The UNIX File System

Every user with a login on the system has a special directory, called a home directory, where personal files can be created and stored. The name of each home directory is the same as its owner's account login.

The diagram below shows part of a typical UNIX file system. Four subdirectories of the root directory are shown. Items ending with a slash, such as j/ and k/, are subdirectories. Items without a trailing slash are files. The home directories are grouped by letters of the alphabet under the home subdirectory.

Understanding Absolute and Relative Paths

/home/k/kay/public_html/index.html

The initial slash in the absolute path represents the root directory. All subsequent directories are written in the form directory_name/. Each slash in the path chows the separation between a directory and a subdirectory or a file contained within it.

public_html/index.html

When writing out a relative path, the sequence .. (two periods in a row) means "parent of current working directory," or "up one directory." (The directory containing a directory is called its parent.) By using this notation, you avoid having to type all the names in a sequence of parent directories. The relative path from kay's home directory to the file index.html in joe's public_html subdirectory is thus:

../../j/joe/public_html/index.html

In plain English, the above path means "up one directory (from kay to k), then up another directory (from k to home), then down from the directory home, through j and joe, to public_html where the file index.html is located.

When specifying relative paths it is imperative that you know the current working directory. If you lose track of your location, the UNIX command pwd (print working directory) will display the absolute path of the current working directory on your screen.

~joe/public_html/index.html

When referring to your own home directory, you don't need to specify a login. A tilde by itself is shorthand for the home directory of the person who is logged in. Thus if kay were logged in to her account, she could refer to her own index.html file with the following path:

~/public_html/index.html

Search Paths: Another Kind of Path

For most users, the search path is set during the login process when the configuration files, known as dot files, are read. You can see your default search path by typing the following command at the UNIX prompt:

printenv PATH

The output of the above command will look something like this:

/usr/local/bin:/bin:/usr/bin:/usr/ccs/bin:/usr/games:/usr/ucb:.

As shown in the sample output above, each directory in the search path is separated from the next by a colon. When the system looks for a program, it searches the leftmost directory first (/usr/local/bin in the above example) and continues to search each directory in the list until it finds a program with the name you typed, at which point the program is executed. If a program by the same name is located in two different directories in your search path, the system will execute the first one it finds. To start a program that isn't in your search path, type out the fell or relative path of the program as described in the previous section.

Typing Commands at the UNIX Prompt

When writing out a command, the various parts of the command must follow a specific order, called the command syntax. The parts include the command itself, any added command options, and the names of files or directories to be acted on by the command. Syntax varies among UNIX commands. For instance, some commands take file or directory names as part of their syntax, whereas others do not. However, file and directory management commands generally have the following syntax:

command [options] file_name(s) or directory_name(s)

Options change how the command is carried out or alter the type of output produced by the command. They take the form x, where the minus sign (-) indicates that the letter immediately following it is an option, and the ldtter (x) represents a particular option. Multiple options can usually be employed simultaneously in the form xyz (rather than writing as xyz).

Many commands take the names of files or directories as part of their syntax. If you type the name of a file or directory by itself (i.e., without its relative or absolute path), it is assumed to be in your current working directory. To refer to files or directories outside of the current working directory, you must specify their relative or absolute paths.

Note: In the remainder of this How-To, the UNIX prompt is shown as $. The prompt may be different on the system you're using.

$ ls
final_paper   homework1.30  mail
homework1.16  homework2.13  public_html
$

Giving the name of a directory as an argument to the ls command causes ls to display the contents of that directory rather than the current working directory. For example:

$ ls public_html index.html $

To view dot files (those files whose names begin with a period), use the -a option, which shows all files:

$ ls -a
.             .kshrc       .profile       homework1.30
..           .mh_profile  .sh_history    homework2.13
.addressbook .nfsB3A1     final_paper    mail
.exrc         .nn           homework1.16   public_html $

Another option, -l, produces a long format listing of the directory contents. This shows detailed information for each file and directory:

$ ls -al
total 106
drwx--l---   5    joe   home     1536 Jan 30 13:10 .
drwxr-xr-x   1482 root  home   27136 Jan 29 14:40 ..
-rw-r-----   1    joe   home        0 Aug 31  1994 .addressbook
-rw-r--r--   1    joe   home     7175 Jan 29 15:11 .kshrc
-rw-r-----   1    joe   home        9 Sep  6  1992 .exrc
-rw-r-----   1    joe   home       11 Jul  2  1992 .mh_profile
-rw-------   1    joe   home      360 Jan 30 13:13 .nfsB3A1
drwx--l---   2    joe   home      512 Jan 11 15:55 .nn
-rw-r--r--   1    joe   home     1004 Jan 29 15:13 .profile
-rw-------   1    joe   home      770 Jan 29 15:13 .sh_history
-rw-------   1    joe   home     1743 Jan  5 14:16 final_paper
-rw-------   1    joe   home     1216 Jan 11 14:41 homework1.16
-rw-------   1    joe   home     1216 Jan 11 14:41 homework1.30
-rw-------   1    joe   home     1216 Jan 16 11:46 homework2.13
drwx------   2    joe   home      512 Jan 11 15:54 mail
drwxr-sr-x   2    joe   home      512 Jan 17 10:05 public_html
$

Example entry:
permissions	links	owner's login	group	size (bytes)	date/time last modified	file name
-rw-------	1	joe	home	1216	Jan 16 11:46	homework2.13

The information you are most likely to use from a long-format listing is labelled in the above example. You may wish to pay special attention to file size, since too many large files can fill up your disk space. Permissions determine who may read, write, or execute each file and directory. Only the owner of a file or directory can change its permissions. In the above example, the user joe is the owner of all the files and directories shown. For a detailed discussion of permissions, see the section "Sharing and Protecting your Files and Directories" later in this How-To.

$ ls .m*
.menueditorprog  .menumotd        .menushellprog
.menumailprog    .menunewsprog    .mh_profile
$

You can use multiple wildcards in the same name. For example, if you have files as shown on the long listing in the previous section above, an easy way to list only the files homework1.16 and homework1.30, is to use the command lsh*1.*. Be very careful when using wildcards to delete files, because your command could remove more files than you originally intended. To ensure that you know what files a wildcard expression represents, it's a good idea to use the wildcard expression in an ls command before using it with a command that can modify or erase your files. Note, for instance, that leaving out the period from the example above produces a list of all the homework files:

$ ls h*1*
homework1.16 homework1.30 homework2.13
$

To see your current disk usage and limits, use the quota -v command. The following is an example of typical quota output on the student cluster:

$ quota -v
Disk quotas for joe (uid 9593):
Filesystem usage quota limit timeleft files quota limit timeleft
/homeb     368     4096     4300          94     2000     2500
/homec     0     1     1          0     1     1
$

In the Filesystem column, two file systems are shown, /homeb and /homec. All student home directories are in the /homeb directory. Faculty and staff home directories are stored in /homea, which does not show in the above output. As a student, joe need only pay attention to the data given for the /homeb file system.

When reading the quota output, your current disk usage (counted in 1 kilboyte blocks) is reflected in the usage column, and your total number of files is shown in the files column. The quota columns indicate the point at which the system will complain that you hare too many files or are using too much of your allotted disk space (also counted in 1 kilobyte blocks). The limit columns indicate the absolute maximum amount of disk space or number of files you can use. Think of the difference between the quota and the limit as a buffer. Once you've surpassed your quota, the relevant timeleft (called grace on ux7) column will start a count-down - you have seven days to reduce your disk usage or number of files. After seven days you will no longer be able to do anything other than delete files.

If you get a message from the system stating that you've exceeded your disk space quota, immediately delete some of the files in your home directory, transfer them to a different system, or compress them so that they take up less space. Various compression programs are available on the CCSO UNIX systems, including compress, gzip, and zip. You can learn more about these compression programs using the UNIX online documentation. For help, see the section below that discusses "Using Online UNIX Documentation."

