I wouldn't recommend changing worker threads. 0 is a good default, it shouldn't need changing. Identifying why you're running out of worker threads should be a priority.

Maybe schedule a job to run every couple minutes that runs various diagnostic scripts and inserts results into a monitoring database.