分类目录归档：数据库

ORA-600 kghstack_free2异常恢复

发表于 2013 年 2 月 20 日 由 惜分飞

今天接触一案例ORA-00600[kghstack_free2],恢复起来很简单,顺便记录下，供其他人遇到类似情况参考.

Wed Feb 20 15:05:04 2013
SMON: enabling cache recovery
Wed Feb 20 15:05:05 2013
Successfully onlined Undo Tablespace 1.
Wed Feb 20 15:05:05 2013
SMON: enabling tx recovery
Wed Feb 20 15:05:06 2013
Errors in file f:\oracle\product\10.2.0\admin\cksoltp\bdump\cksoltp_smon_3556.trc:
ORA-00600: internal error code, arguments: [kghstack_free2], [], [], [], [], [], [], []

Wed Feb 20 15:05:06 2013
Database Characterset is AL32UTF8
replication_dependency_tracking turned off (no async multimaster replication found)
Wed Feb 20 15:05:08 2013
Errors in file f:\oracle\product\10.2.0\admin\cksoltp\bdump\cksoltp_smon_3556.trc:
ORA-07445: exception encountered: core dump [ACCESS_VIOLATION] [_kghstack_err+101] 
[PC:0x603CC77F] [ADDR:0x59004E04] [UNABLE_TO_READ] []
ORA-00600: internal error code, arguments: [kghstack_free2], [], [], [], [], [], [], []

Starting background process QMNC
QMNC started with pid=18, OS id=3504
Wed Feb 20 15:05:13 2013
Errors in file f:\oracle\product\10.2.0\admin\cksoltp\bdump\cksoltp_pmon_4064.trc:
ORA-00474: SMON process terminated with error

这个错误很明显是因为smon在回滚事务的时候，遇到异常从而出现ORA-00600[kghstack_free2]错误，使得数据库不能被正常open，通过分析trace文件发现回滚段36中有事务未提交，需要回滚，针对这样的数据库恢复方法很简单:1.屏蔽事务回滚，2.屏蔽回滚段强制offline等
1) EVENT=”10513 trace name context forever”
2）_offline_rollback_segments= _SYSSMU36$
数据库启动后，通过分析trace文件，找出来异常对象，然后重建该对象,除掉event/_offline_rollback_segments即可

发表在 Oracle备份恢复 | 标签为 , , | 一条评论

通过rowid获取segment header坏块数据

发表于 2013 年 2 月 19 日 由 惜分飞

在上篇(table中各种坏块对select/dml操作影响)中说到如果segment header异常了，不能通过ctas来获得相关数据，在群的讨论中，EZIO说到可以通过rowid方式来获得相关数据，通过测试证明，确实可以通过该方法获得数据，以后遇到此类错误，大家也不必惊慌.
创建测试表

SQL> create table t_xifenfei
  2  as
  3  select * from dba_objects where object_id is not null;


SQL> alter table t_xifenfei
  2  add constraint PK_t_xifenfei primary key (object_id)
  3  ;

Table altered.

SQL> alter system checkpoint;

System altered.

SQL> select count(*) from chf.t_xifenfei;

  COUNT(*)
----------
     74762

SQL> select header_file,header_block from 
  2  DBA_SEGments where segment_name='T_XIFENFEI' AND OWNER='CHF';

HEADER_FILE HEADER_BLOCK
----------- ------------
          4          170

dump block

alter system dump datafile 4 block 170;

Dump of buffer cache at level 4 for tsn=4 rdba=16777386
Block dump from disk:
buffer tsn: 4 rdba: 0x010000aa (4/170)
scn: 0x0b8c.3c0092e4 seq: 0x01 flg: 0x04 tail: 0x91e42301
frmt: 0x02 chkval: 0xa531 type: 0x23=PAGETABLE SEGMENT HEADER

通过header_block和dump block确定block 170即为PAGETABLE SEGMENT HEADER

bbed制造SEGMENT HEADER坏块

BBED> set block 170
        BLOCK#          170

BBED> set offset 8188
        OFFSET          8188

BBED> d
 File: /u01/oracle/oradata/ora11g/users01.dbf (0)
 Block: 170              Offsets: 8188 to 8191           Dba:0x00000000
------------------------------------------------------------------------
 0123e492 

