Few aspects of Unix system administration are more intimate than the relationship between the admin and their chosen shell. After all, the shell is the most fundamental interface to the system, the conduit for all command-line interactions. Thus, it's important to have a solid foundation in whatever shell you choose.
There are many different shells available, from Ash to Zsh. They all differ in many ways, and some (such as Csh and Tcsh) vary wildly from the others. Each has its selling points. However, because Bash has become the de-facto standard shell on most Unix-like operating systems and can generally be found everywhere, it makes an excellent place to start, even if you ultimately choose a different shell for daily use.
[ How to get started with MySQL | How to install Apache on Linux | Prove your expertise with the free OS in InfoWorld's Linux admin IQ test round 1 and round 2. | Track the latest trends in open source with InfoWorld's Open Sources blog and Technology: Open Source newsletter. ]
This is not designed to be an exhaustive tutorial, but is structured to introduce you to shells in general and Bash in particular, touching on many common aspects and elements of the shell. It is intended to provide a basic understanding of Bash, from the shell prompt to beginning scripting.
Command-line essentialsWhen you ssh into a Unix-like system, the system looks up your user account information and spawns a new shell process based on your preferences. This shell process is then connected to the ssh session, and you've logged into the system. The shell then presents you with a prompt:
This prompt is fully configurable, and the different distributions have different ways of presenting it. Without any configuration, the prompt would simply be $. The dollar sign signifies that the shell session has the privileges of a regular user, not a privileged user or administrator. If you log in to a root shell, the dollar sign changes to a hash:
The prompts above contain the username, the @ symbol, and the hostname of the system, followed by the current working directory. In this case, we're in our home directory, which is represented by the tilde. There are all kinds of ways that you can modify your prompt to better suit your workflow. The settings are stored in the PS1 environment variable. The prompt shown above is constructed like this:
This means that the shell expands \u to the username, \h to the hostname, and \W to the working directory. Other special characters can be added to the prompt as well. You can find a list in the Bash man page (more on that later).
But we now have a Bash prompt, and we can start using the shell to navigate around the system.
The basic file system navigation command is cd, short for "change directory." This is fairly self-explanatory, though note that cd .. will move you down a directory in the tree and just cd (when not followed by a directory name) will return you to your home directory. The other most common file system interaction is ls, short for "list." The ls command will show the files and directories within the current working directory.
Bash supports tab completion, which means that if you type only a few characters of a filename or command, then hit the Tab key, Bash will automatically complete the rest of the name. Tab completion often comes in handy, especially with large file names. Bash is smart about this. It knows when you're trying to enter a command or reference a file, and if you don't type enough unique characters to define a single command or file, it will show you the options that fit the characters you entered.
A few command options come in handy when performing normal file system navigation. The -la option causes ls to show more information about the files and directories in the current directory. It will show permissions, owner, size, and the time stamp for each file and directory. Another handy command is -lat, which will show a list of files with the newest files at the top:
[myname@lab1 ~]$ ls -lat
You might find that in some directories there are so many files that the list scrolls off the top of the terminal window. This is where one of the most fundamental aspects of shell interaction comes into play: the Unix pipe.
The pipe is represented by the vertical bar, |. It functions just as you might guess, allowing the output of one command to be "piped" into the input of another command. This allows you to string together many commands at once to modify the output on the screen, to cause certain actions to be taken on a set of files, or something similar.
For example, we might want to use the less command to introduce pagination into our file listing. Thus, we would enter:
[myname@lab1 ~]$ ls -lat | less
This would pass the output of the ls -lat command to less. The less command allows us to view the output a page at a time by pressing the space bar or to advance the text by pressing the Enter key.
Another common example of using a pipe is to manipulate entries in a text file. For instance, we might have a file named names.txt that has hundreds or thousands of names entered, one per line, like so:
Notice that the file contains some duplicates. Let's say we want to alphabetize the list and remove the duplicate entries. This can quickly and easily be done with one line:
[myname@lab1 ~]$ sort names.txt | uniq
This command causes the shell to invoke the sort tool to sort the list, then pipe the output to uniq, which removes all the duplicate lines. Now, we probably want that output to go into another file, so we'd instruct the shell to create a new file and place the output there:
[myname@lab1 ~]$ sort names.txt | uniq > sortednames.txt
Here we've used shell output redirection to create the new file called sortednames.txt with the output of the previous commands. The shell uses < and > as directional markers, as well as << and >>. There's a vital difference between these, as the single > will cause a file to be overwritten, whereas >> will cause the output to be appended to any data already in the file. It's important to get this straight early on, or you may find that you have unwittingly deleted the contents of a file that you intended to add content to.
The < symbol is used to direct text or command output from one file to another or back to a program. For instance, you might have a file that contains some commands that you want another program to run. Thus, you might enter:
[myname@lab1 ~]$ myprogram < ./mycommands.txt
Many other Bash operators can be used in normal shell sessions or in Bash scripting, but we've covered the major ones. A full listing can be found in the Bash man page, which is accessible by typing man bash. It's a lengthy document, but then again, shells do quite a lot, and all of that functionality needs to be documented.
