Fairshare & job priority

Overview

The clusters use a fairshare scheduling system to allocate resources fairly across all users. Job priority reflects historical usage, job size, and requested time — understanding it helps you reduce wait times and maximize throughput.

Key concepts

• Fairshare — users who have consumed fewer resources recently receive a higher fairshare score.
• Scheduling priority — determines the order in which jobs are considered for scheduling based on factors such as queue, requested resources, walltime, and fairshare.
• Priority access — certain queues give specific groups higher scheduling priority when resources are constrained.

Tips for maximizing throughput

• Submit several smaller jobs instead of one large job — multiple smaller jobs often start faster
• Select queues that match your job's walltime and node requirements
• Space out submissions rather than sending many simultaneously
• Monitor your position with squeue -u <username> and fairshare with sshare -u $USER
• Use job arrays for repeated tasks to reduce queue overhead

Balancing job size, queue selection, and fairshare awareness helps ensure faster start times and efficient use of the clusters.