 <32 bytes per line>

BBED> m /x 0123e491
Warning: contents of previous BIFILE will be lost. Proceed? (Y/N) y
 File: /u01/oracle/oradata/ora11g/users01.dbf (0)
 Block: 170              Offsets: 8188 to 8191           Dba:0x00000000
------------------------------------------------------------------------
 0123e491 

 <32 bytes per line>

BBED> sum apply 
Check value for File 0, Block 170:
current = 0xa531, required = 0xa531

BBED> verify
DBVERIFY - Verification starting
FILE = /u01/oracle/oradata/ora11g/users01.dbf
BLOCK = 170

Block 170 is corrupt
Corrupt block relative dba: 0x010000aa (file 0, block 170)
Fractured block found during verification
Data in bad block:
 type: 35 format: 2 rdba: 0x010000aa
 last change scn: 0x0b8c.3c0092e4 seq: 0x1 flg: 0x04
 spare1: 0x0 spare2: 0x0 spare3: 0x0
 consistency value in tail: 0x91e42301
 check value in block header: 0xa531
 computed block checksum: 0x0


DBVERIFY - Verification complete

Total Blocks Examined         : 1
Total Blocks Processed (Data) : 0
Total Blocks Failing   (Data) : 0
Total Blocks Processed (Index): 0
Total Blocks Failing   (Index): 0
Total Blocks Empty            : 0
Total Blocks Marked Corrupt   : 1
Total Blocks Influx           : 2
Message 531 not found;  product=RDBMS; facility=BBED

测试segment header坏块后select操作

SQL> select * from chf.t_xifenfei;
select * from chf.t_xifenfei
                         *
ERROR at line 1:
ORA-01578: ORACLE data block corrupted (file # 4, block # 170)
ORA-01110: data file 4: '/u01/oracle/oradata/ora11g/users01.dbf'

SQL> select /*+index(t PK_T_XIFENFEI)*/  count(rowid) from chf.t_xifenfei t;

COUNT(ROWID)
------------
       74762

基于rowid获取segment header 坏块对象数据

SQL> create table chf.bad_rows (table_name varchar2(60),
   2 row_id rowid, oracle_error_code number);

Table created.

SQL> DECLARE
  2   TYPE RowIDTab IS TABLE OF ROWID INDEX BY BINARY_INTEGER;
  3   CURSOR c1 IS  select /*+index(t PK_T_XIFENFEI)*/ rowid from chf.t_xifenfei t;
  4   r RowIDTab;
  5   rows  NATURAL := 20000;
  6   bad_rows number := 0 ;
  7   errors number;
  8   error_code number;
  9   myrowid rowid;
 10  BEGIN
 11   OPEN c1;
 12   LOOP
 13     FETCH  c1 BULK COLLECT INTO r LIMIT rows;
 14     EXIT WHEN r.count=0;
 15     BEGIN
 16      FORALL i IN r.FIRST..r.LAST SAVE EXCEPTIONS
 17       insert into chf.t_xifenfei_new
 18       select /*+ ROWID(A) */ *
 19       from chf.t_xifenfei A where rowid = r(i);
 20     EXCEPTION
 21     when OTHERS then
 22      BEGIN
 23       errors := SQL%BULK_EXCEPTIONS.COUNT;
 24       FOR err1 IN 1..errors LOOP
 25           error_code := SQL%BULK_EXCEPTIONS(err1).ERROR_CODE;
 26           myrowid := r(SQL%BULK_EXCEPTIONS(err1).ERROR_INDEX);
 27           bad_rows := bad_rows + 1;
 28           insert into chf.bad_rows values('chf.t_xifenfei',myrowid, error_code);
 29       END LOOP;
 30       END;
 31     END;
 32    commit;
 33   END LOOP;
 34   commit;
 35   CLOSE c1;
 36   dbms_output.put_line('Total Bad Rows: '||bad_rows);
 37  END;
 38  /
Total Bad Rows: 0

PL/SQL procedure successfully completed.

