HPC Terminology
Here are some common terms you will encounter as you get accustomed to using our HPC resources:
Batch Scheduler / Job Scheduler: When you submit a job to run on the supercomputer, it gets placed in a queue along with other jobs from other users. A job scheduler (also referred to as a batch scheduler or workload manager) manages this queue and will execute your job once the needed resources become available. OSCER uses SLURM as our job scheduler. More information about SLURM, and how to use it, can be found in our Support article - Running Jobs on Schooner.
Compute nodes: These are the nodes where your jobs are run. OSCER has over 400 compute nodes to meet a variety of your needs. More infomration about our compute nodes can be found here.
Home folder: Every OSCER user is given a personal folder where files can be stored long-term. Home directories typically have a 20GB quota and is backed up nightly. You can access your home folder by going to - /home/username. This is also the default folder location when you log into Schooner.
Login node: This is your entry point into the OSCER supercomputer. Whenever you log into Schooner, you are logging into the login node. Here you can perfrom various tasks, such as - viewing your files & folders, submitting jobs, checking the queue, monitoring your pending and running jobs, etc. A list of all the login nodes (and recommended uses) can be found here.
Queue: All jobs submitted to the supercomputer have to go through a queue. OSCER maintains several public and private queues. Public queues are accessible by everyone, and generally follows a “first in, first out” rule. Private queues are only accessible to the group for whom the queue was set up. A list of all publicly accessible queues can be found here. Please contact us at support@oscer.ou.edu if you have questions regarding which queue offers the best performance for your job.
Schooner: Schooner refers to the current OU supercomputer. You can get more details about our supercomputer by clicking here - http://ou.edu/content/oscer/resources/hpc.html
Scratch folder: scratch is meant to be a space for temporary files, even large ones. It is ideal if you are running jobs that do a lot of file access (reading, writing, etc.). Each user has their own scratch folder, usually - /scratch/username. We strongly recommend users to treat scratch as a temporary storage. We do not back up the scratch folder. While OSCER currently does not enforce quotas on /scratch, we do have a two-week deletion policy. Please see the Schooner Scratch Autodelete Policy for more information. If you think you'll need larger storage capacity than the 20GB on your Home folder, you can find all the storage options here.
SLURM: SLURM is the name of batch scheduler (job scheduler) that OSCER uses on our supercomputer. More information about SLURM, and how to use it, can be found on our Support article - Running Jobs on Schooner.
