《AIXHACMPoracle双机集群项目.doc》由会员分享，可在线阅读，更多相关《AIXHACMPoracle双机集群项目.doc（18页珍藏版）》请在taowenge.com淘文阁网|工程机械CAD图纸|机械工程制图|CAD装配图下载|SolidWorks_CaTia_CAD_UG_PROE_设计图分享下载上搜索。
1、AIX+HACMP+oracle双机集群项目一、集群分类1、按用途分类（1）高可用集群（High Availability Cluster）（2）负载均衡集群（Load Balance Cluster）（3）科学计算集群（High Performance Computing Cluster）2、按结构分类（1）主从模式(active-standby mode)（2）双机互备(mutual backup mode)（3）多点集群(multi node cluster)3、按厂商分类（1）IBM：HACMP(High Availability Cluster Multi-Processing)（2
2、）HP：HPTC(High Performance Technical Computing)（3）Sun：Sun Cluster（4）RedHat：RHCS(RedHat Cluster )HA的工作原理：Boot IP ,standby IP ,service IP(服务IP) ,heart beatStandby Network Service Aservice NetworkService B 二、项目规划：1、项目流程：集群项目步骤（重要）（1）Define HA Topology Structure（定义HA TOP结构） P3021)configure HA Cluster nam
3、e（配置集群名字）2)Add nodes to HA Cluster（加节点）3)configure network types between two nodes（配置网络类型）4)configure Interfaces/devices for network types（配置接口/设备）（2）Define HA resource configure（定义HA资源配置）1)configure 2 Application server (DB start/stop service)（配置2个启停脚本）2)configure 2 service IP Address（配置2个服务IP）(2个V
4、G组免配)3)configure resource Groups & participating nodes（创建资源组和节点配置）4)configure resources in each resource groups （配置每个资源组的资源）2、硬件配置：IBM P640小型机2台，单台配置本地SCSI硬盘2块及双网卡IBM 7133 D40 磁盘阵列一台，配置SSA阵列硬盘8块3、软件配置：AIX 5.3.0.0-0.4操作系统；HACMP 5.3 集群软件；ORACLE 10.2 数据库软件；4、实现目标：1）阵列硬盘分为两组分别实现RAID5,并分别归属于两台小型机作为datavg
5、；2）两台小型机分别运行两个不同的数据库服务，数据存放在datavg上，两机实现双机互备。3）当有任一台小机故障时，另一台小机将自动接管故障机的数据库应用，以保证服务的持续高可用性。三、项目实施：1、硬件连接2、操作系统及配置（1）安装AIX5300-04操作系统，选择64bit内核及JFS2文件系统（2）安装HACMP所需扩展软件包，列表如下：bos.adt.libmbos.adt.syscallsbos.clvm.enhbos.datarsct.basic.hacmprsct.basic.rtersct.basic.sppat.basic.hacmppat.basic.rtepat.bas
6、ic.sppat.clients.hacmppat.clients.hacmppat.clients.rtepat.clients.sp（3）安装oracle所需扩展软件包，列表如下：（可以和上面的包一起装）bos.cifs_fs.rtebos.adt.prof# smit install（4）串口心跳或磁盘心跳测试（二选一）1）串口心跳测试（使用串口线传输心跳信号）B802# lsdev -C | grep ttytty0 Avaliable01-S1-00-00Asynchronous TerminalB801# stty /dev/tty0B802机向TTY0输入信息B802# stty
7、 /dev/tty02）磁盘心跳测试（使用共享磁盘传输心跳信号，AIX5.1开始支持）在B801机上使用独立PV（与数据VG分开）建立enhanced concurrent卷组，专用于传输心跳信号关键点：VOLUME GROUP name设为 heartvgPhysical partition size in megabytes设为128physical volume names设为hdisk9Activate volume group automatically设为noCreate VG concurrent capable设为enhanced concurrent在B802机上将B801机
8、上定义好的心跳卷组import导入关键点：Volume group name设为heartvgPhysical volume name设为hdisk9测试磁盘心跳：B801# /usr/sbin/rsct/bin/dhb_read -p hdisk9 -rB801机等待从心跳磁盘输入信息B802# /usr/sbin/rsct/bin/dhb_read -p hdisk9 -tB802机向心跳磁盘输出信息，两机均应出现link operating normally信息B802# /usr/sbin/rsct/bin/dhb_read -p hdisk9 -r 上述过程反向测试B801# /us
9、r/sbin/rsct/bin/dhb_read -p hdisk9 t（5）网络地址配置：B801服务器：en0:192.168.100.101/24boot IPen0:192.168.99.101/24standby IPB801# smit tcpipB801# hostname B801B802服务器：en0:192.168.100.102/24boot IPen0:192.168.99.102/24standby IPB802# smit tcpipB802# hostname B802（6）datavg配置（可仅在主机上进行，完成后import到备机即可）B801# smit s
