10gClusterwareVotedisk损坏的恢复方法

10gClusterwareVotedisk损坏的恢复方法

votedisk无论是对于RAC(10g Clusterware、11g GI)而言，是非常重要的，我们称它为仲裁盘，当RAC集群中的某个节点发生故障而脱网掉线时，就由它来判断是否将其踢出集群，以保证集群正常运行，当votedisk损坏了，也就会导致集群服务无法启动，集群资源都无法加载，最后导致罢工。那么我们平时就要注意对votedisk的备份，在11g中，由于votedisk和ocr默认就会放进ASM磁盘组，因此可以不用特别关注，但对于10g的Cluster来说，由于不能放到ASM磁盘组，只能以raw的形式使用，因此要特别关注votedisk，定期对其进行备份，如：
用dd命令备份和恢复votedisk的方法： 备份：dd if=/dev/raw/raw3 of=/tmp/votedisk.bak 恢复：dd if=/tmp/votedisk.bak of=/dev/raw/raw3
如果很不幸，之前没有做过备份，且没有做过镜像，当votedisk损坏的时候，就只能对crs进行重建了，下面来演示一下这个过程：
--关闭crs，对votedisk的盘进行破坏，这里是/dev/raw/raw3 [root@rac1 ~]# dd if=/dev/zero of=/dev/raw/raw3 bs=4096 count=12800

再次重启crs，就提示无法启动了，查找ocssd.log日志文件发现，其中有记录，说明了是磁盘损坏 PS:10g Clusterware的日志入口地址是$ORA_CRS_HOME/log/主机名/...
[ CSSD]2015-01-16 09:37:38.327 >USER: Oracle Database 10g CSS Release 10.2.0.1.0 Production Copyright 1996, 2094 Oracle. All rights reserved. [ CSSD]2015-01-16 09:37:38.327 >USER: CSS daemon log for node rac1, number 1, in cluster cluster [ clsdmt]Listening to (ADDRESS=(PROTOCOL=ipc)(KEY=rac1DBG_CSSD)) [ CSSD]2015-01-16 09:37:38.332 [3059615952] >TRACE: clssscmain: local-only set to false [ CSSD]2015-01-16 09:37:38.344 [3059615952] >TRACE: clssnmReadNodeInfo: added node 1 (rac1) to cluster [ CSSD]2015-01-16 09:37:38.352 [3059615952] >TRACE: clssnmReadNodeInfo: added node 2 (rac2) to cluster [ CSSD]2015-01-16 09:37:38.356 [3032808336] >TRACE: clssnm_skgxnmon: skgxn init failed, rc 1 [ CSSD]2015-01-16 09:37:38.356 [3059615952] >TRACE: clssnm_skgxnonline: Using vacuous skgxn monitor [ CSSD]2015-01-16 09:37:38.362 [3059615952] >TRACE: clssnmDiskStateChange: state from 1 to 2 disk (0//dev/raw/raw3) [ CSSD]2015-01-16 09:37:40.381 [3032808336] >TRACE: clssnmvDiskOpen: corrupt kill block on disk (0x09!=0x636c73536b696c4c) [ CSSD]2015-01-16 09:37:40.381 [3032808336] >TRACE: clssnmDiskStateChange: state from 2 to 3 disk (0//dev/raw/raw3)
重建crs很简单，就执行2个脚本：
1.$ORA_CRS_HOME/install/rootdelete.sh
2.$ORA_CRS_HOME/install/rootdeinstall.sh

节点1： [root@rac1 install]# ./rootdelete.sh Shutting down Oracle Cluster Ready Services (CRS): Stopping resources. Error while stopping resources. Possible cause: CRSD is down. Stopping CSSD. Unable to communicate with the CSS daemon. Shutdown has begun. The daemons should exit soon. Checking to see if Oracle CRS stack is down... Oracle CRS stack is not running. Oracle CRS stack is down now. Removing script for Oracle Cluster Ready services Updating ocr file for downgrade Cleaning up SCR settings in '/etc/oracle/scls_scr' [root@rac1 install]# ./rootdeinstall.sh
Removing contents from OCR device 2560+0 records in 2560+0 records out 10485760 bytes (10 MB) copied, 0.590608 seconds, 17.8 MB/s
节点2： [root@rac2 install]# ./rootdelete.sh Shutting down Oracle Cluster Ready Services (CRS): OCR initialization failed with invalid format: PROC-22: The OCR backend has an invalid format Shutdown has begun. The daemons should exit soon. Checking to see if Oracle CRS stack is down... Oracle CRS stack is not running. Oracle CRS stack is down now. Removing script for Oracle Cluster Ready services Updating ocr file for downgrade Cleaning up SCR settings in '/etc/oracle/scls_scr' [root@rac2 install]# ./rootdeinstall.sh
Removing contents from OCR device 2560+0 records in 2560+0 records out 10485760 bytes (10 MB) copied, 0.627909 seconds, 16.7 MB/s [root@rac2 install]# dd if=/dev/zero of=/dev/raw/raw3 bs=4096 count=128000 dd: writing `/dev/raw/raw3': No space left on device 25601+0 records in 25600+0 records out 104857600 bytes (105 MB) copied, 5.40456 seconds, 19.4 MB/s
然后重新在2个节点依次执行$ORA_CRS_HOME/root.sh就可以了，软件的OUI不用重新安装

如果通过脚本无法删除成功，安装顺利重新安装crs，可以手工删除以下目录：
rm /etc/oracle/* rm -f /etc/init.d/init.cssd rm -f /etc/init.d/init.crs rm -f /etc/init.d/init.crsd rm -f /etc/init.d/init.evmd rm -f /etc/rc2.d/K96init.crs rm -f /etc/rc2.d/S96init.crs rm -f /etc/rc3.d/K96init.crs rm -f /etc/rc3.d/S96init.crs rm -f /etc/rc5.d/K96init.crs rm -f /etc/rc5.d/S96init.crs rm -Rf /etc/oracle/scls_scr rm -f /etc/inittab.crs cp /etc/inittab.orig /etc/inittab
总结：
平时我们都会对ocr和votedisk磁盘做多个镜像冗余，另外，如果是裸设备的话，还会通过dd命令单独去备份，通常是不太容易损坏和丢失的，万一发生了无备份情况下的损坏，那么就只能工作重建crs来解决问题了，这就是DBAs们的最后一根救命稻草了。