Unix Tutorial

A Sample Login Session

The IMSS UNIX Cluster is available for logins from the Student Lab (see the Getting Started on the IMSS UNIX Cluster and the Sun Console Guide) or over the campus network or dial-up lines.

The Cluster has several computer which can be logged into from the network of the dial-up lines. We recommend that people connect to ssh.caltech.edu, using either ssh or telnet, rather than using the name of a specific server. The name its rotates between the servers to distribute the load.


Logging On

When you first connect to one of the UNIX computers you will see the prompt:




If you see only the prompt Password: you probably used rlogin. rlogin assumes that your username is the same on all computers and enters it for you. If your username is different, don't worry, just press -- short for "carriage return": the Return or Enter key on your keyboard -- until you see the login: prompt and start from scratch.

At the login: prompt, type in your username. Be careful to type only lower case! The UNIX operating system is "case sensitive." If you type your username in mixed case (Rarmour rather than rarmour, for example) the computer will not recognize it.


Your Password

Once you have typed in your username you will be prompted to type in your password. Type carefully! It won't be displayed on the screen.

When you first log in, you should change your password.

In the interests of self-preservation, don't set your password to any information which people are likely to know about you (your real name, your nickname, your pet dog's name). Because there are programs that people can run to guess at your password, you should not make it relate to any word or name. For more information about choosing a good password, as well as how to change your password using the new IMSS Web Utilities System, see the IMSS reference Password Security Tips.

If you mistype your username or password you will get a suspicious message from the computer and see the login: prompt again.


The motd

If you type your username and password correctly, the computer will begin running the login program. It starts by displaying a special message of the day contained in the /etc/motd file. This file will usually contain information about the computer you are logging onto, maybe a basic message about getting help, and any important system messages from the system manager.


The announce Program

Next the announce program will display any brief one-time messages which might be of interest to cluster users. announce is run by your .login.


Initialization Files

When you log in, the UNIX login program starts up a command "shell." Users do not deal with the operating system directly. Instead they interact with a shell, which is initialized with several pieces of information such as your username, home directory and "path." By default all users use the C-shell (the program /bin/csh) and interact with it.

There are a couple of files read by this shell when your login session starts up. These are the .cshrc file and the .login file. These files are created when your account is created. As you learn more about how UNIX and the C-shell work, you may want to customize these files.

If your files get corrupted for some reason, you can run the ccoconfig program to get new copies of the default files given to all new accounts.


Using the System

Finally you are logged in! You will see a prompt like one of the following three:






just waiting for you to type something. Throughout the UNIX Tutorial we will use % to indicate the computer's "ready" prompt.



Okay, let's try a simple command. Type ls and press . ls lists files in a directory. Right now you may or may not see any files, but not seeing any files doesn't mean that you don't have any! ls by itself won't list hidden files (files whose names start with ".", like .login). Now try typing:


% ls -a


Don't actually type the % symbol! Remember, that's the computer's prompt which indicates it is ready to accept input. The spacing should be exactly as shown: ls followed by a space, followed by -a. The -a is a "flag" which tells the ls program to list all files.

For more about command flags see below.



Just for fun, let's look at the contents of another directory, one with lots of files. Directory names in UNIX are straightforward. They are all arranged in a tree structure from the root directory "/".

For now, use cd to change your directory to the /bin directory. Type:


% cd /bin


and press . Now type ls again. You should see a long list of files. In fact, if you look carefully you will see files with the names of the commands we've been typing (like ls and cd). Note that the /bin in the command we typed above was not a flag to cd. It was a "parameter." Flags tell commands how to act; parameters tell them what to act on.

Now return to your home directory with:


% cd


Entering cd with no parameter returns you to your home directory. You can check to make sure that it worked by entering:


% pwd


which prints your current (or "working") directory. The computer will return a line of words separated by "/" symbols which should look something like:




Whatever it returns, the list should end in your username.


Using the On-Line Manual Pages

Most UNIX commands have very short and sometimes cryptic names like ls. This can make remembering them difficult. Fortunately there are on-line manual pages which allow you to display information on a specific program (to list all the flags of ls, for example) or list all the information available on a certain topic.



To investigate other flags to the command you are interested in type:

% man command


For example, to investigate other flags to the ls command (such as which flags will display file size and ownership) you would type man ls.


man -k

The second way of using the on-line manual pages is with:

% man -k topic


In this case you use a word to describe the topic you are interested in or that you might expect to be in a one-line description of the command you wish to find. For example, to find a program which "lists directory contents" you might type man -k dir. Partial words can be used and this is one of the few places in UNIX where upper and lower case are allowed to match each other.


Using man and more

Try it now. Use man ls to find out how to make the ls program print the sizes of your files as well as their names. After typing man ls and pressing , note how man displays a screenful of text and then waits with a prompt --More-- at the bottom of the screen.

What man is doing is sending everything it wants to display on the screen through a program known as a "pager." The pager program is called more. When you see --More-- (in inverse video) at the bottom of the screen, just press the space bar to see the next screenful. Press to scroll a line at a time.

