日韩性视频-久久久蜜桃-www中文字幕-在线中文字幕av-亚洲欧美一区二区三区四区-撸久久-香蕉视频一区-久久无码精品丰满人妻-国产高潮av-激情福利社-日韩av网址大全-国产精品久久999-日本五十路在线-性欧美在线-久久99精品波多结衣一区-男女午夜免费视频-黑人极品ⅴideos精品欧美棵-人人妻人人澡人人爽精品欧美一区-日韩一区在线看-欧美a级在线免费观看

歡迎訪問 生活随笔!

生活随笔

當前位置: 首頁 > 编程资源 > 编程问答 >内容正文

编程问答

xcat 安装(liunx高性能刀片集群管理软件)

發(fā)布時間:2025/3/15 编程问答 18 豆豆
生活随笔 收集整理的這篇文章主要介紹了 xcat 安装(liunx高性能刀片集群管理软件) 小編覺得挺不錯的,現(xiàn)在分享給大家,幫大家做個參考.

xcat 安裝(liunx高性能刀片集群管理軟件)

標簽:?集群managercomments腳本timezoneservice ?9968人閱讀?評論(4)?收藏?舉報 ?分類: ? hpc(4)?

目錄(?)[+]

xcat非常方便,hp有cmu,當然那個是花錢的,xcat是ibm前兩年開源的一個軟件,功能可以說還是夠強大的。

1、先刪除掉沖突軟件

xCAT2和操作系統(tǒng)的載OpenIPMI-tools和tftpserver不兼容,需要把這兩個包刪除。

同時bind-chroot和dhcp-chroot軟件包如果存在也會導(dǎo)致問題,需要刪除。

# rpm -e --nodeps??tftp-server-0.49-2

2、xCAT2軟件安裝

創(chuàng)建xcat的yum安裝源

把xcat上傳到服務(wù)器,本例是/root /xcat/

# cd /root/xcat/

# ls

xcat-core-2.6.1.tar.bz2??xcat-dep-201105100946.tar.bz2

# tar -xjvfxcat-core-2.6.1.tar.bz2

# tar -xjvfxcat-dep-201105100946.tar.bz2

1)創(chuàng)建xcat-core的yum安裝源

# cd xcat-core

#./mklocalrepo.sh

2)創(chuàng)建xcat-dep的yum安裝源

#cd ../xcat-dep/rh5/x86_64

#./mklocalrepo.sh

?

安裝xCAT

# yum install xCAT

根據(jù)提示進行安裝即可.

默認安裝到/opt/xcat目錄中

測試

#source /etc/profile.d/xcat.sh

#tabdump?將列出xcat所有可配置的表。讀取site表

#key,value,comments,disable

"blademaxp","64",,

"fsptimeout","0",,

"installdir","/install",,

"ipmimaxp","64",,

"ipmiretries","3",,

"ipmitimeout","2",,

"consoleondemand","no",,

"master","172.16.1.1",,

"forwarders","172.16.1.1",,

"nameservers","172.16.1.1",,

"maxssh","8",,

"ppcmaxp","64",,

"ppcretry","3",,

"ppctimeout","0",,

"sharedtftp","1",,

"SNsyncfiledir","/var/xcat/syncfiles",,

"tftpdir","/tftpboot",,

"xcatdport","3001",,

"xcatiport","3002",,

"xcatconfdir","/etc/xcat",,

"timezone","Asia/Shanghai",,

"useNmapfromMN","no",,

"enableASMI","no",,

"db2installloc","/mntdb2",,

"databaseloc","/var/lib",,

"sshbetweennodes","ALLGROUPS",,

"dnshandler","ddns",,

"vsftp","y",,

節(jié)點網(wǎng)絡(luò)安裝配置

設(shè)置site表

執(zhí)行添加命令

#chtab key=ntpserverssite.value=172.16.1.1

#chtab key=dhcpinterfacessite.value='manager1|eth1'

