Linux Administration

Linux administration A.HARISH KUMAR

Introduction Linux is based on Unix Unix philosophy Unix commands Unix standards and conventions Linux is a true 32−bit operating system that runs on a variety of different platforms, including Intel, Sparc, Alpha, and Power−PC (on some of these platforms, such as Alpha, Linux is actually 64−bit).

Topics of Discussion Unix System Architecture Unix Philosophy What is Linux? Using a Linux System Using a Linux System Linux Command Line Logging Out Command Syntax Files Creating Files with cat Displaying Files’ Contents with cat Deleting Files with rm Unix Command Feedback Copying and Renaming Files with cp and mv

Topics Filename Completion Command History Files and Directories Examples of Absolute Paths Current Directory Making and Deleting Directories Relative Paths Special Dot Directories Scripting Backup Hidden Files Paths to Home Directories Running Programs

Topics Hidden Files Paths to Home Directories Looking for Files in the System Running Programs Specifying Files with Wildcards Graphical and Text Interfaces Text Editors Shells The Bash Shell Shell Commands Command-Line Arguments Syntax of Command-Line Options Examples of Command-Line Options

Topics Setting Shell Variables Summary of Bash Editing Keys Working with Text Files Lines of Text Filesystem Objects Directory and File Names File Extensions Going Back to Previous Directories Filename Completion Copying Files with cp Moving Files with mv Deleting Files with rm Making Directories with mkdir

TOPICS Removing Directories with rmdir Standard Files Standard Input Standard Output Pipes Searching Files with grep Pattern Matching Job Control jobs What is a Process? Process Properties Parent and Child Processes About Quoting

TOPICS Quoting: Backslashes Quoting: Double Quotes Shell Programming Filesystems The Unified Filesystem File Types Inodes and Directories Symbolic Links Hard Links Symlinks and Hard Links Illustrated Comparing Symlinks and Hard Links Examining and Creating Hard Links Users and Groups

TOPICS The Superuser: Root Changing File Ownership with chown Changing File Group Ownership with chgrp Changing the Ownership of a Directory and Its Contents Changing Ownership and Group Ownership Simultaneously Basic Concepts: Permissions on Files Basic Concepts: Permissions on Directories Examining Permissions: ls -l Changing File and Directory Permissions: chmod Specifying Permissions for chmod Disks and Partitions Disk Naming

TOPICS Using fdisk Making New Partitions Changing Partition Types Making Filesystems with mkfs Mounting Filesystems Mounting a Filesystem: mount Mounting Other Filesystems Unmounting a Filesystem: umount Configuring mount: /etc/fstab Sample /etc/fstab Filesystem Types Mount Options Other Columns in /etc/fstab

TOPICS Mounting a File Monitoring Space: df Monitoring Inodes: df Monitoring Disk Usage: du du Options Finding and Repairing Filesystem Corruption: fsck Running fsck Unix Filesystem Layout Static and Dynamic Data Overview of the FHS FHS: Installed Software FHS: Other Directories Under /usr FHS: Directories Under /var

TOPICS The type Built-in Command Checking for Shell Builtins with type Updating the locate Data updatedb.conf What are Quotas? Hard and Soft Limits Per-User and Per-Group Quotas Displaying Quota Limits: quota Enabling Quota: quotaon Changing Quota Limits: setquota edquota Repquota Boot Loaders

TOPICS LILO Sample lilo.conf File Selecting What to Boot Other Ways of Starting Linux Boot Messages Understanding Runlevels Typical Runlevels Single-User Mode and sulogin Shutting Down and Restarting the System Setting the Default Runlevel Selecting a Different Runlevel at Bootup Determining the Current Runlevel Switching Runlevel

TOPICS Navigating Within Manual Pages Sections of a Manual Page Printing Manual Pages Configuration Files Shell Configuration Files Changing Environment Variables Changing the Prompt Shell Aliases syslog /etc/syslog.conf syslog’s configuration is in / Reconfiguring syslog Examining Logs: less and grep

TOPICS Running Commands in the Future At Commands Commands Run by the At Dæmon At Command Specification Opening Windows from At Commands Managing At Commands Simple Cron Job Specification Crontab Format Crontab Date & Time Specification More Complex Crontab Dates & Times /etc/crontab User Crontabs Cron Job Output

TOPICS At Command and Cron Job Permissions Reasons for Backup Backup Media Types of Backup Backup Strategy Archiving Files with tar Creating Archives with tar Listing the Files in tar Archives Extracting Files from tar Archives Device Files for Accessing Tapes Using tar for Backups Controlling Tape Drives with mt Deciding What to Backup

TOPICS What Not to Backup Scripting Backup Other Backup Software

The shell and the window environment are programs Programs’ only access to hardware is via the kernel Unix System Architecture

Unix Philosophy Multi-user l A user needs an account to use a computer l Each user must log in l Complete separation of different users’ files and configuration settings Small components l Each component should perform a single task l Multiple components can be combined and chained together for more complex tasks l An individual component can be subsituted for another, without affecting other components

What is Linux? Linux kernel l Developed by Linus Torvalds l Strictly speaking, ‘Linux’ is just the kernel Associated utilities l Standard tools found on (nearly) all Linux systems l Many important parts come from the GNU project Free Software Foundation’s project to make a free Unix Some claim the OS as a whole should be ‘GNU/Linux’ Linux distributions l Kernel plus utilities plus other tools, packaged up for end users l Generally with installation program l Distributors include: Red Hat, Debian, SuSE, Mandrake

Using a Linux System Login prompt displayed When Linux first loads after booting the computer After another user has logged out Need to enter a username and password The login prompt may be graphical or simple text If text, logging in will present a shell If graphical, logging in will present a desktop Some combination of mousing and keystrokes will make a terminal window appear A shell runs in the terminal window

Linux Command Line The shell is where commands are invoked A command is typed at a shell prompt Prompt usually ends in a dollar sign ($) After typing a command press Enter to invoke it The shell will try to obey the command Another prompt will appear Example: $ date Thu Jun 14 12:28:05 BST 2001 $ The dollar represents the prompt in this course — do not type it

Logging Out To exit from the shell, use the exit command Pressing Ctrl+D at the shell prompt will also quit the shell Quitting all programs should log you out If in a text-only single-shell environment, exiting the shell should be sufficient In a window environment, the window manager should have a log out command for this purpose After logging out, a new login prompt should be displayed

