Scripts for controlling job times sbatch

function slurm-job-add-timelimit() {
    timelimit="$1"
    jobids=("${@:2}")
    
    for job in "${jobids[@]}"; do
      echo "$job"
      scontrol show job $job | grep TimeLimit | awk '{print "  Before: " $2}'
      scontrol update jobid=$job TimeLimit+="$timelimit"
      scontrol show job $job | grep TimeLimit | awk '{print "  After:  " $2}'
    done
}

function slurm-job-check-timelimit() {
    jobids=("${@:1}")
    
    for job in "${jobids[@]}"; do
      echo "$job"
      scontrol show job $job | grep TimeLimit | awk '{print "  Time Limit: " $2}'
    done
}

The following code could be useful in niche usecases, but it can also force a job to terminate if the input timelimit is shorter than the currently elapsed time.

function slurm-job-set-timelimit() {
    timelimit="$1"
    jobids=("${@:2}")
    
    for job in "${jobids[@]}"; do
      echo "$job"
      scontrol show job $job | grep TimeLimit | awk '{print "  Before: " $2}'
      scontrol update jobid=$job TimeLimit="$timelimit"
      scontrol show job $job | grep TimeLimit | awk '{print "  After:  " $2}'
    done
}

Start with the three proposed shell functions—slurm-job-add-timelimit, slurm-job-check-timelimit, and slurm-job-set-timelimit—and review their scontrol, grep, and awk commands. Done means determining where these helpers belong in devops-docs and adding or documenting the intended Slurm time-limit controls, including the stated shorter-than-elapsed-time risk.