Print-friendly Version

Automatically Saving Script Output Using the tee Utility

István Bátori, August 2007

Introduction

This tip helps you to save script output by using the tee utility. You can use the script I provide below under "Installation." This tip corresponds to the tip Automatically Saving Script Output Using the nohup Utility.

I have created tons of scripts for administrators, and I was interested in how script output can be saved in an easy, elegant way. Scripts can have their own logging functions, but the output also contains important information. Saving the output is typically solved by output redirection, for example:

myscript.sh 2>&1 | tee /var/log/myscript.log

How do you feel about typing such a long command? Isn't it boring? It's a lot to type. And it's confusing to read the long, uninteresting ballast in an admin document, especially if the script has many arguments and options and the log file contains the start date.

The tee.sh wrapper script simplifies the process and hides the tee command from the user. You only need to insert the following line as the first line of your script:

#!/usr/bin/ksh /opt/bin/tee.sh

That's all! The script is started automatically by tee and the output is redirected.

This method is usable for shell, Perl, Python, or Ruby scripts, and it is easy to extend for other script languages.

First, let me explain the trick I use.

Script File Execution

UNIX executes a script file according to its first line:

#! <exec_command> [<args...>]

The system invokes the first line and passes the script file name and the command-line arguments. <exec_command> needs to be an executable binary file. It cannot be a script file (although some Linux distributions support using a script file).

For example, the foo.sh script file starts with the following line:

#!/usr/bin/ksh

If you type this at the command prompt:

foo.sh /tmp 3 1

The following is invoked:

/usr/bin/ksh foo.sh /tmp 3 1

If foo.sh is started with this:

#!/usr/bin/ksh /opt/bin/tee.sh

The following is invoked:

/usr/bin/ksh /opt/bin/tee.sh foo.sh /tmp 3 1

tee.sh gets the command-line parameters and it calls foo.sh.

tee.sh Installation and Usage

Installation

Copy the tee.txt file, changing the ".txt" suffix to ".sh", and save into the /opt/bin directory. Then give it execute rights.

cp tee.sh /opt/bin
chmod 755 /opt/bin/tee.sh
chown root:admin /opt/bin/tee.sh

Usage

There are two different ways to use the wrapper script:

1. Start the wrapper script directly from the command line:

/opt/bin/tee.sh <command> [<args...>]

For example:

/opt/bin/tee.sh foo.sh

2. Put the following into a script as the first line:

#!/usr/bin/ksh /opt/bin/tee.sh

For example:

cat >/tmp/tee_test.sh <<EOF
#!/usr/bin/ksh /opt/bin/tee.sh
# example script to demonstrate tee.sh usage
echo "hello world"
EOF

Executing tee_test.sh automatically calls tee.sh, which starts the script with the tee utility and redirects the script output into /var/tmp/tee_test_<date>.log.

Configuration

Both usage methods have the same functionality. The functionality can be configured by shell variables, as follows:

• tee.sh starts the script with the interpreter specified in the $TEE_SHELL variable. If the variable is not set, the default interpreter is started according to the script file's extension:
• A .sh script is started with /usr/bin/ksh.
• A .pl script is started with /usr/bin/perl.
• A .rb script is started with /usr/bin/ruby.
• A .py script is started with /usr/bin/python.
• Output is redirected into the directory specified by $TEE_LOGDIR. If $TEE_LOGDIR is not set, the /var/tmp directory is used.
• The log file name is the script name postfixed with the date, for example, foo_2007_03_23_11_11_22.log.

Note: There is a bug (but it's not really a big one): STDERR is redirected to STDOUT, and it is not possible to separate them.

Unless otherwise licensed, code in all technical manuals herein (including articles, FAQs, samples) is provided under this License.