Command Syntax Most commands take parameters Some commands require them Parameters are also known as arguments For example, echo simply displays its arguments: $ echo $ echo Hello there Hello there Commands are case-sensitive Usually lower-case $ echo welcome welcome $ ECHO SHOUT bash: ECHO: command not found

Files Data can be stored in a file Each file has a filename A label referring to a particular file Permitted characters include letters, digits, hyphens (-), underscores (_), and dots (.) Case-sensitive — News Crew .mov is a different file from NewScrew.mov The ls command lists the names of files

Creating Files with cat There are many ways of creating a file One of the simplest is with the cat command: $ cat > shopping_list cucumber bread yoghurts fish fingers Note the greater-than sign (>) — this is necessary to create the file The text typed is written to a file with the specified name Press Ctrl+D after a line-break to denote the end of the file l The next shell prompt is displayed ls demonstrates the existence of the new file

Displaying Files’ Contents with cat There are many ways of viewing the contents of a file One of the simplest is with the cat command: $ cat shopping_list cucumber bread yoghurts fish fingers n Note that no greater-than sign is used n The text in the file is displayed immediately: l Starting on the line after the command l Before the next shell prompt

Deleting Files with rm To delete a file, use the rm (‘remove’) command Simply pass the name of the file to be deleted as an argument: $ rm shopping_list n The file and its contents are removed l There is no recycle bin l There is no ‘unrm’ command n The ls command can be used to confirm the deletion

Unix Command Feedback Typically, succesful commands do not give any output Messages are displayed in the case of errors The rm command is typical l If it manages to delete the specified file, it does so silently l There is no ‘File shopping_list has been removed’ message l But if the command fails for whatever reason, a message is displayed The silence can be be off-putting for beginners It is standard behaviour, and doesn’t take long to get used to

Copying and Renaming Files with cp and mv To copy the contents of a file into another file, use the cp command: $ cp CV.pdf old-CV.pdf To rename a file use the mv (‘move’) command: $ mv commitee_minutes.txt committee_minutes.txt l Similar to using cp then rm For both commands, the existing name is specified as the first argument and the new name as l If a file with the new name already exists, it is overwritten

Filename Completion The shell can making typing filenames easier Once an unambiguous prefix has been typed, pressing Tab will automatically ‘type’ the rest n For example, after typing this: $ rm sho pressing Tab may turn it into this: $ rm shopping_list n This also works with command names l For example, da may be completed to date if no other commands start ‘da’

Command History Often it is desired to repeat a previously-executed command The shell keeps a command history for this purpose l Use the Up and Down cursor keys to scroll through the list of previous commands l Press Enter to execute the displayed command n Commands can also be edited before being run l Particularly useful for fixing a typo in the previous command The Left and Right cursor keys navigate across a command l Extra characters can be typed at any point l Backspace deletes characters to the left of the cursor l Del and Ctrl+D delete characters to the right Take care not to log out by holding down Ctrl+D too long

Files and Directories A directory is a collection of files and/or other directories l Because a directory can contain other directories, we get a directory hierarchy The ‘top level’ of the hierarchy is the root directory Files and directories can be named by a path l Shows programs how to find their way to the file l The root directory is referred to as / l Other directories are referred to by name, and their names are separated by slashes (/) n If a path refers to a directory it can end in / l Usually an extra slash at the end of a path makes no difference

Examples of Absolute Paths An absolute path starts at the root of the directory hierarchy, and names directories under it: /etc/hostname l Meaning the file called hostname in the directory etc in the root directory n We can use ls to list files in a specific directory by specifying the absolute path: $ ls /usr/share/doc/

Current Directory Your shell has a current directory — the directory in which you are currently working Commands like ls use the current directory if none is specified Use the pwd (print working directory) command to see what your current directory is: $ pwd /home/fred n Change the current directory with cd: $ cd /mnt/cdrom $ pwd /mnt/cdrom n Use cd without specifying a path to get back to your home directory

Making and Deleting Directories The mkdir command makes new, empty, directories For example, to make a directory for storing company accounts: $ mkdir Accounts n To delete an empty directory, use rmdir: $ rmdir OldAccounts n Use rm with the -r (recursive) option to delete directories and all the files they contain: $ rm -r OldAccounts n Be careful — rm can be a dangerous tool if misused

Relative Paths Paths don’t have to start from the root directory l A path which doesn’t start with / is a relative path l It is relative to some other directory, usually the current directory For example, the following sets of directory changes both end up in the same directory: $ cd /usr/share/doc $ cd / $ cd usr $ cd share/doc n Relative paths specify files inside directories in the same way as absolute ones

Hidden Files The special . and .. directories don’t show up when you do ls l They are hidden files n Simple rule: files whose names start with . are considered ‘hidden’ n Make ls display all files, even the hidden ones, by giving it the -a (all) option: $ ls -a . .. .bashrc .profile report.doc n Hidden files are often used for configuration files l Usually found in a user’s home directory n You can still read hidden files — they just don’t get listed by ls by default

Paths to Home Directories The symbol ˜ (tilde) is an abbreviation for your home directory l So for user ‘fred’, the following are equivalent: $ cd /home/fred/documents/ $ cd ˜/documents/ The ˜ is expanded by the shell, so programs only see the complete path You can get the paths to other users’ home directories using ˜, for example: $ cat ˜alice/notes.txt The following are all the same for user ‘fred’: $ cd $ cd ˜ $ cd /home/fred

Running Programs Programs under Linux are files, stored in directories like /bin and /usr/bin l Run them from the shell, simply by typing their name Many programs take options, which are added after their name and prefixed with - For example, the -l option to ls gives more information, including the size of files and the date they were last modified: $ ls -l drwxrwxr-x 2 free users 4096 Jan 21 10:57 Accounts -rw-rw-r-- 1 free users 345 Jan 21 10:57 notes.txt -rw-r--r-- 1 free users 3255 Jan 21 10:57 report.txt Many programs accept filenames after the options l Specify multiple files by separating them with spaces

Specifying Files with Wildcards Use the * wildcard to specify multiple filenames to a program: $ ls -l *.txt -rw-rw-r-- 1 free users 108 Nov 16 13:06 report.txt -rw-rw-r-- 1 fred users 345 Jan 18 08:56 notes.txt n The shell expands the wildcard, and passes the full list of files to the program n Just using * on its own will expand to all the files in the current directory: $ rm * l (All the files, that is, except the hidden ones) n Names with wildcards in are called globs, and the process of expanding them is called globbing

