diff --git a/roles/calculateSlurmConf/templates/slurm.conf.j2 b/roles/calculateSlurmConf/templates/slurm.conf.j2 index f8818eca480935c56a5974c7241d5f66f0d99155..66ea9f7fc30dc920dabb1e07580fb4c006f0a1fe 100644 --- a/roles/calculateSlurmConf/templates/slurm.conf.j2 +++ b/roles/calculateSlurmConf/templates/slurm.conf.j2 @@ -30,7 +30,8 @@ SwitchType=switch/none MpiDefault=pmi2 SlurmctldPidFile={{ slurmpiddir }}/slurmctld.pid SlurmdPidFile={{ slurmpiddir }}/slurmd.pid -ProctrackType=proctrack/linuxproc +#ProctrackType=proctrack/linuxproc +ProctrackType=proctrack/cgroup #PluginDir= CacheGroups=0 #FirstJobId= @@ -78,16 +79,16 @@ SelectType={{ slurmselecttype }} SelectTypeParameters=CR_Core_Memory {% endif %} FastSchedule={{ slurmfastschedule }} -#PriorityType=priority/multifactor +PriorityType=priority/multifactor #PriorityFlags=Ticket_Based #PriorityCalcPeriod=5 #PriorityDecayHalfLife=0 #PriorityUsageResetPeriod=14-0 -##PriorityWeightFairshare=10000 -#PriorityWeightAge=10000 -#PriorityWeightPartition=10000 -#PriorityWeightJobSize=10000 -#PriorityMaxAge=14-0 +PriorityWeightFairshare=10000 +PriorityWeightAge=10000 +PriorityWeightPartition=10000 +PriorityWeightJobSize=10000 +PriorityMaxAge=14-0 # # LOGGING {% if slurmctlddebug %} @@ -120,21 +121,21 @@ Epilog={{ slurmjob.epilog }} {% endif %} # # ACCOUNTING -#JobAcctGatherType=jobacct_gather/linux -#JobAcctGatherFrequency=30 +JobAcctGatherType=jobacct_gather/linux +JobAcctGatherFrequency=30 # AccountingStorageType=accounting_storage/slurmdbd AccountingStorageHost={{ slurmdbd }} {% if slurmdbdbackup is defined %} AccountingStorageBackupHost={{ slurmdbdbackup }} {% endif %} -#AccountingStorageEnforce=limits,safe +AccountingStorageEnforce=limits,safe #AccountingStorageLoc= #AccountingStoragePass= #AccountingStorageUser= # #GRES -GresTypes=gpu +#GresTypes=gpu # Fair share {% if slurmfairshare.def %} @@ -155,6 +156,10 @@ MpiParams=ports=12000-12999 NodeName={{ hostvars[node]['ansible_hostname'] }} Procs={{ hostvars[node]['ansible_processor_vcpus'] }} RealMemory={{ hostvars[node].ansible_memory_mb.real.total }} Sockets={{ hostvars[node]['ansible_processor_vcpus'] }} CoresPerSocket=1 ThreadsPerCore={{ hostvars[node].ansible_processor_threads_per_core }} {% if hostvars[node].ansible_hostname.find('vis') != -1 %}Gres=gpu:1{% endif %} {% if hostvars[node]['ansible_processor_vcpus'] == 1 %}Weight=1{% endif %}{% if hostvars[node]['ansible_processor_vcpus'] > 1 and hostvars[node]['ansible_processor_vcpus'] <= 16 %}Weight=3{% endif %}{% if hostvars[node]['ansible_processor_vcpus'] > 16 and hostvars[node]['ansible_processor_vcpus'] <= 20 %}Weight=5{% endif %}{% if hostvars[node]['ansible_processor_vcpus'] > 20 and hostvars[node]['ansible_processor_vcpus'] <= 40 %}Weight=7{% endif %}{% if hostvars[node]['ansible_processor_vcpus'] > 40 and hostvars[node]['ansible_processor_vcpus'] <= 64 %}Weight=8{% endif %}{% if hostvars[node]['ansible_processor_vcpus'] > 64 and hostvars[node]['ansible_processor_vcpus'] <= 128 %}Weight=9{% endif %}{% if hostvars[node]['ansible_processor_vcpus'] > 128 %}Weight=10{% endif %} Feature=stage1 State=UNKNOWN {% endfor %} +#monarch specific to stop stupid warning messages +NodeName={{ hostvars[groups['LoginNodes'][0]]['ansible_hostname'] }} State=DOWN +NodeName={{ slurmctrl }} State=DOWN + {% for queue in slurmqueues %} {% set nodenames = [] %} {% for node in groups[queue.group] %}