1、OCR的备份与恢复ASM管理Oracle11gR2 RAC的备份与恢复真实案例讲述Oracle11gR2 RAC软件(Grid Infrastructure+RDBMS)的备份与恢复本案例涉及三个内容标签:如何恢复Oracle Local Registry(OLR)如何恢复ora.crsd和ora.evmd进程如何建立完备有效的Oracle11gR2 RAC软件备份机制环境:Oracle11gR2 RAC shade01(node1)+shade02(node2),ASM,Grid Infrastructure 11.2.0.1,RDBMS 11.2.0.1,AIX 6.1 TL-04至此,这
2、个故障得以修复。但是,更为重要的问题是,我到底该如何去备份11gR2 RAC的软件才是完备有效的呢,为什么我用之前的备份拿来作恢复,会碰到这么多问题呢,我该怎么完善我的备份机制呢?通过查询各方资料和这个案例的经验教训,我总结了一下,以下三个都是需要备份的:1. GRID_HOME2. RDBMS_HOME3. runinstaller inventory (位于/etc/orainst.loc)例如/etc/init.d/init.ohasd等文件都会在上述三个位置有拷贝,所以上述三个位置的备份是必须要做的。另外,在11gR2中,GRID_HOME有两个registry,OCR和OLR,一个负
3、责shared,一个负责local,这两个2个registry都需要特别注意。当crs资源增加或者删除的时候,这两个registry都会发生变化,因此都需要定期备份。对OCR来说,Oracle会通过ocrconfig命令,每隔4小时自动备份OCR到/cdata /下,而voting file会在每次配置变化的时候备份至OCR,并且可以通过crsctl命令去恢复。使用ocrconfig工具可以在任何时候强制Oracle的crs 去执行一个OCR的备份,当然也可以等待4小时的时间间隔去让crs自动备份OCR,但是这个工具还是非常有用的,比如按照你的需要在对OCR进行配置的 变更前,去执行一个二进制
4、的OCR备份,就很有必要。手工备份OCR的方法:1. root用户登录系统2. # ocrconfig -manualbackup =观察执行后OCR备份文件的变化,如日期OLR则会在grid安装或升级结束后自动备份一次,以后就不会自动备份了,我们只能手工去备份OLR,没有其他选择。当我们决定将OCR从ASM 迁移到其他存储方式,或者从其他存储方式迁移到ASM时,我们需要对OLR创建一个新的备份。OLR默认的存放位置是Grid_home/cdata /host_name/下。手工备份OLR的方法:1. root用户登录系统2. # ocrconfig local manualbackup在了解
5、了11gR2 RAC相对于10g的这些新变化后,我们对应该怎么去备份RAC软件(Grid Infrastructure+RDBMS)也有了比较深入的认识。我重新实验了一次,可以顺利恢复。参考:Oracle Support- How to Troubleshoot Grid Infrastructure Startup Issues ID 1050908.1RAC在OCR磁盘故障且无备份时快速恢复的方法2010-12-26 22:46 问题背景:在刷业务库的过程出现IO访问磁盘失败!检查发现RAC进程全部退出!重新执行/opt/oracrs/bin/crsctl start crs失败出错信息(
6、tail -f /var/log/messages):Oracle Cluster Registry initialization failed with invalid format: PROC-22: The OCR backend has an invalid format怀疑时OCR盘物理损坏(后来发现用作OCR的阵列有磁盘坏告警,立即修复,阵列故障处理略过)这时在一个RAC节点检查OCR是否有备份ocrconfig -showbackup无结果输出(如果有自动备份则有结果,不幸的是,没有!)/opt/oracle/product/11g/db/cdata下无OCR备份(如果有自动备份
7、,则在此目录有备份的文件)ocrcheck无结果输出(这时基本可以定位是OCR盘故障了)GDGZ-DCS-SV01C-RAC01:/opt/oracle/product/11g/db/bin # ./crsctl check bootOracle Cluster Registry initialization failed with invalid format: PROC-22: The OCR backend has an invalid format和前面的启动报错是一致的!问题处理(OCR故障且无备份如何快速重建恢复):一,清除OCR1、停止crs进程(我此次RAC已经自己退出,故不涉
8、及;如果发现有crs进程,则需手动退出)#/opt/oracrs/bin/crsctl stop crs2、备份整个OCR目录#cp /opt/oracrs /opt/oracrs_bak3,分别在每个节点上执行$ORA_CRS_HOME/install/rootdelete.sh (必须root执行)GDGZ-DCS-SV02C-RAC01:/etc/oracle # /opt/oracrs/install/rootdelete.shGetting local node nameNODE = GDGZ-DCS-SV02C-RAC01PRKO-2006 : Invalid node name:
9、 GDGZ-DCS-SV02C-RAC01Oracle Cluster Registry initialization failed with invalid format: PROC-22: The OCR backend has an invalid formatOracle CRS stack is not running.Oracle CRS stack is down now.Removing script for Oracle Cluster Ready servicesUpdatingocr file for downgradeCleaning up SCR settings i
10、n /etc/oracle/scls_scrCleaning up Network socket directories4,在某个RAC节点执行/opt/oracrs/install/rootdeinstall.shshell-init: error retrieving current directory: getcwd: cannot access parent directories: No such file or directoryVerifying existence of ocr.loc fileRemoving contents from OCR device2560+0 re
