Currently, the crc usage output pulls directly from the slurm job accounting database which will only show a usage-per-user that includes jobs that are finished. This can create a situation where users will submit a job that exceeds the limit (their job + the running jobs of others even though it seems like there should be enough SUs for their job to run.
Currently, the crc usage output pulls directly from the slurm job accounting database which will only show a usage-per-user that includes jobs that are finished. This can create a situation where users will submit a job that exceeds the limit (their job + the running jobs of others even though it seems like there should be enough SUs for their job to run.