Managing Your Files

$ mkdir myhomework

$ mv homework homework.cs101

If used incorrectly, mv can be a dangerous command. For instance, if a file already exists with the name homework.cs101, the mv command will remove that existing file, replacing it with the one that's being renamed. This may not be what you want. It is safer to use the interactive option (-i) with mv so that the system prompts you to confirm before removing any files. For example:

$ mv -i homework homework.cs101
remove homework.cs101? (y/n) n
$

The mv command can also be used to move files or directories to other locations. If you want to move a file from one directory to another, the syntax is mv file_name directory_name. For example, to move all the homework files in the current working directory to an existing subdirectory called cs101, type:

$ mv -i homework* cs101

The cs101 directory must already exist for the above command to work. The -i option forces the system to prompt you before removing any files of the same name.

The table below shows the various ways the mv command can be used:

mv Command		Action
mv old_name new_name		Rename a file or directory.
mv file_name directory_name		Move a file to an existing directory.
mv directory_name1 directory_name2		Move a directory (and all files therein) to a different directory. Directory_name2 must already exist, or this command will simply rename directory_name1 to directory_name2.

To make a copy of a file in the current working directory, use the syntax cpfile_namecopy_file_name, as shown in the following example:

$ cp -i final_paper econ101_paper

The table below shows the various uses of the cp command:

cp Command		Action
cp file_name copy_file_name		create a duplicate of the file in the current working directory
cp file_name directory_name		create a duplicate of the file in the directory specified (the copy will have the same name as the original, but reside in a different directory)
cp file_name1 directory/file_name2		create a duplicate of the file in the directory specified, and give it a different name as specified

$ rm -i homework*
rm: remove homework1.16 (y/n)? n
rm: remove homework1.30 (y/n)? y
rm: remove homework2.13 (y/n)? y
$

Remember, rm is forever; -i give you one last chance.

There are four parts to the chmod command-the who option(s), the opcodes (i.e., operation codes), the permission option(s), and the name of the file or directory for which permissions are to be changed:

Who Options		Opcode Options		Permissions
u user		+ add permission		r read
g group		- remove permission		w write
o other		= assign permission		x execute
a all of the above

Who specifies who the permission affects, the user (u), the group (g), others (o), or all three (a). The user is the owner of the file or directory, the group consists of those with whom the owner shares a group id (on CCSO UNIX systems, if you're a student, this generally includes other students; if you're faculty/staff, it includes people in your campus unit), and other represents all othdr users on the system. The opcode part of the chmod command specifies whether a permission is added to (+), removed from (), or simply assigned to (=) the who specified. The permission specifies the type of access that is being added or removed, namely read (r), write (w), or execute (x).

The following example shows the structure of the chmod command:

chmod g+r file_name

change mode command	who	opcode	permission	file to be changed
chmod	g	+	r	file_name

A good way to understand a chmod command is to read it aloud. The previous example reads "change mode for group; add read permission on the file file_name."

The permissions associated with a file or directory can be viewed using the long format list command, lsl. The example below shows the permissions for the file findit.sh:

-rwxrwxrwx     1     joe     home     1216     Jan     13     8:22     findit.sh

type	user	group	other
-	rwx	rwx	rwx

Permissions are expressed by a series of ten characters, consisting of letters and dashes. The first character identifies the object's type; a dash (-) represents a file, a d represents a directory. The next three characters indicate the read (r), write (w), and execute (x) permissions of the user (owner) of the file, the second set of three indicates the group's read, write, and execute permissions, and the last three, all others' read, write and execute permissions. Read, write, and execute are always found in the same position for each category of who. A dash appearing in any one of the nine positions means the corresponding permission is turned off. If a letter (i.e., r, w, x) appears, the permission is turned on.

