[20191206]隐含参数_db_always_check_system_ts.txt

[20191206]隐含参数_db_always_check_system_ts.txt

--//今年年头我做tab$删除恢复时,遇到的问题,就是遇到延迟块清除的问题.参考链接:
http://blog.itpub.net/267265/viewspace-2564716/
http://blog.itpub.net/267265/viewspace-2564717/

--//当时测试如果发生延迟块清除,修复删除记录的块,再读取时如果是系统表空间时报错,一般用户的表空间是没有问题.
--//一直想知道系统表空间有什么隐含参数可以绕过这个问题,毕竟修复数据库如果错误太多,无法一块一块来修复.
--//昨天才知道有一个隐含参数_db_always_check_system_ts可以绕过这个错误.今天测试看看.

1.环境:
SCOTT@book> @ ver1
PORT_STRING                    VERSION        BANNER
------------------------------ -------------- --------------------------------------------------------------------------------
x86_64/Linux 2.4.xx            11.2.0.4.0     Oracle Database 11g Enterprise Edition Release 11.2.0.4.0 - 64bit Production

SYS@book> @ hide _db_always_check_system_ts
NAME                       DESCRIPTION                                                   DEFAULT_VALUE SESSION_VALUE SYSTEM_VALUE ISSES ISSYS_MOD
-------------------------- ------------------------------------------------------------- ------------- ------------- ------------ ----- ---------
_db_always_check_system_ts Always perform block check and checksum for System tablespace TRUE          TRUE          TRUE         FALSE IMMEDIATE
--//缺省_db_always_check_system_ts=true.

2.测试:
SYS@book> create table t tablespace system as select rownum id,'test' name from dual connect by level<=2;
Table created.

SYS@book> select rowid,t.* from t;
ROWID                      ID NAME
------------------ ---------- ----
AAAWEgAABAAAAl5AAA          1 test
AAAWEgAABAAAAl5AAB          2 test

SYS@book> @ rowid AAAWEgAABAAAAl5AAA
    OBJECT       FILE      BLOCK        ROW ROWID_DBA            DBA                  TEXT
---------- ---------- ---------- ---------- -------------------- -------------------- ----------------------------------------
     90400          1       2425          0   0x400979           1,2425               alter system dump datafile 1 block 2425
-//建立在system表空间.

SYS@book>  delete from t where id=1;
1 row deleted.

SYS@book>  alter system flush buffer_cache;
System altered.

SYS@book>  alter system flush buffer_cache;
System altered.

SYS@book> @ bh 1 2425
HLADDR              DBARFIL     DBABLK      CLASS CLASS_TYPE         STATE             TCH CR_SCN_BAS CR_SCN_WRP CR_UBA_FIL CR_UBA_BLK CR_UBA_SEQ BA               OBJECT_NAME
---------------- ---------- ---------- ---------- ------------------ ---------- ---------- ---------- ---------- ---------- ---------- ---------- ---------------- --------------------
0000000084DEACF8          1       2425          1 data block         free                0          0          0          0          0          0 00000000772C2000 T
0000000084DEACF8          1       2425          1 data block         free                0          0          0          0          0          0 00000000772C4000 T
--//确定该块不在数据库缓存.

SYS@book> commit ;
Commit complete.
--//这个时候不会写块提交到块中,因为数据块已经不在数据缓存了.

3.使用bbed修复该记录看看:
BBED> set dba   1,2425
        DBA             0x00400979 (4196729 1,2425)

BBED> x /rnc *kdbr[1]
rowdata[0]                                  @8166
----------
flag@8166: 0x2c (KDRHFL, KDRHFF, KDRHFH)
lock@8167: 0x00
cols@8168:    2

col    0[2] @8169: 2
col    1[4] @8172: test

BBED> x /rnc *kdbr[0]
rowdata[11]                                 @8177
-----------
flag@8177: 0x3c (KDRHFL, KDRHFF, KDRHFD, KDRHFH)
lock@8178: 0x02
cols@8179:    0

--//第1条记录已经删除,flag=0x3c.

BBED> assign offset 8177 =0x2c;
Warning: contents of previous BIFILE will be lost. Proceed? (Y/N) y
ub1 rowdata[0]                              @8177     0x2c

BBED> x /rnc *kdbr[0]
rowdata[11]                                 @8177
-----------
flag@8177: 0x2c (KDRHFL, KDRHFF, KDRHFH)
lock@8178: 0x02
cols@8179:    2

col    0[2] @8180: 1
col    1[4] @8183: test
--//ok,现在已经恢复.lock=0x02,使用itl槽1(从0开始)

BBED> sum apply
Check value for File 1, Block 2425:
current = 0xff20, required = 0xff20

BBED> verify
DBVERIFY - Verification starting
FILE = /mnt/ramdisk/book/system01.dbf
BLOCK = 2425

Block Checking: DBA = 4196729, Block Type = KTB-managed data block
data header at 0x7f344dbba274
kdbchk: the amount of space used is not equal to block size
        used=44 fsc=9 avsp=8028 dtl=8072
Block 2425 failed with check code 6110
--//注:这个报错不必理会在select读取时.