#chtab key=domain site.value=cluster.net

#tapdump site??查看site表

#key,value,comments,disable

"blademaxp","64",,

"fsptimeout","0",,

"installdir","/install",,

"ipmimaxp","64",,

"ipmiretries","3",,

"ipmitimeout","2",,

"consoleondemand","no",,

"master","172.16.1.1",,

"forwarders","172.16.1.1",,

"nameservers","172.16.1.1",,

"maxssh","8",,

"ppcmaxp","64",,

"ppcretry","3",,

"ppctimeout","0",,

"sharedtftp","1",,

"SNsyncfiledir","/var/xcat/syncfiles",,

"tftpdir","/tftpboot",,

"xcatdport","3001",,

"xcatiport","3002",,

"xcatconfdir","/etc/xcat",,

"timezone","Asia/Shanghai",,

"useNmapfromMN","no",,

"enableASMI","no",,

"db2installloc","/mntdb2",,

"databaseloc","/var/lib",,

"sshbetweennodes","ALLGROUPS",,

"dnshandler","ddns",,

"vsftp","y",,

"ntpservers","172.16.1.1",,

"domain","cluster.net",,

"dhcpinterface","manager1|eth1",,

設(shè)置networks表

先查看一下networks表

[root@manager1 yum.repos.d]#tabdump networks

#netname,net,mask,mgtifname,gateway,dhcpserver,tftpserver,nameservers,ntpservers,logservers,dynamicrange,nodehostname,ddnsdomain,vlanid,comments,disable

"172_16_1_0-255_255_255_0","172.16.1.0","255.255.255.0","eth1","172.16.1.253","172.30.1.129","172.16.1.1","172.16.1.1","172.16.1.1",,,,,,,

"169_254_95_0-255_255_255_0","169.254.95.0","255.255.255.0","usb0","<xcatmaster>",,"169.254.95.120","172.16.1.1",,,,,,,,

按環(huán)境改

# chtab net=172.16.1.0networks.dhcpserver=172.16.1.1

# chtab net=172.16.1.0networks.ntpservers=172.16.1.1

# chtab net=172.16.1.0 networks.logservers=172.16.1.1

# chtab net=172.16.1.0networks.tftpserver=172.16.1.1

使用這類命令來設(shè)置,建議

netname,net,mask,mgtifname,dhcpserver,tftpserver,nameservers,ntpservers,logservers這些都設(shè)置上。

注意:

當不設(shè)置dynamicrange這個字段時,系統(tǒng)可以按照mac地址一對一的分配IP地址。不過在makedhcp時會有一條告警,不用管它。(建議方式)

設(shè)置dynamicrange這個字段時,系統(tǒng)不能按照mac地址一對一的分配IP地址。(在集群連接用戶現(xiàn)有網(wǎng)絡(luò)時,可能會產(chǎn)生災(zāi)難性的后果

設(shè)置passwd密碼表

#chtab key=systempasswd.username=root passwd.password=root123(用于每個xcat安裝出來的節(jié)點)

#chtab key=bladepasswd.username=USERID passwd.password=PASSW0RD(對應(yīng)刀箱的AMM管理模塊)

#chtab key=ipmipasswd.username=USERID passwd.password=PASSW0RD(對應(yīng)機架式服務(wù)器的IMM管理模塊)

[root@manager1 yum.repos.d]#tabdump passwd??查看password表

#key,username,password,cryptmethod,comments,disable

"system","root","root123",,,

"blade","USERID","PASSW0RD",,,

設(shè)置nodelist表

#nodeadd ON001-ON128 groups=all,compute

#nodeadd L001-L002 groups=all,compute

#nodeadd manager1-manager2??groups=all,compute

[root@serv ~]# tabdump nodelist??檢查nodelist表

[root@manager1 yum.repos.d]#tabdump nodelist

#node,groups,status,statustime,appstatus,appstatustime,primarysn,hidden,comments,disable

