Task Queue Processing Script Best Practices

I have this project where I have a task queue that holds data for some computationally heavy tasks. I was wondering if there are any best practices when writing the task queue processing script.

As of right now, I was just planning on having a script that runs forever (with some error handling) on a server and perhaps some sort of system process to restart the script if the script does crash for whatever reason.

Is there anything else I should be doing or is it really as simple as that?