Levels can be set in an alarm template to monitor a parameter, such as warning, minor, major, critical (high and low).
There is also the "normal" level and I'm wondering what the practical use is of filling in something in that field?
-When specifying a warning high of 3 (or another high level of 3) without filling in anything in the normal then as soon as the value goes below 3, the alarm severity is normal. When specifying something in the normal level, such as 1, then for a value of 2 the severity will still be normal. If there is minor low specified of -3 and minor high of +3 (no warning specified) then the severity will be normal when value > -3 and value < 3, regardless of the value in the normal level.
-When using a discreet value: if I have possibilities "A" and "B" and I set "B" as critical then value "A" automatically has the normal alarm severity, regardless whether "A" is or isn't specified in the normal level of the alarm template.
-When working with normalization, then the normal level contains [BASELINE] indication, but this could also be derived by looking at the type to know if normalization is used.
This makes me wonder if there is a use case where it does matter that something is filled in in the normal level of an alarm template?
It is exactly the way Thijs explained it. The normal value is not required to determine whether a metric is not in an alarm state (i.e. the state of a metric is always automatically considered not in alarm (green) whenever an alarm state is effectively defined on that metric and it doesn't meet the alarm condition), but it is useful to provide an optional graphical indication for the user on analog readings as to which value is typically to be expected (i.e. it is marked at that specific point with a small green marker).