"BCE1","mm","booting","03-31-201223:58:58",,,,,,

"BCE2","mm","booting","03-31-201223:58:58",,,,,,

"BCE3","mm","booting","03-31-201223:58:58",,,,,,

"BCE4","mm","booting","03-31-201223:58:58",,,,,,

"BCE5","mm","booting","03-31-201223:58:58",,,,,,

"BCE6","mm","booting","03-31-201223:58:58",,,,,,

"BCE7","mm","booting","03-31-201223:58:58",,,,,,

"BCE8","mm","booting","03-31-201223:58:58",,,,,,

"BCE9","mm","booting","03-31-201223:58:58",,,,,,

"BCE10","mm","booting","03-31-201223:58:58",,,,,,

"ON001","all,compute",,,,,,,,

"ON002","all,compute",,,,,,,,

。。。。。。

"ON127","all,compute",,,,,,,,

"ON128","all,compute",,,,,,,,

"L001","all,compute",,,,,,,,

"L002","all,compute",,,,,,,,

"manager001","all,compute",,,,,,,,

"manager002","all,compute",,,,,,,,

設(shè)置noderes表

#chtab node=compute noderes.netboot=pxenoderes.nfsserver=manager1 noderes.monserver=manager1 noderes.installnic=eth1noderes.primarynic=eth1 noderes.xcatmaster=manager1?noderes. servicenode=manager1

[root@manager1 yum.repos.d]#tabdump noderes???查看noders

#node,servicenode,netboot,tftpserver,nfsserver,monserver,nfsdir,installnic,primarynic,discoverynics,cmdinterface,xcatmaster,current_osimage,next_osimage,nimserver,routenames,comments,disable

"ON001",,"pxe",,"172.16.1.1","172.16.1.1",,"eth1","eth1","eth1",,"172.16.1.1","172.16.1.1",,,,,

"ON002",,"pxe",,"172.16.1.1","172.16.1.1",,"eth1","eth1","eth1",,"172.16.1.1","172.16.1.1",,,,,

。。。。。。

"ON126",,"pxe",,"172.16.1.1","172.16.1.1",,"eth1","eth1","eth1",,"172.16.1.1","172.16.1.1",,,,,

"ON127",,"pxe",,"172.16.1.1","172.16.1.1",,"eth1","eth1","eth1",,"172.16.1.1","172.16.1.1",,,,,

"ON128",,"pxe",,"172.16.1.1","172.16.1.1",,"eth1","eth1","eth1",,"172.16.1.1","172.16.1.1",,,,,

"manager1",,"pxe",,"172.16.1.1","172.16.1.1",,"eth1","eth1","eth1",,"172.16.1.1","172.16.1.1",,,,,

"LG01",,"pxe",,"172.16.1.1","172.16.1.1",,"eth1","eth1","eth1",,"172.16.1.1","172.16.1.1",,,,,

"LG02",,"pxe",,"172.16.1.1","172.16.1.1",,"eth1","eth1","eth1",,"172.16.1.1","172.16.1.1",,,,,

servicenode,tftpserver,nfsserver,monserver,installnic,primarynic,xcatmaster

discoverynics可根據(jù)具體環(huán)境,不是必須。建議配置以上服務(wù)

設(shè)置nodetype表

[root@manager1 yum.repos.d]#tabdump nodetype

#node,os,arch,profile,provmethod,supportedarchs,nodetype,comments,disable

"ON001","rhels5u5","x86_64","compute",,,,,

"ON002","rhels5u5","x86_64","compute",,,,,

。。。。。。

"ON126","rhels5u5","x86_64","compute",,,,,

"ON127","rhels5u5","x86_64","jss",,,,,

"ON128","rhels5u5","x86_64","oracle",,,,,

#chtab node=節(jié)點名nodetype.os=rhels5u5 nodetype.arch=x86_64 nodetype.profile=compute nodetype.nodetype=osi

在nodetype表中,定義計算節(jié)點ON001-ON126的操作系統(tǒng)是rhels5.5

x86_64,使用的profile是compute,節(jié)點類型是osi(OS image)

