Hawkular defines its alert triggers in the UI code. This served a purpose as a quick way to get alerting in place but there are several disadvantages to this approach:

• UI session and navigation required to start alerting on discovered resources.
• Nondeterministic start to alerting.
• Unnecessary complexity in the UI code.

Furthermore, the current trigger definition approach is to define the same initial trigger to every resource of a specific type. For example, HeapUsed > 80% of HeapMax, for every JVM. But the triggers are individually managed. To change that to 90% would require either UI navigation to each JVM and a trigger update, or a custom script to make the changes via REST.

As part if the migration it is likely that we will move to using the Hawkular Alerts "Group Trigger" templating mechanism. In this way we can manage the type-level triggers as a group.

This will have some UX/UI impact.