10、saraid查看阵列中物理盘的属性- List SSA physical disk- List HotSpare/Candidate/system diskB801# smit ssaraid- Change use of an SSA phydical diskB801# smit ssaraid选择raid成员创建两个Array(hdisk)- Add an SSA ArrayB801# lspv列出生成的两个物理卷pv(hdisk)B801# smit vg建datavg801和datavg802,各含一个hdisk,PP大小64MB801#lsvg -o 查看两卷组均已varyon.B
11、801# smit chvg分别定义两个datavg的active automatically属性为No关键点：Volume grouop name分别设为datavg801和datavg802Activate volume group automatically设为no(表示此处不要做(由集群软件控制) B801#varyoffvg datavg801在B801机上卸载两激活卷组 B801# varyoffvg datavg802以便备机能将VG信息进行importvgB802#cfgmgrB802机重认设备B802#importvg -y datavg801 hdisk?B802机重认设备
12、B802#importvg -y datavg802 hdisk?B802机重认设备B802# lsvg -o 查看两卷组均已varyonB802# smit chvg分别定义两datavg的active automatically属性为no关键点：Volume grouop name分别设为datavg801和datavg802Activate volume group automatically设为no(表示此处不要做(由集群软件控制)B802# varyoffvg datavg801将所有卷组在所有机器上varyoffB802# varyoffvg datavg802以便交由HA软件控制
13、varyonvg/varyoff（7）应用系统启动脚本配置B801# touch/etc/801dbstart生成B801服务器数据库启动脚本B801# touch/etc/801dbstop生成B801服务器数据库停止脚本B801# touch/etc/802dbstart生成B802服务器数据库启动脚本B801# touch/etc/802dbstop生成B802服务器数据库停止脚本B801# chmod +x /etc/80*B802# touch/etc/801dbstartB802# touch/etc/801dbstopB802# touch/etc/802dbstartB802#
14、 touch/etc/802dbstopB802# chmod +x /etc/80*3、HACMP安装及配置（1）安装HACMP5.3除HAtivoli/Haview；两类软件包外，安装其他所有HA软件包# smitty install（2）双机网络地址配置B801# vi /etc/hosts(最好三个都不同)192.168.100.104B802_service192.168.100.103B802_service192.168.99.102B802_standby192.168.99.101B802_standby192.168.100.102B802_boot192.168.100.
15、101B802_bootB801B802# vi /etc/hosts192.168.100.104B802_service192.168.100.103B802_service192.168.99.102B802_standby192.168.99.101B802_standby192.168.100.102B802_boot192.168.100.101B802_bootB802（3）双机信任关系配置B801# vi /.rhostsB802_serviceB801_serviceB802_standbyB801_standbyB802_bootB801_bootB801# cp /.rh
16、osts/etc/hosts.equivB801# cp /.rhosts/usr/es/sbin/cluster/netmon.cfB802机同上设置（4）HA服务配置B801# cat /etc/inittab确定inittab文件末行已加入启动HA服务.。clcomdES:2:once:startsrc -s clcomdES /dev/console 2&1 B801# lssrc -s clcomdES确定HA服务当前为ACTIVE状态B801#startsrc -s clcomdES否则以命令启动HA服务B802机同上设置（后续HA配置则可仅在单台主机如B801上进行）（5）clu
17、ster配置1）配置cluster逻辑名B801#smit hacmp -Etxtended Configuration -Extended Topology -configgure an HA cluster -Add a hacmp cluster配置cluster名称关键点：Cluster name自己定2）配置cluster节点B801#smit hacmp -Initization -Add node指定cluster名称，并定义NewNode(选择两个boot地址)关键点：New Nodes选择B801_boot和 B802_boot B801#smit hacmp -Extend
18、ed Configuration -Extended Topology -configure HA node -Add a node to HA cluster 分别定义两个node name为两机机器名，及各自的communication Path为自己的boot IP关键点：Node name设为B801，对应communication path为B801_boot；Node name设为B802，则对应communication path为B802_boot。3）配置cluster中网络类型B801#smit hacmp -Extended Configuration -Extended