BBED> p ktbbh.ktbbhitl[1]
struct ktbbhitl[1], 24 bytes                @68
   struct ktbitxid, 8 bytes                 @68
      ub2 kxidusn                           @68       0x000a
      ub2 kxidslt                           @70       0x0013
      ub4 kxidsqn                           @72       0x00004d92
   struct ktbituba, 8 bytes                 @76
      ub4 kubadba                           @76       0x00c00288
      ub2 kubaseq                           @80       0x0f0a
      ub1 kubarec                           @82       0x0e
   ub2 ktbitflg                             @84       0x0002 (NONE)
   union _ktbitun, 2 bytes                  @86
      sb2 _ktbitfsc                         @86       9
      ub2 _ktbitwrp                         @86       0x0009
   ub4 ktbitbas                             @88       0x00000000

--//可以发现ktbitflg=0x0002,表示没有提交.有点奇怪为什么是0x0002,应该是0x0001(因为我仅仅删除1条记录)
--//注:关于这点我在以前blog提到参考链接http://blog.itpub.net/267265/viewspace-2564779/,视乎使用表空间类型是mssm就能看到
--//这样的情况.
--//ktbitbas=0x00000000,也就是没有scn相关信息写入.

--//如果我这时读取该块就会遇到链接测试遇到的情况:链接http://blog.itpub.net/267265/viewspace-2564717/
ORA-00607: Internal error occurred while making a change to a data block
ORA-00600: internal error code, arguments: [kdBlkCheckError], [1], [2537], [6110], [], [], [], [], [], [], [], []
--//注意错误号6110,与bbed的错误号一致.

--//因为延迟块清除,在读取该块时要写入itl槽scn号.设置提交标识.这样对于system表空间这样的块就会报错(bbed verify没有通过)
--//现在修改参数:
SYS@book> alter system set "_db_always_check_system_ts"=false scope=memory ;
System altered.

SYS@book> @ hide "_db_always_check_system_ts"
NAME                       DESCRIPTION                                                   DEFAULT_VALUE SESSION_VALUE SYSTEM_VALUE ISSES ISSYS_MOD
-------------------------- ------------------------------------------------------------- ------------- ------------- ------------ ----- ---------
_db_always_check_system_ts Always perform block check and checksum for System tablespace TRUE          FALSE         FALSE        FALSE IMMEDIATE

--//按照介绍该参数是立即生效ISSYS_MOD=IMMEDIATE.保险起见还是退出会话在登录看看.

SYS@book> select rowid,t.* from t;
ROWID               ID NAME
------------------ --- -----
AAAWEgAABAAAAl5AAA   1 test
AAAWEgAABAAAAl5AAB   2 test

--//OK,现在读取就没有问题.再次通过bbed观察:

BBED> set dba   1,2425
        DBA             0x00400979 (4196729 1,2425)

BBED> p ktbbh.ktbbhitl[1]
struct ktbbhitl[1], 24 bytes                @68
   struct ktbitxid, 8 bytes                 @68
      ub2 kxidusn                           @68       0x000a
      ub2 kxidslt                           @70       0x0013
      ub4 kxidsqn                           @72       0x00004d92
   struct ktbituba, 8 bytes                 @76
      ub4 kubadba                           @76       0x00c00288
      ub2 kubaseq                           @80       0x0f0a
      ub1 kubarec                           @82       0x0e
   ub2 ktbitflg                             @84       0xa000 (KTBFUPB, KTBFCOM)`
   union _ktbitun, 2 bytes                  @86
      sb2 _ktbitfsc                         @86       3
      ub2 _ktbitwrp                         @86       0x0003
   ub4 ktbitbas                             @88       0x17600426

--//可以发现ktbitflg=0xa000(KTBFUPB, KTBFCOM),表示提交.
--//ktbitbas=0x17600426,也就是scn相关信息已经写入.

BBED> verify
DBVERIFY - Verification starting
FILE = /mnt/ramdisk/book/system01.dbf
BLOCK = 2425

Block Checking: DBA = 4196729, Block Type = KTB-managed data block
data header at 0x1f09e74
kdbchk: the amount of space used is not equal to block size
        used=44 fsc=0 avsp=8037 dtl=8072
Block 2425 failed with check code 6110

--//还是报6110错误.但是该块的读取是没有问题的.

