Need help creating an Alert based on Node thresholds

LatteLarry

Can someone advise how I would create an alert based on the thresholds for CPU that are defined in the node properties? I know how to create an alert by setting my own thresholds for CPU in the alert trigger conditions, but I want to be able to customize what triggers the alert at the node level. For example, a SQL server that we know runs high and spikes on CPU compared to all other servers. I'd like to just adjust that node threshold to a different value to prevent it from being in a near constant critical status and triggering an alert. 

m-milligan

It sounds like you want to account for two things: (1) absolute CPU utilization level, and (2) spikes in utilization. Here's what you need to do.

First, set custom thresholds for the node using the "Node Thresholds" settings. You'll want to establish thresholds based on the normal sustained CPU load for this node. For instance, if it's usually around 85% with spikes around 98%, you might want to set the warning threshold at 90% and the critical threshold at 95%.

For the alert trigger, you can use the CPU Thresholds values "Warning Value Reached" and "Critical Value Reached. In the example below, the alert will trigger when the warning threshold is reached but the critical value hasn't yet been reached:

To avoid triggering an alert for spikes in CPU utilization, check the box for "Condition must exist for more then.." and specify how long CPU utilization must remain over the threshold.