Slurm 수동 설치

https://estar987.tistory.com/164

 

이전에 포스팅한 Munge를 설치 완료했다면 이제 Slurm을 설치할 차례이다. 이번 포스팅은 마스터 노드에서만 돌아가는 Slurm을 설정할 것이고 차후 업로드하는 게시물에 마스터 노드에 계산 노드를 연동하여 HPC를 구성한 후 Slurm을 사용하는 방법을 포스팅할 예정이다. 

아래 모든 설정은 마스터 노드에서 작업한다.

Slurm download

cd /engrid/slurm/src/
wget https://download.schedmd.com/slurm/slurm-23.11.6.tar.bz2

Slurm install

tar xvfj slurm-23.11.6.tar.bz2
cd slurm-23.11.6/
./configure --prefix=/engrid/slurm
make && make install

Slurm Setting

# mkdir /engrid/slurm/etc

# vim /engrid/slurm/etc/slurm.conf

---------------------------------------------

SlurmctldAddr=192.168.207.191
MpiDefault=none
ProctrackType=proctrack/pgid
SlurmctldPidFile=/var/run/slurmctld.pid
SlurmdPidFile=/var/run/slurmd.pid
SlurmdSpoolDir=/var/spool/slurmd
StateSaveLocation=/var/spool/slurmctld
SwitchType=switch/none
TaskPlugin=task/affinity
SchedulerType=sched/backfill
SelectType=select/cons_tres
SelectTypeParameters=CR_CPU
SlurmctldLogFile=/var/log/slurmctld.log
SlurmdLogFile=/var/log/slurmd.log

---------------------------------------------

# /engrid/slurm/sbin/slurmd -C | head -n 1 >> /engrid/slurm/etc/slurm.conf
# echo "PartitionName=all.q Nodes=slurm01 Default=Yes OverSubscribe=FORCE:1 MaxTime=UNLIMITED" >> /engrid/slurm/etc/slurm.conf

cat << EOL >> /etc/profile.d/slurm.sh
> SLURM_HOME=/engrid/slurm
> PATH=\{SLURM_HOME}/bin:
> {PATH}
> LD_LIBRARY_PATH=\{SLURM_HOME}/lib:
> {LD_LIBRARY_PATH}
> export SLURM_HOME PATH LD_LIBRARY_PATH
> EOL

 

# cat /engrid/slurm/etc/slurm.conf
---------------------------------------------

ClusterName=Your Cluster Name Setting
SlurmctldHost=slurm01(192.168.207.191)
SlurmctldAddr=192.168.207.191
MpiDefault=none
ProctrackType=proctrack/pgid
SlurmctldPidFile=/var/run/slurmctld.pid
SlurmdPidFile=/var/run/slurmd.pid
SlurmdSpoolDir=/var/spool/slurmd
StateSaveLocation=/var/spool/slurmctld
SwitchType=switch/none
TaskPlugin=task/affinity
SchedulerType=sched/backfill
SelectType=select/cons_tres
SelectTypeParameters=CR_CPU
SlurmctldLogFile=/var/log/slurmctld.log
SlurmdLogFile=/var/log/slurmd.log
NodeName=slurm01 CPUs=12 Boards=1 SocketsPerBoard=6 CoresPerSocket=2 ThreadsPerCore=1 RealMemory=7949
PartitionName=all.q Nodes=slurm01 Default=Yes OverSubscribe=FORCE:1 MaxTime=UNLIMITED

Slurm Run and Check

# /engrid/slurm/sbin/slurmctld
# /engrid/slurm/sbin/slurmd

# cat /etc/rc.local
---------------------------------------------

#!/bin/bash
/engrid/slurm/sbin/slurmctld
/engrid/slurm/sbin/slurmd
exit 0

---------------------------------------------

chmod 755 /etc/rc.local

 cat <<EOL>> /etc/systemd/system/rc-local.service
> [Unit]
> Description=/etc/rc.local Compatibility
> ConditionPathExists=/etc/rc.local
> [Service]
> Type=forking
> ExecStart=/etc/rc.local start
> TimeoutSec=0
> StandardOutput=tty
> RemainAfterExit=yes
> SysVStartPriority=99
> [Install]
> WantedBy=multi-user.target
> EOL

cat /etc/systemd/system/rc-local.service
---------------------------------------------

[Unit]
Description=/etc/rc.local Compatibility
ConditionPathExists=/etc/rc.local
[Service]
Type=forking
ExecStart=/etc/rc.local start
TimeoutSec=0
StandardOutput=tty
RemainAfterExit=yes
SysVStartPriority=99
[Install]
WantedBy=multi-user.target

---------------------------------------------

systemctl daemon-reload
systemctl enable rc-local
systemctl start rc-local

/engrid/slurm/bin/sinfo
PARTITION AVAIL  TIMELIMIT  NODES  STATE NODELIST
all.q*       up   infinite      1   idle slurm01