SYS@book> alter system set "_db_always_check_system_ts"=true scope=memory ;
System altered.

SYS@book> select rowid,t.* from t;
ROWID                      ID NAME
------------------ ---------- ----
AAAWEgAABAAAAl5AAA          1 test
AAAWEgAABAAAAl5AAB          2 test

--//读取没有问题,如果修改该记录现在就会报错(注意"_db_always_check_system_ts"=true),验证看看.

SYS@book> update t set name='TEST' where id=2;
update t set name='TEST' where id=2
       *
ERROR at line 1:
ORA-00600: internal error code, arguments: [kdBlkCheckError], [1], [2425], [6110], [], [], [], [], [], [], [], []

SYS@book> alter system set "_db_always_check_system_ts"=false scope=memory ;
System altered.

SYS@book> update t set name='TEST' where id=2;
update t set name='TEST' where id=2
       *
ERROR at line 1:
ORA-01578: ORACLE data block corrupted (file # 1, block # 2425)
ORA-01110: data file 1: '/mnt/ramdisk/book/system01.dbf'

SYS@book> alter system flush buffer_cache;
System altered.

SYS@book> update t set name='TEST' where id=2;
1 row updated.
--//ok现在没有问题.

SYS@book> commit ;
Commit complete.

SYS@book> select rowid,t.* from t;
ROWID               ID NAME
------------------ --- -----
AAAWEgAABAAAAl5AAA   1 test
AAAWEgAABAAAAl5AAB   2 TEST

--//实际上这个时候使用bbed verify检查还是报错.
BBED> set dba   1,2425
        DBA             0x00400979 (4196729 1,2425)

BBED> verify
DBVERIFY - Verification starting
FILE = /mnt/ramdisk/book/system01.dbf
BLOCK = 2425

Block Checking: DBA = 4196729, Block Type = KTB-managed data block
data header at 0x7f756ceee274
kdbchk: the amount of space used is not equal to block size
        used=44 fsc=0 avsp=8037 dtl=8072
Block 2425 failed with check code 6110

4.总结:
--//设置_db_always_check_system_ts=false,可以绕过一些数据库块错误,实际上遇到这样的情况,最佳的方式设置read only.
--//采用exp或者expdp方式尽快取出数据重新建库.

5.补充如何修复该块:
BBED> verify dba 1,2425
DBVERIFY - Verification starting
FILE = /mnt/ramdisk/book/system01.dbf
BLOCK = 2425

Block Checking: DBA = 4196729, Block Type = KTB-managed data block
data header at 0x15b9e74
kdbchk: the amount of space used is not equal to block size
        used=44 fsc=0 avsp=8037 dtl=8072
Block 2425 failed with check code 6110

--//avsp= dtl-used-fsc = 8072-0 -44 = 8028

BBED> assign kdbh.kdbhavsp=8028
Warning: contents of previous BIFILE will be lost. Proceed? (Y/N) y
sb2 kdbhavsp                                @126      8116

BBED> assign kdbh.kdbhavsp=8028
sb2 kdbhavsp                                @126      8028

BBED> sum apply
Check value for File 1, Block 2425:
current = 0x8014, required = 0x8014

BBED> verify dba 1,2425
DBVERIFY - Verification starting
FILE = /mnt/ramdisk/book/system01.dbf
BLOCK = 2425

Block Checking: DBA = 4196729, Block Type = KTB-managed data block
data header at 0x16d5e74
kdbchk: space available on commit is incorrect
        tosp=8039 fsc=0 stb=0 avsp=8028
Block 2425 failed with check code 6111

--// tosp = avsp+stb+fsc= 8028+0+0 = 8028
BBED> assign kdbh.kdbhtosp=8028
sb2 kdbhtosp                                @128      8028

BBED> sum apply
Check value for File 1, Block 2425:
current = 0x802f, required = 0x802f

BBED> verify dba 1,2425
DBVERIFY - Verification starting
FILE = /mnt/ramdisk/book/system01.dbf
BLOCK = 2425

--//现在设置"_db_always_check_system_ts"=true,在执行dml就不错报错了.

SYS@book> alter system set "_db_always_check_system_ts"=true scope=memory ;
System altered.

SYS@book> update t set name='Tttt' where id=2;
1 row updated.

SYS@book> commit ;
Commit complete.

SYS@book> select rowid,t.* from t;
ROWID                      ID NAME
------------------ ---------- ----
AAAWEgAABAAAAl5AAA          1 test
AAAWEgAABAAAAl5AAB          2 Tttt