An How-To example for beginners.
Connect to the cluster with you user and pass.
To utilize the power of the cluster you to add you code (project files) onto your home directory, then run it.
If you examine you home directory (use the command “ls” on the command line), use see a some files “readMe.1st” and “myJob.sh”. The “readMe.1st” file list some helpful commands to start using the cluster, view the content of the file on command line by running “cat readMe.1st”
In you working (home) directory create your project files for computing. For example if using python create a python file “runSomeCode.py” with a editor. Using VScode might be the easiest way for beginner to work on the cluster, setup. If using the command line you can use the editors vi(or vim) or nano.
After your coding file's are ready you need you test the code try them and resolve any issue, bugs(debug)
When ready send your job(project) to cluster, using sbatch command.
The cluster uses Slurm as a scheduler and workload manager.
To submit your you need to create a file(batch file for Slurm), an example filed is already in your home directory called “myJob.sh”, which can be used with minor adjustments.
But user can also create their own batch file for Slurm, but the content of it must follow slurms ruleset to work and it must be executable.
Assuming your python file is called runSomeCode.py and the batch file for Slurm is myJob.sh
This is the content of myJob.sh
#!/bin/bash
#SBATCH --account=student
#SBATCH --job-name=MyJob
#SBATCH --gpus-per-node=1
#SBATCH --mem-per-cpu=2G
#SBATCH --output=myBatch.log
python3 runSomeCode.py
To submit this job to the cluster you run the command sbatch myJob.sh.
This should be the result user gets
For user to see the status of his job issue the command squeue.
Beware if coping (uploading) files from Windows to the cluster, you might encounter issues because of different handling of character encoding (Linux/Windows)