I've noticed on machines with large event logs (ie > 100mb) that applications manager seems to take excessive amounts of time to poll those machines (ie > 1 hour).
Does it process all events everytime? I can't think why else it would be so non-performant.
Additionally, I noticed that the date on alerted events is incorrect. For example I have an event from today which is 03/05/2006 (Australian format, ie 3rd of May). However, applications manager says the event is from 05/03/2006 (ie 5th of March).