupted files every now and then. Some systems automatically check their file systems at boot time (with the appropriate commands in /etc/rc or /etc/init.d/boot).
The command used to check a file system depends on the type of the file system. For ext2fs file systems (the most commonly used type), this command is e2fsck. For example, the command

checks the ext2fs file system on /dev/hda2 and automatically corrects any errors.
It is usually a good idea to unmount a file system before checking it, and necessary, if e2fsck is to perform any repairs on the file system. The command

unmounts the file system on /dev/hda2. The one exception is that you cannot unmount the root file system. In order to check the root file system when it's unmounted, you should use a maintenance boot/root diskette (see page ). You also cannot unmount a file system if any of the files which it contains are ``busy''--that is, in use by a running process. For example, you cannot unmount a file system if any user's current working directory is on that file system. You will instead receive a ``Device busy'' error message.
Other file system types use different forms of the e2fsck command, like efsck and xfsck. On some systems, you can simply use the command fsck, which automatically determines the file system type and executes the appropriate command.
If e2fsck reports that it performed repairs on a mounted file system, you must reboot the system immediately. You should give the command shutdown -r to perform the reboot. This allows the system to re-synchronize the information about the file system after e2fsck modifies it.
The /proc file system never needs to be checked in this manner. /proc is a memory file system and is managed directly by the kernel.

4.5 Using a swap file.

  Instead of reserving a separate partition for swap space, you can use a swap file. However, you need to install Linux and get everything running before you create the swap file.
With Linux installed, you can use the following commands to create a swap file. The command below creates a swap file of size 8208 blocks (about 8 Mb).

This command creates the swap file, /swap. The ``count='' parameter is the size of the swap file in blocks.

This command initializes the swap file. Again, replace the name and size of the swapfile with the appropriate values.

Now the system is swapping on the file /swap. The sync command ensures that the file has been written to disk.
One major drawback to using a swap file is that all access to the swap file is done through the file system. This means the blocks which make up the swap file may not be contiguous. Performance may not be as good as a swap partition, where the blocks are always contiguous and I/O requests are made directly to the device.
Another drawback of large swap files is the greater chance that the file system will be corrupted if something goes wrong. Keeping the regular file systems and swap partitions separate prevents this from happening.
Swap files can be useful if you need to use more swap space temporarily. If you're compiling a large program and would like to speed things up somewhat, you can create a temporary swap file and use it in addition to the regular swap space.
To remove a swap file, first use swapoff, as in

Then the file can be deleted.


Each swap file or partition may be as large as 16 megabytes, but you may use up to 8 swap files or partitions on your system.

4.6 Managing users.

   Even if you're the only user on your system, it's important to understand the aspects of user management under Linux. You should at least have an account for yourself (other than root) to do most of your work.
Each user should have his or her own account. It is seldom a good idea to have several people share the same account. Security an issue, and accounts uniquely identify users to the system. You must be able to keep track of who is doing what.

4.6.1 User management concepts.

The system keeps track of the following information about each user:


This information is stored in the file /etc/passwd. Each line in the file has the format

An example might be

In this example, the first field, ``kiwi,'' is the user name.
The next field, ``Xv8Q981g71oKK'', is the encrypted password. Passwords are not stored on the system in human-readable format. The password is encrypted using itself as the secret key. In other words, one must know the password in order to decrypt it. This form of encryption is reasonably secure.
Some systems use ``shadow passwords,'' in which password information is stored in the file /etc/shadow. Because /etc/passwd is world-readable, /etc/shadow provides some degree of extra security because its access permissions are much more restricted. Shadow passwords also provide other features, like password expiration.
The third field, ``102'', is the UID. This must be unique for each user. The fourth field, ``100'', is the GID. This user belongs to the group numbered 100. Group information is stored in the file /etc/group. See Section 4.6.5 for more information.
The fifth field is the user's full name, ``Laura Poole''. The last two fields are the user's home directory (/home/kiwi), and login shell (/bin/bash), respectively. It is not required that the user's home directory be given the same name as the user name. It simply helps identify the directory.

4.6.2 Adding users.

When adding users, several steps must be taken. First, the user is given an entry in /etc/passwd, with a unique user name and UID. The GID, full name, and other information must be specified. The user's home directory must be created, and the permissions on the directory set so that the user owns the directory. Shell initialization files must be installed in the home directory, and other files must be configured system-wide (for example, a spool for the user's incoming e-mail).
It is not difficult to add users by hand, but when you are running a system with many users, it is easy to forget something. The easiest way to add users is to use an interactive program which updates all of the system files automatically. The name of this program is useradd or adduser, depending on what software is installed.
The adduser command takes its information from the file /etc/adduser.conf, which defines a standard, default configuration for all new users.
A typical /etc/adduser.conf file is shown below.


In addition to defining preset variables that the adduser command uses, /etc/adduser.conf also specifies where default system configuration files for each user are located. In this example, they are located in the directory /etc/skel, as defined by the SKEL= line, above. Files which are placed in this directory, like a system-wide, default .profile, .tcshrc, or .bashrc file, will be automatically installed in a new user's home directory by the adduser command.

