Slurm: Difference between revisions

Revision as of 22:07, 12 January 2016

Build slurm

rpmbuild -ta slurm*.tar.bz2

Install rpms.

yum -y install munge slurm slurm-plugins slurm-munge

dd if=/dev/random bs=1 count=1024 > /etc/munge/munge.key
chmod 0600 /etc/munge/munge.key
chown munge /etc/munge/munge.key
systemctl start munge

Restore node.

scontrol update nodename=www state=resume

Generate a credential on stdout.

munge -n

Check if a credential can be locally decoded.

munge -n | unmunge

Check if a credential can be remotely decoded.

munge -n | ssh somehost unmunge

Run a quick benchmark.

remunge

how does it work

scontrol show config

check priorities of jobs using the command

scontrol show job".

Submit a job

sbatch /tmp/slurm_test_1

List jobs:

squeue

Get job details:

scontrol show job 106

Suspend a job (root only):

scontrol suspend 135

Resume a job (root only):

scontrol resume 135

Kill a job. Users can kill their own jobs, root can kill any job.

scancel 135

Hold a job

scontrol hold 139

Release a job:

scontrol release 139

List partitions:

sinfo

example job script.

#!/usr/bin/env bash
#SBATCH -p defq
#SBATCH -J simple
sleep 60

@@ Line 4: / Line 4: @@
 Install rpms.
   yum -y install munge slurm slurm-plugins slurm-munge
-configure munge
+==configure munge==
   dd if=/dev/random bs=1 count=1024 > /etc/munge/munge.key
   chmod 0600 /etc/munge/munge.key
   chown munge /etc/munge/munge.key
   systemctl start munge
-Change config to better suit fedora 21
+==troubleshooting==
- SlurmctldPidFile=/var/run/slurm/slurmctld.pid
+Restore node.
- SlurmdPidFile=/var/run/slurm/slurmd.pid
+  scontrol update nodename=www state=resume
- StateSaveLocation=/var/lib/slurm/slurm.state
- SlurmdSpoolDir=/var/lib/slurm/slurm.spool
-Create corresponding directories.
-  DIR=/var/run/slurm;chown slurm:slurm $DIR ; chmod 755 $DIR
-Update systemd config files for slurm daemons to point to new location of pid file.
- vim /usr/lib/systemd/system/slurmctld.service /usr/lib/systemd/system/slurmd.service
-Reread config changes.
- systemctl daemon-reload
 ==test installation==