由于共享磁盘问题导致的两个ORA-600错误。

客户的10.2.0.4 RAC for Linux X86-64，在告警日志中出现了大量的错误信息：

Tue Apr 24 16:15:04 2012
Errors in file /u01/admin/orcl/udump/orcl1_ora_10437.trc:
ORA-00600: internal error code, arguments: [KSFD_DECAIOPC], [0xFC213CBF0], [], [], [], [], [], []
ORA-07445: exception encountered: core dump [<0x9293a0>] [SIGSEGV] [Address not mapped to object] [0x0000007CA] [] []
ORA-07445: exception encountered: core dump [<0xb5814a>] [SIGSEGV] [Address not mapped to object] [0xFFFFFFFFFFFFFFF9] [] []
ORA-00333: redo log read error block 2 count 8192
ORA-00202: control file: '+ASM_DISK1/orcl/controlfile/current.256.757170241'
ORA-15081: failed to submit an I/O operation to a disk
Tue Apr 24 16:15:14 2012
WARNING: kfk failed to open a disk[/dev/oracleasm/disks/DISK4]
Tue Apr 24 16:15:14 2012
Errors in file /u01/admin/orcl/udump/orcl1_ora_10437.trc:
ORA-15025: could not open disk '/dev/oracleasm/disks/DISK4'
ORA-27041: unable to open file
Linux-x86_64 Error: 24: Too many open files
Additional information: 3
ORA-00600: internal error code, arguments: [KSFD_DECAIOPC], [0xFC213CBF0], [], [], [], [], [], []
ORA-07445: exception encountered: core dump [<0x9293a0>] [SIGSEGV] [Address not mapped to object] [0x0000007CA] [] []
ORA-07445: exception encountered: core dump [<0xb5814a>] [SIGSEGV] [Address not mapped to object] [0xFFFFFFFFFFFFFFF9] [] []
ORA-00333: redo log read error block 2 count 8192
ORA-00202: control file: '+ASM_DISK1/orcl/controlfile/current.256.757170241'
ORA-15081: failed to submit an I/O operation to a disk
WARNING: kfk failed to open a disk[/dev/oracleasm/disks/DISK2]

当前的版本10.2.0.4和ORA-600错误信息KSFD_DECAIOPC，都符合bug 8433026的描述，但是当前数据库并未配置STREAM环境，虽然当前库配置了DSG的复制应用，但是毕竟和流应用还是有所区别。

从错误信息可以判断，当前的问题要问题来自ASM磁盘组中部分磁盘存在异常，导致读取时出现错误。

除了KSFD_DECAIOPC错误外，由于底层共享存储的问题，还导致了另外的ORA-600错误：

Tue Jul 10 10:39:53 2012
Errors in file /u01/admin/orcl/udump/orcl1_ora_19546.trc:
ORA-00600: internal error code, arguments: [kfioReapIO00], [0], [52], [], [], [], [], []
ORA-00333: redo log read error block 2 count 8192
ORA-00600: internal error code, arguments: [KSFD_DECAIOPC], [0xFC7D6ADB8], [], [], [], [], [], []
ORA-07445: exception encountered: core dump [<0x9293a0>] [SIGSEGV] [Address not mapped to object] [0x0000007CA] [] []
ORA-07445: exception encountered: core dump [<0xb5814a>] [SIGSEGV] [Address not mapped to object] [0xFFFFFFFFFFFFFFF9] [] []
ORA-00333: redo log read error block 2 count 8192
ORA-00202: control file: '+ASM_DISK1/orcl/controlfile/current.256.757170241'
ORA-15081: failed to submit an I/O operation to a disk

显然这两个ORA-600都是由于底层磁盘错误所引起的，而当硬件人员解决了共享磁盘错误后，ASM实例没有经过重启就恢复了正常，此后也没有类似错误的出现。