Graphical and Text Interfaces Most modern desktop Linux systems provide a graphical user interface (GUI) Linux systems use the X window system to provide graphics X is just another program, not built into Linux Usually X is started automatically when the computer boots Linux can be used without a GUI, just using a command line Use Ctrl+Alt+F1 to switch to a text console — logging in works as it does in X Use Ctrl+Alt+F2, Ctrl+Alt+F3, etc., to switch between virtual terminals — usually about 6 are provided Use Ctrl+Alt+F7, or whatever is after the virtual terminals, to switch back to X

Text Editors Text editors are for editing plain text files l Don’t provide advanced formatting like word processors l Extremely important — manipulating text is Unix’s raison d’être The most popular editors are Emacs and Vim, both of which are very sophisticated, but take time to learn Simpler editors include Nano, Pico, Kedit and Gnotepad Some programs run a text editor for you l They use the $EDITOR variable to decide which editor to use l Usually it is set to vi, but it can be changed l Another example of the component philosophy

Shells A shell provides an interface between the user and the operating system kernel Either a command interpreter or a graphical user interface Traditional Unix shells are command-line interfaces (CLIs) Usually started automatically when you log in or open a terminal

The Bash Shell Linux’s most popular command interpreter is called bash l The Bourne-Again Shell l More sophisticated than the original sh by Steve Bourne l Can be run as sh, as a replacement for the original Unix shell n Gives you a prompt and waits for a command to be entered n Although this course concentrates on Bash, the shell tcsh is also popular l Based on the design of the older C Shell (csh)

Shell Commands Shell commands entered consist of words l Separated by spaces (whitespace) l The first word is the command to run l Subsequent words are options or arguments to the command n For several reasons, some commands are built into the shell itself l Called builtins l Only a small number of commands are builtins, most are separate programs

Command-Line Arguments The words after the command name are passed to a command as a list of arguments Most commands group these words into two categories: l Options, usually starting with one or two hyphens l Filenames, directories, etc., on which to operate n The options usually come first, but for most commands they do not need to n There is a special option ‘--’ which indicates the end of the options l Nothing after the double hyphen is treated as an option, even if it starts with -

Syntax of Command-Line Options Most Unix commands have a consistent syntax for options: Single letter options start with a hyphen, e.g., -B Less cryptic options are whole words or phrases, and start with two hyphens, for example --ignore-backups Some options themselves take arguments Usually the argument is the next word: sort -o output_file A few programs use different styles of command-line options

Summary of Bash Editing Keys These are the basic editing commands by default: l Right — move cursor to the right l Left — move cursor to the left l Up — previous history line l Down — next history line l Ctrl+A — move to start of line l Ctrl+E — move to end of line l Ctrl+D — delete current character There are alternative keys, as for the Emacs editor, which can be more comfortable to use than the cursor keys

Working with Text Files Unix-like systems are designed to manipulate text very well The same techniques can be used with plain text, or text-based formats l Most Unix configuration files are plain text Text is usually in the ASCII character set l Non-English text might use the ISO-8859 character sets l Unicode is better, but unfortunately many Linux command-line utilities don’t (directly) support it yet

Filesystem Objects A file is a place to store data: a possibly-empty sequence of bytes A directory is a collection of files and other directories Directories are organized in a hierarchy, with the root directory at the top The root directory is referred to as /

Directory and File Names Files and directories are organized into a filesystem Refer to files in directories and sub-directories by separating their names with /, for example: /bin/ls /usr/share/dict/words /home/jeff/recipe Paths to files either start at / (absolute) or from some ‘current’ directory

File Extensions It’s common to put an extension, beginning with a dot, on the end of a filename The extension can indicate the type of the file: .txt Text file .gif Graphics Interchange Format image .jpg Joint Photographic Experts Group image .mp3 MPEG-2 Layer 3 audio .gz Compressed file .tar Unix ‘tape archive’ file .tar.gz , .tgz Compressed archive file On Unix and Linux, file extensions are just a convention l The kernel just treats them as a normal part of the name l A few programs use extensions to determine the type of a file

Filename Completion Modern shells help you type the names of files and directories by completing partial names Type the start of the name (enough to make it unambiguous) and press Tab For an ambiguous name (there are several possible completions), the shell can list the options: l For Bash, type Tab twice in succession l For C shells, type Ctrl+D Both of these shells will automatically escape spaces and special characters in the filenames

Copying Files with cp n Syntax: cp [ options ] source-file destination-file n Copy multiple files into a directory: cp files directory n Common options: l -f, force overwriting of destination files l -i, interactively prompt before overwriting files l -a, archive, copy the contents of directories recursively

Moving Files with mv mv can rename files or directories, or move them to different directories n It is equivalent to copying and then deleting l But is usually much faster n Options: l -f, force overwrite, even if target already exists l -i, ask user interactively before overwriting files n For example, to rename poetry.txt to poems.txt : $ mv poetry.txt poems.txt n To move everything in the current directory somewhere else: $ mv * ˜/old-stuff/

Deleting Files with rm rm deletes (‘removes’) the specified files You must have write permission for the directory the file is in to remove it Use carefully if you are logged in as root! Options: -f, delete write-protected files without prompting -i, interactive — ask the user before deleting files -r, recursively delete files and directories For example, clean out everything in /tmp , without prompting to delete each file: $ rm -rf /tmp/*

Making Directories with mkdir Syntax: mkdir directory-names Options: -p, create intervening parent directories if they don’t already exist -m mode , set the access permissions to mode For example, create a directory called mystuff in your home directory with permissions so that only you can write, but eveyone can read it: $ mkdir -m 755 ˜/mystuff n Create a directory tree in /tmp using one command with three subdirectories called one , two and three : $ mkdir -p /tmp/one/two/three

Removing Directories with rmdir rmdir deletes empty directories, so the files inside must be deleted first For example, to delete the images directory: $ rm images/* $ rmdir images For non-empty directories, use rm -r directory The -p option to rmdir removes the complete path, if there are no other files and directories in it l These commands are equivalent: $ rmdir -p a/b/c $ rmdir a/b/c a/b a

Standard Files Processes are connected to three standard files n Many programs open other files as well

Standard Input Programs can read data from their standard input file Abbreviated to stdin By default, this reads from the keyboard Characters typed into an interactive program (e.g., a text editor) go to stdin 39

