`

ORA-04031: unable to allocate 4096 bytes of shared memory ("shared pool","BEGIN

阅读更多

 EXP-00008: ORACLE error 4031 encountered

 ORA-04031: unable to allocate 4096 bytes of shared memory

 ("shared pool","BEGIN :EXEC_STR := SYS.DBMS...","PL/SQL MPCODE","BAMIMA: Bam Buffe

  

  二、错误原因:

  共享内存太小,存在一定碎片,没有有效的利用保留区,造成无法分配合适的共享区。

  

  三、解决步骤:

  1.查看当前环境

  SQL>  show sga

  

  Total System Global Area 566812832 bytes

  Fixed Size          73888 bytes

  Variable Size       28811264 bytes

  Database Buffers     536870912 bytes

  Redo Buffers        1056768 bytes

  

  show parameter shared_pool

  NAME                 TYPE  VALUE

  ------------------------------------ ------- -----

  shared_pool_reserved_size      string 1048576

  shared_pool_size           string 20971520

  

  SQL> select sum(free_space) from v$shared_pool_reserved;

  

  SUM(FREE_SPACE)

  ---------------

  1048576

  

  我们可以看到没有合理利用保留区

  

  SQL> SELECT SUM(RELOADS)/SUM(PINS) FROM V$LIBRARYCACHE;

  

  SUM(RELOADS)/SUM(PINS)

  ----------------------

  .008098188

  

  不算太严重

  SQL> SELECT round((B.Value/A.Value)*100,1) hardpaseperc

  FROM V$SYSSTAT A,

  V$SYSSTAT B

  WHERE A.Statistic# = 171

  AND B.Statistic# = 172

  AND ROWNUM = 1;

  

  hardpaseperc

  ------------------

  26.5

  

  2.查看保留区使用情况

  SQL>  SELECT FREE_SPACE,

  FREE_COUNT,

  REQUEST_FAILURES,

  REQUEST_MISSES,

  LAST_FAILURE_SIZE

  FROM V$SHARED_POOL_RESERVED;

  

  FREE_SPACE FREE_COUNT REQUEST_FAILURES REQUEST_MISSES LAST_FAILURE_SIZE

  ---------- ---------- ---------------- -------------- -----------------

  1048576     1       146       0       4132

  

  最近一次申请共享区失败时该对象需要的共享区大小4132

  

  select name from v$db_object_cache where sharable_mem = 4132;

  name

  ----------------

  dbms_lob

  

  -- dbms_lob正是exp时申请保留区的对象

  3.查看导致换页的应用

  SQL> select * from x$ksmlru where ksmlrsiz>0;

  

  ADDR   INDX  INST_ID KSMLRCOM   KSMLRSIZ KSMLRNUM KSMLRHON KSMLROHV KSMLRSES

  

  50001A88 0     1 BAMIMA: Bam Buffer 4100     64 DBMS_DDL 402745060 730DEB9C

  

  50001ACC 1     1 BAMIMA: Bam Buffer 4108    736 DBMS_SYS_SQL 1909768749 730D0838

  

  50001B10 2     1 BAMIMA: Bam Buffer 4112    1576 STANDARD 2679492315 730D7E20

  

  50001B54 3     1 BAMIMA: Bam Buffer 4124    1536 DBMS_LOB 853346312 730DA83C

  

  50001B98 4     1 BAMIMA: Bam Buffer 4128    3456 DBMS_UTILITY 4041615653 730C5FC8

  

  50001BDC 5     1 BAMIMA: Bam Buffer 4132    3760 begin :1 := dbms_lob.getLeng... 2942875191 730CFFCC

  

  50001C20 6     1 state objects    4184    1088 0 00

  

  50001C64 7     1 library cache    4192    488 EXU8VEW  2469165743 730C1C68

  

  50001CA8 8     1 state objects    4196     16 0 730C0B90

  

  50001CEC 9     1 state objects    4216    3608 0 730D0838

  

  3.分析各共享池的使用情况

  SQL> select KSPPINM,KSPPSTVL

  from x$ksppi,

  x$ksppcv

  where x$ksppi.indx = x$ksppcv.indx

  and KSPPINM = '_shared_pool_reserved_min_alloc';

  

  KSPPINM     KSPPSTVL

  -------------------------------  --------

  _shared_pool_reserved_min_alloc  4400  --(门值)