4.6.3 Deleting users.

Deleting users can be accomplished with the commands userdel or deluser, depending on the software installed on the system.
If you'd like to temporarily ``disable'' a user from logging in to the system without deleting his or her account, simply prepend an asterisk (``*'') to the password field in /etc/passwd. For example, changing kiwi's /etc/passwd entry to

prevents kiwi from logging in.

4.6.4 Setting user attributes.

After you have created a user, you may need to change attributes for that user, like the home directory or password. The easiest way to do this is to change the values directly in /etc/passwd. To set a user's password, use passwd. The command

will change larry's password. Only root may change other users' passwords in this manner. Users can change their own passwords, however.
On some systems, the commands chfn and chsh allow users to set their own full name and login shell attributes. If not, the system administrator must change these attributes for them.

4.6.5 Groups.

  As mentioned above, each user belongs to one or more groups. The only real importance of group relationships pertains to file permissions. As you'll recall from Section 3.10, each file has a ``group ownership'' and a set of group permissions which defines how users in that group may access the file.
There are several system-defined groups, like bin, mail, and sys. Users should not belong to any of these groups; they are used for system file permissions. Instead, users should belong to an individual group like users. You can also maintain several groups for users, like student, staff, and faculty.
The file /etc/group contains information about groups. The format of each line is

Some example groups might be:

The first group, root, is a special system group reserved for the root account. The next group, users, is for regular users. It has a GID of 100. The users mdw and larry are given access to this group. Remember that in /etc/passwd each user was given a default GID. However, users may belong to more than one group, by adding their user names to other group lines in /etc/group. The groups command lists what groups you are given access to.
The third group, guest, is for guest users, and other is for ``other'' users. The user kiwi is given access to this group as well.
The ``password'' field of /etc/group is sometimes used to set a password on group access. This is seldom necessary. To protect users from changing into privileged groups (with the newgroup command), set the password field to ``*''.
The commands addgroup or groupadd may be used to add groups to your system. Usually, it's easier just to add entries in /etc/group yourself, as no other configuration needs to be done to add a group. To delete a group, simply delete its entry in /etc/group.

4.6.6 System administration responsibilities.

Because the system administrator has so much power and responsibility, when some users have their first opportunity to login as root. either on a Linux system or elsewhere, the tendency is to abuse root's privileges. I have known so-called ``system administrators'' who read other users' mail, delete users' files without warning, and generally behave like children when given such a powerful ``toy''.
Because the administrator has such power on the system, it takes a certain amount of maturity and self-control to use the root account as it was intended--to run the system. There is an unspoken code of honor which exists between the system administrator and the users on the system. How would you feel if your system administrator was reading your e-mail or looking over your files? There is still no strong legal precedent for electronic privacy on time-sharing computer systems. On UNIX systems, the root user has the ability to forego all security and privacy mechanisms on the system. It is important that the system administrator develop a trusting relationship with his or her users. I can't stress that enough.

4.6.7 Coping with users.

System administrators can take two stances when dealing with abusive users: they can be either paranoid or trusting. The paranoid system administrator usually causes more harm than he or she prevents. One of my favorite sayings is, ``Never attribute to malice anything which can be attributed to stupidity.'' Put another way, most users don't have the ability or knowledge to do real harm on the system. Ninety percent of the time, when a user is causing trouble on the system (for instance, by filling up the user partition with large files, or running multiple instances of a large program), the user is simply unaware that he or she is creation a problem. I have come down on users who were causing a great deal of trouble, but they were simply acting out of ignorance--not malice.
When you deal with users who cause potential trouble, don't be accusatory. The burden of proof is on you; that is, the rule of ``innocent until proven guilty'' still holds. It is best to simply talk to the user and question him or her about the trouble instead of being confrontational. The last thing you want is to be on the user's bad side. This will raise a lot of suspicion about you--the system administrator--running the system correctly. If a user believes that you distrust or dislike them, they might accuse you of deleting files or breaching privacy on the system. This is certainly not the kind of position you want to be in.
If you find that a user is attempting to ``crack,'' or otherwise intentionally do harm to the system, don't return the malicious behavior with malice of your own. Instead, provide a warning, but be flexible. In many cases, you may catch a user ``in the act'' of doing harm to the system. Give them a warning. Tell them not to let it happen again. However, if you do catch them causing harm again, be absolutely sure that it is intentional. I can't even begin to describe the number of cases where it appeared as though a user was causing trouble, when in fact it was either an accident or a fault of my own.

4.6.8 Setting the rules.

The best way to run a system is not with an iron fist. That may be how you run the military, but Linux is not designed for such discipline. It makes sense to lay down a few simple and flexible guidelines. The fewer rules you have, the less chance there is of breaking them. Even if your rules are perfectly reasonable and clear, users will still at times break them without intending to. This is especially true of new users learning the ropes of the system. It is not patently obvious that you shouldn't download a gigabyte of files and mail them to everyone on the system. Users need help to understand the rules and why they are there.
If you do specify usage guidelines for your system, make sure also that the rationale for a particular guideline is clear. If you don't, users will find all sorts of creative ways to get around the rule, and not know that they are breaking it.

4.6.9 What it all means.

We don't tell you how to run your system down to the last detail. That depends on how you're using the system. If you have many users, things are much different than if you have only a few users, or if you're the only user on the system. However, it's always a good idea--in any situation--to understand what being the system administrator really means.
Being the system administrator doesn't make a Linux wizard. There are many administrators who know very little about Linux. Likewise, many ``normal'' users know more about Linux than any system administrator. Also, being the system administrator does not allow one to use malice against users. Just because the system gives administrators the ability to mess with user files does not mean that he or she has a right to do so.
Being the system administrator is not a big deal. It doesn't matter if your system is a tiny 386 or a Cray supercomputer. Running the system is the same, regardless. Knowing the root password isn't going to earn you money or fame. It will allow you to maintain the system and keep it running. That's it.

4.7 Archiving and compressing files.

Before we can talk about backups, we need to introduce the tools used to archive files on UNIX systems.

4.7.1 Using tar.

The tar command is most often used to archive files. Its command syntax is


where options is the list of commands and options for tar, and files is the list of files to add or extract from the archive.
For example, the command

packs all of the files in /etc into the tar archive backup.tar. The first argument to tar, ``cvf'', is the tar ``command.'' c tells tar to create a new archive file. v forces tar to use verbose mode, printing each file name as it is archived. The ``f'' option tells tar that the next argument, backup.tar, is the name of the archive to create. The rest of the arguments to tar are the file and directory names to add to the archive.
The command

will extract the tar file backup.tar in the current directory.
Old files with the same name are overwritten when extracting files into an existing directory.
Before extracting tar files it is important to know where the files should be unpacked. Let's say that you archive the following files: /etc/hosts, /etc/group, and /etc/passwd. If you use the command

the directory name /etc/ is added to the beginning of each file name. In order to extract the files to the correct location, use

because files are extracted with the path name saved in the archive file.
However, if you archive the files with the command

the directory name is not saved in the archive file. Therefore, you need to ``cd /etc'' before extracting the files. As you can see, how the tar file is created makes a large difference in where you extract it. The command

can be used to display a listing of the archive's files without extracting them. You can see what directory the files in the archive are stored relative to, and extract the archive in the correct location.

4.7.2 gzip and compress.

Unlike archiving programs for MS-DOS, tar does not automatically compress files as it archives them. If you are archiving two, 1-megabyte files, the resulting tar file is two megabytes in size. The gzip command compresses a file (it need not be a tar file). The command

compresses backup.tar and leaves you with backup.tar.gz, a compressed version of the file. The -9 switch tells gzip to use the highest compression factor.
The gunzip command may be used to uncompress a gzipped file. Equivalently, you may use ``gzip -d''.
gzip is a relatively new tool in the UNIX community. For many years, the compress command was used instead. However, because of several factors, including a software patent dispute against the compress data compression algorithm, and the fact that gzip is much more efficient, compress is being phased out.
Files that are output by compress end in ``.Z.'' backup.tar.Z is the compressed version of backup.tar, while backup.tar.gz is the gzipped version. The uncompress command is used to expand a compressed file. It is equivalent to ``compress -d.'' gunzip knows how to handle compressed files as well.

4.7.3 Putting them together.

To archive a group of files and compress the result, use the commands:

The result is backup.tar.gz. To unpack this file, use the reverse commands:

Always make sure that you are in the correct directory before unpacking a tar file.
You can use some Linux cleverness to do this on one command line.

Here, we send the tar file to ``-'', which stands for tar's standard output. This is piped to gzip, which compresses the incoming tar file. The result is saved in backup.tar.gz. The -c option tells gzip to send its output to standard output, which is redirected to backup.tar.gz.
A single command to unpack this archive would be:

Again, gunzip uncompresses the contents of backup.tar.gz and sends the resulting tar file to standard output. This is piped to tar, which reads ``-'', this time referring to tar's standard input.
Happily, the tar command also includes the z option to automatically compress/uncompress files on the fly, using the gzip compression algorithm.
The command

is equivalent to

Just as the command

may be used instead of


Refer to the tar and gzip manual pages for more information.

4.8 Using floppies and making backups.

      Floppies are often used as backup media. If you don't have a tape drive connected to your system, floppy disks can be used (although they are slower and somewhat less reliable).
As mentioned earlier, floppy diskettes must be formatted with the MS-DOS FORMAT.COM or the Linux fdformat program. This lays down the sector and track information that is appropriate to the floppy's capacity.
A few of the device names and formats of floppy disks which are accessible by Linux are given in Table 4.4.
  
Table 4.4: Linux floppy disk formats.

Devices which begin with fd0 are the first floppy diskette drive, which is named the A: drive under MS-DOS. The driver file names of second floppy device begin with fd1. Generally, the Linux kernel can detect the format of a diskette that has already been formatted--you can simply use /dev/fd0 and let the system detect the format. But when you first use completely new, unformatted floppy disks, you may need to use the driver specification if the system can't detect the diskette's type.
A complete list of Linux devices and their device driver names is given in Linux Allocated Devices, by H. Peter Anvin (see Appendix A).
You can also use floppies to hold individual file systems and mount the floppy to access the data on it. See section 4.8.4.