11、cords in2560+0 records out10485760 bytes (10 MB) copied, 1.15907 seconds, 9.0 MB/s5,删除旧OCR信息修改 /etc/inittab, 删除以下三行.h1:2:respawn:/etc/init.evmd run /dev/null 2&1 /dev/null 2&1 /dev/null 2&1 /dev/nullrm -rf /etc/oracle/* rm -f /etc/init.d/init.cssdrm -f /etc/init.d/init.crsrm -f /etc/init.d/init.crsd
12、rm -f /etc/init.d/init.evmdrm -f /etc/inittab.crscp /etc/inittab.orig /etc/inittabrm -rf /var/tmp/.oraclerm -rf /tmp/.oracle (由于执行rootdelete.sh时删除了上面的一些目录或文件,故仅需补充删除存在的即可)6,使用dd清除vote disk和ocr,需要修改成相应的文件dd if=/dev/zero of=/dev/raw/raw1 bs=8192 count=12800dd if=/dev/zero of=/dev/raw/raw2 bs=8192 coun
13、t=12800GDGZ-DCS-SV02C-RAC01:/opt # dd if=/dev/zero of=/dev/raw/raw2 bs=8192 count=128000dd: writing /dev/raw/raw2: No space left on device123500+0 records in123499+0 records out1011709440 bytes (1.0 GB) copied, 67.9911 seconds, 14.9 MB/sGDGZ-DCS-SV02C-RAC01:/opt # dd if=/dev/zero of=/dev/raw/raw1 bs
14、=8192 count=128000dd: writing /dev/raw/raw1: No space left on device123496+0 records in123495+0 records out1011677184 bytes (1.0 GB) copied, 66.8413 seconds, 15.1 MB/s(从第4步结果提示来看,应该ocr已经清除了,可以尝试只清除VOTEDISK,可惜我先执行了dd清除命令)二,重建OCR(这里我独创性的使用图形工具runInstaller,模仿新建的方法来重建OCR)1,由于首次安装CRS时需要新的空的/opt/oracle/op
15、t/oracrs目录,所以先在各个节点执行如下命令:cd /optmv oracle oracle_oldmv oracrsoracrs_oldmkdir oraclemkdiroracrsmkdir /opt/oracle/oraInventorychownoracle:dba /opt/oracle/oraInventorychmod 755 /opt/oracle/oraInventorychown -R oracle:dba /opt/oraclechmod -R 770 /opt/oraclechown -R oracle:dba /opt/oracrschmod -R 770 /o
16、pt/oracrs2,xmanager登陆RAC1节点使用图形工具安装CRS(这里可以参考RAC安装过程的CRS安装部分)oracle用户执行/opt/orabak/clusterware/runInstaller需要注意的是,这里的CRS安装配置需要和第一次安装配置保持完全一致!安装最后需要根据图形界面提示在各个RAC节点执行root.shGDGZ-DCS-SV01C-RAC01:/opt/oracle # /opt/oracrs/root.shChecking to see if Oracle CRS stack is already configured/etc/oracle does
17、not exist. Creating it now.Setting the permissions on OCR backup directorySetting up Network socket directoriesOracle Cluster Registry configuration upgraded successfullySuccessfully accumulated necessary OCR keys.Using ports: CSS=49895 CRS=49896 EVMC=49898 and EVMR=49897.node : node 1: gdgz-dcs-sv0
18、1c-rac01 gdgz-dcs-sv01c-rac01_base gdgz-dcs-sv01c-rac01node 2: gdgz-dcs-sv02c-rac01 gdgz-dcs-sv02c-rac01_base gdgz-dcs-sv02c-rac01node 3: gdgz-dcs-sv03c-rac01 gdgz-dcs-sv03c-rac01_base gdgz-dcs-sv03c-rac01node 4: gdgz-dcs-sv04c-rac01 gdgz-dcs-sv04c-rac01_base gdgz-dcs-sv04c-rac01Creating OCR keys fo
19、r user root, privgrp root.Operation successful.Now formatting voting device: /dev/raw/raw2Format of 1 voting devices complete.Startup will be queued to init within 30 seconds.Adding daemons to inittabExpecting the CRS daemons to be up within 600 seconds.Cluster Synchronization Services is active on
20、these nodes. gdgz-dcs-sv01c-rac01gdgz-dcs-sv02c-rac01gdgz-dcs-sv03c-rac01gdgz-dcs-sv04c-rac01Cluster Synchronization Services is active on all the nodes. Waiting for the Oracle CRSD and EVMD to startOracle CRS stack installed and running under init(1M)Running vipca(silent) for configuring nodeappsCr