Standard Output Programs can write data to their standard output file Abbreviated to stdout Used for a program’s normal output By default this is printed on the terminal

Pipes n A pipe channels the output of one program to the input of another l Allows programs to be chained together l Programs in the chain run concurrently n Use the vertical bar: | l Sometimes known as the ‘pipe’ character n Programs don’t need to do anything special to use pipes l They read from stdin and write to stdout as normal n For example, pipe the output of echo into the program rev (which reverses each line of its input): $ echo Happy Birthday! | rev

Searching Files with grep grep prints lines from files which match a pattern For example, to find an entry in the password file /etc/passwd relating to the user ‘nancy’: $ grep nancy /etc/passwd grep has a few useful options: l -i makes the matching case-insensitive l -r searches through files in specified directories, recursively l -l prints just the names of files which contain matching lines l -c prints the count of matches in each file l -n numbers the matching lines in the output l -v re v erses the test, printing lines which don’t match

Pattern Matching Use grep to find patterns, as well as simple strings Patterns are expressed as regular expressions Certain punctuation characters have special meanings For example this might be a better way to search for Nancy’s entry in the password file: $ grep ’ˆnancy’ /etc/passwd l The caret (ˆ) anchors the pattern to the start of the line In the same way, $ acts as an anchor when it appears at the end of a string, making the pattern match only at the end of a line

Job Control Most shells offer job control The ability to stop, restart, and background a running process The shell lets you put & on the end of a command line to start it in the background Or you can hit Ctrl+Z to suspend a running foreground job Suspended and backgrounded jobs are given numbers by the shell These numbers can be given to shell job-control built-in commands Job-control commands include jobs, fg, and bg

Jobs n The jobs builtin prints a listing of active jobs and their job numbers: $ jobs [1]- Stopped vim index.html [2] Running netscape & [3]+ Stopped man ls n Job numbers are given in square brackets l But when you use them with other job-control builtins, you need to write them with percent signs, for example %1 n The jobs marked + and - may be accessed as %+ or %- as well as by number l %+ is the shell’s idea of the current job — the most recently active job l %- is the previous current job

What is a Process? The kernel considers each program running on your system to be a process A process ‘lives’ as it executes, with a lifetime that may be short or long A process is said to ‘die’ when it terminates The kernel identifies each process by a number known as a process id, or pid The kernel keeps track of various properties of each process

Process Properties A process has a user id (uid) and a group id (gid) which together specify what permissions it has A process has a parent process id (ppid) — the pid of the process which created it l The kernel starts an init process with pid 1 at boot-up l Every other process is a descendant of pid 1 n Each process has its own working directory, initially inherited from its parent process n There is an environment for each process — a collection of named environment variables and their associated values l A process’s environment is normally inherited from its parent process

Parent and Child Processes n The init process is the ancestor of all other processes: n (Apache starts many child processes so that they can serve HTTP requests at the same time)

About Quoting The shell actually has three different quoting mechanisms: l Single quotes l Backslashes l Double quotes

Quoting: Single Quotes Putting single quotes round something protects it from special interpretation by the shell: $ xmms ’Tom Lehrer - Poisoning Pigeons in the Park.mp3’ $ rm ’b*lls and whistles’ But single quotes (obviously) don’t protect single quotes themselves l So you can’t quote something like She said, "Don’t go." with only single quotes

Quoting: Backslashes n You can put a backslash \ in front of any single character to turn off its special meaning: $ echo M\&S $ xmms Suzanne\ Vega\ -\ Tom\’s\ Diner.mp3 $ mail -s C:\\MSDOS.SYS

Quoting: Double Quotes Putting double quotes round something protects most things within it from interpretation by the shell A dollar sign $ retains its special interpretation As do backticks “ can’t be escaped in double quotes A backslash can be used within double quotes to selectively disable the special interpretation of $, ‘ and \: $ mail -s "C:\\MSDOS.SYS" windows-user@example.com $ echo "It cost $price US\$" n Putting a backslash in front of anything else gives you both characters: $ echo "\*/"

Shell Programming The shell is designed to be both: A convenient environment to type commands into A simple programming language Any command that can be typed at the command line can be put into a file Programming features include variables, loops (including for), and even shell functions The Unix component approach makes it very easy to write shell scripts to perform fairly complex tasks Common application domains for shell scripting include: Text processing Automation of system administration tasks

Filesystems Some confusion surrounds the use of the term ‘filesystem’ Commonly used to refer to two distinct concepts 1. The hierarchy of directories and files which humans use to organise data on a system (‘unified filesystem’) 2. The formatting system which the kernel uses to store blocks of data on physical media such as disks (‘filesystem types ’)

Symbolic Links A symbolic link (or symlink) is a pseudo-file which behaves as an alternative name for some other file or directory The ‘contents’ of the symlink are the real name pointed to When you try to use a file name including a symlink, the kernel replaces the symlink component with its ‘contents’ and starts again Symlinks allow you to keep a file (or directory) in one place, but pretend it lives in another For example, to ensure that an obsolete name continues to work for older software Or to spread data from a single filesystem hierarchy over multiple disk partitions

Examining and Creating Symbolic Links ls -l shows where a symbolic link points to: $ ls -l /usr/tmp lrwxrwxrwx 1 root root 30 Sep 26 2000 /usr/tmp -> /var/tmp n ls can also be made to list symlinks in a different colour to other files, or to suffix their names with ‘@’ A symlink is created with the ln -s command Its syntax is similar to cp — the original name comes first, then the name you want to create: $ ln -s real-file file-link $ ln -s real-dir dir-link $ ls -l file-link dir-link lrwxrwxrwx 1 bob bob 9 Jan 11 15:22 file-link -> real-file lrwxrwxrwx 1 bob bob 8 Jan 11 15:22 dir-link -> real-dir

Hard Links Where symlinks refer to other files by name, a hard link refers to another file by inode number l An inode is the data structure that describes a file on disk l It contains information about the file, including its type (file/directory/device), modification time, permissions, etc. n A directory entry contains a name and an inode number l So a file’s name is not considered to be part of the file itself n You get a hard link when different directory entries on a filesystem refer to the same inode number

Symlinks and Hard Links Illustrated A symbolic link refers to filename, which in turn refers to an inode: A hard link is a normal directory entry, referring directly to an inode:

Examining and Creating Hard Links Use the ln command to create a hard link Don’t use the -s option when creating hard links n As when creating symlinks, the order of the arguments to ln mimics cp: $ ls -l *.dtd -rw-r--r-- 1 anna anna 11170 Dec 9 14:11 module.dtd $ ln module.dtd chapter.dtd $ ls -l *.dtd -rw-r--r-- 2 anna anna 11170 Dec 9 14:11 chapter.dtd -rw-r--r-- 2 anna anna 11170 Dec 9 14:11 module.dtd n Notice that the link count in the listing increases to 2 n The two names are now indistinguishable l Deleting or renaming one doesn’t affect the other

Users and Groups Anyone using a Linux computer is a user The system keeps track of different users, by username Security features allow different users to have different privileges Users can belong to groups, allowing security to be managed for collections of people with different requirements Use su to switch to a different user Quicker than logging off and back on again su prompts you for the user’s password: $ su - bob Password: The - option makes su behave as if you’ve logged in as that user

The Superuser: Root Every Linux system has a user called ‘root’ The root user is all-powerful Can access any files The root user account should only be used for system administration, such as installing software When logged in as root, the shell prompt usually ends in # Usually best to use su for working as root: $ whoami fred $ su - Password: # whoami root

Changing File Ownership with chown The chown command changes the ownership of files or directories Simple usage: # chown aaronc logfile.txt Makes logfile.txt be owned by the user aaronc Specify any number of files or directories Only the superuser can change the ownership of a file This is a security feature — quotas, set-uid

Changing File Group Ownership with chgrp The chgrp command changes the group ownership of files or directories Simple usage: # chgrp staff report.txt Makes staff be the group owner of the file logfile.txt As for chown, specify any number of files or directories The superuser may change the group ownership of any file to any group The owner of a file may change its group ownership But only to a group of which the owner is a member

Basic Concepts: Permissions on Files Three types of permissions on files, each denoted by a letter A permission represents an action that can be done on the file: Permission Letter Description Read r Permission to read the data stored in the file Write w Permission to write new data to the file, to truncate the file, or to overwrite existing data Execute x Permission to attempt to execute the contents of the file as a program n Occasionally referred to as ‘permission bits’ n Note that for scripts, you need both execute permission and read permission l The script interpreter (which runs with your permissions) needs to be able to read the script from the file

Basic Concepts: Permissions on Directories The r, w, x permissions also have a meaning for directories The meanings for directories are slightly different: Permission Letter Description Read r Permission to get a listing of the directory Write w Permission to create, delete, or rename files (or subdirectories) within the directory Execute x Permission to change to the directory, or to use the directory as an intermediate part of a path to a file The difference between read and execute on directories is specious — having one but not the other is almost never what you want

Concepts: Disks and Partitions A hard disk provides a single large storage space Usually split into partitions Information about partitions is stored in the partition table Linux defaults to using partition tables compatible with Microsoft Windows For compatibility with Windows , at most four primary partitions can be made But they can be extended partitions, which can themselves be split into smaller logical partitions Extended partitions have their own partition table to store information about logical partitions

Disk Naming The device files for IDE hard drives are /dev/hda to /dev/hdd hda and hdb are the drives on the first IDE channel, hdc and hdd the ones on the second channel The first drive on each channel is the IDE ‘master’, and the second is the IDE ‘slave’ Primary partitions are numbered from 1–4 Logical partitions are numbered from 5 The devices /dev/hda , etc., refer to whole hard disks, not partitions Add the partition number to refer to a specific partition

Using fdisk The fdisk command is used to create, delete and change the partitions on a disk Give fdisk the name of the disk to edit, for example: # fdisk /dev/hda fdisk reads one-letter commands from the user Type m to get a list of commands Use p to show what partitions currently exist Use q to quit without altering anything Use w to quit and write the changes Use with caution, and triple-check what you’re doing!

Making New Partitions Create new partitions with the n command Choose whether to make a primary, extended or logical partition Choose which number to assign it fdisk asks where to put the start and end of the partition The default values make the partition as big as possible The desired size can be specified in megabytes, e.g., +250M Changes to the partition table are only written when the w command is given

Changing Partition Types Each partition has a type code, which is a number The fdisk command l shows a list of known types The command t changes the type of an existing Partition Enter the type code at the prompt Linux partitions are usually of type ‘Linux native’ (type 83) Other operating systems might use other types of partition, many of which can be understood by Linux

Making Filesystems with mkfs The mkfs command initializes a filesystem on a new partition Warning: any old data on the partition will be lost For example, to make an ext2 filesystem on /dev/hda2: # mkfs -t ext2 -c /dev/hda2 -t sets the filesystem type to make, and -c checks for bad blocks on the disk mkfs uses other programs to make specific types of filesystem, such as mke2fs and mkdosfs

Mounting Filesystems As far as many parts of a Linux system are concerned, a partition contains entirely arbitrary data When installing, you set things up so that a partition contains a filesystem — a way of organising data into files and directories n One filesystem is made the root filesystem: the root directory on that filesystem becomes the directory named / Other filesystems can be mounted: the root directory of that filesystem is grafted onto a directory of the root filesystem This arranges for every file in every mounted filesystem to be accessible from a single unified name space n The directory grafted onto is called the mount point

Mounting a Filesystem: mount ‘ Important’ filesystems are mounted at boot-up; other filesystems can be mounted or unmounted at any time The mount command mounts a filesystem You usually need to have root permission to mount a filesystem mount makes it easy to mount filesystems configured by the system administrator For example, many systems are configured so that $ mount /mnt/cdrom will mount the contents of the machine’s CD-ROM drive under the directory /mnt/cdrom

Mounting Other Filesystems mount /dev/sdb3 /mnt/extra mounts the filesystem stored in the /dev/sdb3 device on the mount point /mnt/extra You may occasionally need to specify the filesystem type explicitly: # mount -t vfat /dev/hdd1 /mnt/windows l Allowable filesystem types are listed in the mount(8) manpage n To see a list of the filesystems currently mounted, run mount without any options

Unmounting a Filesystem: umount A filesystem can be unmounted with umount umount /mnt/extra unmounts whatever is on the /mnt/extra mount point umount /dev/sdb3 unmounts the filesystem in the /dev/sdb3 device, wherever it is mounted You normally need to have root permission to unmount a filesystem It’s also impossible to unmount a ‘busy’ filesystem A filesystem is busy if a process has a file on it open Or if a process has a directory within it as its current directory

Configuring mount: /etc/fstab The /etc/fstab file contains information about filesystems that are known to the system administrator Specifying a filesystem in /etc/fstab makes it possible to use its mount point as the only argument to mount /etc/fstab also configures which filesystems should be mounted at boot-up Each line in /etc/fstab describes one filesystem Six columns on each line

Sample /etc/fstab A sample /etc/fstab file: # device mount-point type options (dump) pass-no /dev/hda3 / ext2 defaults 1 1 /dev/hda1 / boot ext2 defaults 1 2 /dev/hda5 / usr ext2 defaults 1 2 /dev/hdb1 / usr/local ext2 defaults 1 2 /dev/hdb2 / home ext2 defaults 1 2 none / proc proc defaults 0 0 /dev/scd0 / mnt/cdrom iso9660 noauto, users, 0 0 /dev/fd0 /mnt/floppy auto noauto, users 0 0

Filesystem Types The most common filesystem types are: Type Usage ext2 The standard Linux filesystem iso9660 The filesystem used on CD-ROMs proc Not a real filesystem, so uses none as the device. Used as a way for the kernel to report system information to user processes vfat The filesystem used by Windows 95 auto Not a real filesystem type. Used as a way of asking the mount command to probe for various filesystem types, particularly forremovable media Networked filesystems include nfs (Unix-specific) and smbfs (Windows or Samba) Other, less common types exist; see mount(8)

Mount Options Comma-separated options in /etc/fstab Alternatively, use comma-separated options with -o on the mount command line Common mount options: Option Description noauto In /etc/fstab , prevents the filesystem being mounted at bootup. Useful for removable media ro Mount the filesystem read-only users Let non-root users mount and unmount this filesystem user Like users, but non-root users can only unmount filesystems that they themselves mounted n Other less common mount options exist, as well as many options for individual filesystem types

Filesystem Concepts The files stored on a disk partition are organised into a filesystem There are several filesystem types; the common Linux one is called ext2 A filesystem contains a fixed number of inodes An inode is the data structure that describes a file on disk It contains information about the file, including its type (file/directory/device), modification time, permissions, etc. A file name refers to an inode, not to the file directly This allows hard links: many file names referring to the same inode

Monitoring Space: df Run df with no arguments to get a listing of free space on all mounted filesystems Usually better to use the -h option, which displays space in human-readable units: $ df -h Filesystem Size Used Avail Use% Mounted on /dev/hda8 248M 52M 183M 22 % / /dev/hda1 15M 5.6M 9.1M 38% /boot /dev/hda6 13G 5.0G 7.4G 41% /home /dev/hda5 13G 4.6G 7.8G 37% /usr /dev/hda7 248M 125M 110M 53% /var n The Use% column shows what percentage of the filesystem is in use n You can give df directories as extra arguments to make it show space on the filesystems those directories are mounted on

Monitoring Inodes: df Filesystems rarely run out of inodes, but it would be possible if the filesystem contains many small files Run df -i to get information on inode usage on all mounted filesystems: $ df -i Filesystem Inodes I Used I Free Use% Mounted on /dev/hda8 65736 8411 57325 13% / /dev/hda1 4160 30 4130 1% /boot /dev/hda6 1733312 169727 1563585 10% /home /dev/hda5 1733312 138626 1594686 8% /usr /dev/hda7 65736 1324 64412 2% /var n In this example, every filesystem has used a smaller percentage of its inodes (IUse%) than of its file space

Monitoring Disk Usage: du df shows a summary of the free space on a partition du, on the other hand, shows information about disk space used in a directory tree Takes one or more directories on the command line: $ du /usr/share/vim 2156 /usr/share/vim/vim58/doc 2460 /usr/share/vim/vim58/syntax 36 /usr/share/vim/vim58/tutor 16 /usr/share/vim/vim58/macros/hanoi 16 /usr/share/vim/vim58/macros/life 40 /usr/share/vim/vim58/macros/maze 20 /usr/share/vim/vim58/macros/urm 156 /usr/share/vim/vim58/macros 100 /usr/share/vim/vim58/tools 5036 /usr/share/vim/vim58 5040 /usr/share/vim

du Options Option Description -a Show all files, not just directories -c Print a cumulative total for all directories named on the command line -h Print disk usage in human-readable units -s Print only a summary for each directory named on the command line -S Make the size reported for a directory be the size of only the files in that directory, not the total including the sizes of its subdirectories

Finding and Repairing Filesystem Corruption: fsck Sometimes filesystems do become corrupted Perhaps there was a power failure Or maybe your kernel version has a bug in it The fsck program checks the integrity of a filesystem And can make repairs if necessary Actually has two main parts: A ‘driver program’, fsck, which handles any filesystem type One ‘backend program’ for each specific filesystem type The backend program for ext2 is e2fsck, but it is always invoked through fsck

Running fsck fsck is normally run at system startup So it gets run automatically if the system was shut down uncleanly n It can also be run manually: # fsck /dev/sdb3 Interactively asks whether to fix problems as they are found Use -f to force checking the filesystem, even if fsck thinks it was cleanly umounted Use -y to automatically answer ‘yes’ to any question Usually a bad idea to run fsck on a mounted filesystem!

Unix Filesystem Layout Many common directory names are abbreviated versions of real words Traditional structure which has developed over many years Most system files have their proper place Programs rely on them being in the correct place Users familiar with Unix directory structure can find their way around any Unix or Linux system But a user’s home directory can be structured however they want

The Filesystem Hierarchy Standard Started as an attempt to standardise Linux filesystem layouts Called the FSSTND when the first version was published in 1994 Widely accepted by distributors But few Linux systems are 100% compliant yet Intended to avoid fragmentation of Linux distributions Renamed to the File Hierarchy Standard, or FHS Now intended to apply to all Unix-like operating systems

Static and Dynamic Data Some files hardly ever need to be changed, while others change all the time It can help to store static files separately from those which regularly change: The static files can be on a partition mounted read-only (such as a CD-ROM) Programs and libraries are usually static (except when new software is installed) Home directories and status files are usually more variable

FHS: Installed Software Programs are usually found in the bin and sbin directories These are found in / , /usr and /usr/local sbin is used for programs likely to be useful to system administrators rather than to general users (mail dæmon, web server, etc.) These directories are named after binaries Most programs in them are binaries (compiled programs), although some are human-readable scripts n Libraries are stored in directories called lib , found in the same places as bin l These directories should be listed in /etc/ld.so.conf

