ORA-04031错误导致宕机案例分析

今天遇到一起ORACLE数据库宕机案例，下面是对这起数据库宕机案例的原因进行分析、解读。分析过程中顺便记录一下这个案例的前因后果，攒点经验值，培养一下分析、解决问题的能力。

 

案例环境：

 

   操作系统 ：Oracle Linux Server release 5.7 64 bit

   数据库版本：Oracle Database 10g Release 10.2.0.4.0 - 64bit Production

 

案例分析：

收到告警去检查数据库时，发现实例已经宕机。检查告警日志，发现下面错误信息：

ORA-00604: error occurred at recursive SQL level 1
ORA-04031: unable to allocate 32 bytes of shared memory ("shared pool","select count(*) from sys.job...","sql area","tmp")
Mon Nov  2 11:43:00 2015
Errors in file /u01/app/oracle/admin/SCM2/bdump/scm2_cjq0_6571.trc:
ORA-00604: error occurred at recursive SQL level 1
ORA-04031: unable to allocate 32 bytes of shared memory ("shared pool","select job, nvl2(last_date, ...","sql area","tmp")
Mon Nov  2 11:43:00 2015
Errors in file /u01/app/oracle/admin/SCM2/bdump/scm2_cjq0_6571.trc:
ORA-00604: error occurred at recursive SQL level 1
ORA-04031: unable to allocate 32 bytes of shared memory ("shared pool","select count(*) from sys.job...","sql area","tmp")
Mon Nov  2 11:43:05 2015
Errors in file /u01/app/oracle/admin/SCM2/bdump/scm2_cjq0_6571.trc:
ORA-00604: error occurred at recursive SQL level 1
ORA-04031: unable to allocate 32 bytes of shared memory ("shared pool","select job, nvl2(last_date, ...","sql area","tmp")
Mon Nov  2 11:43:05 2015
Errors in file /u01/app/oracle/admin/SCM2/bdump/scm2_cjq0_6571.trc:
ORA-00604: error occurred at recursive SQL level 1
ORA-04031: unable to allocate 32 bytes of shared memory ("shared pool","select count(*) from sys.job...","sql area","tmp")
Mon Nov  2 11:43:08 2015
Errors in file /u01/app/oracle/admin/SCM2/bdump/scm2_reco_6569.trc:
ORA-04031: unable to allocate 32 bytes of shared memory ("shared pool","select host,userid,password,...","sql area","tmp")
Mon Nov  2 11:43:08 2015
RECO: terminating instance due to error 4031
Mon Nov  2 11:43:08 2015
Errors in file /u01/app/oracle/admin/SCM2/bdump/scm2_pmon_6555.trc:
ORA-04031: unable to allocate  bytes of shared memory ("","","","")
Instance terminated by RECO, pid = 6569

从告警日志我们可以看到ORA-00604与ORA-04031错误导致了这次宕机事故（RECO: terminating instance due to error 4031）：

$ oerr ora 4031

04031, 00000, "unable to allocate %s bytes of shared memory ("%s","%s","%s","%s")"

// *Cause: More shared memory is needed than was allocated in the shared

// pool.

// *Action: If the shared pool is out of memory, either use the

// dbms_shared_pool package to pin large packages,

// reduce your use of shared memory, or increase the amount of

// available shared memory by increasing the value of the

// INIT.ORA parameters "shared_pool_reserved_size" and

// "shared_pool_size".

// If the large pool is out of memory, increase the INIT.ORA

// parameter "large_pool_size".

 

一般出现ORA-04031错误可能由两个原因引起：

1：内存中存在大量碎片，导致在分配内存的时候，没有连续的内存可存放，此问题一般是需要在开发的角度上入手，比如增加绑定变量，减少硬解析来改善和避免；

2.内存容量不足，需要扩大内存。

这台机器分配的物理内存为8G，结果检查发现SGA只分配了1168M，不到2G，瞬时碉堡了。此时真是很无语。ASH Report分析宕机前后的Buffer Cache和Shared Pool大小如下所示。

 

 

查看跟踪文件，可以看到SGA: allocation forcing component growth等待事件，可以确认的是由于SGA无法增长导致，也就是SGA被撑爆了，结合ASH Report我们可以看到当时Shared Pool的大小已经接近SGA的69.6%大小。

SO: 0xa617d9c0, type: 4, owner: 0xa8a26c68, flag: INIT/-/-/0x00
  (session) sid: 932 trans: (nil), creator: 0xa8a26c68, flag: (51) USR/- BSY/-/-/-/-/-
            DID: 0001-000A-00000003, short-term DID: 0000-0000-00000000
            txn branch: (nil)
            oct: 0, prv: 0, sql: (nil), psql: (nil), user: 0/SYS
  last wait for 'SGA: allocation forcing component growth' blocking sess=0x(nil) seq=51324 wait_time=10714 seconds since wait started=0
          =0, =0, =0
  Dumping Session Wait History
   for 'SGA: allocation forcing component growth' count=1 wait_time=10714
          =0, =0, =0
   for 'SGA: allocation forcing component growth' count=1 wait_time=10512
          =0, =0, =0
   for 'latch: shared pool' count=1 wait_time=892
          address=600e7320, number=d6, tries=0
   for 'latch: shared pool' count=1 wait_time=28
          address=600e7320, number=d6, tries=0
   for 'latch: shared pool' count=1 wait_time=51
          address=600e7320, number=d6, tries=0
   for 'latch: shared pool' count=1 wait_time=114
          address=600e7320, number=d6, tries=0
   for 'latch: shared pool' count=1 wait_time=120
          address=600e7320, number=d6, tries=0
   for 'latch: library cache' count=1 wait_time=33
          address=a3fa46e8, number=d7, tries=1

结合上面的一些分析，可以断定SGA的不合理设置导致shared pool的内存被全部耗尽，SGA被撑爆了。于是调整SGA的参数才是解决问题的正确对策。另外考虑到这个数据库也正常运行了较长一段时间，也分析了一下awr、addm报告，发现系统的硬解析相当严重。另外通过下面脚本观察了一段时间shared pool的变化，发现其收缩、增长较频繁。

SELECT start_time,
       component,
       oper_type,
       oper_mode,
       initial_size / 1024 / 1024 "INITIAL",
       final_size / 1024 / 1024   "FINAL",
       end_time
FROM   v$sga_resize_ops
WHERE  component IN ( 'DEFAULT buffer cache', 'shared pool' )
       AND status = 'COMPLETE'
ORDER  BY start_time,
          component;

这个可以通过设置数据库参数SHARED_POOL_SIZE，保证SHARED_POOL_SIZE大小不会由于内存紧张而低于这个大小，另外可以设置SGA resize的时间间隔

ALTER SYSTEM SET “_memory_broker_stat_interval”=n SCOPE=SPFILE;

问题虽然解决了，但是真正需要反思的是为什么这个SGA_MAX_SIZE设置为1168M大小的事情！而且没有在巡检当中被发现。

 

参考资料：

http://blog.csdn.net/wenzhongyan/article/details/29866845

http://blog.chinaunix.net/uid-20802110-id-4188357.html

http://www.oraclefreebase.com/blog/2015/10/%E6%95%B0%E6%8D%AE%E5%BA%93ora-4031%E6%95%B0%E6%8D%AE%E5%BA%93crash/