> I personally found that runbooks were even worse for small size teams (like our four person reddit team) because they would get out of data even quicker than

> I personally found that runbooks were even worse for small size teams (like our four person reddit team) because they would get out of data even quicker than at the bigger places due to the rapidly changing environment. My experience is that you can get regularly updated runbooks, but they're at the wrong level of abstraction. They'll discuss some odd one-off failure that happened to trigger a given alert, rather than the general class of problems that this alert is trying to catch. These days I consider looking at a runbook to be an act of desperation, that'll I'll only perform after attempting to debug from first principles.