Diagnostic method and root-cause thinking, as a topic area in its own right.
Why reading logs from the moment of failure backward, instead of scrolling forward from an arbitrary starting point, is faster and catches the real cause more reliably.
A step-by-step method for isolating whether a slow or failing request is a network problem, a proxy/load-balancer problem, or an application problem — before you start reading application logs.