记录一则ORA-600 [13011]错误

环境：Solaris 10 + Oracle 11.2.0.1

现象：alert告警日志定期出现ORA-600 [13011]错误

• 1.故障现象
• 2.初步分析
• 3.匹配MOS
• 4.定位解决

1.故障现象

数据库alert 日志：

Fri Jul 13 02:00:00 2018
Clearing Resource Manager plan via parameter
Fri Jul 13 02:00:46 2018
Errors in file /u01/app/oracle/diag/rdbms/prod/prod/trace/prod_j000_1757.trc  (incident=42249):
ORA-00600: internal error code, arguments: [13011], [67896], [8421830], [44], [8421830], [17], [], [], [], [], [], []
Incident details in: /u01/app/oracle/diag/rdbms/prod/prod/incident/incdir_42249/prod_j000_1757_i42249.trc
Fri Jul 13 02:00:49 2018
Trace dumping is performing id=[cdmp_20180713020049]
Fri Jul 13 02:00:49 2018
Sweep [inc][42249]: completed
Sweep [inc2][42249]: completed
Fri Jul 13 03:00:55 2018
Errors in file /u01/app/oracle/diag/rdbms/prod/prod/trace/prod_j000_2053.trc  (incident=42250):
ORA-00600: internal error code, arguments: [13011], [67896], [8421830], [44], [8447731], [0], [], [], [], [], [], []
Incident details in: /u01/app/oracle/diag/rdbms/prod/prod/incident/incdir_42250/prod_j000_2053_i42250.trc
Fri Jul 13 03:00:56 2018
Trace dumping is performing id=[cdmp_20180713030056]
Fri Jul 13 03:01:48 2018
Sweep [inc][42250]: completed
Sweep [inc2][42250]: completed
Fri Jul 13 04:00:03 2018
Errors in file /u01/app/oracle/diag/rdbms/prod/prod/trace/prod_j000_2338.trc  (incident=42251):
ORA-00600: internal error code, arguments: [13011], [67896], [8421830], [44], [8421830], [0], [], [], [], [], [], []
Incident details in: /u01/app/oracle/diag/rdbms/prod/prod/incident/incdir_42251/prod_j000_2338_i42251.trc
Fri Jul 13 04:00:05 2018
Trace dumping is performing id=[cdmp_20180713040005]
Fri Jul 13 04:00:05 2018
Sweep [inc][42251]: completed
Sweep [inc2][42251]: completed
省略部分相似输出..
Fri Jul 13 09:00:50 2018
Errors in file /u01/app/oracle/diag/rdbms/prod/prod/trace/prod_j000_3816.trc  (incident=42273):
ORA-00600: internal error code, arguments: [13011], [67896], [8421830], [44], [8421827], [0], [], [], [], [], [], []
Incident details in: /u01/app/oracle/diag/rdbms/prod/prod/incident/incdir_42273/prod_j000_3816_i42273.trc
Fri Jul 13 09:00:53 2018
Trace dumping is performing id=[cdmp_20180713090053]
Fri Jul 13 09:01:39 2018
Sweep [inc][42273]: completed

2.初步分析

从/u01/app/oracle/diag/rdbms/prod/prod/trace/prod_j000_1757.trc文件中没有过多信息：

Incident 42249 created, dump file: /u01/app/oracle/diag/rdbms/prod/prod/incident/incdir_42249/prod_j000_1757_i42249.trc
ORA-00600: internal error code, arguments: [13011], [67896], [8421830], [44], [8421830], [17], [], [], [], [], [], []

从/u01/app/oracle/diag/rdbms/prod/prod/incident/incdir_42249/prod_j000_1757_i42249.trc文件中可以看到Current SQL:

Dump continued from file: /u01/app/oracle/diag/rdbms/prod/prod/trace/prod_j000_1757.trc
ORA-00600: internal error code, arguments: [13011], [67896], [8421830], [44], [8421830], [17], [], [], [], [], [], []
========= Dump for incident 42249 (ORA 600 [13011]) ========
*** 2018-07-13 02:00:46.871
dbkedDefDump(): Starting incident default dumps (flags=0x2, level=3, mask=0x0)
----- Current SQL Statement for this session (sql_id=11p815z8hkfms) -----
DELETE MGMT_SYSTEM_PERFORMANCE_LOG WHERE TIME < :B2 AND ROWNUM <= :B1

3.匹配MOS

搜索MOS可以找到以下两篇文档：
- ORA-00600 [13011] (文档 ID 1547827.1)
- ORA-600 [13011] "Problem occurred when trying to delete a row" (文档 ID 28184.1)

从文档 ID 1547827.1中可以匹配到现象Current SQL一致：

DELETE MGMT_SYSTEM_PERFORMANCE_LOG WHERE TIME < :B2 AND ROWNUM <= :B1

MOS中给出的解决方案，

SOLUTION

1)As a possible solution, please apply patch set 11.2.0.3

• OR -

2)Drop and recreate the index(es) on SYSMAN.MGMT_SYSTEM_PERFORMANCE_LOG.