21、eating VIP application resource on (4) nodes.Creating GSD application resource on (4) nodes.Creating ONS application resource on (4) nodes.Starting VIP application resource on (4) nodes.Starting GSD application resource on (4) nodes.Starting ONS application resource on (4) nodes.Done.点击OK后检查有一个失败,根据
22、提示:/opt/oracrs/cfgtoollogs下有出错命令日志/opt/oracrs/bin/cluvfy stage -post crsinst -n GDGZ-DCS-SV01C-RAC01,GDGZ-DCS-SV02C-RAC01,GDGZ-DCS-SV03C-RAC01,GDGZ-DCS-SV04C-RAC01GDGZ-DCS-SV01C-RAC01:/opt/oracrs/cfgtoollogs # vi configToolFailedCommandsGDGZ-DCS-SV01C-RAC01:/opt/oracrs/cfgtoollogs # /opt/oracrs/bin/
23、cluvfy stage -post crsinst -n GDGZ-DCS-SV01C-RAC01,GDGZ-DCS-SV02C-RAC01,GDGZ-DCS-SV03C-RAC01,GDGZ-DCS-SV04C-RAC01这是root信任关系导致,可以不理!3,此时CRS的3个核心进程资源已经online,但需要先在各节点执行/opt/oracrs/bin/crsctl stop crs,退出CRS进程。4,恢复原来的/opt/oracle(各节点都要执行)cd /optmv oracle oracle_newmv oracle_old oracle5,在各节点执行/opt/oracrs/
24、bin/crsctl start crs启动CRS6,将数据库、实例等信息手工注册到OCR中有ASM需要向OCR中注册srvctl add asm -n gdgz-dcs-sv01c-rac01 -i +ASM1 -o $ORACLE_HOMEsrvctl add asm -n gdgz-dcs-sv02c-rac01 -i +ASM2 -o $ORACLE_HOMEsrvctl add asm -n gdgz-dcs-sv03c-rac01 -i +ASM3 -o $ORACLE_HOMEsrvctl add asm -n gdgz-dcs-sv04c-rac01 -i +ASM4 -o $
25、ORACLE_HOME(参考深圳的ASM信息)NAME=ora.gdsz-dcs-sv01c-rac01.ASM1.asmTYPE=applicationTARGET=ONLINESTATE=ONLINE on gdsz-dcs-sv01c-rac01注册数据库srvctl add database -d ORA -o $ORACLE_HOME (单独启动srvctl start database -d ORA,执行这个将下面加的实例也拉起来)(参考深圳的db信息)NAME=ora.ORA.dbTYPE=applicationTARGET=ONLINESTATE=ONLINE on gdsz-
26、dcs-sv01c-rac01注册实例srvctl add instance -d ORA -i ORA1 -n gdgz-dcs-sv01c-rac01srvctl add instance -d ORA -i ORA2 -n gdgz-dcs-sv02c-rac01srvctl add instance -d ORA -i ORA3 -n gdgz-dcs-sv03c-rac01srvctl add instance -d ORA -i ORA4 -n gdgz-dcs-sv04c-rac01(参考深圳的inst信息)注册服务srvctl add service -d ORA -s ser
27、vice_ora -r ORA1,ORA2,ORA3,ORA4 -P BASIC(单独启动srvctl start service -d ORA -s service_ora,可以后面统一启动)NAME=ora.ORA.ORA1.instTYPE=applicationTARGET=ONLINESTATE=ONLINE on gdsz-dcs-sv01c-rac01添加LISTENER(建议用netca配置,需要先删除原来的监听再添加新的监听)7,重新启动RAC/opt/oracrs/bin/crsctl stop crs/opt/oracrs/bin/crsctl start crs8,检查
28、RAC是否起来oracleGDGZ-DCS-SV04C-RAC01:crs_stat -tName Type Target State Host -ora.A1.inst application ONLINE ONLINE gdgz.ac01 ora.A2.inst application ONLINE ONLINE gdgz.ac01 ora.A3.inst application ONLINE ONLINE gdgz.ac01 ora.A4.inst application ONLINE ONLINE gdgz.ac01 ora.ORA.db application ONLINE ONLI
29、NE gdgz.ac01 ora.RA1.srv application ONLINE ONLINE gdgz.ac01 ora.RA2.srv application ONLINE ONLINE gdgz.ac01 ora.RA3.srv application ONLINE ONLINE gdgz.ac01 ora.RA4.srv application ONLINE ONLINE gdgz.ac01 ora._ora.cs application ONLINE ONLINE gdgz.ac01 ora.SM1.asm application ONLINE ONLINE gdgz.ac01 ora.01.lsnr application ONLINE ONLINE gdgz.ac01 ora.c01.gsd application ONLINE ONLINE gdgz.ac01 ora.c01.ons application ONLINE ONLINE gdgz.ac01 ora.c01.vip application ONLINE ONLINE gdgz.ac01 ora.SM2.asm application ONLINE ONLINE gdgz.ac01 ora.01.
copyright@ 2008-2022 冰豆网网站版权所有
经营许可证编号:鄂ICP备2022015515号-1