I can’t count the number of times that I’ve started working with a new service and discovered that they had an alert configured to page based on the number of HTTP 404s served. Every single time I’ve found a 404s-too-high alert, I’ve deleted it entirely from monitoring. It’s too noisy of a signal; there’s just too much random traffic on the Internet probing for security vulnerabilities to make 404 monitoring even slightly useful. Most of the time, at best we end up with a graph of...