SQL> select count(*) from chf.t_xifenfei_new;

  COUNT(*)
----------
     74762

通过上面pl/sql，基于rowid成功获得segment header 异常对象中的所有数据记录.如果没有主键的表出现该问题,可以参考:使用plsql抢救数据

发表在 Oracle备份恢复 | 标签为 | 评论关闭

table中各种坏块对select/dml操作影响

发表于 2013 年 2 月 18 日 由 惜分飞

在春节前写过table中各种类型block坏块是否能被跳过,本来准备节前写完它的姊妹篇关于table中各种blog如果出现坏块,对select/dml操作影响,因为回家一些事情给耽误了,今天补上该文章,这篇文章主要基于试验测试为主,没有从相关block原理上进行分析,如果有时间,后续文章从原理上来分析为什么这些select/dml操作不能执行
创建测试表

SQL> create table t_xifenfei as
  2  select * from dba_objects where rownum<10;

Table created.

SQL> select count(*) from chf.t_xifenfei;

  COUNT(*)
----------
         9

SQL> select
  2  dbms_rowid.rowid_relative_fno(rowid)rel_fno,
  3       max(dbms_rowid.rowid_block_number(rowid)) max_block,
  4       min(dbms_rowid.rowid_block_number(rowid)) min_block
  5       from chf.t_xifenfei
  6       group by dbms_rowid.rowid_relative_fno(rowid);

   REL_FNO  MAX_BLOCK  MIN_BLOCK
---------- ---------- ----------
         4        171        171

SQL> select EXTENT_ID,FILE_ID,BLOCK_ID,blocks from dba_extents where owner='CHF'
  2   AND SEGMENT_NAME='T_XIFENFEI';

 EXTENT_ID    FILE_ID   BLOCK_ID     BLOCKS
---------- ---------- ---------- ----------
         0          4        168          8

SQL> select SEGMENT_NAME,HEADER_FILE,HEADER_BLOCK,blocks,extents from DBA_SEGMENTS
  2  WHERE OWNER='CHF' AND SEGMENT_NAME='T_XIFENFEI';

SEGMENT_NAME                   HEADER_FILE HEADER_BLOCK     BLOCKS    EXTENTS
------------------------------ ----------- ------------ ---------- ----------
T_XIFENFEI                               4          170          8          1

通过alter system dump datafile 4 block n得出相关block数据块类型
168为FIRST LEVEL BITMAP BLOCK
169为SECOND LEVEL BITMAP BLOCK
170为PAGETABLE SEGMENT HEADER
171为trans data

处理block 168

--制造坏块
BBED> set block 168
        BLOCK#          168

BBED> set offset 8188
        OFFSET          8188

BBED> d   
 File: /u01/oracle/oradata/ora11g/users01.dbf (0)
 Block: 168              Offsets: 8188 to 8191           Dba:0x00000000
------------------------------------------------------------------------
 0320d14f 

 <32 bytes per line>

BBED> m /x 0320d14e
Warning: contents of previous BIFILE will be lost. Proceed? (Y/N) y
 File: /u01/oracle/oradata/ora11g/users01.dbf (0)
 Block: 168              Offsets: 8188 to 8191           Dba:0x00000000
------------------------------------------------------------------------
 0320d14e 

 <32 bytes per line>

BBED> sum apply
Check value for File 0, Block 168:
current = 0xf60b, required = 0xf60b

BBED> verify
DBVERIFY - Verification starting
FILE = /u01/oracle/oradata/ora11g/users01.dbf
BLOCK = 168

Block 168 is corrupt
Corrupt block relative dba: 0x010000a8 (file 0, block 168)
Fractured block found during verification
Data in bad block:
 type: 32 format: 2 rdba: 0x010000a8
 last change scn: 0x0b8c.3bff4fd1 seq: 0x3 flg: 0x04
 spare1: 0x0 spare2: 0x0 spare3: 0x0
 consistency value in tail: 0x4ed12003
 check value in block header: 0xf60b
 computed block checksum: 0x0


DBVERIFY - Verification complete

Total Blocks Examined         : 1
Total Blocks Processed (Data) : 0
Total Blocks Failing   (Data) : 0
Total Blocks Processed (Index): 0
Total Blocks Failing   (Index): 0
Total Blocks Empty            : 0
Total Blocks Marked Corrupt   : 1
Total Blocks Influx           : 2
Message 531 not found;  product=RDBMS; facility=BBED

--select操作
SQL> select count(*) from chf.t_xifenfei;

  COUNT(*)
----------
         9
--dml操作
SQL> delete from chf.t_xifenfei where rownum<3;

2 rows deleted.