注意這里的nodetype.os、nodetype.profile這兩個字段和kickstart文件名關(guān)聯(lián),如上設(shè)置,則:

kickstart文件名為:compute.rhels5u5.tmpl

kickstart軟件包列表的文件名為:compute.rhels5u5.pkglist

compute.tmpl+compute.pkglist?存放位置

/opt/xcat/share/xcat/install/rh

設(shè)置postscripts表

這個表指定在kickstart安裝前,rpm包安裝后和安裝完成后第一次啟動時執(zhí)行的腳本,不同集群的環(huán)境定制全都在這里做。

[root@manager1 etc]# tabdumppostscripts

#node,postscripts,postbootscripts,comments,disable

"xcatdefaults","syslog,remoteshell,syncfiles,setupntp","otherpkgs",,

"service","servicenode,xcatserver,xcatclient",,,

"compute","hardeths,configRSH,configService",,,

postscripts這列表示rpm包安裝后后執(zhí)行。

postbootscripts這列表示安裝完成后第一次啟動時執(zhí)行的腳本。

其中,xcatdefaults和service這兩段為系統(tǒng)默認就有

xcatdefaults這個段代表所有節(jié)點都要執(zhí)行的腳本

service是服務(wù)節(jié)點需要執(zhí)行的腳本,服務(wù)節(jié)點是在較大(也許256節(jié)點以上)的集群環(huán)境中配置的,主要是分擔管理節(jié)點的壓力。

所有腳本位于/install/postscripts目錄下

xcatdefaults默認的postscripts為"syslog,remoteshell,syncfiles"。

1)syslog腳本把各節(jié)點的log重定向到管理節(jié)點。

2)remoteshell腳本設(shè)置ssh無密碼互通。注意xcat默認情況下root從管理節(jié)點ssh到各被管節(jié)點ssh無密碼,但是從被管節(jié)點ssh到管理節(jié)點是要密碼的。

3)syncfiles腳本指定安裝過程中要同步的文件。但還需要其他相關(guān)配置,詳細后面說明。

xcatdefaults默認的postscripts為"otherpkgs",指定了要安裝的非操作系統(tǒng)光盤里的標準的rpm包,這也需要其他相關(guān)配置才行。

更改如下:

#chtab node=xcatdefaultspostscripts.postscripts="syslog,remoteshell,syncfiles,setupntp"

這里增加一個setupntp,指定所有的節(jié)點都要設(shè)置ntp時間同步。

#chtab node=compute??postscripts.postscripts="hardeths,configRSH,configService,configNFS"

這里自定義了compute這一段,他代表計算節(jié)點(必須要和nodelist表中g(shù)roups字段的定義一致),

定義了rpm包安裝后后執(zhí)行hardeths,configRSH,configService,configNFS幾個腳本。

這幾個腳本中hardeths為系統(tǒng)默認就有,它指定節(jié)點設(shè)置固定IP地址,否則默認是DHCP的。

其他腳本要自己寫的, configRSH,configService,configNFS是自己寫的腳本。

configRSH是節(jié)點在安裝時配置RSH無密碼。

configService是節(jié)點在安裝時配置需要起停的系統(tǒng)服務(wù)。

configNFS是節(jié)點在安裝時需要在/etc/fstab定義的NFS文件系統(tǒng),此次實施沒有配置

這些腳本的內(nèi)容請看附錄。

注意這些腳本要放在/install/postscripts目錄下,別忘了給他們加上可執(zhí)行屬性。

#chtab node=compute??postscripts.postbootscripts="setupGang,reboot"

定義了安裝完成后第一次啟動執(zhí)行setupGang,reboot兩個腳本。

setupGang是配置節(jié)點安裝ganglia客戶端,腳本內(nèi)容參看附錄,此次實施沒有執(zhí)行

reboot就是重啟,默認就有這個腳本。

再檢查一下:

#tabdump postscripts

#node,postscripts,postbootscripts,comments,disable

"xcatdefaults","configRSH,configService,setupntp,syslog,remoteshell,syncfiles","otherpkgs",,

