Every integration(component) has a set of KPI metrics that are set as default by PacketAI. These are the metrics that are responsible for raising incidents. The users can adjust the sensitivity of these metrics and even turn them off so they don't raise incidents.
Default list of KPI metrics for a Linux host component
The KPI metric filters for a targeted component can be accessed by going to the Integrations pageand clicking on the Info button of each component integration card. They are located under the metric tab.
KPI metric filters table consists of:
Active toggle - allows you to disable the KPI. If an anomaly is detected on the KPI it will still be registered but it wont raise an issue
Stats - the number of anomalies of the KPI in the last 14 days.
Sensitivity - allows you to change the threshold of the anomaly score for which an anomaly will be considered important and raise incidents. The default ML anomaly score threshold is set at 90%.
Expected - the expected number of anomalies of the KPI you will get by changing the sensitivity.
Globalize - allows you to generalize the KPI filter options to any component of the same type in your infrastructure (Linux, MySQL, Apache...). Examples:
If you turn the active filter off and choose the global option for a specific Linux host metric it means that all of your other Linux hosts will have it turned off.
Same is also applied if you keep the active filter on and change the sensitivity. That sensitivity will be applied to all other Linux hosts.
After adjusting the options for a component click on the Update filters buttons to save them.