OPENHPC搭建集群
许亦
目前搭建HPC的开源工具主要有RocksCluster, Xcat，Open-HPC。其中OpenHPC 是Linux基金开源的超算项目，致力于为高性能计算构建一个开源框架，适应学术研究的需求，为 HPC 环境创建一个开源框架，降低成本。
本次采用虚拟机实验，没有Infiniband,故先不介绍Infiniband的配置
主节点系统（CentOS-7.7）配置
- 配置参数说明
| 环境变量 | 变量名称说明 | 模版参数 |
|---|---|---|
| ${sms_name} | 主机名 | server |
| ${sms_ip} | 主节点内网地址 | 192.168.130.100 |
| $sms_eth_internal | 主节点内网网卡名称 | ens33 |
| ${eth_provision} | 主节点与计算节点通信网卡名称 | ens33 |
| ${internal_netmask} | 内网子网掩码 | 255.255.255.0 |
| ${ntp_server} | 本地时间同步服务器 | 192.168.130.100 |
| ${c_ip[0]}, {c_ip[1]} ,… | 计算节点名称 | 192.168.130.[1-5] |
| ${c_name[0]}, ${c_name[1]} ,… | 计算节点名称 | 192.168.130.[1-5] |
| ${sms_ipoib} | 主节点Inifiniband网络地址 | 192.168.120.100 |
| ${ipoib_netmask} | 主节点Infiniband网络子网掩码 | 255.255.255.0 |
| ${c_ipoib[0]},&{c_ipoib[1]},… | 计算节点Infiniband网络IP地址 | 192.168.120.[1-5] |
| $compute_regex | 计算节点名称正则表达式 | node[1-5] |
| ${compute_prefix} | 计算节点名称前缀 | node |
| ${kargs} | Linux内核启动参数 | net.ifnames=1 |
- 添加hosts文件信任
[root@server ~]#echo "192.168.130.1 node1" >> /etc/hosts
[root@server ~]#echo "192.168.130.2 node2" >> /etc/hosts
[root@server ~]#echo "192.168.130.3 node3" >> /etc/hosts
[root@server ~]#echo "192.168.130.4 node4" >> /etc/hosts
[root@server ~]#echo "192.168.130.5 node5" >> /etc/hosts
- 关闭防火墙
[root@server ~]#systemctl stop firewaled.sevice
[root@server ~]#systemctl disabled firewaled.sevice
- 关闭SELinux ；
[root@server ~]#vi /etc/selinux/deisabled
- 配置NTP服务 ；
[root@server ~]#echo "server 192.168.130.100" >> /etc/ntp.conf
[root@server ~]#systemctl enable ntpd.service
- 配置内网网卡 ；
[root@server ~]# vi /etc/sysconfig/network-scripts/ifcfg-ens33
TYPE=Ethernet
PROXY_METHOD=none
BROWSER_ONLY=no
BOOTPROTO=static
DEFROUTE=yes
IPV4_FAILURE_FATAL=no
IPV6INIT=yes
IPV6_AUTOCONF=yes
IPV6_DEFROUTE=yes
IPV6_FAILURE_FATAL=no
IPV6_ADDR_GEN_MODE=stable-privacy
NAME=ens33
UUID=******************************
DEVICE=ens33
ONBOOT=yes
IPADDR=192.168.130.100
PREFIX=24
IPV6_PRIVACY=no
- ohpc-base,warewulf,pbs ；
安装open-HPC yum源 ohpc-release；
[root@server ~]#yum install http://build.openhpc.community/OpenHPC:/1.3/CentOS_7/x86_64/ohpc-release-1.3-1.el7.x86_64.rpm
