Description: This class provides an overview of the scheduler system used to schedule the execution of jobs on the ACCRE High Performance Computing cluster environment. Details will be provided regarding how to track your own job submissions and how to take optimal advantage of some of the scheduler options.
Required for ACCRE Use: Yes. This class is required of all new users for continued use of the cluster (please read the training requirements).
Prerequisites: Yes. Fluency with Unix/Linux is a necessary prerequisite for this class. If you are not fully comfortable working in the Unix/Linux environment, you must also attend the Introduction to Unix/Linux. You should also have already attended the required Introduction to the Compute Cluster. In addition, the class is aimed at users who have already begun submitting jobs for execution on the cluster. Therefore, we advise you wait until the following series (or later) after first attending the Introduction to the Compute Cluster. If you do not yet have an account, please feel free to request one.
Format: "Job Scheduler Details" is conducted in one 2-hour session of hands-on, instructor-led presentations and demonstrations. Participants are strongly encouraged to bring their laptops or use a laptop available in the training classroom to follow along with the demonstrations. If you already have a cluster account, know your username and password.
Cost: None
Course Content: This is a beginning level class on the scheduler. Topics covered include:
- Components of the scheduler (Torque and Moab) and their functions
- Submitting a job for execution
- Changing scheduler options after a job has been submitted
- Checking the status of a submitted job
- Checking your account's status (fairshare)
- Using information about the overall cluster status (node availability) to optimize the timing of your job's execution
Scheduler Details Presentation (PDF)