19、 Topology -configure HA networks -show HA networks确定系统中已有两种网络类型：net_ether_01(系统自定IP网络，其中包含bootIP和standbyIP的网段)；net_rs232_01(系统自定串口心跳网络)或net_hdiskb_01(系统自定磁盘心跳网络)，并将net_ether_01属性enable IP address takeover via IP aliases 定义为NO关键点：Enable IP address takeover via IP Aliases设为NOB801#smit hacmp -Extended
20、configuration -Extended Topology -configure HA networks -add a network to the HA cluster若上步中无net_ether_01和net_rs232_01信息，则需手动指定添加两种网络类型:IP网络类型ether;rs232（串口心跳网络）或diskhb(磁盘心跳网络)关键点：(串口心跳rs232和磁盘diskhb只要配其中一个就行了)a.选择#pre-defined IP-based Network Types下几行的etherb.Enable IP address takeover via IP Aliase
21、s设为NOc.选择#pre-defined IP-serial Device Types下几行的ers232(串口心跳)d.选择#pre-defined IP-serial Device Types下几行的diskhb(磁盘心跳)4）配置不同类型网络中的具体接口或设备B801#smit hacmp -Extended Configuration -Extended Topology -configure communication -Add -Predefined -Interfaces关键点：a.IP Label/Address设为B801_boot Node Name设为B801b.IP
22、Label/Address设为B801_standby Node Name设为B801c.IP Label/Address设为B802_boot Node Name设为B801d.IP Label/Address设为B802_standby Node Name设为B801B801#smit hacmp -Extended Configuration -Extended Topology -configure communication/devices -Add -Predefined -devices(配串口（或者配磁盘心跳）a.选择net_ether_01；B802 tty0/dev/tty
23、0b.选择net_diskhb_01；Device Name设为diskhb_B801；Device Path设为/dev/hdisk9；Node Name设为B801;Device Name设为diskhb_B802；Device Path设为/dev/hdisk9；Node Name设为B802。5）配置HA应用B801#smit hacmp -Extended Configuration-Extended Resources-Resources Configuration -Application-servers-add app server定义两机各自不同的APP名，及各自的服务启停脚
24、本关键点：a.Server Name设为application_B801;Start Script设为/etc/801dbstart;Stop Script设为/etc/801dbstopb.Server Name设为application_B802;Start Script设为/etc/802dbstart;Stop Script设为/etc/802dbstop6）配置cluster服务IPB801#smit hacmp -Extended Configuration -Extended Resources -Resources Configure -IP lables/Addresses
25、-add a service IP -configurable -net_ethernet_01定义两机的IP lables/Addresses为各自机器的service IP关键点：a.IP Label/Address设为B801_serviceb.IP Label/Address设为B801_service7）配置cluster资源组 B801#smit hacmp -Extended Configuration-Extended Resources -Resources Group Configure -Add a resource group关键点：a.Resource Group N
26、ame设为resource_B801;Participating Nodes设为B801 B802(注意801,802前后顺序)b.Resource Group Name设为resource_B802;Participating Nodes设为B802 B801(注意801,802前后顺序)8）配置cluster资源组的资源B801#smit hacmp -Extended Configuration-Extended Resources -Resources Group Configure -Change attributes -resource_B801分别确定两Resources Gro
27、up的service IP/Address,ApplicationServers和各自所属VGs关键点：a.Service IP Label/Address设为B801_service;Application Servers设为application_B801;Volume Groups设为datavg801b.Service IP Label/Address设为B802_service;Application Servers设为application_B802;Volume Groups设为datavg8029）同步及校验clusterB801#smit hacmp -Extended Co