4.8.1 Using floppies for backups.

      The easiest way to make a backup using floppies is with tar. The command

will make a complete backup of your system using the floppy drive /dev/fd0. The ``M'' option to tar allows the backup to span multiple volumes; that is, when one floppy is full, tar will prompt for the next. The command

restores the complete backup. This method can also be used with a tape drive connected to your system. See section 4.8.3.
    Several other programs exist for making multiple-volume backups; the backflops program found on tsx-11.mit.edu may come in handy.
Making a complete backup of the system with floppies can be time- and resource-consuming. Many system administrators use an incremental backup policy. Every month, a complete backup is made, and every week only those files which have been modified in the last week are backed up. In this case, if you trash your system in the middle of the month, you can simply restore the last full monthly backup, and then restore the last weekly backups as needed.
    The find command is useful for locating files which were modified after a certain date. Several scripts for managing incremental backups can be found on sunsite.unc.edu.    

4.8.2 Backups with a Zip drive.

  Making backups to a Zip drive is similar to making floppy backups, but because Zip disks commonly have a capacity of 98 Kb, it is feasible to use a single, mounted Zip disk for a single backup archive.
Zip drives are available with three different hardware interfaces: a SCSI interface, an IDE interface and a parallel port PPA interface. Zip drive support is not included as a pre-compiled Linux option, but it can be specified when building a custom kernel for your system. Page  describes the installation of an Iomega Zip device driver.
The SCSI and PPA interface Zip drives use the SCSI interface and follow the naming conventions for other SCSI devices, which are described on page .
Zip disks are commonly pre-formatted with a MS-DOS file system. You can either use the existing MS-DOS filesystem, which must be supported by your Linux kernel, or use mke2fs or a similar program to write a Linux file system to the disk.
A Zip disk, when mounted as the first SCSI device, is /dev/sda4.


It is often convenient to provide a separate mount point for Zip file systems; for example, /zip. The following steps, which must be executed as root, would create the mount point:

Then you can use /zip for mounting the Zip file system.
Writing archives to Zip disks is similar to archiving to floppies. To archive and compress the /etc directory to a mounted Zip drive, the command used would be


This command could be executed from any directory because it specifies absolute path names. The archive name etc.tgz is necessary if the Zip drive contains a MS-DOS file system, because any files written to the disk must have names which conform to MS-DOS's 8+3 naming conventions; otherwise, the file names will be truncated.
Similarly, extracting this archive requires the commands


To create, for example, an ext2 file system on a Zip drive, you would give the command (for an unmounted Zip disk)


With a Zip drive mounted in this manner, with an ext2 file system, it is possible to back up entire file systems with a single command.


Note that backing up with tar is still preferable in many cases to simply making an archival copy with the cp -a command, because tar preserves the original files' modification times.

4.8.3 Making backups to tape devices.

  Archiving to a streaming tape drive is similar to making a backup to a floppy file system, only to a different device driver. Tapes are also formatted and handled differently that floppy diskettes. Some representative tape device drivers for Linux systems are listed in Table 4.5.
  
Table 4.5: Tape device drivers.

Floppy tape drives use the floppy drive controller interface and are controlled by the ftape device driver, which is covered below. Installation of the ftape device driver module is described on page . SCSI tape devices are listed in Table 4.3.
To archive the /etc directory a tape device with tar, use the command


Similarly, to extract the files from the tape, use the commands


These tapes, like diskettes, must be formatted before they can be used. The ftape driver can format tapes under Linux. To format a QIC-40 format tape, use the command

Other tape drives have their own formatting software. Check the hardware documentation for the tape drive or the documentation of the Linux device driver associated with it.
Before tapes can be removed from the drive, they must be rewound and the I/O buffers written to the tape. This is analogous to unmounting a floppy before ejecting it, because the tape driver also caches data in memory. The standard UNIX command to control tape drive operations is mt. Your system may not provide this command, depending on whether it has tape drive facilities. The ftape driver has a similar command, ftmt, which is used to control tape operations.
To rewind a tape before removing it, use the command

Of course, substitute the correct tape device driver for your system.
It is also a good idea to retension a tape after writing to it, because magnetic tapes are susceptible to stretch. The command


To obtain the status of the tape device, with a formatted tape in the drive, give the command

4.8.4 Using floppies as file systems.

        You can create a file system on a floppy as you would on a hard drive partition. For example,

creates a file system on the floppy in /dev/fd0. The size of the file system must correspond to the size of the floppy. High-density 3.5" disks are 1.44 megabytes, or 1440 blocks, in size. High-density 5.25" disks are 1200 blocks. It is necessary to specify the size of the file system in blocks if the system cannot automatically detect the floppy's capacity.   In order to access the floppy, you must mount the file system contained on it. The command

will mount the floppy in /dev/fd0 on the directory /mnt. Now, all of the files on the floppy will appear under /mnt on your drive.
The mount point, the directory where you're mounting the file system, must exist when you use the mount command. If it doesn't exist, create it with mkdir as described on page .
See page  for more information on file systems, mounting, and mount points.
    Note that any I/O to the floppy is buffered the same as hard disk I/O is. If you change data on the floppy, you may not see the drive light come on until the kernel flushes its I/O buffers. It's important that you not remove a floppy before you unmount it with the command

Do not simply switch floppies as you would on a MS-DOS system. Whenever you change floppies, umount the first floppy and mount the next.

