But the point is that warnings are often not good predictors of problems. There's a good argument that "warning" on anything at all is an anti-pattern, because you're just attracting valuable attention towards something that may or may not be anything to worry about.

Where warnings are useful are when "something is actually now going wrong" - they provide a valuable context that can help an engineer figure out what the issue might be, which is exactly how it worked in this case.