Special Bash filesThere are a few special files in your home directory that Bash reads when you start a new shell. These are .bash_profile, .bashrc, .bash_history, and .bash_logout. Note that the file names start with a period, which designates them as hidden files. To see them in a directory listing, you have to use the -a switch with ls, or ls -a.
The .bash_history file contains the commands you've entered into the shell. These can be viewed by running the history command. The .bash_profile and .bashrc files contain special instructions to create your environment by setting aliases, environment variables such as your path, and any other variables you wish to set or commands you want to run. For instance, you might add this line to .bashrc:
export PS1='[\u@\h \W - \d]\
It would set your prompt to show the date:
[myname@lab1 ~ - Mon Feb 25]$
You can also set aliases in your .bashrc file. Aliases are a handy way to reduce typing for common commands. For instance, if you find that you're constantly having to type something like cd ~/myfiles/documents/project1/data, you might want add an alias to your .bashrc like so:
alias cdp1=' cd ~/myfiles/documents/project1/data'
Then you would only have to type cdp1 to navigate to the directory.
The purpose of the .bash_logout file is to store commands to be run when you terminate the shell, but it's not used often.
File globbingGlobs are Bash's method of pattern matching. Bash doesn't have regular expressions, so globs are used to make it easier to work on multiple files that match a single pattern.
A simple glob would be file.*. The asterisk matches anything, so this expression would match files named file.1 and file.reallylongnamehere. Globs can get more complex and thus more useful. A glob such as file.? narrows the options. This would match file.1 or file.2 or file.a, but would not match file.11 or file.ab or file.reallylongnamehere. This is because the ? denotes a single character.
You can also use square brackets to reference multiple options, so file.[ab] would match file.a and file.b but not file.c. Further, you can use brackets to select based on ranges of the alphabet or numbers. Thus, file.[a-d] would match file.a, file.b, file.c, and file.d but not file.e. You can also negate a match with the ^ character. Thus, file.[^ab] would match neither file.a nor file.b, but it would match file.c.
File globs are very different from regular expressions, but they are very handy when working with many files or in scripting.
Standard output and standard errorThe two main methods that Bash uses to pass output from programs or scripts back to your session are called stdout and stderr. These are used for normal output and error reporting, respectively. If a program is running normally, it will use stdout to communicate with the session, but if something fails, it will send errors through stderr instead. Among other things, this allows you to redirect normal or error output from any program to a file instead of the screen.
You generally don't have to worry too much about stdout and stderr when you're just getting started, but understanding what they are and what they do will be important when you get deeper into Bash.
Background processing and job controlAlthough the shell might seem one-dimensional, in fact you can run many different jobs at the same time. For instance, say you wanted to run a program that you knew would take a long time to complete, and you had other things to do in the shell session in the meantime -- no need to just sit and wait. Rather, tell Bash to run the process in the background. You do this by adding an ampersand after the command line:
[myname@lab1 ~]$ ./longprocess.bin &
The response below the command is Bash's notification that you've started job No. 1, and its process ID is 3104. The process will continue to run in the background until it completes. When it's done, Bash will let you know:
+ Done ./longprocess.bin
If you had not added the & when you ran longprocess.bin, you would not get a Bash prompt back until the job completed.
You can bring a background job back into the foreground by using the fg command. Likewise, you can place a foreground job into the background by using the bg command. You can stop a process that's running in the foreground by hitting Ctrl-Z, then place it in the background by entering bg %1:
[myname@lab1 ~]$ ./longprocess.bin
+ Stopped ./longprocess.bin[myname@lab1 ~]$ bg %1+ ./longprocess.bin[myname@lab1 ~]$
You can view a list of running jobs by using the jobs command:
[myname@lab1 ~]$ jobs
+ Running ./longprocess1.bin+ Running ./longprocess2.bin -afG+ Running ./longprocess3.bin -rT
You can use the fg command to bring one of them back into the foreground:
[myname@lab1 ~]$ fg %1./longprocess1.bin
Note that any output produced by background processes will be shown in the shell, which can get messy if there's a lot of it. Fortunately, we can use our operators to tell Bash to send the output to /dev/null, which will prevent it from displaying in our session:
[myname@lab1 ~]$ ./longprocess1.bin > /dev/null &
This will still show us error messages sent to stderr, but not output sent to stdout. If we wanted to redirect stderr as well, we would type this:
[myname@lab1 ~]$ ./longprocess1.bin > /dev/null 2>&1 &
This tells Bash to send stderr (2) to the same place as stdout (1).
Bash loopsOne of the major functions that Bash provides is the ability to run commands in a loop. Loops allow you to quickly perform many functions on a file or set of files or just for general output.
For instance, you might use a loop to systematically and, very quickly, rename a large number of files. Let's say we have a directory full of files like so:
These files have served their purpose, and now we need to rename each of them to Info#.txt.old. Rather than manually typing in each name and using the mv (or move) command, we would be better off using a for loop:
for i in 'ls *.txt'; do mv $i $i.old; done
This is a very simple one-line loop command that will rename all the files ending in .txt to their original name with .old appended at the end. Thus, our files are now named Info1.txt.old, Info2.txt.old, and so forth.