We monitor a system that has tasks. One of the metrics we monitor is failed tasks. It is a value that can only go up until the system is rebooted. E.g. the current value is 1162, if another tasks fails, it will be 1163.
Users want to be alerted if the value increases (goes from 1162 to 1163) and want to keep receiving alerts until the events are acknowledged. After acknowledgement, users no longer want to receive any alerts until the value increases again.
We created an alert that compares the previous value, but the alert stops after the initial increase.