Currently, lots of unnecessary updates are executed, as the update loop runs on a tight schedule.
Adding dynamic reconfiguration of the service interval, to a lowest common denominator of all the checked out repositories, should help to avoid excessive check executions.