[root@server ~]#yum -y install ohpc-base
[root@server ~]#yum -y install ohpc-warewulf
[root@server ~]#yum -y install pbspro-server-ohpc
管理节warewulf点配置
将/etc/warewulf/provision.conf文件中的 network device = eth1 改成内网网卡ens33
[root@server ~]# vi /etc/warewulf/provision.conf
# What is the default network device that the master will use to
# communicate with the nodes?
network device = ens33
# Which DHCP server implementation should be used?
dhcp server = isc
# What is the TFTP root directory that should be used to store the
# network boot images? By default Warewulf will try and find the
# proper directory. Just add this if it can't locate it.
#tftpdir = /var/lib/tftpboot
# Automatically generate and manage a dynamnic_host virtual file
# object in the datastore? This is useful for provisioning this
# out to nodes so they always have a current /etc/hosts file.
generate dynamic_hosts = yes
# Should we manage and overwrite the local hostfile file on this
# system? This will cause all node entries to be added
# automatically to /etc/hosts.
update hostfile = yes
# If no cluster/domain is set on a node, should we add 'localdomain'
# as the default domain
use localdomain = yes
# The default kernel arguments to pass to the nodes boot kernel
default kargs = "net.ifnames=0 biosdevname=0 quiet"
编辑/etc/xinetd.d/tftp文件, 将disable = yes改为 disable = no
[root@server ~]# vi /etc/xinetd.d/tftp
# default: off
# description: The tftp server serves files using the trivial file transfer \
# protocol. The tftp protocol is often used to boot diskless \
# workstations, download configuration files to network-aware printers, \
# and to start the installation process for some operating systems.
service tftp
{
socket_type = dgram
protocol = udp
wait = yes
user = root
server = /usr/sbin/in.tftpd
server_args = -s /var/lib/tftpboot
disable = no
per_source = 11
cps = 100 2
flags = IPv4
}
启动xinetd,mariadb,http服务
[root@server ~]# systemctl restart xinetd.service
[root@server ~]# systemctl enable xinetd.service
[root@server ~]# systemctl restart mariadb.service
[root@server ~]# systemctl enable mariadb.service
[root@server ~]# systemctl enable httpd.service
[root@server ~]# systemctl restart httpd.service
[root@server ~]# systemctl restart dhcpd.service
重启管理节点！
配置计算节点
- 配置DHCP服务
[root@server ~]# cp /etc/warewulf/dhcpd-template.conf /etc/dhcp/dhcpd.conf
[root@server ~]# vi /etc/dhcp/dhcpd.conf
# DHCPD Configuration written by Warewulf. Do not edit this file, rather
# edit the template: /etc/warewulf/dhcpd-template.conf
allow booting;
allow bootp;
ddns-update-style interim;
authoritative;
option space ipxe;
# Tell iPXE to not wait for ProxyDHCP requests to speed up boot.
option ipxe.no-pxedhcp code 176 = unsigned integer 8;
option ipxe.no-pxedhcp 1;
option architecture-type code 93 = unsigned integer 16;
if exists user-class and option user-class = "iPXE" {
filename "http://192.168.130.100/WW/ipxe/cfg/${mac}";
} else {
if option architecture-type = 00:0B {
filename "/warewulf/ipxe/bin-arm64-efi/snp.efi";
} elsif option architecture-type = 00:0A {
filename "/warewulf/ipxe/bin-arm32-efi/placeholder.efi";
} elsif option architecture-type = 00:09 {
filename "/warewulf/ipxe/bin-x86_64-efi/snp.efi";
} elsif option architecture-type = 00:07 {
filename "/warewulf/ipxe/bin-x86_64-efi/snp.efi";
} elsif option architecture-type = 00:06 {
filename "/warewulf/ipxe/bin-i386-efi/snp.efi";