FHS: Other Directories Under /usr /usr/include contains header files used by C/C++ programs /usr/X11R6 contains files used by the X Window system, including programs, libraries, configuration files and documentation /usr/local is where software is installed when it is compiled from source code rather than installed as a package /usr/share contains files which are not specific to the architecture of a machine, e.g., fonts and icons Theoretically could be shared between different types of machine over a network n /usr/src often contains the source code for the Linux kernel l Usually kept in a directory such as linux-2.2.20 , with a symbolic link to it called linux

FHS: Directories Under /var /var/run contains pid files (process-id files for currently-running dæmon programs) Also contains utmp , a record of user logins /var/mail or /var/spool/mail is where each user’s email is queued up until it is deleted or saved /var/log contains logs produced by various programs, including syslog /var/cache contains data generated by programs which is cached to save time Cached data can be regenerated if deleted

FHS: Other Directories /proc provides access to information from the kernel, particularly about running processes /home contains directories which belong to each user Use echo ˜ to find out where your home directory is /root is the home directory of the root user

The type Built-in Command type is like which, but is built into the shell It tells you about shell aliases and functions Not available in the C Shell type -p is just like which type -a shows all the commands of the name given Useful for detecting duplicate programs, or aliases which are hiding real programs See help type for full details

Updating the locate Database Use the updatedb program to refresh the database of files used by locate Modern versions are configured by giving options to updatedb -e provides a list of directories which will not be searched -f gives the names of filesystem types to miss out See the manpage for full details updatedb is usually run by cron every night l Look in /etc/cron.daily for the script which runs it

updatedb.conf Older versions of GNU updatedb used the configuration file /etc/updatedb.conf For compatibility, some modern versions still read it The configuration is done by setting environment variables For example, to ignore certain filesystems: $ PRUNEPATHS="/tmp /usr/tmp /var/tmp /mnt /var/spool" $ export PRUNEPATHS The $PRUNEFS variable lists the names for filesystems which should be ignored (e.g., nfs,iso9660, etc.) These variables are equivalent to the -e and -f options

What are Quotas? Quotas are a way of limiting the amount of disk space that users may take up Some organisations (perhaps those with untrusted external users) absolutely need to ensure that: No user can prevent other users from using a reasonable amount of disk space No user can impede the correct functioning of the system Some organisations don’t need to worry about this — their users can be trusted not to abuse the system Unfortunately, quota management is unnecessarily hard on Linux Could user education avoid the need for quotas? Disk space is cheap!

Displaying Quota Limits: quota The quota command displays quota limits Specifying a username or the name of a group will show information about their quotas: # quota fred n The -v option will show full information about all quotas, even where there are no limits

Options in /etc/fstab The options in /etc/fstab specify which filesystems should have quota enabled Add the option usrquota to enable user quotas Use grpquota to enable group quotas Either or both can be used for each filesystem: /dev/hda1 / ext2 defaults /dev/hdb1 /home ext2 defaults,usrquota /dev/hdb2 /work/shared ext2 defaults,usrquota,grpquota The filesystems with quota enabled should have files called quota.user and quota.group in their root directories The following commands will create them: # touch /partition/quota.{user,group} # chmod 600 /partition/quota.{user,group}

edquota edquota allows quotas to be edited interactively, in a text editor The file in the text editor will be a temporary file edquota will read it back in when the editor terminates Use the -g option to edit group quotas Some versions of Red Hat have a bug where you need to delete an extraneous space before the time unit when doing edquota -t

repquota The repquota command prints information about the quota limits assigned to each user Also shows actual number of blocks and inodes used Use the -a option for information on all filesystems, or specify the filesystem on the command line n Use -g to show group quotas n Use -v for more complete information

Boot Loaders When Linux boots, the kernel is loaded into memory by a boot loader Passes parameters to the Linux kernel Allows one of several operating systems to be loaded Multiple versions of the Linux kernel Dual-booting with Windows and other OSes The most popular boot loader is LILO (the Linux loader) Full user documentation provided Look for the a directory called something like /usr/share/doc/lilo/ or /usr/doc/lilo-0.21/

LILO LILO runs when the system is booted The lilo command configures how LILO will next run The file /etc/lilo.conf specifies the configuration to be set by the lilo command Need to run the lilo command for changes to have affect Manual page lilo.conf(5) lilo.conf has options in the form name = value Options for specifix OSes are indented Linux kernels to install are introduced with image= Other OSes are introduced with other=

Sample lilo.conf File boot = /dev/hda # put loader on the MBR root = /dev/hda1 # device to mount as / delay = 40 # 4 second delay compact # may make booting faster read-only # needed to allow root to be fscked image = /vmlinuz-2.2.20 # stable kernel (default because it’s 1st) label = linux-2.2.20 alias = linux # shorter label vga = ask # let us choose the console size image = /vmlinuz-2.5.1 # cutting edge kernel label = linux-2.5.1 other = /dev/hda3 # Windows is installed on a different partition label = windows table = /dev/hda

Selecting What to Boot When LILO runs it displays the prompt LILO: If only some of the letters appear, the boot process failed at some point It waits the specified delay for something to start being typed Load a particular kernel or OS by entering its label or alias Just press Enter for the default Pressing Tab lists the available labels Some versions of LILO present a menu to select from with the cursor keys If no key has been pressed by the end of the delay, the first kernel or OS is loaded

Other Ways of Starting Linux Grub — complex boot loader which includes a shell and support for accessing filesystems LoadLin — a Dos program which can start Linux from within Dos Occasionally used to start Linux after a Dos driver has configured some hardware

Boot Messages When the kernel starts up it prints a flurry of information This can often be useful in finding problems A log of this information is kept in /var/log/dmesg The dmesg command can print the most recent messages This can show problems which have occurred since boot After boot, most log messages are handled by syslog

Understanding Runlevels A Linux system runs in one of several runlevels — modes providing different features and levels of functionality Linux systems normally have seven runlevels, numbered from 0–6: Three are mandatory (0 = halt, 6 = reboot, 1 = single-user) Four are user-defined (2–5) No consensus between administrators or distributions about how to organise the user-defined runlevels Some rely (partly) on runlevels to define which major subsystems are running Others prefer the flexibility of starting and stopping subsystems individually, without changing runlevel l In every common distribution, there is at least one user-defined runlevel which has the same services as another

Typical Runlevels Runlevel Description 0 A ‘transitional’ run-level, used to tell the system to shut itself down safely. Once the system has shut down, it needs a manual reboot to reactivate. 1 Single-user mode, used for maintenance. Users may not log in, and any services (usually including all networking facilities) are disabled. There is only one terminal active, on which root is automatically logged in. 2–5 Multi-user modes. Some systems make all of these identical. Others disable networking (or NFS file-sharing) in runlevel 2, and/or enable a graphical login in runlevel 5 (but not in other runlevels). 6 A ‘transitional’ run-level, used to tell the system to reboot.

