I look on things from a game theory perspective. Is worrying about the problem helping you to solve it? Usually the answer is no, but what does help instead? I find that bisecting the issues involved is a good start. It's practically impossible to control for many different variables all at once. And you can't control what you cannot measure, so adding better means of measurement is good too. Adding extra logging of information to your system for example.
Also sometimes the problem that needs resolution isn't what you think at the outset, so keep an open mind. Maybe your implicit assumptions are at fault so you may need to go back several steps to identify and validate those first.
Indeed, the 'solution' may be to do nothing at all but rather to think about the world in a new way. The 'problem' to be solved might then disappear like mist.
Also sometimes the problem that needs resolution isn't what you think at the outset, so keep an open mind. Maybe your implicit assumptions are at fault so you may need to go back several steps to identify and validate those first.
Indeed, the 'solution' may be to do nothing at all but rather to think about the world in a new way. The 'problem' to be solved might then disappear like mist.