"service","servicenode,xcatserver,xcatclient",,,

"compute","hardeths,configRSH,configService,setupntp","reboot",,

設(shè)置配置文件同步目錄

對于redhat系統(tǒng)來說,創(chuàng)建一個/install/custom/install/rh目錄,這里放要安裝的額外rpm包和同步列表文件。

在本目錄寫一個同步列表文件,這個文件的文件名是有規(guī)矩的,和nodetype表的定義有關(guān),格式如下:

nodetype.profile+nodetype.os+nodetype.arch+synclist。

本例中,文件名為:compute.rhels5u5.x86_64.synclist,文件內(nèi)容如下:

/etc/hosts -> /etc/hosts

設(shè)置刀片中心

如果計算節(jié)點是刀片

1)給每個管理模塊配置IP地址

2)把這些AMM的IP加入管理節(jié)點的hosts文件

3)nodeadd把每個AMM加入xcat管理,使用和刀片計算節(jié)點不同的groups。

4)配置每個管理模塊的ssh和snmp

#rspconfig $管理模塊名字 snmpcfg=enable sshcfg=enable

rspconfig???BCE1?? snmpcfg=enable? sshcfg=enable

rspconfig???BCE2?? snmpcfg=enable? sshcfg=enable

rspconfig???BCE3?? snmpcfg=enable? sshcfg=enable

rspconfig???BCE4?? snmpcfg=enable? sshcfg=enable

rspconfig???BCE5? snmpcfg=enable? sshcfg=enable

rspconfig???BCE6? snmpcfg=enable? sshcfg=enable

rspconfig???BCE7?? snmpcfg=enable? sshcfg=enable

rspconfig???BCE8?? snmpcfg=enable? sshcfg=enable

rspconfig???BCE9?? snmpcfg=enable? sshcfg=enable

rspconfig???BCE10? snmpcfg=enable? sshcfg=enable

#rpower?管理模塊名字 reset

rpower??BCE1??? reset

在執(zhí)行rspconfig命令時,如果遇到Cannot communicate with XXXX的提示,需要用瀏覽器登到管理模塊,啟用SNMPv1

設(shè)置nodehm表

#chtab node=ON001??nodehm.mgt=blade

[root@manager1 postscripts]#tabdump nodehm??查看nodehm表

#node,power,mgt,cons,termserver,termport,conserver,serialport,serialspeed,serialflow,getmac,comments,disable