4.9 Upgrading and installing new software.

 
    Another duty of the system administrator is the upgrading and installation of new software.
Linux system development is rapid. New kernel releases appear every few weeks, and other software is updated nearly as often. Because of this, new Linux users often feel the need to upgrade their systems constantly to keep up the the rapidly changing pace. This is unnecessary and a waste of time. If you kept up with all of the changes in the Linux world, you would spend all of your time upgrading and none of your time using the system.
Some people feel that you should upgrade when a new distribution release is made; for example, when Slackware comes out with a new version. Many Linux users completely reinstall their system with the newest Slackware release every time.
The best way to upgrade your system depends on the Linux distribution you have. Debian, S.u.S.E., Caldera and Red Hat Linux all have intelligent package management software which allows easy upgrades by installing a new package. For example, the C compiler, gcc, comes in a pre-built binary package. When it is installed, all of the files of the older version are overwritten or removed.
For the most part, senselessly upgrading to ``keep up with the trend'' is not important at all. This isn't MS-DOS or Microsoft Windows. There is no important reason to run the newest version of all of the software. If you find that you would like or need features that a new version offers, then upgrade. If not, don't upgrade. In other words, upgrade only what you must, when you must. Don't upgrade for the sake of upgrading. This wastes a lot of time and effort.

4.9.1 Upgrading the kernel

  Upgrading the kernel is a matter of obtaining the kernel sources and compiling them. This is generally a painless procedure, but you can run into problems if you try to upgrade to a development kernel, or upgrade to a new kernel version. The version of a kernel has two parts, the kernel version and patchlevel. As of the time of this writing, the latest stable kernel is version 2.0.33. The 2.0 is the kernel version and 33 is the patch level. Odd-numbered kernel versions like 2.1 are development kernels. Stay away from development kernels unless you want to live dangerously! As a general rule, you should be able to upgrade easily to another patch level, but upgrading to a new version requires the upgrade of system utilities which interact closely with the kernel.
  The Linux kernel sources may be retrieved from any of the Linux FTP sites (see page  for a list). On sunsite.unc.edu, for instance, the kernel sources are found in /pub/Linux/kernel, organized into subdirectories by version number.
Kernel sources are released as a gzipped tar file. For example, the file containing the 2.0.33 kernel sources is linux-2.0.33.tar.gz.
Kernel sources are unpacked in the /usr/src directory, creating the directory /usr/src/linux. It is common practice for /usr/src/linux to be a soft link to another directory which contains the version number, like /usr/src/linux-2.0.33. This way, you can install new kernel sources and test them out before removing the old kernel sources. The commands to create the kernel directory link are


When upgrading to a newer patchlevel of the same kernel version, kernel patch files can save file transfer time because the kernel source is around 7MB after being compressed by gzip. To upgrade from kernel 2.0.31 to kernel 2.0.33, you would download the patch files patch-2.0.32.gz and patch-2.0.33.gz, which can be found at the same FTP site as the kernel sources. After you have placed the patches in the /usr/src directory, apply the patches to the kernel in sequence to update the source. One way to do this would be

After the sources are unpacked and any patches have been applied, you need to make sure that three symbolic links in /usr/include are correct for your kernel distribution. To create these links use the commands

After you create the links, there is no reason to create them again when you install the next kernel patch or a newer kernel version. (See Section 3.11 for more about symbolic links.)
  In order to compile the kernel, you must have the gcc C compiler installed on your system. gcc version 2.6.3 or a more recent version is required to compile the 2.0 kernel.
First cd to /usr/src/linux. The command make config prompts you for a number of configuration options. This is the step where you select the hardware that your kernel will support. The biggest mistake to avoid is not including support for your hard disk controller. Without the correct hard disk support in the kernel, the system won't even boot. If you are unsure about what a kernel option means, a short description is available by pressing ? and Enter.
Next, run the command make dep to update all of the source dependencies. This is an important step. make clean removes old binary files from the kernel source tree.
  The command make zImage compiles the kernel and writes it to /usr/src/linux/arch/i386/boot/zImage. Linux kernels on Intel systems are always compressed. Sometimes the kernel you want to compile is too large to be compressed with the compression system that make zImage uses. A kernel which is too large will exit the kernel compile with the error message: Kernel Image Too Large. If this happens, try the command make bzImage, which uses a compression system that supports larger kernels. The kernel is written to /usr/src/linux/arch/i386/boot/bzImage.
Once you have the kernel compiled, you need to either copy it to a boot floppy (with a command like ``cp zImage /dev/fd0'') or install the image so LILO will boot from your hard drive. See page  for more information.

4.9.2 Adding a device driver to the kernel.

 
Page  describes how to use an Iomega Zip drive to make backups. Support for the Iomega Zip drive, like many other devices, is not generally compiled into stock Linux distribution kernels--the variety of devices is simply too great to support all of them in a usable kernel. However, the source code for the Zip parallel port device driver is included as part of the kernel source code distribution. This section describes how to add support for an Iomega Zip parallel port drive and have it co-exist with a printer connected to a different parallel port.
You must have installed and sucessfully built a custom Linux kernel, as described in the previous section.
Selecting the Zip drive ppa device as a kernel option requires that you answer Y to the appropriate questions during the make config step, when you determine the configuration of the custom kernel. In particular, the ppa device requires answering `` Y'' to three options:


