Database triggers do not run specifically on a primary or secondary node. This is because App Services triggers use change streams, which subscribe to all data changes on a single collection. A change is only committed after the data change event has been committed by a majority of the nodes and this is not configurable.