In the above example, the user joe, his group, and all other users have read, write and execute access to the file findit.sh. The initial dash indicates that the item is a file. If it were a directory, the first bharacter would be a d instead of a dash.

$ chmod o-rwx homework1.16

To check that the permissions have been changed, list the file with ls -l:

$ ls -l homework1.16
total 8
-rwxrwx--- 1 joe home 1216 Jan 13 8:22 homework1.16

Adding, or granting, permissions can be accomplished using the add (+) or assign (=) opcodes. For example, to give others read and write permissions again, you could use the command:

$ chmod o=rw homework1.16

Multiple who, opcodes, and permissions can be used together, as long as a comma (but no spaces) separates each combination (if they are different), and a space separates each file or directory name if more than one is specified. For example, you can simultaneously change permissions on two files at once, granting read, write, and execute permission to your group, but only read and write permission to all others, as follows:

$ chmod g=rwx,o=rw homework1.1 econ101_paper

Or, if the same opcode and permissIons are being used for each who, they can be combined without commas. The following example adds read and write permissions for the group and others to the file homework1.16:

$ chmod go+rw homework1.16

The chmod command works exactly the same way for directories. Important: a directory must be made executable in order for anyone, including yourself, to be able to make it the current directory (i.e., to cd into it) and read a file in it. Because of this, if you want people to be able to read your Web pages (from UIUC or from Tokyo), you must add croup and other executable permissions to both your home and public_html directories (i.e., chmodgo+x~public_html).

Using Online UNIX Documentation

$ man nn

If you cannot recall the exact name of the command you want to look up, or you don't get the desired output using the form mancommand, try using the -k option to search the man pages for a particular keyword, as in mankkeyword. This displays a listing of summaries for all man pages that contain the keyword (sometimes the resulting output can be quite long). >From this list you can choose the appropriate man pages to browse. For example:

$ man -k remove
prunehistory (8) - remove file names from Usenet history file
prunehistory (8) - remove file names from Usenet history file
rm, rmdir (1) - remove (unlink) files or directories
rmdir (2) - remove a directory file
rmdir, rm (1) - remove (unlink) directories or files
rmf (1) - remove an MH folder
[output omitted]...
$

Sometimes you will see a number in parentheses after the name of the program or command. This is the volume number of the manual set. By default, when you type man nn, you are looking in volume 1. If you would like to see the manual pages for a command in a different volume, you need to specify the volume number. For example, to see the man page for 'prunehistory' shown above, type:

$ man 8 prunehistory

Viewing Files

As you page through a file with more, an indicator in the lower left corner of the screen shows you how much of the file has been displayed. To advance one screen at a time, press <SPACE>, or proceed one line at a time by pressing <RETURN>. Several other navigational commands are also available within more, including a command to search for a particular word or phrase. Here is a summary of useful more keystrokes:

more Keystroke		Result
<SPACE>		display next page
<RETURN>		display next line
d		Scroll forward (down) one-half page
f		skip forward one page
b		skip backward one page (on students and staff)
/string		search for occurrence of string
n		search for next occurrence of specified string
h		help
q		exit from more

Printing Files

out -dest printer_name file_name

Here are some publicly accessible printer destinations for out:

Printer Name		Type and Location of Printer
3800		IBM 3800 Laser printer, L440 DCL
ag3812		IBM 3812 Laser printer, N-120 Turner Hall
comm		high-speed printer, 70 Comm West
dcl		IBM 3203 printer, L441 DCL
fax		outbound fax service
lincoln		HP LaserJet 3si, OCCSS at 212 Lincoln Hall
ps3812		IBM 3812 Laser printer, 493 Lincoln Hall
psych		IBM 3262 Line printer, 453 Psychology

The out command can take a number of additional options (some depending on the destination of the print job). The syntax of the out command with several of the more common options is shown below:

out -dest printer_name -bin value -banner string file_name

The -bin option takes a numeric value (i.e., -binvalue) that references a numbered bin that is used to store the print job at the printer site until it is picked up by the sender. The -banner option takes a character string, such as a name or file identification (i.e., -bannerstring), that is printed in large lettdrs on the banner (first) page of the print job. If -banner is not specified (or not active for a particular printer), publicly accessible destinations will print a banner page showing the sender's login.