After you have sucessfully run make config with all of the support options you want included in the kernel, then run make dep, make clean, and make zImage to build the kernel, you must tell the kernel how to install the driver. This is done via a command line to the LILO boot loader. As described in section 4.2.1, the LILO configuration file, /etc/lilo.conf has ``stanzas'' for each operating system that it knows about, and also directives for presenting these options to the user at boot time.
Another directive that LILO recognizes is ``append='', which allows you to add boot-time information required by various device drivers to the command line. In this case, the Iomega Zip ppa driver requires an unused interrupt and I/O port address. This is exactly analogous to specifying separate printer devices like LPT1: and LPT2: under MS-DOS.
For example, if your printer uses the hexadecimal (base 16) port address 0x378 (see the installation manual for your parallel port card if you don't know what the address is) and is polled (that is, it doesn't require an IRQ line, a common Linux configuration), you would place the following line in your system's /etc/lilo.conf file:

It is worth noting that Linux automatically recognizes one /dev/lp port at boot time, but when specifying a custom port configurations, the boot-time instructions are needed.
The ``0'' after the port address tells the kernel not to use a IRQ (interrupt request) line for the printer. This is generally acceptable because printers are much slower than CPUs, so a slower method of accessing I/O devices, known as polling, where the kernel periodically checks the printer status on its own, still allows the computer to keep up with the printer.
However, devices that operate at higher speeds, like serial lines and disks, each require an IRQ, or interrupt request, line. This is a hardware signal sent by the device to the processor whenever the device requires the processor's attention; for example, if the device has data waiting to be input to the processor. The processor stops whatever it is doing and handles the interrupt request of the device. The Zip drive ppa device requires a free interrupt, which must correspond to the interrupt that is set on the printer card that you connect the Zip drive to. At the time of this writing, the Linux ppa device driver does not support ``chaining'' of parallel port devices, and separate parallel ports must be used for the Zip ppa device and each printer.
To determine which interrupts are already in use on your system, the command

displays a list of devices and the IRQ lines they use. However, you also need to be careful not to use any automatically configured serial port interrupts as well, which may not be listed in the /proc/interrupt file. The Linux Documentation Project's Serial HOWTO, available from the sources listed in Appendix A, describes in detail the configuration of serial ports.
You should also check the hardware settings of various interface cards on your machine by opening the machine's case and visually checking the jumper settings if necessary, to ensure that you are not co-opting an IRQ line that is already in use by another device. Multiple devices fighting for an interrupt line is perhaps the single most common cause of non-functioning Linux systems.
A typical /proc/interrupt file looks like

The first column is of interest here. These are the numbers of the IRQ lines that are in use on the system. For the ppa driver, we want to choose a line which is not listed. IRQ 7 is often a good choice, becuase it is seldom used in default system configurations. We also need to specify the port address which the ppa device will use. This address needs to be physically configured on the interface card. Parallel I/O ports are assigned specific addresses, so you will need to read the documentation for your parallel port card. In this example, we will use the I/O port address 0x278, which corresponds to the LPT2: printer port under MS-DOS. Adding both the IRQ line and port address to our boot-time command line, above, yields the following statement as it would appear in the appropriate stanza of the /etc/lilo.conf file:


These statements are appended to the kernel's start-up parameters at boot time. They ensure that any printer attached to the system does not interfere with the Zip drive's operation. Of course, if your system does not have a printer installed, the ``lp='' directive can, and should, be omitted.
After you have installed the custom kernel itself, as described in section 4.2.1, and before you reboot the system, be sure to run the command

to install the new LILO configuration on the hard drive's boot sector.

4.9.3 Installing a device driver module.

 
Page  describes how to back up files to a tape drive. Linux provides support for a variety of tape drives with IDE, SCSI, and some proprietary interfaces. Another common type of tape drive connects directly to the floppy drive controller. Linux provides the ftape device driver as a module.
At the time of this writing, the most recent version of ftape is 3.04d. You can retrieve the package from the sunsite.unc.edu FTP archive (see Appendix B for instructions). The ftape archive is located in /pub/Linux/kernel/tapes. Be sure to get the most recent version. At the time of this writing, this is ftape-3.04d.tar.gz.
After unpacking the ftape archive in the /usr/src directory, typing make install in the top-level ftape directory will compile the ftape driver modules and utilities, if necessary, and install them. If you experience compatibility problems with the ftape executable distribution files and your system kernel or libraries, executing the commands make clean and make install will ensure that the modules are compiled on your system.
To use this version of the ftape driver, you must have module support compiled into the kernel, as well as support for the kerneld kernel daemon. However, you must not include the kernel's built-in ftape code as a kernel option, as the more recent ftape module completely replaces this code.
make install also installs the device driver modules in the correct directories. On standard Linux systems, modules are located in the directory

If your kernel version is 2.0.30, the modules on your system are located in /lib/modules/2.0.30. The make install step also insures that these modules are locatable by adding the appropriate statements to the modules.dep file, located in the top-level directory of the module files, in this case /lib/modules/2.0.30. The ftape installation adds the following modules to your system (using kernel version 2.0.30 in this example):


The instructions to load the modules also need to be added to the system-wide module configuration file. This is the file /etc/conf.modules on many systems. To automatically load the ftape modules on demand, add the following lines to the /etc/conf.modules file:

The first statement loads all of the ftape related modules if necessary when a device with the major number 27 (the ftape device) is accessed by the kernel. Because support for the zftape module (which provides automatic data compression for tape devices) requires the support of the other ftape modules, all of them are loaded on demand by the kernel. The second line specifies load-time parameters for the modules. In this case, the utility /sbin/swapout, which is provided with the ftape package, ensures that sufficient DMA memory is available for the ftape driver to function.
To access the ftape device, you must first place a formatted tape in the drive. Instructions for formatting tapes and operation of the tape drive are given in section 4.8.3.