Have you found the flag yet? The -s flag should display the size in kilobytes. You don't need to continue paging once you have found the information you need. Press q and more will exit.


Listing File Sizes

Now type ls -as. You can stack flags together like this. The command ls -as lists all files, and lists their sizes in kilobytes.


Logging Off

When you are finished you should be sure to log out! You need to be careful that you've typed logout correctly. The UNIX operating system is not forgiving of mistyped commands. Mistyping logout as "logotu", pressing return and then leaving without glancing at the screen can leave your files at anyone's mercy.


The UNIX Shell Syntax

As mentioned earlier, user commands are parsed by the shell. There are many shells other than the the C-shell which allow different types of shortcuts. We will only discuss the C-shell here, but some alternate shells installed on the IMSS UNIX Cluster include the Bourne shell (/bin/sh), the Bourne-Again Shell (bash), zsh and tcsh (a C-shell variant). The last three all reside in /usr/local/bin. While you are welcome to experiment with any of these shells, realize that C-shell is our default shell and that some IMSS UNIX Cluster utilities assume you are using it (and may not work if you're not).


Changing Your Shell

To change your current shell to another use the chsh command (for change shell). You will then be prompted for your password and the new shell's path (e.g. /usr/local/bin/tcsh for the tcsh shell). After enter ing these you will be informed that the login shell has been changed. To change to this shell logout and then re-login. For more information on changing your shell see the Changing Shell FAQ page.


The Path

One of the most important elements of the shell is the path. Whenever you type something at the % prompt, the C-shell first checks to see if this is an alias you have defined and, if not, searches all the directories in your path to determine the program to run.

The path is just a list of directories, delimited by colons, which are searched in order. Your default .cshrc will have a path defined for you. If you want other directories (such as a directory of your own programs) to be searched for commands, add them to your path by editing your .cshrc file. This list of directories is stored in the PATH environment variable. We will discuss how to manipulate environment variables later.


Flags and Parameters

Most commands expect or allow parameters (usually files or directories for the command to operate on) and many provide option flags. A flag, as we saw before, is a character or string with a - before it such as the -s we used with the ls command.

Some commands, such as cp and mv require file parameters. Not surprisingly, cp and mv (the copy and move commands) each require two: one for the original file and one for the new file or location.

It would seem logical that if ls by itself just lists the current directory then cp filename should copy a file to the current directory. Instead you must enter

% cp filename .


where the "." tells cp to place the file in the current directory. filename in this case would be a long filename with a complete directory specification.

Not surprisingly, ls . and ls are almost the same.


Directory and File Structure


Directories in UNIX start at the root directory "/". Files are fully specified when you list each directory branch needed to get to them.





Your Home Directory

A home directory can always be specified with ~username (~ is commonly called "twiddle," derived from the proper term "tilde.") If you needed to list files in someone else's home directory, you could do so by issuing the command:


% ls ~username


substituting in their username. You can do the same with your own directory if you've cd'd elsewhere. Please note: many people consider looking at their files an invasion of privacy, even if the files are not protected. Just as some people leave their doors unlocked but do not expect random passers-by to walk in, other people leave their files unprotected without intending to invite browsers.



If you have many files or multiple things to work on, you probably want to create subdirectories in your home directory. This allows you to place files which belong together in one distinct place.


Creating Subdirectories

The program to make a subdirectory is mkdir. If you are in your home directory and wish to create a directory, type the command:


% mkdir directory-name


Once this directory has been created you can copy or move files to it (with the cp or mv commands) or you can cd to the directory and start creating files there.


File Manipulation


Copying Files

To copy a file use the cp command, specifying the name of the file you wish to be copied and the filename you wish to give the copied file:


% cp filename copy-filename


You can copy a file from the current directory into another subdirectory by entering:


cpfilename directory-name
copy file, filename will be the same as original
cpfilename directory-name/new-filename
copy file, give it a new name


Or cd into the new directory and move the file from elsewhere:


% cd directory-name

% cp ../filename .


copies the file from the directory above (represented by "..") to the current directory (represented by "."), giving it the same filename.


Moving Files

Files can be renamed or moved to different subdirectories using the mv command. The mv command works in a similar fashion to the cp command:


mvold-filename new-filename
renames original file with new-filename
mvfilename directory-name/filename
move file to another directory, keeping the same filename
mvfilename directory-name/new-filename
move file to another directory and renaming it new-filename



Removing Files and Directories

To remove files use the rm command, specifying the name of the file you want to remove (and the path if it is in another directory):

removes file from current directory
removes file from another directory
removes file from directory above


To remove a directory the command rmdir is used. NB: This command will only work if the directory speficied is empty. To remove a directory with files in it use the command rm -r directory-name (for recursive).


Interactive File Handling

To avoid unwanted deletion of precious files the mv and rm commands can be made interactive using the -i flag. For example rm -i filename would return a prompt asking if you are certain you want to delete tha t file. We recommend that you alias these commands to be interactive in your .cshrc file. For more help on aliasing see the aliases section later.


Finding Files

You can search for files using the find command, using the -name flag. The directory you wish to be searched must also be specified.

