Monitoring Processes on the Command Line

Updated 2012-11-06 with an alternative.

Sometimes you have a long running process that is firing off many commands in a row and you don’t get much output from the command. I like to have an idea of what’s happening, so I employ a simple Bash script that I’ve hand typed a million times. It keeps just the data I want on screen and auto-updates ever couple of seconds, so I can keep an eye on it while I’m working.

while true
do
    # Some command goes here
    sleep 2
    clear
done

If you have any modern shell, you can type this in with the line breaks. Now I have a clear console that is showing me just the data I want.

How do I use this? Right now, I have an apt-mirror post mirror script that rsyncs multiple directories. This is an automated process that runs quietly, so I can use ps -ef | grep rsync to monitor the process. I’ve also used this to monitor massive file creation / deletions, so the command could be ls -l | wc -l to count the files in the director. I use Hylafax, so I can monitor the modems with faxstat -a.

An Alternative

After posting this article, it was brought to my attention that I could just use the watch command. As with a lot of complex technologies, you tend to learn to do something one way and stay oblivious to other possible ways. watch is part of the procps project, so instead of the above loop, you could use:

watch "ps -ef | grep rsync"

This by default runs the given command every 2 seconds. Check out the man page for more options.

Unfortunately, watch is not part of the default UNIX stack on OS X. It seems to be available in Homebrew, but since it isn’t already on my system, I’ll probably stick with the bash script. Old habits die hard.