适用范围
Oracle Database 19.3 + SUSE Linux Enterprise Server 12 SP4 EMC存储
问题概述
Oracle 19c RAC 节点2 操作系统由SUSE Linux 12 SP4升级到SP5后，重启操作系统，该节点存储无法识别，cssd进程无法启动到real time 模式，crs无法正常启动。
问题原因
1、操作系统安全加固产品导致19c集群的CSSD进程无法启动real time模式加固产品调整了操作系统CPU Accounting。请参考 Oracle 12c RAC CSSD进程无法启动real time模式。
解决方案
1、关闭操作系统安全加固软件的服务； 2、安装SUSE 12 SP5匹配的EMC存储管理软件DellEMCPower.LINUX-7.1.0.00.00-075.SLES12SP5.x86_64.rpm
分析过程
1、启动节点2 crs启动时的alert日志
2025-03-27 20:17:50.802 [OCSSD(14861)]CRS-1714: Unable to discover any voting files, retrying discovery in 15 seconds; Details at (:CSSNM00070:)in /u01/app/grid/diag/crs/host01/crs/trace/ocssd.trc
节点2的crs启动时alert日志显示无法找到vote盘，15秒重试，仍然无法找到vote盘。 2、节点2 ocssd日志
2025-03-27 20:17:50.801 : CSSD:3324163840: clssnmReadDiscoveryProfile: voting file discovery string(/dev/emcpower) 2025-03-27 20:17:50.801 : CSSD:3324163840: clssnmvDDiscThread: using discovery string /dev/oracleasm/disks for initial discovery 2025-03-27 20:17:50.801 : SKGFD:3324163840: Discovery with str:/dev/emcpower: 2025-03-27 20:17:50.801 : SKGFD:3324163840: UFS discovery with :/dev/emcpower: 2025-03-27 20:17:50.801 : SKGFD:3324163840: Execute glob on the string /dev/emcpower 2025-03-27 20:17:50.801 : SKGFD:3324163840: OSS discovery with :/dev/emcpower 2025-03-27 20:17:50.802 : SKGFD:3324163840: Discovery skipping bad :ASM:: 2025-03-27 20:17:50.802 : CSSD:3324163840: clssnmvDiskVerify: Successful discovery of 0 disks 2025-03-27 20:17:50.802 : CSSD:3324163840: clssnmCompleteInitVFDiscovery: Completing initial voting file discovery 2025-03-27 20:17:50.802 : CSSD:3324163840: clssnmvFindInitialConfigs: No voting files found 2025-03-27 20:17:50.802 : CSSD:3324163840: (:CSSNM00070:)clssnmCompleteInitVFDiscovery: Voting file not found. Retrying discovery in 15 seconds 2025-03-27 20:17:51.687 : CSSD:3558553344: clsssc_CLSFAInit_CB: System not ready for CLSFA initialization 2025-03-27 20:17:52.687 : CSSD:3558553344: clsssc_CLSFAInit_CB: System not ready for CLSFA initialization 2025-03-27 20:17:53.687 : CSSD:3558553344: clsssc_CLSFAInit_CB: System not ready for CLSFA initialization 2025-03-27 20:17:54.687 : CSSD:3558553344: clsssc_CLSFAInit_CB: System not ready for CLSFA initialization 2025-03-27 20:17:55.688 : CSSD:3558553344: clsssc_CLSFAInit_CB: System not ready for CLSFA initialization 2025-03-27 20:17:56.688 : CSSD:3558553344: clsssc_CLSFAInit_CB: System not ready for CLSFA initialization 2025-03-27 20:17:57.688 : CSSD:3558553344: clsssc_CLSFAInit_CB: System not ready for CLSFA initialization
cssd日志显示无法找到/dev/emcpower存储，clsssc_CLSFAInit_CB初始化失败，cssd进程无法正常启动。 3、集群状态检查
$crsctl stat res -t -init -------------------------------------------------------------------------------- NAME TARGET STATE SERVER STATE_DETAILS -------------------------------------------------------------------------------- Cluster Resources -------------------------------------------------------------------------------- ora.asm 2 ONLINE OFFLINE ora.cluster_interconnect.haip 2 ONLINE OFFLINE ora.crf 2 ONLINE ONLINE fadb02 ora.crsd 2 ONLINE OFFLINE ora.cssd 2 ONLINE OFFLINE STARTING ora.cssdmonitor 2 ONLINE ONLINE fadb02 ora.ctssd 2 ONLINE OFFLINE ora.diskmon 2 OFFLINE OFFLINE ora.evmd 2 ONLINE OFFLINE ora.gipcd 2 ONLINE ONLINE fadb02 ora.gpnpd 2 ONLINE ONLINE fadb02 ora.mdnsd 2 ONLINE ONLINE fadb02复制
cssd进程是starting状态，因为vote无法找到，所以cssd进程无法正常启动。 4、存储检查 节点2：
$ls -l /dev/emcpower*
没有返回值，节点2无法识别存储。
节点1：
$ls -l /dev/emcpower* /dev/emcpowera /dev/emcpowerb /dev/emcpowerc /dev/emcpowere
节点1的存储盘正常。 4、检查节点2的存储
$powermt display dev=all initialization error $rpm -qa | grep EMCpower EMCpower.LINUX-7.0.0.00.00-064.suse12sp4.x86_64
节点2操作系统升级到SUSE12 SP5后EMCpower软件仍然是sp4的，初步判断EMCpower与操作系统不兼容导致EMC存储无法识别 5、安装EMC匹配的软件 下载并安装DellEMCPower.LINUX-7.1.0.00.00-075.SLES12SP5.x86_64.rpm，节点2 存储可以正常识别，重新启动crs可以正常启动。
总结
数据库基础环境生产环境变更，一定要经过严格测试后在实施，OS和数据库升级等操作，应考虑该服务器上齐滔软件的兼容性。 -the end-