} elsif option architecture-type = 00:00 {
filename "/warewulf/ipxe/bin-i386-pcbios/undionly.kpxe";
}
}
subnet 192.168.130.0 netmask 255.255.255.0 {
not authoritative;
# option interface-mtu 9000;
# option subnet-mask 255.255.255.0;
}
# Node entries will follow below
- 创建BOS镜像
[root@server ~]#export CHROOT=/opt/ohpc/admin/images/centos7.7
[root@server ~]#wwmkchroot centos-7 $CHROOT
[root@server ~]#yum -y --installroot=$CHROOT install ohpc-base-compute
-
配置NFS服务
初始化warewulf database和ssh_keys
[root@server ~]#wwinit database
[root@server ~]#wwinit ssh_keys
[root@server ~]#echo "192.168.130.100:/home /home nfs nfsvers=3,nodev,nosuid 0 0" >> $CHROOT/etc/fstab
[root@server ~]#echo "192.168.130.100:/opt/ohpc/pub /opt/ohpc/pub nfs nfsvers=3,nodev 0 0" >> $CHROOT/etc/fstab
[root@server ~]#echo "/home 192.168.130.0/24(rw,no_subtree_check,fsid=10,no_root_squash)" >> /etc/exports
[root@server ~]#echo "/opt/ohpc/pub 192.168.130,0/24(ro,no_subtree_check,fsid=11)" >> /etc/exports
[root@server ~]#exportfs -a
[root@IMP-Cluster ~]# showmount -e
Export list for server:
/opt/ohpc/pub 192.168.130.0/24
/home 192.168.130.0/24
[root@server ~]#systemctl restart nfs-server.service
[root@server ~]#systemctl enable nfs-server.service
[root@server ~]#chroot $CHROOT systemctl enable ntpd.service
[root@server ~]# echo "server 192.168.130.100" >> $CHROOT/etc/ntp.conf
- 配置PBSPro
[root@server ~]# yum -y --installroot=$CHROOT install pbspro-execution-ohpc
[root@server ~]# vi $CHROT/etc/pbs.conf
PBS_EXEC=/opt/pbs
PBS_SERVER=server !!!修改为管理节点的名称server
PBS_START_SERVER=0
PBS_START_SCHED=0
PBS_START_COMM=0
PBS_START_MOM=1
PBS_HOME=/var/spool/pbs
PBS_CORE_LIMIT=unlimited
PBS_SCP=/bin/scp
[root@server ~]#echo "PBS_LEAF_NAME=server" >> /etc/pbs.conf
PBS_EXEC=/opt/pbs
PBS_SERVER=IMP-Cluster
PBS_START_SERVER=1
PBS_START_SCHED=1
PBS_START_COMM=1
PBS_START_MOM=0
PBS_HOME=/var/spool/pbs
PBS_CORE_LIMIT=unlimited
PBS_SCP=/bin/scp
PBS_LEAF_NAME=server !!增加这一行
[root@server ~]#chroot $CHROOT opt/pbs/libexec/pbs_habitat
[root@server ~]#echo "\$usecp *:/home /home" >> $CHROOT/var/spool/pbs/mom_priv/config
[root@server ~]#vi $CHROOT/var/spool/pbs/mom_priv/config
$clienthost server
$usecp *:/home /home
[root@server ~]# chroot $CHROOT systemctl enable pbs
#NTP网络服务
[root@server ~]# yum -y --installroot=$CHROOT install ntp
#内核驱动
[root@server ~]# yum -y --installroot=$CHROOT install kernel
# module加载环境
[root@server ~]# yum -y --installroot=$CHROOT install lmod-ohpc
- 取消镜像中计算节点内存资源限制
[root@server ~]#vi $CHROOT/etc/security/limits.conf
#<domain> <type> <item> <value>
#
#* soft core 0
#* hard rss 10000
#@student hard nproc 20
#@faculty soft nproc 20
#@faculty hard nproc 50
#ftp hard nproc 0
#@student - maxlogins 4
* soft memlock unlimited !!修改
* hard memlock unlimited !!修改
# End of file
- 配置ganglia监控
[root@server ~]# yum -y install ohpc-ganglia
[root@server ~]# yum -y --installroot=$CHROOT install ganglia-gmond-ohpc
[root@server ~]# cp /opt/ohpc/pub/examples/ganglia/gmond.conf /etc/ganglia/gmond.conf
[root@server ~]# vi /etc/ganglia/gmond.conf
/* The host section describes attributes of the host, like the location */
host {
location = "unspecified"
}
/* Feel free to specify as many udp_send_channels as you like. Gmond
used to only support having a single channel */
udp_send_channel {
#bind_hostname = yes # Highly recommended, soon to be default.
