Hi,
Recently we start to use the MultyFactorPriority plugin in our cluster,
however some of our users had found a strange behavior in the JobSize. It
happens that when a user submit a Job to the queue in the first minutes the
JobSize is larger than later. For example ("sprio" reports):
Immediatly after submmisionJOBSIZE = 175
After few minutes it suddenly drops to JOBSIZE = 63, after the drop it will
keep in that number.
I did not found any reason for this to be happening. Any ideas? Our versions
are:
slurm-slurmdbd-2.2.5-1
slurm-2.2.5-1
slurm-devel-2.2.5-1
slurm-sjstat-2.2.5-1
slurm-munge-2.2.5-1
slurm-plugins-2.2.5-1
slurm-sql-2.2.5-1
slurm-pam_slurm-2.2.5-1
slurm-slurmdb-direct-2.2.5-1
io-watchdog-slurm-0.7-1
slurm-perlapi-2.2.5-1
slurm-torque-2.2.5-1
and the MultiFactorPriority configuration is:
PriorityType=priority/multifactor
PriorityDecayHalfLife=7
PriorityCalcPeriod=5
PriorityFavorSmall=NO
PriorityMaxAge=7-0
PriorityWeightAge=1000
PriorityWeightFairshare=5000
PriorityWeightJobSize=1000
PriorityWeightPartition=1000
PriorityWeightQOS=5000
Thanks,
Daniel