28、nfiguration-Verification and synchronization将HA配置信息同步到从机如B802上，并进行校验关键点：Automatically correct errors found during设为Yes.10）启停HAB801#smit clstart (now)B802#smit clstart (now)两机上同时启动HA服务关键点：Start now,on system restart or boot设为nowB801#lssrc -g clusterB802#lssrc -g cluster两机上同时查看HA服务是否正常（至少得有clstrmgrES有
29、的有clsmuxpdES）B801#ifconfig -aB802#ifconfig -a两机上同时查看bootIP已切换为serviceIPB801#lsvg -oB802#lsvg -o两机上同时查看datavg已varyonB801#smit clstopB802#smit clstop两机上同时停止HA服务关键点：Stop now,on system restart or boot设为nowShutdown mode设为graceful(停节点但不接管)或Force(停节点但不释放资源)或Takeover(停节点，释放资源，且可以接管)Stop HA的3种模式：Graceful：停节点
30、，释放资源，活动节点不接管Force：停节点，不释放资源Takeover：停节点，释放资源，活动节点接管资源B801#lssrc -g clusterB802#lssrc -g cluster两机上同时查看HA服务是否已停B801#ifconfig -aB802#ifconfig -a两机上同时查看serviceIP已切换为bootIPB801#lsvg -oB802#lsvg -o两机上同时查看datavg已varyoff11）HA测试A.两机clstart(now),可见datavg801和datavg802已在两机上分别varyon，然后各自拔掉bootIP/service IP网卡的网
31、线，可见serviceIP切换到本机的standbyIP网卡上，两datavg不受影响。C.接好上一步中拔掉的网线，在B801机上clstop(now)并确定参数takeover，则B801机的serviceIP会切换回bootIP，且datavg801在B801机上自动varyoff,随后B802机的standbyIP网卡自动接管B801机的serviceIP,同时datavg801在B802机上被自动varyon,此时来那个datavg均在B802上被varyon。C.在B801机上重新clstart(now)，则B801机会把自己的serviceIP从B802机上收回，B801机的boo
32、tIP切换为serviceIP,datavg80被重新varyon到B801上，同时B802机的standbyIP网卡恢复standbyIP。D.在B802上再进行上述两步测试。上述测试正常，clstop(now)关闭两机的HA服务，以便进行后续配置4、oracle安装及配置（两台服务器上均需进行）（1）oracle安装环境检查1）mem1G#lsattr -El mem03、swapmem*1.2#lsps -a4、/tmp400M#df -k(df -m或df -g都可)5、Hard Disk Space: 4G(安装所需空间)+3G(安装源文件)#df -k （2）AIX扩展软件包及补丁
33、需求#lslpp -l bos.adt.lib#lslpp -l bos.adt.libm#lslpp -l bos.perf.libbperfstat#lslpp -l bos.perf.libperfstat#lslpp -l bos.perf.proctools#instfix -ik IY58143#instfix -ik IY59386#instfix -ik IY60930#instfix -ik IY66513#instfix -ik IY70159#instfix -ik IY68989（3）其他参数限制#smit system -change/show characteris
34、tics of OS关键点：Maximum number of PROCESSES allowed per user设为2048#smit aio -change/show characteristics of AIO关键点：STATE to be configured at system restart设为avaliable(更改异步IO（在oracle中必须打开（AIX6默认打开)（4）增加用户和组#smit group -add group -oinstall/dba#smit user -add user -oracle#passwd oracle关键点：User Name设为orac
35、lePrimary Group设为dba,haguserSoft file size,soft cpu time,soft data segment,soft stack size和soft core file size均设为-1。软限：(-1表示没有限制)（5）修改环境变量#vi /home/oracle/.profileUmask 022Export ORACLE_SID=ora101(B801设置)Export ORACLE_SID=ora102(B802设置)Export ORACLE_BASE=/oracleExport ORACLE_HOME=$ORACLE_BASE/produc
36、t/10.2.0/db_1Export ORACLE_TERM=xtermExport ORA_NLS33=$ORACLE_HOME/nls/dataExport LS_LIBRARY_PATH=$ORACLE_HOME/lib:$ORACLE_HOME/network/lib:.Export NLS_LANG=American_america.zhs16gbkExport SHLIB_PATH=$ORACLE_HOME/lib:.:$PATH:/usr/local/bin:/usr/ccs/binExport CLASSPATH=$ORACLE_HOME/jlib:.Export LIBPA