# This option tells gmond to use a source address
# that resolves to the machine's hostname. Without
# this, the metrics may appear to come from any
# interface and the DNS names associated with
# those IPs will be used to create the RRDs.
host = server !!!修改成当前主机名
port = 8649
ttl = 1
}
[root@server ~]# cp /etc/ganglia/gmond.conf $CHROOT/etc/ganglia/gmond.conf
[root@server ~]# echo "gridname MyCluster" >> /etc/ganglia/gmetad.conf
# 设置ganglia服务开机启动
[root@server ~]# systemctl enable gmond.service
[root@server ~]# systemctl enable gmetad.service
[root@server ~]# systemctl start gmond.service
[root@server ~]# systemctl start gmetad.service
[root@server ~]# chroot $CHROOT systemctl enable gmond.service
#重启HTTP服务
[root@server ~]# systemctl try-restart httpd.service
在浏览器中输入http://192.168.130.100/ganglia就可以查看ganglia监控界面。
- 安装节点健康检查资源管理包NHC
# 管理节点和计算节点分别安装NHC
[root@server ~]# yum -y install nhc-ohpc
[root@server ~]# yum -y --installroot=$CHROOT install nhc-ohpc
- 同步用户信息
[root@server ~]# wwsh file import /etc/passwd
[root@server ~]# wwsh file import /etc/group
[root@server ~]# wwsh file import /etc/shadow
创建计算节点启动镜像
[root@server ~]# wwbootstrap `uname -r`
组装Virtual Node File System (VNFS) 镜像
[root@server ~]# wwvnfs --chroot $CHROOT
添加节点
- 设置主节点与计算节点默认网络通讯网卡
# Set provisioning interface as the default networking device
[root@server ~]# echo "GATEWAYDEV=ens33" > /tmp/network.$$
[root@server ~]# wwsh -y file import /tmp/network.$$ --name network
[root@server ~]# wwsh -y file set network --path /etc/sysconfig/network --mode=0644 --uid=0
- 设置计算节点网卡名称
[root@server ~]# export kargs="${kargs} net.ifnames=1,biosdevname=1"
[root@server ~]# wwsh provision set --postnetdown=1 "node[1-5]"
- 设置计算节点安装镜像，启动项与用户信息
[root@server ~]# wwsh -y provision set "node[1-5]" --vnfs=centos7.7 --bootstrap=`uname -r` --files=dynamic_hosts,passwd,group,shadow,network
- 添加节点
[root@server ~]# wwsh -y node new node1 --netdev=ens33 --ipaddr=192.168.130.1 --hwaddr=00:00:00:00:00:01 -D ens33
[root@server ~]# wwsh -y node new node2 --netdev=ens33 --ipaddr=192.168.130.2 --hwaddr=00:00:00:00:00:02 -D ens33
[root@server ~]# wwsh -y node new node3 --netdev=ens33 --ipaddr=192.168.130.3 --hwaddr=00:00:00:00:00:03 -D ens33
[root@server ~]# wwsh -y node new node4 --netdev=ens33 --ipaddr=192.168.130.4 --hwaddr=00:00:00:00:00:04 -D ens33
[root@server ~]# wwsh -y node new node5 --netdev=ens33 --ipaddr=192.168.130.5 --hwaddr=00:00:00:00:00:05 -D ens33
- 设置计算节点安装在本地硬盘并安装grub引导
[root@server ~]# yum -y --installroot=$CHROOT install grub2
[root@server ~]# wwvnfs --chroot $CHROOT
[root@server ~]# cp /etc/warewulf/filesystem/examples/gpt_example.cmds /etc/warewulf/filesystem/gpt.cmds
[root@server ~]# wwsh provision set --filesystem=gpt "node[1-5]"
[root@server ~]# wwsh provision set --bootloader=sda "node[1-5]"
- 更新DHCP和PXE
[root@server ~]# wwsh dhcp update
[root@server ~]# wwsh pxe update
这是一个简单的脚本，可以自动完成计算节点添加与配置
