HiPerGator Basics

HiPerGator is the University of Florida's high-performance computing (HPC) cluster, managed by UF Research Computing. It gives you access to far more computational power than a personal computer, which is essential for steps like running the nf-core/rnaseq pipeline. This section covers everything you need to know to get started on HiPerGator for this workshop.

For more comprehensive documentation, see the UF Research Computing docs.

Logging In

You connect to HiPerGator via SSH from your terminal. If you haven't set up your terminal yet, see the Command Line Basics section. Replace username with your GatorLink username:

ssh username@hpg.rc.ufl.edu

You will be prompted for your password and then a Duo two-factor authentication push. After authenticating you will land on a login node, which looks like this:

[username@login1 ~]$

Do not run work on login nodes

Login nodes are shared resources for everyone on the cluster. All real work must be submitted as SLURM jobs. Running computationally intensive work on login nodes is against HiPerGator policy and can result in account suspension.

The HiPerGator Filesystem

HiPerGator has three main storage locations, each with a different purpose:

Location	Path	Quota	Use for
Home	`/home/username`	40 GB	Scripts, configuration files, important small files
Blue	`/blue/group/username`	Varies	All active research work and data
Orange	`/orange/group/username`	Varies	Long-term archival storage

Always run your analyses from /blue — it is a high-performance storage system designed to handle the input/output demands of HPC workloads. /home and /orange are not suitable for running analyses.

Already have a HiPerGator account?

If you already have an active HiPerGator account, do not use your personal /blue directory for this workshop. Instead, create and use a personal folder inside the shared workshop directory:

mkdir /blue/bioinf_workshop/$USER
cd /blue/bioinf_workshop/$USER

For this workshop, your working directory will be:

/blue/bioinf_workshop/username

To navigate there:

cd /blue/bioinf_workshop/$USER

You can check your storage quotas with:

home_quota

blue_quota

orange_quota

Open OnDemand

Open OnDemand (OOD) is a web interface for HiPerGator available at https://ood.rc.ufl.edu. It provides a graphical file browser for navigating, uploading, downloading, and editing files, a dashboard for monitoring your jobs, and a browser-based terminal if you prefer not to use SSH. For this workshop, OOD is particularly useful for uploading input files and downloading results without needing to use scp on the command line.

Loading Software with Modules

HiPerGator uses a module system to manage software. Software is not available by default — you need to load it first.

module spider — search for available software

module spider nextflow

type q to exit the documentation

module load — load a module

module load nextflow

module list — see what you have loaded

module list

module unload — unload a module

module unload nextflow

Some software requires loading a prerequisite module first. module spider will tell you if this is the case and what to load.

Writing and Submitting SLURM Jobs

HiPerGator uses the SLURM scheduler to manage jobs. Rather than running commands directly on the login node, you write a job script that specifies the resources you need and the commands to run, then submit it to the scheduler with sbatch.

A SLURM script is just a regular bash script with a block of #SBATCH directives at the top that tell the scheduler what resources to allocate. To illustrate this, here is a simple bash script — each echo command just prints a line of text to the screen:

#!/bin/bash
echo "Hello from HiPerGator!"
echo "My username is: $(whoami)"
echo "Today's date is: $(date)"
echo "I am running on node: $(hostname)"

You can run this directly on the login node by copying this text in a file that you create and name hello.sh and run with bash hello.sh.

Now here is the exact same script as a SLURM job — the only difference is the #SBATCH header:

#!/bin/bash
#SBATCH --job-name=hello
#SBATCH --nodes=1
#SBATCH --ntasks=1
#SBATCH --cpus-per-task=1
#SBATCH --mem=1gb
#SBATCH --time=00:05:00
#SBATCH --account=bioinf_workshop
#SBATCH --qos=bioinf_workshop
#SBATCH --output=hello_%j.out
#SBATCH --error=hello_%j.err
#SBATCH --mail-type=END,FAIL
#SBATCH --mail-user=$USER@ufl.edu

echo "Hello from HiPerGator!"
echo "My username is: $(whoami)"
echo "Today's date is: $(date)"
echo "I am running on node: $(hostname)"

Now create this file on HiPerGator. Open a new file with nano:

nano hello.sbatch

Need a nano refresher?

Type or paste your content
Ctrl+O then Enter to save
Ctrl+X to exit

Paste the script above into nano, save, and exit. Then verify the file was created:

cat hello.sbatch

The key #SBATCH flags are summarized here:

Flag	Description
`--cpus-per-task`	Number of CPU cores. Only request multiple if your software can use them.
`--mem`	Total memory, e.g. `16gb`. If your job exceeds this it will be killed.
`--time`	Maximum runtime. Job will be cancelled if it exceeds this. Request ~20% more than you expect.
`--partition`	The queue to submit to. Use `[training-partition]` for this workshop.
`--account` and `--qos`	Must match your group. Use `bioinf_workshop` for this workshop.

Using burst QOS for your own group

Most HiPerGator groups have access to a burst QOS that allows you to temporarily use more resources than your group's allocation. To use it, append -b to your group name for the --qos flag:

#SBATCH --account=mygroup
#SBATCH --qos=mygroup-b

Burst jobs are lower priority and can be preempted (cancelled and requeued) if the resources are needed by higher-priority jobs, so it's best used for jobs that can be restarted or that you don't need to complete by a specific time.

Now submit it:

sbatch hello.sbatch

Submitted batch job 12345678

SLURM will give you a job ID — keep track of it. You can check the status of your job with squeue. The ST column shows job state: PD = pending, R = running, CG = completing:

squeue -u $USER

If you need to cancel the job:

scancel 12345678

Once the job completes, check the output log (replace 12345678 with your actual job ID). SLURM writes output and errors to separate log files — these are the first place to look if something goes wrong:

cat hello_12345678.out

Hello from HiPerGator!
My username is: username
Today's date is: Thu Apr 3 10:42:11 EDT 2026
I am running on node: c0709a-s30

If you want to follow the log in real time while a job is still running, use tail -f (press Ctrl+C to stop):

tail -f hello_12345678.out

To check the error log:

tail hello_12345678.err

To see a summary of your recent completed jobs:

sacct

To see overall resource usage for your group:

slurmInfo