计算节点部署

@master
nodeset cn1 osimage=centos8-x86_64-install-compute

@cn1
reboot

配置cn1的yum源
@cn1 (备份)

cd /etc/yum.repos.d/
mkdir bak
mv * ./bak/

@master

xdcp cn1 /etc/yum.repos.d/centos.repo /etc/yum.repos.d/centos.repo
xdcp cn1 /etc/yum.repos.d/epel.repo /etc/yum.repos.d/epel.repo
xdcp cn1 /etc/yum.repos.d/OpenHPC-local.repo /etc/yum.repos.d/OpenHPC-local.repo

nfs挂载(修改网卡ip地址nmcli;nmtui)
@master

xdcp cn1 /etc/hosts /etc/hosts

@cn1:修改/etc/fstab

master-ib0:/home /home nfs nfsvers=3,rdma,port=20049,nodev,noauto,soft,noatime 0 0
master-ib0:/opt/intel /opt/intel nfs nfsvers=3,rdma,port=20049,nodev,noauto,soft,noatime 0 0
master-ib0:/opt/pub /opt/pub nfs nfsvers=3,rdma,port=20049,nodev,noauto,soft,noatime 0 0
master-ib0:/opt/repos/ohpc/CentOS_8/updates/x86_64/ /opt/repos/ohpc/CentOS_8/updates/x86_64/ nfs defaults 0 0

配置cn1的munge和slurm

yum install munge

修改munge的uid和gid

id munge

usermod -u 972 munge
groupmod -g 971 munge

scp -p /etc/munge/munge.key root@cn1:/etc/munge

chown -R munge:munge /var/log/munge
chown -R munge:munge /etc/munge
chown -R munge:munge /var/run/munge
chown -R munge:munge /var/lib/munge


###chown munge: /etc/munge/munge.key

###chmod 400 /etc/munge/munge.key

systemctl status  munge
systemctl enable --now munge
systemctl status  munge

安装slurm

#cd /opt/repos/ohpc/CentOS_8/updates/x86_64/
#yum localinstall slurm-torque-ohpc-20.11.7-5.1.ohpc.2.3.x86_64.rpm -y
#yum localinstall slurm-ohpc-20.11.7-5.1.ohpc.2.3.x86_64.rpm -y
#yum localinstall slurm-perlapi-ohpc-20.11.7-5.1.ohpc.2.3.x86_64.rpm -y
#yum localinstall slurm-slurmd-ohpc-20.11.7-5.1.ohpc.2.3.x86_64.rpm -y
#yum localinstall slurm-slurmdbd-ohpc-20.11.7-5.1.ohpc.2.3.x86_64.rpm -y
yum install slurm-ohpc* -y
yum install slurm-slurmdbd-ohpc* -y
yum install slurm-slurmd-ohpc* -y
yum install slurm-perlapi-ohpc* -y
yum install slurm-torque-ohpc* -y

id slurm
mkdir -p /etc/slurm

@master上同步slurm配置文件
/root/gengzi-script/xcatsync all

@cn1
chown -R slurm:slurm /etc/slurm/

测试效果

sinfo
qstat
scontrol update nodename=cn1 state=resume

安装module
cn1

yum install environment-modules
yum install -y kernel-modules*(4.18.0-305.25.1.el8_4.x86_64)

@master

xdcp cn1 /etc/profile.d/lmod.csh /etc/profile.d/lmod.csh
xdcp cn1 /etc/profile.d/lmod.sh /etc/profile.d/lmod.sh

@cn1

cd /opt/ohpc/
mv pub pub_bak
ln -s /opt/pub pub

语法解释:

ln -s
其中:

-s 选项表示创建一个符号链接(软链接)。
是要指向的目标文件或目录的路径。
是要创建的符号链接的名称。


转载请注明来源,欢迎对文章中的引用来源进行考证,欢迎指出任何有错误或不够清晰的表达。