4.9.4 Upgrading the libraries.

    As mentioned before, most of the software on the system is compiled to use shared libraries, which contain common subroutines shared among different programs. If you see the message

when attempting to run a program, then you need to upgrade to the version of the libraries which the program requires. Libraries are backwardly compatible. A program compiled to use an older version of the libraries should work with the new version of the libraries installed. However, the reverse is not true.
The newest version of the libraries can be found on Linux FTP sites. On sunsite.unc.edu, they are located in /pub/Linux/GCC. The ``release'' files there should explain what files you need to download and how to install them. Briefly, you should get the files image-version.tar.gz and inc-version.tar.gz where version is the version of the libraries to install, such as 4.4.1. These are tar files compressed with gzip. The image file contains the library images to install in /lib and /usr/lib. The inc file contains include files to install in /usr/include
The release-version.tar.gz should explain the installation procedure in detail (the exact instructions vary with each release). In general, you need to install the library's .a and .sa files in /usr/lib. These are the libraries used at compilation time.
In addition, the shared library image files, libc.so.version are installed in /lib. These are the shared library images loaded at run time by programs using the libraries. Each library has a symbolic link using the major version number of the library in /lib.
The libc library version 4.4.1 has a major version number of 4. The file containing the library is libc.so.4.4.1. A symbolic link of the name libc.so.4 is also placed in /lib pointing to the library. You must change this symbolic link when upgrading the libraries. For example, when upgrading from libc.so.4.4 to libc.so.4.4.1, you need to change the symbolic link to point to the new version.
You must change the symbolic link in one step, as described below. If you delete the symbolic link libc.so.4, then programs which depend on the link (including basic utilities like ls and cat) will stop working. Use the following command to update the symbolic link libc.so.4 to point to the file libc.so.4.4.1:

You also need to change the symbolic link libm.so.version in the same manner. If you are upgrading to a different version of the libraries, substitute the appropriate file names, above. The library release notes should explain the details. (See page  for more information about symbolic links.)

4.9.5 Upgrading gcc.

    The gcc C and C++ compiler is used to compile software on your system, most importantly the kernel. The newest version of gcc is found on the Linux FTP sites. On sunsite.unc.edu, it is found in the directory /pub/Linux/GCC (along with the libraries). There should be a release file for the gcc distribution detailing what files you need to download and how to install them. Most distributions have upgrade versions that work with their package management software. In general, these packages are much easier to install than ``generic'' distributions.

4.9.6 Upgrading other software.

Upgrading other software is often simply a matter of downloading the appropriate files and installing them. Most software for Linux is distributed as compressed tar files that include sources, binaries, or both. If binaries are not included in the release, you may need to compile them yourself. This means at least typing make in the directory where the sources are located.
  Reading the Usenet newsgroup comp.os.linux.announce for announcements of new software releases is the easiest way to find out about new software. Whenever you are looking for software on an FTP site, downloading the ls-lR index file from the FTP site and using grep to find the files you want is the easiest way to locate software. If you have archie available to you, it can be of assistance as well. There are also other Internet resources which are devoted specifically to Linux. See Appendix A for more details.
   

4.10 Miscellaneous tasks.

Believe it or not, there are a number of housekeeping tasks for the system administrator which don't fall into any major category.

4.10.1 System startup files.

      When the system boots, a number of scripts are executed automatically by the system before any user logs in. Here is a description of what happens.             At bootup time, the kernel spawns the process /etc/init. Init is a program which reads its configuration file, /etc/inittab, and spawns other processes based on the contents of this file. One of the important processes started from inittab is the /etc/getty process started on each virtual console. The getty process grabs the VC for use, and starts a login process on the VC. This allows you to login on each VC. If /etc/inittab does not contain a getty process for a certain VC, you will not be able to login on that VC.
        Another process executed from /etc/inittab is /etc/rc, the main system initialization file. This file is a simple shell script which executes any initialization commands needed at boot time, such as mounting the file systems (see page ) and initializing swap space. On some systems, init executes the file /etc/init.d/rc.
Your system may execute other initialization scripts as well. For example /etc/rc.local which usually contains initialization commands specific to your own system, such as setting the host name (see the next section). rc.local may be started from /etc/rc or from /etc/inittab directly.

