Cluster Quick Start#
This guide contains the minimal steps to get started running computational workflows, with links to further reading included. If you're a first time user, please follow the links and review the full documentation.
Need help getting started? Send us an email at help-rcc@mcw.edu.
Getting an account#
You need an RCC account to get started. Please see Getting an Account for details.
Logging in#
Login is available both on and off campus via SSH and Open OnDemand. Please see the Login guide for details.
Clusters are shared resources.
Please be respectful of all other users. Do not start resource-intensive scripts on a login node. Do not request more resources than your job can use. For more info see User Etiquette.
Cluster Storage#
Your account has access to a set of storage directories by default. You can easily find your available storage paths and current utilization on the cluster with the mydisks command. Please see the storage guide for details.
$ mydisks
=====My Lab=====
Size Used Avail Use% File
47G 29G 19G 61% /home/user
932G 158G 774G 17% /group/pi
4.6T 0 4.6T 0% /scratch/g/pi
Transferring Files#
Most users will need to transfer files to cluster storage. Several methods are available depending on your need. These include Open OnDemand, SSH (command-line), and desktop client software. Please see the file transfer guide for details.
Using Software#
Software is managed by modules using Lmod. The modules contain information about an application's version, executable, libraries, and documentation. Using module commands, users can add, remove, or switch versions of a application. Please see the cluster software guide for more information.
Running a SLURM Job#
The clusters use SLURM to manage jobs. Most jobs are scheduled using batch job scripts. A batch job script includes a request for cluster resources, and the commands to run in the job. Each batch job is submitted by the user for remote execution on a compute node. The SLURM scheduler decides where the job should run, based on the requested and available resources. Please see the SLURM job guide for details.
Write a job script#
The SLURM job script syntax is shown below. Just like previous clusters, every job needs a job script, which tells the scheduler how and when to run your workload. The following is a simple SLURM batch job script example.
#!/bin/bash
#SBATCH --job-name=test-job
#SBATCH --ntasks=1
#SBATCH --mem-per-cpu=1gb
#SBATCH --time=00:01:00
#SBATCH --account=PI_NetID
#SBATCH --output=%x-%j.out
#SBATCH --mail-type=ALL
#SBATCH --mail-user=NetID@mcw.edu # NetID is your username
echo "Starting at $(date)"
echo "Job name: ${SLURM_JOB_NAME}, Job ID: ${SLURM_JOB_ID}"
echo "I have ${SLURM_CPUS_ON_NODE} CPUs on compute node $(hostname -s)"
Submit the job#
Now I can submit my example job:
sbatch test-job.slurm
Then I can check my job status:
squeue
More info#
This guide has links to additional information about each topic. We strongly recommend to review all documentation.
Need help getting started? Send us an email at help-rcc@mcw.edu.