I once received some sage advice from another System Administrator I worked with years ago regarding working with potentially troublesome servers.
It was back in the Exchange 5.5 days. I had a cranky server with a potentially unsolved hardware problem in the disk subsystem. Every time I powered off the server, it damaged the OS and I was forced to restore Exchange from tape. The manufacturer always replaced a part when I called for support, but I had ended up rebuilding it several times and had not yet confirmed that the latest hardware replacement resolved the issue.
My co-worker was on-site to help me set up a new server room after we relocated the office. Because of the history of the server, I was very anxious about possibly having to restore Exchange again. It was approaching lunch time and we were at the point where it was time to power on the mail server.
He turns to me and says, “We are going to press the power button and then walk out to eat without looking back.” His theory was that if the server was going to be fine, it would be fine without us watching it boot. If was going to have a problem, the problem would still be there when we returned. At the very least we would have had a relaxing lunch break and would be better able to solve a problem without the additional stress of hunger pains.
Turns out the server was fine.
To this day, I still heed that advice. If I’m about to do something to a system that has the potential to backfire, I make sure I’ve already enjoyed my lunch.