There are a multitude of possible reasons why your job might appear to be stuck in the queue. Sometimes the cluster is simply very busy so wait times can become very long. Keep in mind that the job scheduler utilizes an algorithm that takes into account how long a job has been waiting, the amount of resources a job is requesting, and the user's recent level of cluster usage when deciding which jobs to prioritize. In general, jobs that request fewer nodes and a shorter wall time will be prioritized over jobs that request more nodes and a longer wall time.

Sometimes your job queue is empty (as indicated by the command: sinfo), but your job still doesn't run. This is to be expected. Certain sets of queues utilize the same set of nodes so you may have to wait for jobs in another queue to run before yours can.

You can find info about your job and its wait times using the squeue command.

For more information on the different queues available on SeaWulf, please see this FAQ item.