偶遇问题之ORA-00600: internal error code, arguments: [kcblasm_1], [103], [], [], [], [], [], []
今天生产库上产生600错误，记录下。
查看alert
ORA-00600: internal error code, arguments: [kcblasm_1], [103], [], [], [], [], [], []
查看trc文件
/opt/oracle/admin/pridb/udump/pridb1_ora_14941.trc
Oracle Database 10g Enterprise Edition Release 10.2.0.5.0 - 64bit Production
With the Partitioning, Real Application Clusters, OLAP, Data Mining
and Real Application Testing options
ORACLE_HOME = /opt/oracle/product/10.2.0/rac_db
System name: xxxxx
Release: 5.10
Version: Generic_142900-12
Machine: sun4u
Instance name: pridb1
Redo thread mounted by this instance: 1
Oracle process number: 60
Unix process pid: 14941, image: oracle@racdb1
*** ACTION NAME:() 2012-02-23 14:48:25.145
*** MODULE NAME:(DDTEK ODBC Oracle) 2012-02-23 14:48:25.145
*** SERVICE NAME:(db_1) 2012-02-23 14:48:25.145
*** SESSION ID:(6307.53684) 2012-02-23 14:48:25.145
WARNING:Could not lower the asynch I/O limit to 320 for SQL direct I/O. It is set to -1
WARNING:Could not lower the asynch I/O limit to 576 for SQL direct I/O. It is set to -1
WARNING:Could not lower the asynch I/O limit to 640 for SQL direct I/O. It is set to -1
WARNING:Could not lower the asynch I/O limit to 576 for SQL direct I/O. It is set to -1
*** 2012-02-23 14:48:35.868
ksedmp: internal or fatal error
ORA-00600: internal error code, arguments: [kcblasm_1], [103], [], [], [], [], [], []
Current SQL statement for this session:
SELECT sheng.region_code,sheng.region_desc,--省
shi.region_code,shi.region_desc, --市
xian.region_code,xian.region_desc,--区县
school.xx_jbxx_id,replace(replace(school.xxmc,' ',''),' ',''),--学校
jzw.jzwid,replace(replace(jzw.jzwmc,'（','('),'）',')'),--建筑物
下面的语句不放出来了，总之语句很长很长。
METALINK上的相应资料
Applies to:
Oracle Server - Enterprise Edition - Version: 10.2.0.5.0 and later [Release: 10.2 and later ]
Information in this document applies to any platform.
Symptoms
When executing a statement using a SELECT getting:
ORA-00600: internal error code, arguments: [kcblasm_1], [103], [], [], [], [], [], []
The execution plan from the ORA-600 trace file is showing hash join is used.
Call Stack is including:
kcblasm1 kcblasm kxhfFndFreeSlot kxhfNewBuffer qerhjGetNewBuffer ksxb1bqb kxhrPack
Cause
The issue was investigated in:
Bug 9800302 - 10.2.0.5 GETTING ORA-00600 [KCBLASM_1]
closed as duplicate of:Bug 7612454 - Abstract: DSS:PERF REGRESSIONS IN SERIAL DIRECT READS fixed in 11.2.
As per development team the number of slots available for direct I/Os (limited to 4096) forced the hash-join algorithm to operate on fewer number of slots and resulted in more spills to disk. This caused:
direct path IO to perform worse in 10.2.0.5 than earlier releases with more "direct path read" operations or
ORA-600 [kcblasm_1] errors.
Solution
1. Upgrade the database to 11.2.
OR
2. Apply &incFamilyProds=false&flag=search))" target=_blank>Patch 7612454 available on MOS. If a patch is not currently available on top of your database version and/or platform please raise a Service Request to request for it.
Please be sure that your database version qualifies for getting a new patch as per Note 209768.1 and Note 742060.1.
OR
3. Use the workaround of setting:
"_hash_join_enabled"= false
升级现在做不了，语句是一个中间件自动生成，因此只能是改参数了。