SLURM: Basics of Workload Manager

DataScience Workbook / 06. High-Performance Computing (HPC) / 5. Introduction to Job Scheduling / 5.1 SLURM: Basics of Workload Manager

---

Introduction

SLURM: Simple Linux Utility for Resource Management

• A simple text file with all the requirements for running your job
  • Memory requirement
  • Desired number of processors
  • Length of time you want to run the job
  • Type of queue you want to use (optional)
  • Where to write output and error files
  • Name foryour job while running on HPC

Job Script Basics

A typical job script will look like this:

#!/bin/bash
#SBATCH --nodes=1
#SBATCH --cpus-per-task=8
#SBATCH --time=02:00:00
#SBATCH --mem=128G
#SBATCH --mail-user=netid@gmail.com
#SBATCH --mail-type=begin
#SBATCH --mail-type=end
#SBATCH --error=JobName.%J.err
#SBATCH --output=JobName.%J.out

cd $SLURM_SUBMIT_DIR

module load modulename

your_commands_goes_here

Lines starting with #SBATCH are for SLURM resource manager to request resources for HPC. Some important options are as follows:

Option	Examples	Description
--nodes	#SBATCH --nodes=1	Number of nodes
--cpus-per-task	#SBATCH --cpus-per-task=16	Number of CPUs per node
--time	#SBATCH --time=HH:MM:SS	Total time requested for your job
--output	#SBATCH -output filename	STDOUT to a file
--error	#SBATCH --error filename	STDERR to a file
--mail-user	#SBATCH --mail-user user@domain.edu	Email address to send notifications

Job Management Commands

Job Status	Commands
sinfo -a	list all queues
squeue	list all jobs
squeue -u userid	list jobs for userid
squeue -t R	list running jobs

Let’s go ahead and give these job management commands a try.

sinfo -a
squeue
squeue -t R
#pick a name you saw when you typed squeue and specify all the jobs by that person with the following option
squeue -u first.lastname

There can be a lot of information using those two commands. I have created some useful alias’ that change the output to something more informative.

alias sq='squeue -o "%8i %12j %4t %10u %20q %20a %10g %20P %10Q %5D %11l %11L %R"'
alias si='sinfo -o "%20P %5D %14F %8z %10m %10d %11l %16f %N"'

Where (A/I/O/T) = available/idle/other/total

You can place those alias’ into your ~/.bashrc file and it will automatically load every time you log in.

Exercise: Add these two alias’ above to your ~/.bashrc file

nano ~/.bashrc

Job scheduling commands

Commands	Function	Basic Usage	Example
sbatch	submit a slurm job	sbatch [script]	$ sbatch job.sub
scancel	delete slurm batch job	scancel [job_id]	$ scancel 123456

Interactive Session

To start a interactive session execute the following:

# this command will give 1 Node with 1 cpu in the brief-low queue for a time of 00 hours: 01 minutes: 00 seconds

salloc -N 1 -n 1 -p brief-low -t 00:01:00

# You can exit out of an interactive node by typing exit and hitting return

Interactive sessions are very helpful when you need more computing power than your laptop or desktop to wrangle the data or to test new software prior to submitting a full batch script.

---

Further Reading

• 5.1.1 Introduction to SLURM
• 5.1.2 Creating SLURM Job Submission Scripts
• 5.1.3 Submitting Dependency Jobs using SLURM

• SLURM Commands CheatSheet

• 5.2 PBS: Portable Batch System

• 6. Introduction to GNU Parallel
• 7. Introduction to Containers

---

Homepage Section Index Previous Next top of page