% find . -name myfile
looks for the file myfile in the current directory
% find directory/ -name myfile
looks for the file myfile in the directory given


File Names

Unlike other operating systems, filenames are not broken into a name part and a type part. Names can be many characters long and can contain most characters. Some characters such as * and ! have special meaning to the shell. They should not be used in filenames. If you ever do need to use such a symbol from the shell, they must be specified sneakily, by "escaping" them with a backslash (\\). For example:

% rm \\!badfile


C-shell would have interpreted rm !badfile differently. In that case, !badfile would have been replaced with the last command beginning with "badfile." Chances are no such command would have existed, resulting in the error message badfile: Event not found. See the section on history for more information.


Specifying Files

There are two ways to specify files:

  • fully, in which case the name of the file includes all of the directories, starting from the root director, "/", or
  • relatively, in which case the filename starts with the name of a subdirectory or consists solely of its own name.


When Charlotte Lennox (username lennox) created her directory arabella, all of the following sets of commands could be used to display the same file:


% more ~lennox/arabella/chapter1


% cd ~lennox

% more arabella/chapter1


% cd ~lennox/arabella

% more chapter1


The full file specification, beginning with a "/" is very system dependent. On the IMSS UNIX Cluster, all user directories are on the /home partition. This means that ~lennox on the IMSS UNIX Cluster would be the same as /home/lennox and that chapter1 would be fully specified by:




Disk Space Maintenance

It's important to keep track of how much disk space you are using. The command du displays the disk usage of the current directory and all of its subdirectories. It displays the usage, in 512-byte units, for each directory, including any subdirectories it contains, and ends by displaying the total.


% du
Display disk usage of the current directory and its subdirectories.
% du -s
Display only total disk usage.
% du -s -k
Some versions of UNIX, such as Solaris, need -k to report kilobytes.

The df Program

To examine what disks and partitions exist and are mounted, you can type the df command at the % prompt. This should display partitions which have names like /dev/sd3g: 3 for disk 3, g for partition g. It will also display the space used and available in kilobytes and the "mount point" or directory of the partition.


Scratch Space

Users have home directories for storing permanent files. At various busy times of the year there may be shortages of disk space on the IMSS UNIX Cluster. You should use the du command to stay aware of how much space you are using and not try to exceed the system limits. For more information on user disk quotas see the FAQ - Quotas document.

Temporary scratch space is available in the event of a disk crunch, however, and can be used to store files which are extremely large or relatively unimportant. The scratch space is located in the /ccovol/suntmp directory. You should use the mkdir program to create a directory for yourself on either of these partitions.

Remember, because the scratch space is for temporary storage, you should delete the files as soon as you are done with them. This is particularly important if your files are large. If you forget to remove your files, they will be removed for you, but not until seven or more days after you last access them.


Displaying owner, group and permissions

The command ls -l filename will list the long directory list entry (which includes owner and permission bits) and the group of a file.

The display looks something like:


	permission  owner       group      filename
-rw-r-----  hamilton    ug         munster_village

Protecting Files and Directories

When created, all files have an owner and group associated with them. The owner is the same as the username of the person who created the files and the group is the name of the creator's default login group, such as faculty, grads, ug, etc.

Most users belong to one group on the IMSS UNIX computers, such as ug or faculty. If the owner of the file belongs to more than one group (you can display the groups to which you belong with the groups command) then the owner can change the group of the file between these groups. Otherwise, only the root account can change the group of a file.

Only the root account can change the ownership of a file.


The Permission Bits

The first position (which is not set) specifies what type of file this is. If it were set, it would probably be a d (for directory) or l (for link). The next nine positions are divided into three sets of binary numbers and determine permissions for three different sets of people.


	u      g      o
421    421    421
rw-    r--    ---
 6      4      0

The file has "mode" 640. The first bits, set to "r + w" (4+2=6) in our example, specify the permissions for the user who owns the files (u). The user who owns the file can read or write (which includes delete) the file.

The next trio of bits, set to "r" (4) in our example, specify access to the file for other users in the same group (g) as the group of the file. In this case the group is ug -- all members of the ug group can read the file (print it out, copy it, or display it using more).

Finally, all other users (o) are given no access to the file.

The one form of access which no one is given, even the owner, is "x" (for execute). This is because the file is not a program to be executed. It is probably a text file which would have no meaning to the computer. The x would appear in the third position if it was an excutable file.


Changing the Group and the Permission Bits

The group of a file can be changed with the chgrp command. Again, you can only change the group of a file to a group to which you belong. You would type as follows:


% chgrp groupname filename


You can change the protection mode of a file with the chmod command. This can be done relatively or absolutely. The file in the example above had the mode 640. If you wanted to make the file readable to all other users, you could type:


% chmod 644 filename


% chmod o+r filename


% chmod +4 filename


For more information see the man page for chmod.


Default Permissions: Setting the umask

All files are assigned a default set of permissions. To set the default, you must set the value of the variable umask. umask must be defined once per login (usually in the .cshrc file). Common umask values include 022, giving read and execute (or directory search) but not write permission to the group and others and 077 giving no access to group or other users for all new files you create. Note that the umask bits represent permissions not to be given (i.e. the opposite of what ls -l would show).


