I'm trying to create a filter policy to generate incidents only for devices in a specific global collection. It works but only for alarms of the device itself. It does not generate alarms for the device interfaces or any "application model". Even though I understand this behavior it makes the filter useless.
I know I could create another global collection for all the interfaces of the devices, but how about all the other possible "submodels" of the device? Do I need to map all the possible submodels and create global collections? This is very inconvenient and does not scale well.