Hi All, With the introduction of GPU queues in various clusters, traditional SLURM scripts need several changes and those changes are not consistent with each cluster. Because of that, we need to override SLURM script parameters depending on the cluster configuration. This PR [1] provides the flexibility for users to either override existing SLURM parameters or remove them from the scripts. For example, GPU queues on Expanse do not accept scripts with #SBATCH -n parameter and you can remove it by adding a pre-job command RM #SBATCH -n.
I have rebased and merged this as it was extensively tested on Ultrascan environment. [1] https://github.com/apache/airavata/pull/277 Thanks Dimuthu