4.10.2 Setting the host name.

      In a networked environment, the host name is used to uniquely identify a particular machine, while on a standalone machine, the host name simply gives the system personality and charm. It's like naming a pet: you can always address to your dog as ``The dog,'' but it's much more interesting to assign the dog a name such as spot or woofie. Setting the system's host name is a simple matter of using the hostname command. If you are on a network, your host name should be the full host name of your machine, such as goober.norelco.com. If you are not on a network of any kind, you can choose an arbitrary host and domain name, such as loomer.vpizza.com, shoop.nowhere.edu, or floof.org.
The host name must appear in the file /etc/hosts, which assigns an IP address to each host. Even if your machine is not on a network, you should include your own host name in /etc/hosts. If you are not on a TCP/IP network, and your host name is floof.org, simply include the following line in /etc/hosts:

This assigns your host name, floof.org, to the loopback address 127.0.0.1. The loopback interface is present whether the machine is connected to a network or not. The localhost alias is always assigned to this address.
If you are on a TCP/IP network, your actual IP address and host name should appear in /etc/hosts. For example, if your host name is goober.norelco.com, and your IP address is 128.253.154.32, add the following line to /etc/hosts:


To set your host name, simply use the hostname command. For example, the command

sets the host name to goober.norelco.com. In most cases, the hostname command is executed from one of the system startup files, like /etc/rc or /etc/rc.local. Edit these two files and change the hostname command found there to set your own host name. Upon rebooting, the system will use the new name.

4.11 What to do in an emergency.

    On some occasions, the system administrator will be faced with the problem of recovering from a complete disaster, such as forgetting the root password or trashing file systems. The best advice is, don't panic. Everyone makes stupid mistakes--that's the best way to learn about system administration: the hard way.
Linux is not an unstable version of UNIX. In fact, I have had fewer problems with system hangs than with commercial versions of UNIX on many platforms. Linux also benefits from a strong complement of wizards who can help you out of a bind. (The double entendre is intended.)
The first step to fixing any problem youself is finding out what it is. Poke around, and see how things work. Much of the time, a system administrator posts a desperate plea for help before he or she looks into the problem at all. You'll find that fixing problems yourself is actually very easy. It is the path to enlighenment and guruhood.
There are a few times when reinstalling the system from scratch is necessary. Many new users accidentally delete some essential system file, and immediately reach for the installation disks. This is not a good idea. Before taking such drastic measures, investigate the problem and ask others for help. In many cases, you can recover your system from a maintenance diskette.

4.11.1 Recovery with a maintenance diskette.

                One indispensable tool of the system administrator is the so-called ``boot/root disk,'' a floppy that can be booted for a complete Linux system, independent of your hard drive. Boot/root disks are actually very simple--you create a root file system on the floppy, place all of the necessary utilities on it, and install LILO and a bootable kernel on the floppy. Another technique is to use one floppy for the kernel and another for the root file system. In any case, the result is the same: you are running a Linux system completely from the floppy drive.
The canonical example of a boot/root disk is the Slackware boot disks. These diskettes contain a bootable kernel and a root file system, all on floppy. They are intended to be used to install the Slackware distribution, but come in handy when doing system maintenance.
The H.J Lu boot/root disk, available from /pub/Linux/GCC/rootdisk on sunsite.unc.edu, is another example of a maintenance disk. If you're ambitious, you can create your own. In most cases, however, a ready-made boot/root disk is much easier to use and probably will be more complete.
Using a boot/root disk is very simple. Boot the disk on your system, and login as root (usually with no password). In order to access the files on the hard drive, you will need to mount the file systems by hand. For example, the command

will mount an ext2fs file system on /dev/hda2 under /mnt. Remember that / is now on the boot/root disk itself; you need to mount your hard drive file systems under some directory in order to access the files. Therefore, /etc/passwd on your hard drive is now /mnt/etc/passwd if you mount your root file system on /mnt.

4.11.2 Fixing the root password.

    If you forget your root password, this is not a problem, surprisingly. Boot the boot/root disk, mount the root file system on /mnt, and blank out the password field for root in /mnt/etc/passwd, as in this example:

Now root has no password. When you reboot from the hard drive you should be able to login as root and reset the password using passwd.
Aren't you glad that you learned how to use vi? On your boot/root disk, editors like Emacs probably aren't available, but vi should be.

4.11.3 Trashed file systems.

      If you somehow trash a file systems, you can run e2fsck or the appropriate form of fsck for the file system type. (See page .) In most cases, it is safest to correct any damaged data on the hard drive file systems from floppy.
    One common cause of file system damage is a damaged super block. The super block is the ``header'' of the file system that contains information about its status, size, free blocks, and so forth. If you damage the super block, by accidentally writing data directly to the file system's partition table for example, the system probably will not recognize the file system at all. Attempt to mount the file system will fail, and e2fsck won't be able to fix the problem.
Happily, an ext2fs file system type saves copies of the superblock at ``block group'' boundaries on the drive, usually every 8K blocks. To tell e2fsck to use a copy of the superblock, use a command like

where partition is the partition on which the file system resides. The -b 8193 option tells e2fsck to use the copy of the superblock stored at block 8193 in the file system.

4.11.4 Recovering lost files.

  If you accidentally delete an important file on your system, there's no way to ``undelete'' it. However, you can copy the relevant files from the floppy to your hard drive. If you delete /bin/login, for example, which allows you to login, simply boot the boot/root floppy, mount the root file system on /mnt, and use the command

The -a option tells cp to preserve the permissions on the file(s) being copied.
Of course, if the files you deleted aren't essential system files that have counterparts on the boot/root floppy, you're out of luck. If you make backups however, you can always restore them.

4.11.5 Trashed libraries.

  If you accidentally trash your libraries or symbolic links in /lib, more than likely the commands which depend on the libraries will no longer work (see page ). The easiest solution is to boot your boot/root floppy, mount your root file system, and fix the libraries in /mnt/lib. Page  describes how install run time libraries and their symbolic links.