Best options are usually using multiple job queues and placing jobs for different applications into different job queues. This way if jobs are stuck in one queue they don't affect other queues and applications. Now there could billions of reasons why jobs get stuck, such as network outages, file locks, resource leaking, etc... In 99% of all cases running jobs synchronous and detaches in multiple queues ensures optimal performance, application independence and solid stability. Recycling scheduling server every week or two also helps to improve overall system stability and performance. : I was using a single queue, : but based on other postings I have setup queues : for different report categories. : Will see how that works going forward. : Generally use notification on error only, : and use send e-mail message if any notification is used.
|