To drop and recreate the indexes:

A. Get the DDL for recreating the indexes:

set heading off

set echo off

set flush off

set pagesize 9999

set linesize 9999

set long 9999

SQL> select dbms_metadata.get_ddl('INDEX', 'MGMT_SYSTEM_PERF_LOG_IDX_01', 'SYSMAN') from dual;

SQL> select dbms_metadata.get_ddl('INDEX', 'MGMT_SYSTEM_PERF_LOG_IDX_02', 'SYSMAN') from dual;

B. Drop the index:

drop index sysman.MGMT_SYSTEM_PERF_LOG_IDX_01;

drop index sysman.MGMT_SYSTEM_PERF_LOG_IDX_02;

C. Recreate the indexes using the output from step 2A.

3)Run validate against the table and its indexes once more:

SQL> analyze table sysman.MGMT_SYSTEM_PERFORMANCE_LOG validate structure cascade;

4.定位解决

这里选择MOS中第二种解决方案重建MGMT_SYSTEM_PERFORMANCE_LOG索引：

--4.1 查看MGMT_SYSTEM_PERFORMANCE_LOG的索引
SQL> select index_name from dba_indexes where table_name = 'MGMT_SYSTEM_PERFORMANCE_LOG';
MGMT_SYSTEM_PERF_LOG_IDX_01
MGMT_SYSTEM_PERF_LOG_IDX_02
--4.2 获取MGMT_SYSTEM_PERFORMANCE_LOG的2个索引的DDL语句
SQL> select dbms_metadata.get_ddl('INDEX', 'MGMT_SYSTEM_PERF_LOG_IDX_01', 'SYSMAN') from dual;
SQL> select dbms_metadata.get_ddl('INDEX', 'MGMT_SYSTEM_PERF_LOG_IDX_02', 'SYSMAN') from dual;
--4.3 删除之前的2个索引
SQL> drop index sysman.MGMT_SYSTEM_PERF_LOG_IDX_01;
SQL> drop index sysman.MGMT_SYSTEM_PERF_LOG_IDX_02;
--4.4 重新创建2个索引
SQL>  CREATE INDEX "SYSMAN"."MGMT_SYSTEM_PERF_LOG_IDX_01" ON "SYSMAN"."MGMT_SYSTEM_PERFORMANCE_LOG" ("JOB_NAME", "TIME", "NAME")
PCTFREE 10 INITRANS 2 MAXTRANS 255 COMPUTE STATISTICS COMPRESS 3
STORAGE(INITIAL 65536 NEXT 1048576 MINEXTENTS 1 MAXEXTENTS 2147483645
PCTINCREASE 0 FREELISTS 1 FREELIST GROUPS 1 BUFFER_POOL DEFAULT FLASH_CACHE DEFAULT CELL_FLASH_CACHE DEFAULT)
TABLESPACE "SYSAUX";
SQL>  CREATE INDEX "SYSMAN"."MGMT_SYSTEM_PERF_LOG_IDX_02" ON "SYSMAN"."MGMT_SYSTEM_PERFORMANCE_LOG" ("TIME")
PCTFREE 10 INITRANS 2 MAXTRANS 255 COMPUTE STATISTICS
STORAGE(INITIAL 65536 NEXT 1048576 MINEXTENTS 1 MAXEXTENTS 2147483645
PCTINCREASE 0 FREELISTS 1 FREELIST GROUPS 1 BUFFER_POOL DEFAULT FLASH_CACHE DEFAULT CELL_FLASH_CACHE DEFAULT)
TABLESPACE "SYSAUX";
--4.5 重新分析表和索引
SQL> analyze table sysman.MGMT_SYSTEM_PERFORMANCE_LOG validate structure cascade;
Table analyzed.

至此，完成索引重建，后续定期观察alert日志。