Creating Files

The cat Program

cat is one of most versatile commands. The simplest use of cat:


% cat .cshrc


displays your .cshrc file to the screen. Unix allows you to redirect output which would otherwise go to the screen by using a > and a filename. You could copy your .cshrc, for example, by typing:


% cat .cshrc > temp


This would have the same effect as:


% cp .cshrc temp


More usefully, cat will append multiple files together.


% cat .cshrc .login > temp


will place copies of your .cshrc and .login into the same file. Warning! Be careful not to cat a file onto an existing file! The command:


% cat .cshrc > .cshrc


may destroy the file .cshrc.

If you fail to give cat a filename to operate on, cat expects you to type in a file from the keyboard. You must end this with a -D on a line by itself. -D is the end-of-file character.

By combining the above two concepts, leaving off the name of a file to input to cat and telling cat to direct its output to a file with > filename, you can create files. For example:


	% cat > temp


This will create a new file temp, containing the lines of garbage shown above. Note that this creates a new file. If you want to add things on to the end of an existing file you must use cat slightly differently. Instead of > you'd use >> which tells the shell to append any output to an already existing file. If you wanted to add a line onto your .cshrc, you could type


	% cat >> .cshrc
echo "blah blah blah"

This would append the line echo "blah blah blah" onto your .cshrc. Using > here would be a bad idea; it might obliterate your original .cshrc file.


Files as Output and Log Files

Ordinarily there are two types of output from commands: output to standard output (stdout) and to standard error (stderr). The > and >> examples above directed only standard output from programs into files. To send both the standard output and error to a file when using the C-shell, you should type >&:


% command >& filename


Logging Your Actions to a File

Sometimes you may wish to log the output of a login session to a file so that you can show it to somebody or print it out. You can do this with the script command. When you wish to end the session logging, type exit.

When you start up you should see a message saying script started, file is typescript and when you finish the script, you should see the message script done. You may want to edit the typescript file: visible ^M's get placed at the end of each line because linebreaks require two control sequences for a terminal screen but only one in a file.


Text Editors

cat is fine for files which are small and never need to have real changes made to them, but a full-fledged editor is necessary for typing in papers, programs and mail messages. Among the editors available on the IMSS UNIX computers are pico, vi and emacs.

Be careful: not all UNIX editors keep backup copies of files when you edit them.



pico is a simple, friendly editor. Type setup pine to set it up, pico filename to start it and man pine for more information about how to use it.



vi is an editor which has a command mode and a typing mode. When you first startup vi (with the command vi filename) it expects you to enter commands. If you actually want to enter text into your file, you must type the insert command i. When you need to switch back to command mode, hit the escape key, usually in the upper left corner of your keyboard.

To move around you must be in command mode. You can use the arrow keys or use j, k, h, l to move down, up, left and right, respectively.

For more information type man vi.



Emacs is a large editing system.

To use emacs, type:


	% setup emacs
% emacs

Searching Through Files

The grep program can be used to search a file for lines containing a certain string:


% grep string filename

% grep -i string filename (case insensitive match)


or not containing a certain string:


% grep -v string filename


See the man page for grep. It has many useful options.

more and the vi editor can also find strings in files. The command is the same in both: type /string when at the --More-- prompt or in vi command mode. This will scroll through the file so that the line with "string" in it is placed at the top of the screen in more or move the cursor to the string desired in vi. Although vi is a text editor there is a version of vi called view, which lets you read through files but does not allow you to change them.


Comparing Files

The basic commands for comparing files are:


states whether or not the files are the same
lists line-by-line differences
three column output displays lines in file 1 only, file 2 only, and both files

See the man pages on these for more information.


File Types

When you list files in UNIX, it can be very hard to tell what kind of files they are. The default behavior of the ls program is to list the names of all the files in the current directory without giving any additional information about whether they are text files, executable files or directories. This is because the meaning of the contents of each file is imposed on it by how you use the file. To the operating system a file is just a collection of bytes.

There is a program file which will tell you information about a file (such as whether it contains binary data) and make a good guess about what created the file and what kind of file it is.


The System and Dealing with Multiple Users

Most UNIX commands which return information about how much CPU time you've used and how long you've been logged in use the following meanings for the words "job" and "process."

When you log in, you start an interactive "job" which lasts until you end it with the logout command. Using a shell like C shell which has "job control" you can actually start jobs in addition to your login job. But for the purposes of the most information returning programs, "job" refers to your login session.

Processes, on the other hand, are much shorter lived. Almost every time you type a command a new process is started. These processes stay "attached" to your terminal displaying output to the screen and, in some cases (interactive programs like text editors and mailers), accepting input from your keyboard.

Some processes last a very long time -- for example, the /bin/csh (C-shell) process, which gets started when you log in, lasts until you log out.


Information About Your Processes

You can get information about your processes by typing the ps command.


   835 pts/1    0:03 csh

The columns indicate terminal (TTY) on which the process is running, the process identification numbers (PID), and the amount of CPU time the process has accumulated (TIME).


