Yes, and maybe distribute the report creation around the hour. If you have a report with numerous nbr of tables involved, a heave reading of the db occurs. Even if there is no writing to the db, a number of threads are started. Since we do not really know where the limit of concurrent running threads are (especially inside a VM), I think you can adjust the upper limit until we see the "old" problem coming back.But I guess I may have to go down on the frequency of logging in the end.
Since the Q goes up that high, I fear that there are some threads that are not terminating correctly (could be threads used for report creation).
But spreading the execution a bit would not harm...