#!/bin/bash
NODE=node3 !!修改为节点名
IPADDR=192.168.130.3 !!计算节点IP地址
MAC=xx:xx:xx:xx:xx:xx !!计算节点MAC地址
wwsh -y node new $NODE --netdev=enp2s0 --ipaddr=$IPADDR --netmask=255.255.255.0 --hwaddr=$MAC -D enp2s0
sleep 1s
export kargs="enp2s0 net.ifnames=1,biosdevname=1"
wwsh -y provision set --postnetdown=1 "$NODE"
wwsh -y provision set "$NODE" --vnfs=centos7.7 --bootstrap=`uname -r` --files=dynamic_hosts,passwd,group,shadow,network
wwsh -y provision set --filesystem=gpt "$NODE"
wwsh -y provision set --bootloader=sda "$NODE"
sleep 1s
wwsh dhcp update
wwsh pxe update
计算节点完成安装后设置
- 计算节点安装成功后设置从本地硬盘启动
[root@server ~]# wwsh provision set --bootlocal=normal "node[1-5]"
- 查看安装节点
#### node1 ####################################################################
node1: BOOTSTRAP = 3.10.0-1062.el7.x86_64
node1: VNFS = centos7.7
node1: FILES = dynamic_hosts,group,network,passwd,shadow
node1: PRESHELL = FALSE
node1: POSTSHELL = FALSE
node1: CONSOLE = UNDEF
node1: PXELINUX = UNDEF
node1: SELINUX = DISABLED
node1: KARGS = "net.ifnames=1,biosdevname=1"
node1: FS = "select /dev/sda,mklabel gpt,mkpart primary 1MiB 3MiB,mkpart primary ext4 3MiB 513MiB,mkpart primary linux-swap 513MiB 50%,mkpart primary ext4 50% 100%,name 1 grub,name 2 boot,name 3 swap,name 4 root,set 1 bios_grub on,set 2 boot on,mkfs 2 ext4 -L boot,mkfs 3 swap,mkfs 4 ext4 -L root,fstab 4 / ext4 defaults 0 0,fstab 2 /boot ext4 defaults 0 0,fstab 3 swap swap defaults 0 0"
node1: BOOTLOADER = sda
node1: BOOTLOCAL = NORMAL
#### node2 ####################################################################
node2: BOOTSTRAP = 3.10.0-1062.el7.x86_64
node2: VNFS = centos7.7
node2: FILES = dynamic_hosts,group,network,passwd,shadow
node2: PRESHELL = FALSE
node2: POSTSHELL = FALSE
node2: CONSOLE = UNDEF
node2: PXELINUX = UNDEF
node2: SELINUX = DISABLED
node2: KARGS = "net.ifnames=1,biosdevname=1"
node2: FS = "select /dev/sda,mklabel gpt,mkpart primary 1MiB 3MiB,mkpart primary ext4 3MiB 513MiB,mkpart primary linux-swap 513MiB 50%,mkpart primary ext4 50% 100%,name 1 grub,name 2 boot,name 3 swap,name 4 root,set 1 bios_grub on,set 2 boot on,mkfs 2 ext4 -L boot,mkfs 3 swap,mkfs 4 ext4 -L root,fstab 4 / ext4 defaults 0 0,fstab 2 /boot ext4 defaults 0 0,fstab 3 swap swap defaults 0 0"
node2: BOOTLOADER = sda
node2: BOOTLOCAL = NORMAL
#### node3 ####################################################################
node3: BOOTSTRAP = 3.10.0-1062.el7.x86_64
node3: VNFS = centos7.7
node3: FILES = dynamic_hosts,group,network,passwd,shadow
node3: PRESHELL = FALSE
node3: POSTSHELL = FALSE
node3: CONSOLE = UNDEF
node3: PXELINUX = UNDEF
node3: SELINUX = DISABLED
node3: KARGS = "net.ifnames=1,biosdevname=1"
node3: FS = "select /dev/sda,mklabel gpt,mkpart primary 1MiB 3MiB,mkpart primary ext4 3MiB 513MiB,mkpart primary linux-swap 513MiB 50%,mkpart primary ext4 50% 100%,name 1 grub,name 2 boot,name 3 swap,name 4 root,set 1 bios_grub on,set 2 boot on,mkfs 2 ext4 -L boot,mkfs 3 swap,mkfs 4 ext4 -L root,fstab 4 / ext4 defaults 0 0,fstab 2 /boot ext4 defaults 0 0,fstab 3 swap swap defaults 0 0"
