Module 4: Scheduler

Slurm is the workload manager for the CoSTAR cluster. This module bridges the gap between theory and practice, showing you how to request compute resources appropriately, submit jobs, and monitor them as they run.

Learning Goals

By the end of this module, you will be able to:

• Explain the core mechanics of the Slurm workload manager and why it is required.

• Compose, submit, monitor, and cancel batch jobs.

• Estimate and request the compute resources your workloads require, including CPUs, GPUs, memory, and time.

• Apply efficiency best practices to improve queue priority and minimise idle wall time.

Module Navigation

4.1 Scheduler Basics & Slides

Learn the core Slurm commands, partition layouts, and resource request workflow.

Scheduler

4.2 Module Quiz

Check your understanding of basic job submission and queue states.

Scheduler Self-Check