Hello, I'm trying to keep this (for now) as simple as I can make it
for testing purposes. Get one thing working and move on to the next.
I have a 3 node cluster - the head node or ControlMachine is running
munge, slurmctld, and slurmd just fine as far as I can tell.
Other two nodes run munge and slurmd. All version 16.05.
I've managed to get the first 50 or so of the test programs in
/root/rpmbuild/BUILD/slurm-16.05.2/testsuite/expect to run with
success on the head node. Test test1.52, however requires at least
3 nodes and fails with "WARNING: system must have at least 3 nodes to run this
test on. 1 999999"
Output from sinfo and slurmd -C:
[root@xf0 ~]# sinfo
PARTITION AVAIL TIMELIMIT NODES STATE NODELIST
slurm-test* up infinite 1 idle xf0
[root@xf0 ~]# slurmd -C
ClusterName=(null) NodeName=xf0 CPUs=16 Boards=1 SocketsPerBoard=2
CoresPerSocket=4 ThreadsPerCore=2 RealMemory=24017 TmpDisk=999511
UpTime=56-20:18:47
Similar output on the other two nodes.
I see it reports ClusterName=(null) on all 3 nodes.
In my slurm.conf file below - minus all comments (all nodes use the exact same
slurm.conf)
I have the parameter ClusterName=dorfman yet it appears not to be recognized.
I'm a bit at a loss in understanding what may be missing. Something simple
I've overlooked, obviously.
Thanks!
- Larry
ControlMachine=xf0
ControlAddr=xf0
MpiDefault=none
ProctrackType=proctrack/pgid
ReturnToService=1
SlurmctldPidFile=/var/run/slurmctld.pid
SlurmdPidFile=/var/run/slurmd.pid
SlurmdSpoolDir=/tmp/slurmd
SlurmdUser=root
StateSaveLocation=/tmp
SwitchType=switch/none
TaskPlugin=task/none
FastSchedule=1
SchedulerType=sched/backfill
SelectType=select/linear
AccountingStorageType=accounting_storage/slurmdbd
ClusterName=dorfman
JobAcctGatherFrequency=30
JobAcctGatherType=jobacct_gather/linux
NodeName=xf0 CPUs=8 RealMemory=16000 State=UNKNOWN
NodeName=xf1 CPUs=8 RealMemory=16000 State=UNKNOWN
NodeName=xn00 CPUs=4 RealMemory=8000 State=UNKNOWN
PartitionName=slurm-test Nodes=xf0 Default=YES MaxTime=INFINITE State=UP
PartitionName=slurm-test Nodes=xf1 Default=YES MaxTime=INFINITE State=UP
PartitionName=slurm-test Nodes=xn00 Default=YES MaxTime=INFINITE State=UP
--
P. Larry Nelson (217-244-9855) | IT Administrator
457 Loomis Lab | High Energy Physics Group
1110 W. Green St., Urbana, IL | Physics Dept., Univ. of Ill.
MailTo: [email protected] | http://hep.physics.illinois.edu/home/lnelson/
------------------------------------------------------------------------------
"Information without accountability is just noise." - P.L. Nelson