Single-User Mode and sulogin Many Linux distributions use a program called sulogin to restrict access to single-user mode sulogin is run when the system enters single-user mode It requires the root password to be typed on the console before switching to single-user mode If the password is not typed, sulogin returns the system to the normal runlevel Why is sulogin needed? Untrusted users may have access to the system’s keyboard during bootup In many configurations, this would enable them to boot the system up in single-user mode

Shutting Down and Restarting the System To safely shut down the system, run the halt command as root This is a safe shutdown: it stops all services, disables all network interfaces, and unmounts all filesystems To safely reboot the system, run reboot as root Most systems also let you hit Ctrl+Alt+Del on the console Alternatively, the shutdown command allows you to schedule the power-down or reboot, to give users warning to save their work l Halt at 6pm: # shutdown -h 18:00 l Reboot thirty minutes from now:

Setting the Default Runlevel The system’s default runlevel on bootup is configured in /etc/inittab To configure a default runlevel of 3, /etc/inittab should contain the line: id:3:initdefault There should be only one initdefault line in /etc/inittab

Selecting a Different Runlevel at Bootup Most bootloaders (including LILO) give you the ability to type in a kernel command line Naming a runlevel on the kernel command line selects that runlevel for use on system startup To start in single-user mode: linux 1 To start in emergency mode: linux -b Emergency mode provides nothing beyond a shell to type into — useful for repairing serious system corruption

Manual Pages Most Linux commands have an associated manual page Often known as a manpage Viewed with the man command: $ man ls Press q to quit and return to the shell prompt

Navigating Within Manual Pages man uses the less viewer Use the cursor keys for scrolling Other common keystrokes: space jump down a page b jump back up a page / word search for the next occurrence of “word” n repeat the previous search g go to the top

Sections of a Manual Page Manpages have a traditional format Manpages for user commands tend to have some or all of these sections: NAME — name and single-line reason for the command SYNOPSIS — possible arguments DESCRIPTION — fuller explanation of the command OPTIONS FILES — any files the command needs ENVIRONMENT — pertinent environment variables BUGS AUTHOR SEE ALSO

syslog Many events that occur on a Linux system should be logged for administrative purposes Linux has a facility called syslog that allows any service or part of the system to log such events syslog can be configured to log different events to different places Events can be selected based on severity (‘level’) and/or on the service that encountered the event (‘facility’) Messages can go to files, to the system console, or to a centralised syslog server running on another machine

/etc/syslog.conf syslog’s configuration is in /etc/syslog.conf ; each line looks like: facility.level destination The facility is the creator of the message — one of auth, authpriv, cron, daemon, kern, lpr, mail, news, syslog, user, or local0 through local7 The level is a severity threshold beyond which messages will be logged — one of (from lowest to highest): debug, info, notice, warning, err, crit, alert, emerg The destination indicates where messages selected by the facility and level will be sent Normally the name of a log file (under /var/log ), or /dev/console to send messages to the system console

Examining Logs: less and grep You sometimes need to manually scan log files for notable activity Since logs are plain text, you can use standard text-processing tools like to examine them To review the entire contents of a log file: # less /var/log/messages Note: you may need to be root to do this To look for messages on a certain topic: # grep -i sshd /var/log/messages

At Commands At commands are defined using at: $ at 16:30 at> pstree > processes at> <EOT> The time the command should run is given as a parameter to at at then prompts for the command itself Command(s) exactly as they would be typed in the shell Press Ctrl+D to finish The at dæmon will run the command at the specified time In this example, the output of running pstree at 16:30 will be saved in the file processes

Managing At Commands atq lists any pending at commands: $ atq 38 2002-01-16 11:00 a The number at the start of each line identifies that at command A particular at command can be displayed with at -c: $ at -c 38 #!/bin/sh umask 2 cd /home/simon || { echo ’Bad directory’ >&2; exit 1 } echo ’Check the download has completed.’ l Real at commands include the environment too

Simple Cron Job Specification The simplest method for specifying cron jobs is to save each job as a separate file in an appropriate directory: /etc/cron.daily/ is for jobs to be run daily Once a day, each file in that directory is run The files are typically shell scripts There are equivalent directories for monthly, weekly, and possibly hourly jobs Each job is run with root permissions Normally only root can set up cron jobs this way Any required environment variables must be set explicitly

Crontab Format Blank lines are ignored Comments are lines starting with a hash (#) Environment variables can be set: PATH=/usr/local/bin n Example cron job specification 30 9 * * * root /usr/local/bin/check_logins l At 09:30 l On all days l For the root user

/etc/crontab The /etc/crontab file is an older way of specifying cron jobs Each job in that file is like a file from /etc/cron.d/ Having many unrelated cron jobs in a single file is much harder to manage This file may be the mechanism by which your system runs the contents of /etc/cron.daily/ and Friends n There is no need to use this file for anything else

User Crontabs Sometimes non-root users need to set up cron jobs Each user has a crontab file This is not edited directly The crontab command manipulates it Use crontab -e option to edit the crontab The editor in the $EDITOR variable is invoked for this Use crontab -l to display the crontab The format is very similar to that of /etc/rc.d/ crontabs But there is no username field All commands run as the owner of the crontab

Cron Job Output Cron jobs do not run in a terminal window Generally they are administrative tasks designed not to produce any output when run successfully Any output that is generated by a cron job is mailed: The recipient can be specified in the $MAILTO environment variable Otherwise mail is sent to the job’s owner Jobs in /etc/cron.daily et al are owned by root

Linux Administration

More Related Content

What's hot (20)

Viewers also liked (20)

Similar to Linux Administration (20)

Recently uploaded (20)

Linux Administration