37、TH=$ORACLE_HOME/JDK/JRE/BIN:$ORACLE_HOME/jdk/jre/bin/classic:$ORACLE_HOME/lib32Export DISPLAY=192.168.100.101:0.0 (B801设置)Export DISPLAY=192.168.100.102:0.0 (B802设置)#vi /home/oracle/.pprofileDTSOURCEPROFILE=true（6）创建oracle文件系统及处理oracle安装文件#mklv -y lv_oracle rootvg 8G#smit fs#mount /oracle#chown -R o
38、racle:oinstall /oracle#chmod -R 775 /oracle#su - oracle$cd /oracle$gunzip 10gr2_aix5l64_database.cpio.gz$cpio -icdmv 10gr2_5l64_database.cpio（7）以oracle用户重新登录并开始安装oracle$su - root#/oracle/Disk1/rootpre/rootpre.sh(完毕后关闭当前窗口)（系统环境检查脚本，也可能）$cd /oracle/Disk1$./runInstaller（8）建立裸设备并创 建实例（可仅在主机上，完成后import到
39、备机即可）B801#varyonvg datavg801B801#varyonvg datavg802B801#mklv -y lv_ora1_spfile datavg801 1B801#mklv -y lv_ora1_ctrl01 datavg801 1B801#mklv -y lv_ora1_ctrl02 datavg801 1B801#mklv -y lv_ora1_ctrl03 datavg801 1B801#mklv -y lv_ora1_redo01 datavg801 1B801#mklv -y lv_ora1_redo02 datavg801 1B801#mklv -y lv
40、_ora1_redo03 datavg801 1B801#mklv -y lv_ora1_system datavg801 10B801#mklv -y lv_ora1_sysaux datavg801 10B801#mklv -y lv_ora1_undo01 datavg801 10B801#mklv -y lv_ora1_temp01datavg801 10B801#mklv -y lv_ora1_user01 datavg801 20B801#mklv -y lv_ora2_spfile datavg801 1B801#mklv -y lv_ora2_ctrl01 datavg801
41、1B801#mklv -y lv_ora2_ctrl02 datavg801 1B801#mklv -y lv_ora2_ctrl03 datavg801 1B801#mklv -y lv_ora2_redo01 datavg801 1B801#mklv -y lv_ora2_redo02 datavg801 1B801#mklv -y lv_ora2_redo03 datavg801 1B801#mklv -y lv_ora2_system datavg802 10B801#mklv -y lv_ora2_sysaux datavg802 10B801#mklv -y lv_ora2_und
42、o01 datavg802 10B801#mklv -y lv_ora2_temp01 datavg802 10B801#mklv -y lv_ora2_user01 datavg802 20B801#hown oracle /dev/*lv*B801#varyoffvg datavg801B801#varyoffvg datavg802B802#importvg -y datavg801 hdisk?B802#importvg -y datavg802 hdisk?B802#chown oracle /dev/*lv*B802#smitty chvgDatavg801和datavg802在B
43、802上经importvg后，属性恢复为默认设置，应重新定义两数据VG的active automativally属性为NO关键点：a.Volume group name设为datavg801 Active volume group automativally设为nob.Volume group name设为datavg802 Active volume group automativally设为noB802#varyoffvg datavg801在B802上卸载datavg801并在801上激活B801#varyoffvg datavg801以便后续分别在两机上各自建立不同实例在B801/B8
44、02上分别以oracle用户身份开始创建实例ora101和ora102，实例ora101的所有数据库文件都建立在逻辑卷裸设备rlv_ora1_xxxxx上，实例ora102所有数据库文件都建立在逻辑卷裸设备rlv_ora2_xxxxx上。在两机上分别执行DBCA关键点A.Create a databaseB.general purposeC.global database name设为ora101或ora102 SID设为ora101或ora102D.configure the database with enterprise manager不选E.use the same passwd fo
45、r all accountsF.Raw DevicesG.specify flash recovery area; 选enable archiving 编辑edit archive mode parameters中archive log file format中%t_%s_%r.arc 存储位置自己写H.chracter sets选择choose from the list of character sets:ZHS16GBKI.connection mode选dedicated server modeJ.sizing 设server parameter filename为/dev/rlv_ora1_spfile(或/dev/rlv_ora2_spfile)K.controlfileFile NameFile Drectoryrlv_ora1_ctrl01/dev/rlv_ora1_ctrl02/dev/rlv_ora1_ctrl