The gap in traditional monitoring
Typical monitoring is good at catching errors but struggles with other failure conditions such as:
Process not starting
Process getting stuck half way
Process starting late
Process taking longer than expected to complete
Process completing incorrectly
Process Warden works differently, it monitors for success:
Configure your processes to send success emails/notifications to Process Warden
Tell Process Warden what 'success' looks like (e.g. for a backup, an hourly email with 'Succesfully Backed Up' in the subject)
Process Warden will alert you if it fails to recieve the 'success' notification for any reason
The end result is that not receiving an alert from Process Warden means that your processes are definitely working. You can have peace of mind that things are completing as expected without any additional email noise.
Why the model is different
Anything apart from success is failure
At the heart of our new model is the idea that anything apart from a confirmed success should be considered a failure. Do you rely on error reporting? What if the process never runs at all? What does an amber status even mean? Should you worry about it? Did that process complete or did it just start and get stuck?
By modelling anything but success as failure you get peace of mind that a quiet process is a working process. Anything else will light up and show you that something is wrong - that success has not been confirmed.
Only spend time on what needs attention
The gist of it is simple - if nothing is lit up, nothing needs attention. If something is lit up then it clearly didn't confirm success so it needs some sort of attention to get it running again. You can colour code the indicator lights to reflect how you should feel about it lighting up. For example, our backup processes light up in red but our contact forms or successful deployments light up in green.
Set and forget by default
The point of monitoring is to find problems and fix them, not to spend time monitoring. If everything is working then the time spent monitoring should be minimised as much as possible. Simply configure the indicator rules for your process and you never have to think about it again until it lights up. If the process stops working (or stops sending positive success confirmations) then you'll learn about it. If you never see it light up then congratulations, the process is working and you haven't even had to think about it.
Do you use a ticketing workflow? Simply configure Process Warden to send emails to the appropriate support address and you don't even have to look at the dashboard - you can just sit back and wait for anything that needs attention to drop straight into your inbox. It won't spam you with useless positive reports and noisy log output; you'll simply get an email if a process fails.