Information About Other People's Processes


The simplest and quickest information you can get about other people is a list of which users are logged in and at which "terminals" (terminal here is either a terminal device line or telnet or rlogin session). The command to do this is who and it responds quickest of all the commands discussed here because it simply examines a file which gets updated every time someone logs in or out.

Be careful though! This file, /etc/utmp, can get out of date if someone's processes die unexpectedly on the system. Any program which uses utmp to report information might occasionally list users who are not really logged in!


	% who
rem        pts/0        Aug 16 14:40    (shada)
wjzhou     pts/2        Aug 19 00:59    (relax.che.caltech.edu)
aji        pts/13       Aug 23 18:22    (hq.rainfinity.com)
lara       pts/1        Aug 23 12:51    (wolf.eql.caltech.edu)
avayona    pts/3        Aug 22 21:49    (gondor.submm.caltech.edu)
kmu        pts/5        Aug 19 18:37    (ken.gg.caltech.edu)
shachi     pts/7        Aug 22 22:01    (braun-103.caltech.edu)
auddess    pts/73       Aug 23 16:20    (charter-DHCP-59.caltech.edu)
htyu       pts/6        Aug 18 19:36    (afar.carl.rhno.columbia.edu)
dswang     pts/44       Aug 23 17:49    (kuppermac2.caltech.edu)
surpi      pts/8        Aug 23 10:28    (garfield.tapir.caltech.edu)
justa      pts/87       Aug 23 16:47    (30-233-44-208.user.darwin.net)
jasonmc    pts/23       Aug 19 15:01    (fleming-154.caltech.edu)
mkallis    pts/16       Aug 23 18:13    (avery-70.caltech.edu)
marat      pts/28       Aug 20 18:20    (hpl3cit2.cern.ch)
tanja      pts/20       Aug 21 19:00    (
vineet     pts/29       Aug 23 18:18    (DHCP-108-222.caltech.edu)
duello     pts/37       Aug 22 20:26    (ruddock-190.caltech.edu)
qing       pts/17       Aug 23 18:25    (Longbeard.library.caltech.edu)
schwab     pts/33       Aug 23 07:27    (lps-133.umd.edu)


The w command is slower than the who command because it returns more information such as details about what programs people are running. It also returns a line containing the number of users and the system load average. The load average is the average number of processes ready to be run by the CPU and is a rough way of estimating how busy a system is.


	% w
  6:25pm  up 7 day(s),  3:54,  70 users,  load average: 1.56, 1.66, 1.70
User     tty           login@  idle   JCPU   PCPU  what
rem      pts/0        16Aug00         3:28      1  jsh
wjzhou   pts/2        Sat12am  1:35     13         -csh
aji      pts/13        6:22pm                      elm
lara     pts/1        12:51pm  5:29                -tcsh
avayona  pts/3        Tue 9pm  8:13                -tcsh
kmu      pts/5        Sat 6pm     5     12         pine -i
shachi   pts/7        Tue10pm  6:22      4         -tcsh
auddess  pts/73        4:20pm  1:12                pine
htyu     pts/6        Fri 7pm    43     11      7  pine
dswang   pts/44        5:49pm    36                mail
surpi    pts/8        10:28am     5     18      2  -csh
justa    pts/87        4:47pm    12      3      3  pine -z
jasonmc  pts/23       Sat 3pm  4:59     13         -csh
mkallis  pts/16        6:13pm                      pine
marat    pts/28       Sun 6pm 3days                -tcsh
tanja    pts/20       Mon 7pm            2         -csh
vineet   pts/29        6:18pm                      pine
duello   pts/37       Tue 8pm    59      1      1  pine


The ps command used earlier to list your own processes can be used to list other users' processes as well. who and w list logins but not individual processes on the system. They don't list any of the running operating system processes which start when the computer is booted and which don't have logins.

Since ps doesn't use utmp, it is the program to use when you really want to find out what processes you might have accidentally left on the system or if another user is running any processes. Note that although ps might report processes for a user, it might be because that user has left a "background job" executing; the user is not really logged in. In this case you should see a "?" in the TTY field.

To get this fuller listing, use ps -ef. For more information on the uses of ps, type man ps.



top is an interactive command that displays and periodically updates the top cpu processes, ranked by raw cpu percentage. The default number of processes displayed is 15 but you can specify the number of processes with top number (e.g. to specify the top 10 processes enter top 10). You should see something like the following:


	% top
load averages:  0.86,  0.39,  0.27                                15:01:51
452 processes: 401 sleeping, 38 zombie, 11 stopped, 2 on cpu
CPU: 47.2% idle, 50.0% user,  2.8% kernel,  0.0% iowait,  0.0% swap
Memory: 512M real, 14M free, 419M swap in use, 131M swap free

