Web Admin Blog Real Web Admins. Real World Experience.

8Jan/101

Stupid Unix Trick – Command Mashups

I've been a *nix Administrator in some form or fashion for about 10 years now.  I remember back when I was first learning commands and how the OS works and every once in a while I'd come across something stupidly simple yet extremely useful to put in my bag of tricks.  Yesterday I was reminded about one of those things and I figured I'd share it here so that you can throw it in your bag of tricks as well if it's not already in there.

To start out, let me illustrate the problem.  You are writing a shell script or running a series of commands on the CLI.  Let's just say it's something simple like creating a new directory, changing to that directory, and then creating a file.  When I first started out, that command would look something like this:

mkdir newdirectory; cd newdirectory; touch newfile

The problem with this is that each command is executed on it's own regardless of whether or not the previous command was successful.  So if, for example, my mkdir and cd failed (permissions maybe?), I would be creating that newfile in whatever directory I started out in.  At best, I just created a new file in the wrong directory.  At worst, if the file which I'm creating was the same name as another file already in the current directory, I just overwrote it.  Not good!

The way to fix this is to add a dependency so that each command will not execute without the successful return of the command before that.  The way you do this is by putting an "&&" between them instead of the semi-colon.  So now the command string above should look like this:

mkdir newdirectory && cd newdirectory && touch newfile

Now you have guaranteed that the new file will not be created with the touch command unless both the mkdir and cd commands before it are successful.  Stupid simple, right?  Enjoy!

About Josh

Josh graduated in 2002 from the University of Texas at Austin with a BS in Computer Science. He was formerly a member of the Internet Systems team at AMD, a Systems Administrator with BearingPoint, and worked on a contract for the US Army before becoming a member of the Web Systems team at National Instruments in January of 2007. He recently attained his CISSP certification and specializes in the area of web application security.
Comments (1) Trackbacks (0)
  1. Relying on non-zero-exit is such a great idiom and its consistent use will add up to more reliable and predictable outcomes. It’s especially important when one script relies on calls to another.

    Along with the “&&” (and) expression, I also like to use the the “OR” expression to manage error cases. For example this usage:

    statement1 || or-statement

    This will cause the “or-statement” to run if the “statement1″ command fails (ie, exits non-zero).

    One can combine the “OR” and “AND” expressions with code blocks for a short hand if/then/else syntax

    statement1 &|| {
    else-statement1; else-statement2
    }

    Here’s a concrete example that checks if an error string is found in a log file

    grep ERROR /path/to/log &|| {
    echo “error found on `hostname`:/path/log/log|mail -s “error detected”
    exit 1
    }


Leave a comment


No trackbacks yet.