  

  我们看到INDX=5,DBMS_LOB造成换页(就是做exp涉及到lob对象处理造成的换页情况),换出

  最近未使用的内存,但是换出内存并合并碎片后在共享区仍然没有合适区来存放数据,说明共享

  区小和碎片过多,然后根据_shared_pool_reserved_min_alloc的门值来申请保留区,而门值为4400,

  所以不符合申请保留区的条件,造成4031错误。我们前面看到保留区全部为空闲状态,所以我们可以

  减低门值,使更多申请共享内存比4400小的的对象能申请到保留区,而不造成4031错误。

  

  4.解决办法:

  1).增大shared_pool (在不DOWN机的情况下不合适)

  2).打patch  (在不DOWN机的情况下不合适)

  3).减小门值 (在不DOWN机的情况下不合适)

  因为LAST_FAILURE_SIZE<_shared_pool_reserved_min_alloc所以表明没有有效的使用保留区

  SQL> alter system set "_shared_pool_reserved_min_alloc" = 4000;

  alter system set "_shared_pool_reserved_min_alloc"=4000

  *

  ERROR at line 1:

  ORA-02095: specified initialization parameter cannot be modified

  

  -- 9i的使用方法alter system set "_shared_pool_reserved_min_alloc"=4000 scope=spfile;

  

  4).使用alter system flush shared_pool; (不能根本性的解决问题)

  5).使用dbms_shared_pool.keep

  

  5.由于数据库不能DOWN机,所以只能选择3)和4)

  运行dbmspool.sql

  SQL> @/home/oracle/products/8.1.7/rdbms/admin/dbmspool.sql

  找出需要keep到共享内存的对象

  

  SQL> select a.OWNER,

  a.name,

  a.sharable_mem,

  a.kept,

  a.EXECUTIONS ,

  b.address,

  b.hash_value

  from v$db_object_cache a,

  v$sqlarea b

  where a.kept = 'NO' and

  (( a.EXECUTIONS > 1000

  and a.SHARABLE_MEM > 50000)

  or a.EXECUTIONS > 10000)

  and SUBSTR(b.sql_text,1,50) = SUBSTR(a.name,1,50);

  OWNER  NAME            SHARABLE_MEM KEP EXECUTIONS ADDRESS HASH_VALUE

  ------- ----------------------—--- ------------ --- ---------- -------- ----------

  SELECT COUNT(OBJECT_ID)   98292    NO  103207  74814BF8 1893309624

  FROM ALL_OBJECTS

  WHERE OBJECT_NAME = :b1

  AND OWNER = :b2

  

  STANDARD          286632    NO  13501

  DBMS_LOB          98292  NO  103750

  DBMS_LOB        47536    NO  2886542

  DBMS_LOB        11452    NO  2864757

  DBMS_PICKLER        10684    NO  2681194

  DBMS_PICKLER        5224     NO  2663860

  

  SQL> execute dbms_shared_pool.keep('STANDARD');

  SQL> execute dbms_shared_pool.keep('74814BF8,1893309624','C');

  SQL> execute dbms_shared_pool.keep('DBMS_LOB');

  SQL> execute dbms_shared_pool.keep('DBMS_PICKLER');

  SQL> select OWNER, name, sharable_mem,kept,EXECUTIONS from v$db_object_cache where kept = 'YES' ORDER BY sharable_mem;

  SQL> alter system flush shared_pool;

  System altered.

  

  SQL> SELECT POOL,BYTES FROM V$SGASTAT WHERE NAME ='free memory';

  

  POOL       BYTES

  ----------- ----------

  shared pool  7742756

  large pool   614400

  java pool    32768

  

  [oracle@ali-solution oracle]$ sh /home/oracle/admin/dbexp.sh

  

  [oracle@ali-solution oracle]$ grep ORA- /tmp/exp.tmp

  未发现错误,导出数据成功

  

  四、建议:

  由于以上解决的方法是在不能DOWN机的情况下,所以没能动态修改初始化参数,

  但问题的本质是共享区内存过小,需要增加shared pool,使用绑定变量,才能根本

  的解决问题,所以需要在适当的时候留出DOWN机时间,对内存进行合理的配置。

分享到:
评论

相关推荐

Global site tag (gtag.js) - Google Analytics