19760 root       1  40    0 3240K 1872K sleep   0:00  0.81% sshd1
19586 marting    1  43    0 1984K 1648K cpu2    0:02  0.70% top.sun4u
19751 jasonmc    1  58    0 5360K 2944K sleep   0:00  0.14% pine
19763 samuelt    1  30    0 1352K 1120K sleep   0:00  0.12% csh
11406 avayona    1  58    0 2760K 1168K sleep   2:56  0.09% xbiff
19713 milamber   1  58    0 5384K 2968K sleep   0:00  0.07% pine
19683 rogero     1  58    0 5392K 3104K sleep   0:00  0.05% pine
18227 root       1  58    0 3248K 1816K sleep   0:00  0.05% sshd1
26085 tpkmc      1  58    0 5976K 4136K sleep   0:04  0.05% pine
19628 milley     1  58    0 2192K 1936K sleep   0:00  0.04% trn

As top is an interactive command the cursor will remain blinking above the PID column, waiting for the next command. If you only want to see a list of jobs by a specific user type the letter u. You will then be asked which username to you wish to show. To only show non-idle jobs use the -I flag.

To quit or cancel a job that is still running the kill command can be used within top. Simply enter the letter k and the cursor will prompt you with kill . Then enter the PID of the job you wish to cancel and . The process will then be killed. NB: If the job does not immediately disappear from the top listing, don't worry. Wait a few seconds until the screen is updated and your job should be removed.

To quit top simply type the letter q.



The finger program returns information about other users on the system who may or may not be logged in. finger by itself returns yet another variation of the list of currently logged in users. finger followed by a username or an e-mail -style address will return information about one or more users, the last time they logged into the system where you are fingering them, their full name, whether or not they have unread mail and the contents of two files they may have created: .plan and .project.

For more information about using finger or ways to provide information about yourself to others, type man finger.


	% finger
Login       Name               TTY         Idle    When    Where
rem      Roger E. Murray       pts/0          1 Wed 14:40  shada
wjzhou   Weijun Zhou           pts/2       1:36 Sat 00:59  relax.che.caltech.ed
lara     Lara Hughes           pts/1       5:31 Wed 12:51  wolf.eql.caltech.edu
avayona  Anastasios Vayonakis  pts/3       8:14 Tue 21:49  gondor.submm.caltech
kmu      Ken Museth           *pts/5          7 Sat 18:37  ken.gg.caltech.edu
shachi   Shachi S. Gosavi      pts/7       6:24 Tue 22:01  braun-103.caltech.ed
auddess  Audrey Carstensen    *pts/73      1:14 Wed 16:20  charter-DHCP-59.calt
htyu     Haitao Yu             pts/6         45 Fri 19:36  afar.carl.rhno.colum
dswang   Deshen Wang           pts/44        37 Wed 17:49  kuppermac2.caltech.e
surpi    Gabriela C. Surpi     pts/8          6 Wed 10:28  garfield.tapir.calte
justa    Vikki Kowalski        pts/87        14 Wed 16:47  30-233-44-208.user.d
jasonmc  Jason S. McIlhaney    pts/23      5:00 Sat 15:01  fleming-154.caltech.
mkallis  Michelle Kristin All *pts/16           Wed 18:13  avery-70.caltech.edu
marat    Marat Gataullin       pts/28        2d Sun 18:20  hpl3cit2.cern.ch
devang   Devang Shah          *pts/13           Wed 18:26  node-64-249-49-50.ds
tanja    Tanja Bosak          *pts/20         2 Mon 19:00
vineet   Vineet Gupta         *pts/29           Wed 18:18  DHCP-108-222.caltech
duello   Amy Duello           *pts/37      1:00 Tue 20:26  ruddock-190.caltech.
qing     Qing He               pts/17         1 Wed 18:25  Longbeard.library.ca

Sending Messages and Files to Other Users

Electronic mail programs run on almost all the computers at Caltech and usually have two parts: a user interface which lets users read and send messages and a system mailer which talks to mailers on other computers. This mailer receives outgoing messages from the user interface programs and delivers incoming messages to the user mailbox (which the interface program reads).



There are many user interfaces available on the UNIX computers, all of which provide similar functionality. The program supplied with most UNIX computers is /usr/ucb/mail (or mailx). To read messages type mailx, to send messages type:


% mailx address


For information about valid mail addresses, please see the Electronic Mail Guide.

You should next see a Subject: prompt. If you don't see a prompt, don't worry, just type in your one line subject anyway and press return. Most likely you will now see a CC: prompt. If you wish to send copies of your message to someone besides the recipient you would enter the address or addresses (separated by commas) and press return. Otherwise press return without entering an address.

You can now start typing your message (but you will be unable to correct errors on lines after you have pressed to move to the next line) or you may specify a file to include, with ~r filename. If instead you only wish to send a file you can use the command mailx usr < filename instead.

You may invoke a text editor like vi by typing ~v. If you wish regularly to use an editor other than vi you should see the information later in the section about environment variables.

There are many other commands you may enter at this point -- see the Mail man page for all of them. When you are finished typing in your message (if you have used ~v to run a text editor, you should exit from it) either press -D or enter a period on a line by themselves.


Other Mail Programs

Pine is a full-screen interactive mailer that is very straightforward to use. It is currently installed on the IMSS UNIX Cluster Suns. If you have the command setup pine in your .cshrc then you should type pine. If the pine command fails you probably need to type setup pine first. For more information type man pine.