"BCE1","blade","blade",,,,,,,,,,

"BCE2","blade","blade",,,,,,,,,,

"BCE3","blade","blade",,,,,,,,,,

"BCE4","blade","blade",,,,,,,,,,

"BCE5","blade","blade",,,,,,,,,,

"BCE6","blade","blade",,,,,,,,,,

"BCE7","blade","blade",,,,,,,,,,

"BCE8","blade","blade",,,,,,,,,,

"BCE9","blade","blade",,,,,,,,,,

"BCE10","blade","blade",,,,,,,,,,

"ON001","blade","blade",,,,,,,,,,

"ON002","blade","blade",,,,,,,,,,

。。。。。。

"ON127","blade","blade",,,,,,,,,,

"ON128","blade","blade",,,,,,,,,,

"manager2","blade","blade",,,,,,,,,,

"manager1","blade","blade",,,,,,,,,,

"L001","blade","blade",,,,,,,,,,

"L002","blade","blade",,,,,,,,,,

設(shè)置mp表

[root@manager1 xcat_conf]#

#chtab node=ON005??mp.mpa=BCE1? mp.id=5

#tabdump mp??查看mp表

#node,mpa,id,comments,disable

"BCE1","BCE1",,,

"BCE2","BCE2",,,

"BCE3","BCE3",,,

"BCE4","BCE4",,,

"BCE5","BCE5",,,

"BCE6","BCE6",,,

"BCE7","BCE7",,,

"BCE8","BCE8",,,

"BCE9","BCE9",,,

"BCE10","BCE10",,,

"ON001","BCE1","1",,

"ON002","BCE1","2",,

。。。。。。

"ON125","BCE9","13",,

"ON126","BCE9","14",,

"ON127","BCE10","2",,

"ON128","BCE10","3",,

"manager1","BCE10","14",,

"manager2","BCE10","12",,

"L001","BCE10","13",,

"L002","BCE10","4",,

設(shè)置mpa表

# chtab mpa.mpa=BCE1mpa.username=USERID mpa.password=PASSW0RD

# tabdump mpa

#mpa,username,password,comments,disable

"BCE1","USERID","PASSW0RD",,

"BCE2","USERID","PASSW0RD",,

"BCE3","USERID","PASSW0RD",,

"BCE4","USERID","PASSW0RD",,

"BCE5","USERID","PASSW0RD",,

"BCE6","USERID","PASSW0RD",,

"BCE7","USERID","PASSW0RD",,

"BCE8","USERID","PASSW0RD",,

"BCE9","USERID","PASSW0RD",,

"BCE10","USERID","PASSW0RD",,

設(shè)置dns

#makedns –n

#makedns compute

#chkconfig named on

測試一下

#nslookup

注意:

管理節(jié)點的DNS服務(wù)一定要正常,否則白搭。

設(shè)置dhcp

#makedhcp -n

#makedhcp compute?或者makedhcp all

如果只設(shè)置某個節(jié)點,可以如下:

#makedhcp node01

修改kickstar文件

進入/opt/xcat/share/xcat/install/rh

#cp compute.rhels6.tmpl compute.rhels5u5.tmpl

#cp compute.rhels6.pkglistmycomp.rhels5u5.pkglist

(針對jss服務(wù)器為jss.rhel5u5.tmpl; jss.rhel5u5.pkglist.針對oracle服務(wù)器為oracle.rhel5u5.tmpl; oracle.rhel5u5.pkglist.)

其中.tmpl是kickstart文件,.pkglist是指定需要需要裝哪些軟件包

?

修改mycomp.rhels5u5.tmpl文件,

1)改分區(qū)劃分,注意rhel5不支持ext4

[root@manager1 rh]# morecompute.rhels5u5.tmpl

#egan@us.ibm.com

#cmdline

lang en_US

#

# Where's the source?

# nfs --serverhostname.of.server or IP --dir /path/to/RH/CD/image

#

#nfs --server#XCATVAR:INSTALL_NFS# --dir #XCATVAR:INSTALL_SRC_DIR#

url --urlhttp://#TABLE:noderes:$NODE:nfsserver#/install/#TABLE:nodetype:$NODE:os#/#TABLE:nodetype:$NODE:arch#

#device ethernet e100

keyboard "us"

#

# Clear the MBR

#

zerombr yes

#

# Wipe out the disk

#

clearpart --all --initlabel

#clearpart --linux

key 0206063501adfc06

?

#

# Customize to fit your needs

#

#No RAID

#/boot really significant forthis sort of setup nowadays?

