When working with Slurm, the job environment variables available to your job depend on when and how the job is submitted. Here’s how you can access and understand Slurm job environment variables from the master node (the login node where you submit jobs):

---

1. Slurm Environment Variables Available During Job Execution

When Slurm runs your job, it sets several environment variables that describe the job’s context. These are available inside the job script on the compute nodes.

Common Slurm Environment Variables

---

2. How to Access These Variables

From a Job Script

If you submit a job using a script (e.g., sbatch), you can access these variables directly in your script:

bash
Copy
#!/bin/bash
#SBATCH --job-name=my_job
#SBATCH --nodes=1
#SBATCH --ntasks=1
#SBATCH --cpus-per-task=4
#SBATCH --mem=8G
#SBATCH --time=01:00:00

echo "Job ID: $SLURM_JOB_ID"
echo "Job Name: $SLURM_JOB_NAME"
echo "Node List: $SLURM_JOB_NODELIST"
echo "Number of Nodes: $SLURM_JOB_NUM_NODES"
# ... and so on

From the Command Line (Master Node)

You can also check the environment variables of a running job from the master node using:

bash
Copy
scontrol show job <JOB_ID> -o

or

bash
Copy
squeue -j <JOB_ID> --format="%i %j %N %P %T %M %D %m %c"

---

3. Submitting Jobs with Environment Variables

If you want to pass environment variables from the master node to the job, you can use:

bash
Copy
export MY_VAR="value"
sbatch --export=ALL,MY_VAR my_script.sh

or

bash
Copy
sbatch --export=MY_VAR1,MY_VAR2 my_script.sh

---

4. Accessing Variables from Within a Running Job

If you want to inspect the environment of a running job, you can use:

bash
Copy
srun --jobid=<JOB_ID> env

---

Summary Table: Key Slurm Environment Variables

---

Example: Print All Slurm Variables

bash
Copy
#!/bin/bash
#SBATCH --job-name=env_test
#SBATCH --nodes=1
#SBATCH --ntasks=1

echo "All Slurm environment variables:"
env | grep SLURM

---

Let me know if you want to see the output of a specific command or need help with a concrete example!