[collectd] enhancment for threshold (make a notice when value is in range)

[Peter Holik]

>> > yes, I'am implementing a monitor system, which opens a ticket in our
>> > ibm tivoli enterprise console, so our on-call duty can react on it.
>> > If for example the load reach a high value, a ticket is opened.
>> > But when the load falls down after a short time, my monitoring system
>> > can close the ticket, before anybody has to react on it.
>>
>> Okay. Imho it makes more sense then to get some kind of notification, if
>> a value turns back to state "OKAY" when it was it any other state
>> before. That notification could include information about how long the
>> value stayed in that other state. This could even be more generic if a
>> notification is generated for each state change.

Usually i'm interested if a threshold (ex. ping) stays for N sec time
and not a single peak.

Maybe this idea gets included into collectd?

cu Peter