From 47f5d0a129b13b513d6ad148b2a75d869e0ccaa7 Mon Sep 17 00:00:00 2001
From: Arif Ali <arif.ali@canonical.com>
Date: Mon, 11 Dec 2023 15:40:01 +0000
Subject: [PATCH] Fine tuning of creation of cluster

---
 create_cluster.sh |  5 -----
 lxd_profile.yaml  | 45 +++++++++++++++++++++++++++++++++++++++++++++
 2 files changed, 45 insertions(+), 5 deletions(-)

diff --git a/create_cluster.sh b/create_cluster.sh
index 129a15d..2f1e310 100755
--- a/create_cluster.sh
+++ b/create_cluster.sh
@@ -40,15 +40,10 @@ do
 
   ${lxc} launch ubuntu:jammy ${node}
   sleep 5
-  ${lxc_exec} ${node} -- add-apt-repository -y ${ppa}
-  ${lxc_exec} ${node} -- apt -y install slurmd
   ${lxc} file push munge.key ${node}/etc/munge/
   ${lxc_exec} ${node} -- systemctl restart munge
   ${lxc_exec} ${node} -- systemctl enable munge
 
-  ${lxc} file push slurm.conf ${node}/etc/slurm/
-  ${lxc_exec} ${node} -- systemctl start slurmd
-  ${lxc_exec} ${node} -- systemctl enable slurmd
 done
 
 tmp_hosts=$(mktemp)
diff --git a/lxd_profile.yaml b/lxd_profile.yaml
index 45f83eb..e472712 100644
--- a/lxd_profile.yaml
+++ b/lxd_profile.yaml
@@ -7,3 +7,48 @@ devices:
     path: /
     pool: virtual
     type: disk
+config:
+  cloud-init.user-data: |
+    #cloud-config
+    sources:
+      source: "ppa:ubuntu-hpc/slurm-wlm-23.02"
+    packages:
+      - slurmd
+    write_files:
+      - name: /etc/slurm/slurm.conf
+        permission: '0644'
+        owner: root:root
+        contents: |
+          ClusterName=cluster
+          SlurmctldHost=slurm-master
+          ProctrackType=proctrack/cgroup
+          ReturnToService=1
+          SlurmctldPidFile=/var/run/slurmctld.pid
+          SlurmctldPort=6817
+          SlurmdPidFile=/var/run/slurmd.pid
+          SlurmdPort=6818
+          SlurmdSpoolDir=/var/spool/slurmd
+          SlurmUser=slurm
+          StateSaveLocation=/var/spool/slurmctld
+          SwitchType=switch/none
+          TaskPlugin=task/affinity,task/cgroup
+          InactiveLimit=0
+          KillWait=30
+          MinJobAge=300
+          SlurmctldTimeout=120
+          SlurmdTimeout=300
+          Waittime=0
+          SchedulerType=sched/backfill
+          SelectType=select/cons_tres
+          JobAcctGatherFrequency=30
+          JobAcctGatherType=jobacct_gather/none
+          SlurmctldDebug=info
+          SlurmctldLogFile=/var/log/slurmctld.log
+          SlurmdDebug=info
+          SlurmdLogFile=/var/log/slurmd.log
+          NodeName=slurm-master,slurm-comp0[1-5] CPUs=1 State=UNKNOWN
+          PartitionName=debug Nodes=slurm-comp0[1-5] Default=YES MaxTime=INFINITE State=UP
+
+    runcmd:
+      - [ systemctl, enable, slurmd ]
+      - [ systemctl, start, slurmd ]