by reconstruction the core servers with more than 5000 Sensors, we went to set intervalls on sensors with a value setting for warning / error.
Like this example: CPU - Error by 95 % ; Warning by 75 %; Intervall 90s, go up to an Error state after 3 intervals (270s) to notify.
Acutally it is so, that this kind of sensors only check the Sensor configured on the Root or in der Group settings for denpending intervals or directly in the Sensor.
But the measurement are only on one time and if the result is "warning", the shortly following second measurement goes to an error state, when it is further in a ongoing warning state.
We need this intervals too, for Errors & Warnings with defined intervall options. functional like a connection sensor, which is supporting manual interval definitions.