#part /boot --size 50 --fstypeext3

part /????--size40960 --fstype ext3 --asprimary

part /boot --size 1024 --fstype ext3

part /tmp --size 1024 --fstype ext3

part swap --size 24576

part /var --size 2048 --fstype ext3

part /wglogs --size 2048 --fstype ext3

part /local1/scr --size 174080 --fstype ext3

part /local2/scr --size 309600 --fstype ext3

#part / --size 1 --grow--fstype ext4

#RAID 0 /scr for performance

#part / --size 1024 --ondisksda

#part swap --size 512 --ondisksda

#part /var --size 1024 --ondisksdb

#part swap --size 512 --ondisksdb

#part raid.01 --size 1 --grow--ondisk sda

#part raid.02 --size 1 --grow--ondisk sdb

#raid /scr --level 0 --devicemd0 raid.01 raid.02

#Full RAID 1 Sample

#part raid.01 --size 50--ondisk sda

#part raid.02 --size 50--ondisk sdb

#raid /boot --level 1 --devicemd0 raid.01 raid.02

#

#part raid.11 --size 1024--ondisk sda

#part raid.12 --size 1024--ondisk sdb

#raid / --level 1 --device md1raid.11 raid.12

#

#part raid.21 --size 1024--ondisk sda

#part raid.22 --size 1024--ondisk sdb

#raid /var --level 1 --devicemd2 raid.21 raid.22

#

#part raid.31 --size 1024--ondisk sda

#part raid.32 --size 1024--ondisk sdb

#raid swap --level 1 --devicemd3 raid.31 raid.32

#

#part raid.41 --size 1 --grow--ondisk sda

#part raid.42 --size 1 --grow--ondisk sdb

#raid /scr --level 1 --devicemd4 raid.41 raid.42

#

# bootloader config

# --append <args>

# --useLilo

# --md5pass <crypted MD5password for GRUB>

#

bootloader

#

# install or upgrade

#

install

#

# text mode install (default isgraphical)

#

text

#

# firewall

#

firewall --disabled

#

# Select a zone

# Add the --utc switch if yourhardware clock is set to GMT

#

#timezone US/Hawaii

#timezone US/Pacific

#timezone US/Mountain

#timezone US/Central

#timezone US/Eastern

timezone --utc"#TABLE:site:key=timezone:value#"

#

# Don't do X

#

skipx

#

# To generate an encrypted rootpassword use:

#

# perl -e 'printcrypt("blah","Xa") . "\n";'p

# openssl passwd -apr1 -saltxxxxxxxx password

#

# where "blah" isyour root password.

#

#rootpw --iscryptedXaLGAVe1C41x2

#rootpw XaLGAVe1C41x2--iscrypted

rootpw --iscrypted#CRYPT:passwd:key=system,username=root:password#

#

# NIS setup: auth --enablenis--nisdomain sensenet

# --nisserver neptune--useshadow --enablemd5

#

# OR

auth --useshadow --enablemd5

#

# SE Linux

#

selinux --disabled

#

# Reboot after installation

#

reboot

#

#end of section

#

%packages

#INCLUDE_DEFAULT_PKGLIST#

%pre

#INCLUDE:#ENV:XCATROOT#/share/xcat/install/scripts/pre.rh#

%post

#INCLUDE:#ENV:XCATROOT#/share/xcat/install/scripts/post.rh#

?

2)把cmdline注銷掉,這樣節(jié)點安裝時以傳統(tǒng)的網(wǎng)絡(luò)安裝字符界面進行,比較順眼,這不是必須的。

修改compute.rhels5u5.pkglist

如果需要安裝所有的rpm包,這個文件只需要寫兩行,如下:

[root@manager1 rh]# morecompute.rhels5u5.pkglist

kickstar_config/compute.rhels5u5.pkglist.201204174

@包名

注意:

1)在rhel5.5之后,不再支持everything這個詞了,rhel6也不行。

2)如果選擇安裝所有包,默認節(jié)點會從虛擬化kernel啟動。可以在postscripts里自己寫一個腳本,讓系統(tǒng)從標準kernel啟動。

?

復(fù)制安裝文件

cp *.iso?/install

copycds *.iso

mv rhels5.5? rhels5u5 (因為nodetype定義的是rhels5u5; /install/rhels5u5 )

設(shè)置安裝

#rbootseq ON001-0N128??n,h,c,f??設(shè)置啟動順序

#rbootseq ON001-0N128??stat

然后,啟動計算節(jié)點,從網(wǎng)絡(luò)啟動即可。

rpower compute??off

rinstall ON001-ON126

nodestat??ON001-ON126 (查看各節(jié)點狀態(tài))

補充:

configRSH腳本

#!/bin/sh

#

for i in rlogin rsh rexec;

do

????chkconfig $i on;