Other mailers include ELM and MH. Both require setup commands. The command to start ELM is elm. MH is actually a set of several separate programs. See man mh for more information.



The write program can be used to send messages to other users logged onto the system. It's not the best way of having a conversation, but it's simple to use. Enter:


% write username


and you can start writing lines to the terminal of the person you want to send messages to. The person must be logged in, and, if they are logged in more than once, you must specify the terminal to write to; write melville ttyh1, for example.



ytalk is a program which allows two users to hold a conversation. Unlike write, it can be used between different computers; and, unlike write, it divides the screen so that the things you type appear in the top half and the things written to you appear in the bottom half.

To ytalk to users on the same computer:


% ytalk username


To ytalk to users on another computer use the address format of username@nodename:


% ytalk brunton@jarthur.claremont.edu


Addressing Remote Nodes

ytalk can only be used to other Internet nodes: usually computers which have ending names such as .edu, .com, .org, .gov, or .mil. Not all computers with these names are attached directly to the Internet. finger and ytalk won't work with computers which are only attached by mail gateways.



If you use certain command flags regularly (-lga for ls) you can alias them to shorter commands. You can use wildcard symbols to refer to files with very long names. You can easily repeat commands you have already executed or modify them slightly and re-execute them.



As mentioned above, you can alias longer commands to shorter strings. For example, ls -F will list all the files in the current directory followed by a trailing symbol which indicates if they are executable commands (*) or directories (/). If you wanted this to be the default behavior of ls you could add the following command to your .cshrc:


% alias ls ls -F


This command will then be aliased the next time you login. To make the changes you have made effective in your current shell you must first source your .cshrc file by typing:

% source .cshrc


NB: You must be in the directory that contains your .cshrc file or else specify the correct path to your .cshrc file.

To list the aliases which are set for your current process, type:


% alias


without any parameters.



Wildcards are special symbols which allow you to specify matches to letters or letter sequences as part of a filename.

Some examples:


matches zero or more characters
ls *.dat
lists all files ending in .dat
ls r*
lists all files starting with r

Beware of the rm *command!


matches one character
ls ?.dat
lists 5.dat, u.dat, but not 70.dat
matches one of the characters inside the brackets
ls *.[ch]
lists all .hand .cfiles
more [Rr][Ee][Aa][Dd][Mm][Ee]
mores the files README, readme, ReadMeand Readme, among others

Directory Specifications

You've already met the ~ shortcut. The two other important directory symbols are "." for the current directory and ".." for the previous (parent) directory.


% cd ..


moves you out of a subdirectory and into its parent directory.


Environment Variables

Environment variables are pieces of information used by the shell and other programs. A very important one is the PATH variable mentioned earlier. Other important variables you can set include:


To see what environment variables are set and what they are set to, type the command printenv. To set a variable, use the setenv command as in the example below.


% setenv TERM vt100

% setenv EDITOR emacs


Many programs mention environment variables you may want to set for them in their man pages. Look at the csh man page for some of the standard ones.



Most shells allow "command line editing" of some form or another: editing one of the previous few lines you've typed in and executing the changed line. You can set the history variable to determine how many previous command lines you will have access to. set history=40 will let you search the last 40 commands.


Repeating and Modifying the Previous Command

The simplest form of command line editing is to repeat the last command entered or repeat the last command entered with more text appended.

If the last command you typed was:


% ls agreen


Then you can repeat this command by typing:


% !!


This will return a list of files. If you saw a directory leavenworth in the list returned and you wanted to list the files it contained, you could do so by typing:


% !! leavenworth


If you mistype leavenworth as leaveworth you can correct it with the following command:


% ^leave^leaven


This substitutes leaven for leave in the most recently executed command. Beware: this substitutes for the first occurrence of leave only!


Repeating Commands From Further Back in History

You can type history at any time to get a list of all the commands remembered. This list is numbered and you can type !number to repeat the command associated with number. Alternatively you can type ! and a couple of letters of the command to repeat the last line starting with the characters you specify: !ls to repeat your last ls command, for example.


The .login and .cshrc Files

The .cshrc file is run whenever a C shell process is started. Then, if this is a login process, the .login file is executed. If you are using a Sun console and you have the default setup, any xterm windows which you start up will not execute the .login.


Job Control

It is very easy to do many things at once with the UNIX operating system. Since programs and commands execute as independent processes you can run them in the background and continue on in the foreground with more important tasks or tasks which require keyboard entry.

For example, you could set a program running in the background while you edit a file in the foreground.


The fg and bg Commands

When you type -Z (by holding down the Control key and tapping the Z key) whatever you were doing will pause. If you want the job to go away without finishing, then you should kill it with the command kill %. If you don't want it paused but want it to continue in the foreground -- that is, if you want it to be the primary process to which all the characters you type get delivered -- type fg. If you want it to continue processing in the background while you work on something else, type bg.

You should not use bg on things which accept input such as text editors or on things which display copious output like more or ps.


What to Do When You've Suspended Multiple Jobs

If you've got several processes stopped -- perhaps you are editing two files or you have multiple telnet or rlogin sessions to remote computers -- you'll need some way of telling fg which job you want brought to the foreground.