Below is example of the out command with multiple options, followed by the system message that is returned:

$ out -dest lincoln -banner JOSEPH econ101_paper
200 CSO Print Server v3.2 (22 Aug 14:20). Say your request.
200 Cleared as filed, squawk 2435.
Remember the above squawk code; you'll need it to identify your job if you need to cancel or inquire about it later.
$

Part of the system message is the squawk code-an identification number assigned to each print job. If you must cancel your print job, or if it gets lost or never prints out, you need to know the squawk code in order to have the problem investigated.

Each out destination may have additional options for customizing print jobs. Some destinations also have restrictions. To find out what they are, use the command:

out -info printer_name

The example below shows information for the printer at Commerce West:

$ out -info comm
200 CSO Print Server v3.2 (22 Aug 14:20). Say your request.
Information about dest 'comm':
100 One of a family of medium-speed impact printers at RJE sites.
100 Special options [defaults] that are recognized by this driver:
100 -forms name Specify name of form to be loaded [STAN].
100 Additional information:
[output omitted]...
$

Quitting Programs Properly

Occasionally, you may find yourself unable to quit a program in the preferred manner, or you may need to temporarily interrupt a process in order to do something else. Most programs can be temporarily suspended by typing <CONTROL>z. When a program is suspended, it stops processing, but doesn't actually close and quit; therefore, it continues to use system memory. A suspended job can be resumed. Another key combination, <CONTROL>c, interrupts a program permanently. An interrupted program stops processing and using system memory, but open files may not close properly, and any work in progress may not be saved.

If two or more programs are suspended concurrently, you need to specify the process number of the program you want to resume. As previously stated, the system assigns a process number to each program or command that is started. You can review the processes you're running using the command ps at the UNIX prompt. For example, if you're running a small program you wrote, plus Pine and oed2, your ps output might look something like this:

$ ps PID TTY TIME CMD
1058 pts/33 0:02 ksh
3670 pts/33 0:00 my_program
4577 pts/33 0:00 oed2
5432 pts/33 0:00 pine
6022 pts/33 0:00 ps
$

The output shows the three programs that are running, plus your shell (ksh) and the ps command. The process number for each is found in the column labeled PID. To bring any one of these programs to the foreground, use the command fg process_number. For example, if you wanted to resume oed2, you would type fg 4577.

If you're trying to log out of your account, and you get the system message "there are stopped jobs," it means a suspended program is running. In this instance, you should resume the program and exit it in the preferred manner. If you cannot do this because you don't know how to quit or because something is wrong with the program, you can terminate the process with the kill command (i.e., kill process_number). Before killing a process, verify its process number with the ps command. In the above example, if you wanted to terminate oed2, you would type kill 4577.

Sometimes, a simple kill command may not work (i.e., the ps output shows that the process is still running). When this happens, you can add a signal (a command modifier similar to an option) to your kill command. The safest signal is 3 (i.e., kill -3process_number), which tells the system to quit the process by closing all open files and exiting. If 3 doesn't work, use the 9 signal (i.e., kill -9process_number), which is the absolute kill command and has the same effect as an interrupt. Only use kill -9 when nothing else works, since files may not close properly and you may lose any work in progress.

Changing Your Password

$ passwd
Changing Cluster password for joe
Old password: (password not shown)
New password: (password not shown)
Re-enter new password: (password not shown)
Password changed successfully on Cluster server.
Cluster account data will be updated at approximately 11:00
$

New passwords do not take effect immediately on the CCSO UNIX clusters. If you change your password on the cluster, log out, and then log in again immediately, you may need to use your old password again. On other systems, however, passwords may be updated immediately.

Logging Out

UNIX Command Quick Reference