I've testing SiteStats and have found a "curious" issue. When an exception is launched while SST is processing an event, SAKAI_EVENT processing is stop and SST_JOB_RUN row is inserted with end_event_id set to zero and job_end_date set to null. This makes the following execution to fail as it tries to start at SAKAI_EVENT.ID = 1. To solve it I attach a patch that skips problematic events, that is, events launching any kind of exception. In this way, SAKAI_EVENT mapping keeps working though you find a problematic event. At least for UPV this behaviour is preferred than to abort job run due to the exception.
I attach the patch to correct this behaviour.