By default fg will return you to the process you most recently suspended. If you wanted to switch processes you would have to identify it by its job number. This number can be displayed with the jobs command. For example:


	% jobs
[1]          Stopped     vi .login
[2]    +     Stopped     rn
[3]          Running     cc -O -g test.c

The most recently suspended job is marked with a + symbol. If you wanted to return to job one instead, you would type:


% fg %1


You can type %1 as a shortcut.


Starting Jobs in the Background

Some jobs should start in the background and stay there -- long running compilations or programs, for example. In this case you can direct them to the background when you start them rather than after they have already begun. To start a job in the background rather than the foreground, append an & symbol to the end of your command.

You should always run background processes at a lower priority by using the nice command. Non-interactive jobs are usually very good at getting all the resources they need. Running them at a lower priority doesn't hurt them much, but it really helps the interactive users: people running programs that display to terminal screens or that require input from the keyboard.

The IMSS UNIX Cluster has an explicit policy about the proper running of CPU-intensive background jobs. Be sure to read over this information in the UNIX Cluster Policies. You should man nice and man renice for more information about how to alter the priority of your processes.


Suspend, z and -Z

Some programs provide you with special ways of suspending them. If you started another shell by using the csh command, you would have to use the suspend command to suspend it.

If you wish to suspend a telnet or rlogin session you must first get past the current login to get the attention of the telnet or rlogin program.

Use ~ (immediately after pressing a return) to get rlogin's attention. -Z will suspend an rlogin session.

Use -] to get telnet's attention. -]z will suspend a telnet session. Watch out, though, if you are connected from a PC through Kermit! -] is Kermit's default escape sequence. You'll need to type -]-]z or define Kermit's escape sequence to something else.


Getting Help

If you have a general computing question please contact us at http://help.caltech.edu (request type IMSS-->Desktop Support-->Other)


Learning More About UNIX

For more information about UNIX you could consult:


Christian & Richter
"The UNIX Operating System", 3rd edition
ISBN 0-471-58684-6

This book is now a few of years old -- other books might be more up to date, but the above remains useful for informational purposes.

Another good place to start is the O'Reilly series of computing guidebooks.

Online launching points include:
UNIX help for Users
Systems Administrators' Guild


Summary of Common and Useful UNIX Commands For Files


The cp command allows you to create a new file from an existing file. The command line format is:


% cp input-file-spec output-file-spec


where input-file-spec and output-file-spec are valid UNIX file specifications. The file specifications indicate the file(s) to copy from and the file or directory to copy to. Any part of input-file-spec may be replaced by a wildcard symbol (* or ?) and you may specify either a filename or a directory for the output-file-spec. If you do not specify a directory, you should be careful that any wildcard used in the input-file-spec does not cause more than one file to get copied.


	% cp new.c old.c
% cp new.* OLD (where OLD is a directory)


The ls command allows the user to get a list of files in the current directory. The command line format is:


% ls file-spec-list


where file-spec-list is an optional parameter of zero or more UNIX file specifications (separated by spaces). The file specifications supplied (if any) indicates which directories are to be listed and the files within the directories to list.



The lp command tells the system that one or more files are to be printed on the default printer. If the printer is busy with another user's file, an entry will be made in the printer queue and the file will be printed after other lp requests have been satisfied. The command line format is:


% lp file-spec-list


where file-spec-list is one or more UNIX files to be printed on the default printer. Any part of the filenames may be replaced by a wild card.

For more information about where the printers actually are and what kind of printers are available, please see our listing of IMSS Printers.



The man command is a tool that gives the user brief descriptions of UNIX along with a list of all of the command flags that the command can use. To use man, try one of the following formats:


% man command % man -k topic


The more command will print the contents of one or more files on the user's terminal. The command line format is:


% more file-spec-list


more displays a page at a time, waiting for you to either press the space bar at the end of each screen or for single line scrolling. At any time you may type q to quit or h to get a list of other commands that more understands.



The mv command is used to move files to different names or directories. The command line syntax is:


% mv old-file-spec new-file-spec


where old-file-spec is the file or files to be renamed or moved. As with cp, if you specify multiple files, the new-file-spec file should be a directory. Otherwise new-file-spec may specify the new name of the file. Any or all of the old filenames may be replaced by a wild card to abbreviate it or to allow more than one file to be moved.

For example:


% mv data.dat ./research/datadat.old


will change the name of the file data.dat to datadat.old and place it in the subdirectory research. Be very careful when copying or moving multiple files.



The rm command allows you to delete one or more files from a disk. The command line format is:


% rm file-spec-list


where file-spec-list is one or more UNIX file specifications, separated by spaces, listing which files are to be deleted. For example:


% rm *.dat able.txt


will delete the file able.txt and all files in your current working directory which end in .dat. Getting rid of unwanted subdirectories is a little more difficult. You can delete an empty directory with the command rmdir directory-name but you cannot use rmdir to delete a directory that still has files in it.

To delete a directory with files in it, use rm with the -r flag (for recursive).

Beware of the rm * command!