node3: BOOTLOADER = sda
node3: BOOTLOCAL = NORMAL
#### node4 ####################################################################
node4: BOOTSTRAP = 3.10.0-1062.el7.x86_64
node4: VNFS = centos7.7
node4: FILES = dynamic_hosts,group,network,passwd,shadow
node4: PRESHELL = FALSE
node4: POSTSHELL = FALSE
node4: CONSOLE = UNDEF
node4: PXELINUX = UNDEF
node4: SELINUX = DISABLED
node4: KARGS = "net.ifnames=1,biosdevname=1"
node4: FS = "select /dev/sda,mklabel gpt,mkpart primary 1MiB 3MiB,mkpart primary ext4 3MiB 513MiB,mkpart primary linux-swap 513MiB 50%,mkpart primary ext4 50% 100%,name 1 grub,name 2 boot,name 3 swap,name 4 root,set 1 bios_grub on,set 2 boot on,mkfs 2 ext4 -L boot,mkfs 3 swap,mkfs 4 ext4 -L root,fstab 4 / ext4 defaults 0 0,fstab 2 /boot ext4 defaults 0 0,fstab 3 swap swap defaults 0 0"
node4: BOOTLOADER = sda
node4: BOOTLOCAL = NORMAL
#### node5 ####################################################################
node5: BOOTSTRAP = 3.10.0-1062.el7.x86_64
node5: VNFS = centos7.7
node5: FILES = dynamic_hosts,group,network,passwd,shadow
node5: PRESHELL = FALSE
node5: POSTSHELL = FALSE
node5: CONSOLE = UNDEF
node5: PXELINUX = UNDEF
node5: SELINUX = DISABLED
node5: KARGS = "net.ifnames=1,biosdevname=1"
node5: FS = "select /dev/sda,mklabel gpt,mkpart primary 1MiB 3MiB,mkpart primary ext4 3MiB 513MiB,mkpart primary linux-swap 513MiB 50%,mkpart primary ext4 50% 100%,name 1 grub,name 2 boot,name 3 swap,name 4 root,set 1 bios_grub on,set 2 boot on,mkfs 2 ext4 -L boot,mkfs 3 swap,mkfs 4 ext4 -L root,fstab 4 / ext4 defaults 0 0,fstab 2 /boot ext4 defaults 0 0,fstab 3 swap swap defaults 0 0"
node5: BOOTLOADER = sda
node5: BOOTLOCAL = NORMAL
[root@server ~]# wwsh node list
NAME GROUPS IPADDR HWADDR
================================================================================
node1 UNDEF 192.168.130.1 xx:xx:xx:xx:xx:01
node2 UNDEF 192.168.130.2 xx:xx:xx:xx:xx:02
node3 UNDEF 192.168.130.3 xx:xx:xx:xx:xx:03
node4 UNDEF 192.168.130.4 xx:xx:xx:xx:xx:04
node5 UNDEF 192.168.130.5 xx:xx:xx:xx:xx:05
- 可查看/etc/hosts文件，通过warewulf添加成功的hosts
### ALL ENTRIES BELOW THIS LINE WILL BE OVERWRITTEN BY WAREWULF ###
#
# See provision.conf for configuration paramaters
# Node Entry for node: node1 (ID=1)
192.168.130.1 node1.localdomain node1 node1-ens33.localdomain node1-ens33
# Node Entry for node: node2 (ID=2)
192.168.130.2 node2.localdomain node2 node2-ens33.localdomain node2-ens33
# Node Entry for node: node3 (ID=3)
192.168.130.3 node3.localdomain node3 node3-ens33.localdomain node3-ens33
# Node Entry for node: node4 (ID=4)
192.168.130.4 node4.localdomain node4 node4-ens33.localdomain node4-ens33
# Node Entry for node: node5 (ID=5)
192.168.130.5 node5.localdomain node5 node5-ens33.localdomain node5-ens33
至此，整个集群搭建完成，可以部署相关计算应用环境。
类似资料：