Dear SLURM,
We have a user who has thousands of jobs submitted, each one of those jobs will only run for about 30 seconds consistently. Given the following backfill parameters I was hoping I would be able to backfill significant of these waiting jobs. I do have quite a number of nodes idle but only about 100 jobs are scheduled for the user every minute. Would anybody have an idea as to how one can debug what is going on with the scheduler? One message I see often in the scheduler log is: sched: loop taking too long, breaking out I am not sure if this has anything to do with backfill, but wanted to point out. Please advise, Regards, Amit PriorityType=priority/multifactor PriorityDecayHalfLife=14-0 PriorityFavorSmall=NO PriorityMaxAge=2-0 #PriorityUsageResetPeriod= PriorityWeightAge=1000 PriorityWeightFairshare=10000000 PriorityWeightJobSize=1000 PriorityWeightPartition=1000 MaxJobCount=250000 SchedulerParameters=bf_window=43200,bf_resolution=600,bf_max_job_user=10,bf_max_job_start=10000,bf_continue,defer
