The MM server can monitor key performance indicators provided in the metric charts and tables, and generate notifications (alerts) when a specified condition is met.
See Working with Panels and Panes for some examples of charts and tables.
Alert conditions are met by comparing a specified metric value with a reference value. The reference value can be a constant, or it can be another value in the same pane. As an example of a constant reference, you can configure an alert to trigger when the number of deadlocked threads exceeds a certain number. As an example of a reference that is another value in a pane, you could configure an alert to trigger when Used memory exceeds 95% of Max memory.
Each alert has an alert level (critical,
warning, and
normal), and a configurable message. Messages use the Java Message Format syntax. You can, as desired, set up three alerts of different severity for the same metric, using different threshold values.
See Alert Configuration Reference for more details on the settings referenced in this procedure.