Hi - I need to look at the node placement info from the slurm accounting to try to identify for a given job any jobs on adjacent nodes that may have caused lower than expected performance (noisy neighbors), and to try to identify particularly poor node placements due to fragmentation caused by a mix of jobs with varying node counts.
Before I starting writing a script, I thought I would check if anybody else has done something similar and if there might be existing open source scripts. Thanks, Gary
