If we restart while there are problems, the error count restarts at 0. This causes alerts to clear and be reset after the retry count.
Unfortunately, this will require much more complicated restart logic.
Basically, on startup, we need to get the current critical and set the failure count on checks to retry count +1.