Sometimes it's less important to monitor and alert on every single host than it is to know the overall state of a collection of hosts. Cluster alerts allow you to define this kind of alerting, for example, "Alert me when more than 5 of my batch servers are more than 80% busy".

To create Cluster alerts, select a group in the host tab, click the tool icon for the datasource you want to create a cluster alert for, and select Manage Cluster Alerts:

A form will show, listing any existing cluster alerts. To create a new one, click the Add button.

A form displays allowing you to define a new cluster alert:

The fields on this form mean:

Datapoint: The datapoint on the selected datasource to define the cluster alert for.

Enable: Enable or disable this cluster alert

Suppress Individual Alerts: if selected, then only the cluster alerts will be sent - the individual host alerts will be suppressed from escalation.

Evaluate Alerts based on: the options are either the number of hosts in alert, or the number of instances in alert (regardless of how they are spread across the hosts.)

Min Alert level: Define whether a minimum of warning, error or critical alerts should trigger the cluster alert

Threshold: a standard alert threshold, specifying the warning, error  or critical levels for the cluster alert

Values in threshold: you can select whether the defined threshold is an absolute number, or a percentage of cluster members.

For example, to define an alert to trigger a warning when more than 20 servers in a group are in an error state for CPU load, you would define the cluster alert like this:

 

Cluster alerts will show in the alert tab as being from the virtual host "cluster", and this can also be used as the host name to route them for alert rules.