When the acquisition thread fails to lock a certain number of jobs during one acquisition cycle, it backs off, meaning that it will wait a certain amount of time before making the next acquisition query, yet requesting a greater number of jobs then.

AT:

• for a process engine there is a configuration option for the initial backoff time and the backoff factor
• the acquisition thread no longer executes jobs itself
• the number of jobs rejected by the execution queue is logged
• default is no backoff (so that the acquisition thread constantly acquires new jobs and sleeps once it can't submit new jobs to the queue or there aren't any jobs left)