ÁÔÆæÖØ¿Ú

When creating your job's slurm submission script(sbatch) in addition to the specific amounts of resources (RAM, CPUs, etc) you will need to specify which partition(group of nodes) that you're requesting those resources from. As a majority of our nodes are grant funded most of the nodes will have priority access given to the PIs of their grants, those lab groups will be able to submit jobs to the partitions of said grants and these will preempt(end and requeue lower priority jobs). Priority tiers are 1-3 with 1 being lowest priority able to be preempted by partitions of higher numbers. Partitions listed under (safe) do not have a specific grant/PI group associated with them and are not at risk of being preempted, however you should still aim to checkpoint your jobs.

Quality of Service queues are currently in the works and it will change the way that resources are requested, while partitions will largely stay consistent. Specific resource limitations will also be implemented.

The Partitions,the purposes, and their Priorities are as follows(note, while most do not yet have maximum time-limits implemented, jobs running longer than 7 days are at risk of being ended early for maintenance or other issues):

Partitions: