Operating system: centos6.5

Database version: 11.2.0.4 RAC

Patch status: no patch has been updated

Phenomenon: the database suddenly goes down. After startup, it will still happen for some time

Wed Apr 18 11:41:44 2012
Exception [type: SIGSEGV, SI_KERNEL(general_protection)] [ADDR:0x0] [PC:0x1495A37, opiaba()+639] [flags: 0x0, count: 1]
Errors in file /u01/.../SID_ora_26561.trc (incident=272318): ORA-7445: exception encountered: core dump [opiaba()+639] [SIGSEGV][ADDR:0x0] [PC:0x1495A37] [SI_KERNEL(general_protection)] []
Incident details in: /u01/.../incdir_272318/SID_ora_26561_i272318.trc
Errors in file /u01/.../SID_ora_26561.trc (incident=272319):
ORA-600: internal error code, arguments: [kgh_heap_sizes:ds],[0x81BF26440], [], [], [], [], [], [], [], [], [], []
ORA-7445: exception encountered: core dump [opiaba()+639] [SIGSEGV][ADDR:0x0] [PC:0x1495A37] [SI_KERNEL(general_protection)] []
Incident details in: /u01/...t/incdir_272319/SID_ora_26561_i272319.trc
Wed Apr 18 11:41:49 2012
Trace dumping is performing id=[cdmp_20120418114149]
...
Wed Apr 18 11:51:51 2012
Process 0x0x9e0c8d0d0 appears to be hung while dumping <-----
Current time = 746796356, process death time = 746736136 interval = 60000
Attempting to kill process 0x0x9e0c8d0d0 with OS pid = 26561
OSD kill succeeded for process 0x9e0c8d0d0
Errors in file /u01/.../SID_pmon_6020.trc (incident=265950):
ORA-600: internal error code, arguments: [17147], [0x81BF26428], [], [],[], [], [], [], [], [], [], []
Incident details in: /u01/.../incdir_265950/SID_pmon_6020_i265950.trc
Wed Apr 18 11:51:55 2012
Trace dumping is performing id=[cdmp_20120418115155]
Errors in file /u01/.../SID_pmon_6020.trc:
ORA-600: internal error code, arguments: [17147], [0x81BF26428], [], [],[], [], [], [], [], [], [], []
PMON (ospid: 6020): terminating the instance due to error 472

In combination with the error message, check MOS and find that 1466343.1 is consistent. According to the prompt, this bug will be triggered if the binding variable of a single SQL exceeds 65535. This problem will occur in versions 10.2.0.5 to 12.1.0.1.

Contact the application to determine whether such a situation exists. After the application was upgraded last night, multiple update statements were specified in the PLSQL executed today. The total number of bound variables was 7W more than the threshold value of 65535, triggering a bug.

Patch address

https://support.oracle.com/epmos/faces/PatchSearchResults?_ adf.ctrl-state=oj76x1cv0_ 148&_ afrLoop=17411389381759

bug id:12578873

solve:

1. Patch

After patching, the problem that the instance is terminated is fixed, and the binding variable still cannot exceed 65535

2. Notify the application to split the problem PLSQL