Hi Thanks for your reply. I enabled both of them but still the oldest job gets suspended. Here is my slurm.conf if you can see something I can't.
0 MpiDefault=none 1 ProctrackType=proctrack/linuxproc 2 ReturnToService=1 3 SlurmctldPidFile=/var/run/slurmctld.pid 4 SlurmdPidFile=/var/run/slurmd.pid 5 SlurmdSpoolDir=/var/spool/slurmd 6 SlurmUser=slurm 7 StateSaveLocation=/var/spool/slurmSave 8 SwitchType=switch/none 9 TaskPlugin=task/affinity 10 # # SCHEDULING 11 FastSchedule=1 12 SchedulerType=sched/builtin 13 SchedulerParameters=preempt_youngest_first,preempt_strict_order 14 PriorityType=priority/multifactor 15 16 SelectType=select/cons_res 17 SelectTypeParameters=CR_CPU_Memory 18 DefMemPerCPU=1000 19 PreemptType=preempt/partition_prio 20 PreemptMode=suspend,gang 21 SlurmctldParameters=preempt_send_user_signal 22 23 JobSubmitPlugins=lua 24 # # LOGGING AND ACCOUNTING 25 AccountingStorageHost=localhost 26 AccountingStorageType=accounting_storage/slurmdbd 27 ClusterName=simulation 28 # #JobAcctGatherFrequency=30 29 JobAcctGatherType=jobacct_gather/linux 30 SlurmctldLogFile=/var/log/slurmLog/slurmctld.log 31 SlurmdLogFile=/var/log/slurmLog/slurmd.log 32 JobCompType=jobcomp/filetxt # # COMPUTE NODES 33 NodeName=localhost CPUs=2 CoresPerSocket=1 ThreadsPerCore=1 State=UNKNOWN RealMemory=7500 34 PartitionName=DEFAULT Nodes=localhost PreemptMode=suspend OverSubscribe=FORCE:1 Shared=FORCE:1 MaxTime=INF INITE State=UP 36 PartitionName=LowPrio Default=YES PriorityTier=1 PreemptMode=requeue 37 PartitionName=HighPrio MaxTime=240 PriorityTier=2 PreemptMode=suspend Default=NO 38 PartitionName=Express MaxTime=30 PriorityTier=3 PreemptMode=off Default=NO Oytun Peksel oytun.pek...@semcon.com <mailto:oytun.pek...@semcon.com> +46739205917 From: slurm-users <slurm-users-boun...@lists.schedmd.com> On Behalf Of Paul Edmon Sent: den 25 oktober 2019 16:16 To: slurm-users@lists.schedmd.com Subject: Re: [slurm-users] Preemption Priority You might want to look at these options for SchedulerParamtesr: preempt_strict_order If set, then execute extra logic in an attempt to preempt only the lowest priority jobs. It may be desirable to set this configuration parameter when there are multiple priorities of preemptable jobs. The logic to support this option is only available in the select/cons_res and select/cons_tres plugins. preempt_youngest_first If set, then the preemption sorting algorithm will be changed to sort by the job start times to favor preempting younger jobs over older. (Requires preempt/partition_prio or preempt/qos plugins.) -Paul Edmon- On 10/25/19 7:21 AM, Oytun Peksel wrote: Hi, Let's say I have two partitions assigned to the same single load in the cluster. LowPrio with PreemptMode=suspend Priority=1 HighPrio with PreemtMode=off Priority=5 I have 4 identical jobs requiring 1/2 of resources in the cluster. So only 2 jobs can run at the same time. I queue up two of these jobs to LowPrio partition one after the other. Let's say one have jobid=10 and the next one has jobd=11 Then I queue up a third job to HighPrio partition. One of the LowPrio jobs immediately get suspended. Q1. How can I make sure the one submitted latest(jobid=11 in this case) gets suspended and not jobid=10? Then before any job is finished I submit the final job to LowPrio partition. So the queue looks like this [cid:image001.png@01D58E25.DD245E60] Q2. When HighPrio job (12) finishes how can I make sure suspended job(10) would start instead of the pending job(13)? Thanks Oytun Peksel When you communicate with us or otherwise interact with Semcon, we will process personal data that you provide to us or we collect about you, please read more in our Privacy Policy<https://semcon.com/data-privacy-policy/>.