How to define the amount of time for an alarm to be generated in a device?

Document ID : KB000111749
Last Modified Date : 11/09/2018
Show Technical Document Details
Sometimes, due to business reasons, it can be necessary to adjust the amount of time before an alarm is asserted on in a device/host model.
This article explains how to calculate/define it.

Spectrum 10.x
In OneClick Console, select the desired device in Topology -> go to "Component Detail" section -> "Information" tab -> "CA Spectrum Modeling Information".

You will see the following values : 

Poll Interval(sec)
DCM Timeout(ms)
DCM Retry Count 

Device polling parameters

The formula to get the amount of time is:

poll interval sec + (DCM Timeout x DCM retry count)


Let's consider the following values:

Poll Interval(sec) = 300
DCM Timeout(ms) = 30000 (30 seconds)
DCM Retry Count = 3

The alarm will be generated after 300 + (30x3)  seconds = 390 seconds

IMPORTANT: This amount of time may not represent the real total time since the device has stopped responding to polls.

The amount of 390 seconds mentioned above only would be real if the device stopped to respond right after (less than one second) the last successful polling.

Let's consider the following scenario, using the same parameters above (Polling interval=300, DCM timeout=30000, DCM retry count=3):

The device was successfully polled at 12:00 AM. So, the next polling will happen at 12:05 AM (300 seconds after). But the device stopped responding at 12:02 AM.

Based on that, after the device stopped responding, Spectrum will still wait for more 3 minutes before polling the device again.

As a result, the alarm will be raised at 12:06:30 AM (390 seconds after the latest successful polling, 270 seconds after the device stopped responding to polls).