I call it debugging the Sherlock style

Once something unexpected has happened, evaluating the evidence to find the actual story

This is particularly true for long automation batch runs where a script passes in a single run, but consistently failing in the batch

I’ve observed people mostly just hope running the script again and again will reveal the problem itself

Taking a moment to understand what might have gone wrong, and doing some targeted debugging can save a lot of time

My thoughts on why to analyze and how to do it:

Debugging the Sherlock Style