----注意update操作
SQL> update chf.t_xifenfei set object_name='www.xifenfei.com';

7 rows updated.

SQL> insert into chf.t_xifenfei select * from dba_objects where rownum=1;
insert into chf.t_xifenfei select * from dba_objects where rownum=1
                *
ERROR at line 1:
ORA-01578: ORACLE data block corrupted (file # 4, block # 168)
ORA-01110: data file 4: '/u01/oracle/oradata/ora11g/users01.dbf'

这里证明对于FIRST LEVEL BITMAP BLOCK,在delete,select操作正常，insert操作异常，update操作待定(update操作不一定能够立马展示效果)

处理block 169

--标记坏块
BBED> set block 169
        BLOCK#          169

BBED> set offset 8188
        OFFSET          8188

BBED> d
 File: /u01/oracle/oradata/ora11g/users01.dbf (0)
 Block: 169              Offsets: 8188 to 8191           Dba:0x00000000
------------------------------------------------------------------------
 0221ce4f 

 <32 bytes per line>

BBED> m /x 0221ce4e
Warning: contents of previous BIFILE will be lost. Proceed? (Y/N) y
 File: /u01/oracle/oradata/ora11g/users01.dbf (0)
 Block: 169              Offsets: 8188 to 8191           Dba:0x00000000
------------------------------------------------------------------------
 0221ce4e 

 <32 bytes per line>

BBED> sum apply
Check value for File 0, Block 169:
current = 0x9d2f, required = 0x9d2f

BBED> verify
DBVERIFY - Verification starting
FILE = /u01/oracle/oradata/ora11g/users01.dbf
BLOCK = 169

Block 169 is corrupt
Corrupt block relative dba: 0x010000a9 (file 0, block 169)
Fractured block found during verification
Data in bad block:
 type: 33 format: 2 rdba: 0x010000a9
 last change scn: 0x0b8c.3bff4fce seq: 0x2 flg: 0x04
 spare1: 0x0 spare2: 0x0 spare3: 0x0
 consistency value in tail: 0x4ece2102
 check value in block header: 0x9d2f
 computed block checksum: 0x0


DBVERIFY - Verification complete

Total Blocks Examined         : 1
Total Blocks Processed (Data) : 0
Total Blocks Failing   (Data) : 0
Total Blocks Processed (Index): 0
Total Blocks Failing   (Index): 0
Total Blocks Empty            : 0
Total Blocks Marked Corrupt   : 1
Total Blocks Influx           : 2
Message 531 not found;  product=RDBMS; facility=BBED

--select操作
SQL> select count(*) from chf.t_xifenfei;

  COUNT(*)
----------
         9

--dml操作
SQL> delete from chf.t_xifenfei where rownum<2;

1 row deleted.

----注意update操作
SQL> update chf.t_xifenfei set object_name='www.xifenfei.com';

9 rows updated.

SQL> alter table t_xifenfei modify EDITION_NAME varchar2(4000);

Table altered.

SQL> update t_xifenfei set EDITION_NAME=lpad('www.xifenfei.com', 4000, '0');
update t_xifenfei set EDITION_NAME=lpad('www.xifenfei.com', 4000, '0')
       *
ERROR at line 1:
ORA-01578: ORACLE data block corrupted (file # 4, block # 169)
ORA-01110: data file 4: '/u01/oracle/oradata/ora11g/users01.dbf'

SQL> insert into chf.t_xifenfei 
  2  select * from dba_objects where rownum<2;
insert into chf.t_xifenfei
                *
ERROR at line 1:
ORA-01578: ORACLE data block corrupted (file # 4, block # 169)
ORA-01110: data file 4: '/u01/oracle/oradata/ora11g/users01.dbf'

