How do you guys handle applying them to production machines, this has been a bit of a debate latley in my lug and in some of the classes /I take.
Do you test before applying them, how much?

In most professional environments there are test systems on which things should be done before attempted in Production. The theory being it is better to crash the test system than the Production system and you can find the things to look out for in Production implementation.

And for this purpose a test system can also be a productive system if you use something like vmware or vbox whatever virtualization. There you can test the Updates w/o any problems.

I'm sure some folks are doing it that way but if it were me I'd want the physical hardware to be separate. More than once in my career I've been stymied on moving ahead because tie ins between dev/test and prod that seemed trivial at first kept me from doing things (e.g. rebooting the physical server for some reason).

Of course there is always a limit to this idea. (e.g. you can usually not have separate data centers for Prod and Dev/Test but at at least one company I worked at we actually did for a while have Prod in a different location than Dev/Test.)

You could always take a look at a Patch management guide.. such as Linux Patch Management by Michael Jang. Maybe your local library has a copy of this ?

It should cover recommended procedures such as testing etc..