????echo $i >> /etc/securetty;

done;

service xinetd restart;

cat > /root/.rhosts<<EOF

節(jié)點名稱

EOF

cat > /etc/hosts.equiv<<EOF

節(jié)點名稱

EOF

logger -t xCAT "configRSH:autostart RSH successfully"

exit 0

configService腳本

#!/bin/sh

#

services="avahi-daemon gpmhidd hplip isdn mcstrans pcscd restorecond rpcgssd rpcidmapd setroubleshootsendmail smartd bluetooth cups firstboot iptables ip6tables iscsi iscsid rhnsdyum-updatesd";

logger -t xCAT"confService: starting config services"

for srv in $services

do

????chkconfig $srv off > /dev/null 2>&1

done

if [ -f /etc/cron.daily/logrotate] ; then

????rm -f /etc/cron.daily/logrotate

????logger -t xCAT "confService: remove lorotate crontabdaily"

fi

logger -t xCAT"confService: config services successfully"

exit 0

?

xcat管理節(jié)點配置命令

Xcat服務(wù)啟停

service xcatd (status?、stop、start、restart)

示例:

service xcatd status

構(gòu)建dhcp服務(wù)

makedhcp

示例:

makedhcp compute (compute為節(jié)點組名)

構(gòu)建dns服務(wù)

makedns

示例:

makedns compute (compute為節(jié)點組名)

查看xcat信息

tabdump

示例:

tabdump nodelist

編輯xcat數(shù)據(jù)庫

tabedit

示例:

tabedit passwd

改變表項的值

chtab

示例:

chtab mpa.mpa=testmpa.username=admin mpa.password=123456

獲得計算節(jié)點mac地址

getmacs

示例:

getmacs all

nodeadd

nodeaddnoderangegroups=groupnames?[table.column=value][...]

nodeadd ON001-ON128 groups=all,compute

lsdef??ON001

[root@manager1 ~]# lsdef ON001

Object name: ON001

????arch=x86_64

????groups=all,compute

????id=1

????installnic=eth1

????mac=34:40:B5:81:7F:5A

????mgt=blade

????monserver=172.16.1.1

????mpa=BCE1

????netboot=pxe

????nfsserver=172.16.1.1

????os=rhels5u5

????postbootscripts=otherpkgs

???postscripts=syslog,remoteshell,syncfiles,setupntp,hardeths,configRSH,configService

????power=blade

????primarynic=eth1

????profile=compute

????xcatmaster=172.16.1.1

計算節(jié)點并行命令

并行執(zhí)行命令

psh

Usage: psh [-i<interface>] [-l <user>] <noderange> <command>

示例:

psh ON001-ON006??date

從節(jié)點范圍中剔除某個節(jié)點可以如下使用

示例:

psh ON001-ON006,-ON001 date

并行ping

pping

示例:

pping ON001-ON126

并行拷貝文件

pscp

Usage: pscp [-i <SUFFIX>][SCP OPTIONS...] FILE... <NODERANGE>:<DESTINATION>

示例:

pscp??example.txt? ON001-ON006,-ON005,-ON004:/root/

并行開關(guān)機,查看電源狀態(tài)

rpower

Usage: rpower <noderange>[--nodeps] [on|onstandby|off|suspend|reset|stat|state|boot]

示例:

rpower ON001-ON126 stat

Rpower ON001-ON126 off

并行修改引導(dǎo)順序

rbootseq

rbootseq <noderange>[hd0|hd1|hd2|hd3|net|iscsi|usbflash|floppy|none],

示例:

rbootseq ON001-ON126 h,n,c,f

并行安裝命令

rinstall

示例:

rinstall ON001-ON126

查看節(jié)點狀態(tài)

nodestat

nodestat [noderange][-m|--usemon] [-p|powerstat] [-u|--updatedb]

示例:

nodestat ON001-ON126

歡迎加入
database群:119224876(db china聯(lián)盟) 虛擬化方面群:229845401(虛擬化-云計算-物聯(lián)網(wǎng))

與50位技術(shù)專家面對面20年技術(shù)見證,附贈技術(shù)全景圖

總結(jié)

以上是生活随笔為你收集整理的xcat 安装(liunx高性能刀片集群管理软件)的全部內(nèi)容,希望文章能夠幫你解決所遇到的問題。

如果覺得生活随笔網(wǎng)站內(nèi)容還不錯,歡迎將生活随笔推薦給好友。