SECOND LEVEL BITMAP BLOCK在delete,select操作正常，insert操作异常，update操作分情况(如果更新的列字符串交短，可能不报错，如果更新的字符串较长可能报错)

处理block 170

--标记坏块
BBED> SET BLOCK 170
        BLOCK#          170

BBED> set offset 8188
        OFFSET          8188

BBED> d
 File: /u01/oracle/oradata/ora11g/users01.dbf (0)
 Block: 170              Offsets: 8188 to 8191           Dba:0x00000000
------------------------------------------------------------------------
 0223b91b 

 <32 bytes per line>

BBED> m /x  0223b91a
 File: /u01/oracle/oradata/ora11g/users01.dbf (0)
 Block: 170              Offsets: 8188 to 8191           Dba:0x00000000
------------------------------------------------------------------------
 0223b91a 

 <32 bytes per line>

BBED> sum apply 
Check value for File 0, Block 170:
current = 0xb7d4, required = 0xb7d4

BBED> verify
DBVERIFY - Verification starting
FILE = /u01/oracle/oradata/ora11g/users01.dbf
BLOCK = 170

Block 170 is corrupt
Corrupt block relative dba: 0x010000aa (file 0, block 170)
Fractured block found during verification
Data in bad block:
 type: 35 format: 2 rdba: 0x010000aa
 last change scn: 0x0b8c.3c001bb9 seq: 0x2 flg: 0x04
 spare1: 0x0 spare2: 0x0 spare3: 0x0
 consistency value in tail: 0x1ab92302
 check value in block header: 0xb7d4
 computed block checksum: 0x0


DBVERIFY - Verification complete

Total Blocks Examined         : 1
Total Blocks Processed (Data) : 0
Total Blocks Failing   (Data) : 0
Total Blocks Processed (Index): 0
Total Blocks Failing   (Index): 0
Total Blocks Empty            : 0
Total Blocks Marked Corrupt   : 1
Total Blocks Influx           : 2
Message 531 not found;  product=RDBMS; facility=BBED

--select操作
SQL> select count(*) from chf.t_xifenfei;
select count(*) from chf.t_xifenfei
                         *
ERROR at line 1:
ORA-01578: ORACLE data block corrupted (file # 4, block # 170)
ORA-01110: data file 4: '/u01/oracle/oradata/ora11g/users01.dbf'

--dml操作
SQL> update chf.t_xifenfei where object_name='www.xifenfei.com';
update chf.t_xifenfei where object_name='www.xifenfei.com'
                      *
ERROR at line 1:
ORA-00971: missing SET keyword


SQL> update chf.t_xifenfei set object_name='www.xifenfei.com';
update chf.t_xifenfei set object_name='www.xifenfei.com'
           *
ERROR at line 1:
ORA-01578: ORACLE data block corrupted (file # 4, block # 170)
ORA-01110: data file 4: '/u01/oracle/oradata/ora11g/users01.dbf'


SQL> delete from chf.t_xifenfei where rownum<2;
delete from chf.t_xifenfei where rownum<2
                *
ERROR at line 1:
ORA-01578: ORACLE data block corrupted (file # 4, block # 170)
ORA-01110: data file 4: '/u01/oracle/oradata/ora11g/users01.dbf'

PAGETABLE SEGMENT HEADER异常的时候，数据库包括select,update,delete,insert操作都不能进行

结果汇总
1.BITMAP BLOCK异常的时候，select/delete操作可以正常进行，insert操作异常，update操作可能异常也可能正常
2.SEGMENT HEADER异常的时候，数据库包括select,update,delete,insert操作都不能进行
3.对于这些特殊的block出现坏块，如果有rman备份，从10g开始可以通过rman blockrecover来修复
4.如果没有rman备份,可以BITMAP BLOCK可以类似ctas重建,SEGMENT HEADER可以通过dul scan extent抽取数据
5.对于trans data太过于常见，而且event就可以跳过,在以前的文章中说过，不再讲述

发表在 Oracle | 评论关闭