solved the oversubscription problem. Everything was doubling, so halv…

…ed the forcing oversubscribe
hariseldon99 · Mar 2, 2023 · 59fadd3 · 59fadd3
1 parent 144f5a5
commit 59fadd3
Show file tree

Hide file tree

Showing 2 changed files with 5 additions and 4 deletions.
diff --git a/shavak_root/etc/slurm-llnl/cgroup.conf b/shavak_root/etc/slurm-llnl/cgroup.conf
@@ -2,4 +2,5 @@ CgroupMountpoint="/sys/fs/cgroup"
 CgroupAutomount=yes
 AllowedDevicesFile="/etc/slurm/cgroup_allowed_devices_file.conf"
 TaskAffinity=no
+ConstrainCores=yes
 ConstrainDevices=yes
diff --git a/shavak_root/etc/slurm-llnl/slurm.conf b/shavak_root/etc/slurm-llnl/slurm.conf
@@ -29,7 +29,7 @@ Epilog=/etc/slurm-llnl/epilog.d/*
 #KillOnBadExit=0
 #LaunchType=launch/slurm
 #Licenses=foo*4,bar
-#MailProg=/bin/mail
+MailProg=/usr/local/bin/goslmailer
 #MaxJobCount=5000
 #MaxStepCount=40000
 #MaxTasksPerNode=128
@@ -38,7 +38,7 @@ MpiDefault=pmix
 #PluginDir=
 #PlugStackConfig=
 #PrivateData=jobs
-ProctrackType=proctrack/pgid
+ProctrackType=proctrack/cgroup
 #Prolog=
 Prolog=/etc/slurm-llnl/prolog.d/*
 #PrologFlags=
@@ -170,6 +170,6 @@ GresTypes=gpu,mps
 NodeName=shavak-DIT400TR-55L CPUs=64 SocketsPerBoard=2 CoresPerSocket=16 ThreadsPerCore=2 RealMemory=95311 Gres=gpu:1,mps:100
 #PartitionName=debug Nodes=ALL Default=YES MaxTime=INFINITE State=UP
 
-PartitionName=CPU Nodes=ALL Default=Yes QOS=nogpu MaxTime=INFINITE OverSubscribe=FORCE:2 State=UP 
+PartitionName=CPU Nodes=ALL Default=Yes QOS=nogpu MaxTime=INFINITE OverSubscribe=FORCE:1 State=UP 
 
-PartitionName=GPU Nodes=ALL Default=NO MaxTime=INFINITE OverSubscribe=FORCE:2 State=UP
+PartitionName=GPU Nodes=ALL Default=NO MaxTime=INFINITE OverSubscribe=FORCE:1 State=UP