Various stored objects in OA4MP such as clients, admin clients, ersatz clients and virtual organizations will have their access times logged and updated. Given the nature of the server, this is done by keeping a running log of these accesses and periodically doing a batch update on persistent storage. This keeps the server load to a minimum, since a typical use pattern is that a client is accessed very often at the beginning of a flow then may go a very long time (say awaiting an exchange) before being accessed again. Caching all of the intial accesses into a single call is a much less impactful way of doign it.
The server tag contains three attributes, enableMonitor, monitorInterval and monitorAlarms that control the behavior. You must enable this component to use it, then either set an interval so that the monitor wakes up every interval and runs, or you set specific alarms to run. Alarms are given preferences, so if you set both, you will get alarms. A plus with alarms is that you can schedule them to run when you know server load is low.
You may look at the last accessed time for the object in the CLI. If it has been accessed, then it will have the property last_accessed. You can use this to schedule unused client cleanups with various parameters for standard clients. This unused client cleanup is also available in the CLI,