[20190918]shrink space与ORA-08102错误.txt

[20190918]shrink space与ora-08102错误.txt

1.环境:
scott@test01p> @ ver1
port_string                    version        banner                                                                               con_id
------------------------------ -------------- -------------------------------------------------------------------------------- ----------
ibmpc/win_nt64-9.1.0           12.2.0.1.0     oracle database 12c enterprise edition release 12.2.0.1.0 - 64bit production              0

2.再现ora-08102错误:

scott@test01p> create table t(x int, pad varchar2(100)) enable row movement;
table created.

scott@test01p> insert /*+ append*/  into t select level, lpad('x', 100, 'x') from dual connect by level<=1e4;
10000 rows created.

scott@test01p> alter table t add y int default 10 not null;
table altered.

scott@test01p> create index i_t_xy on t(x,y);
index created.

scott@test01p> delete t where x<=5000;
5000 rows deleted.

scott@test01p> commit ;
commit complete.

scott@test01p> alter table t shrink space;
alter table t shrink space
*
error at line 1:
ora-08102: index key not found, obj# 27979, file 11, block 2445 (2)

scott@test01p> host  oerr ora 8102
08102, 00000, "index key not found, obj# %s, file %s, block %s (%s)"
// *cause:  internal error: possible inconsistency in index
// *action:  send trace file to your customer support representative, along
//           with information on reproducing the error

3.10046跟踪看看.
scott@test01p> alter session set events '10046 level 12';
session altered.

scott@test01p> alter table t shrink space;
alter table t shrink space
*
error at line 1:
ora-08102: index key not found, obj# 27979, file 11, block 2445 (2)

scott@test01p> alter session set events '10046 off';
session altered.

--//检查转储发现:
oer 8102.2 - obj# 27979, rdba: 0x02c0098d(afn 11, blk# 2445)
kdk key 8102.2:
  ncol: 3, len: 12
  key: (12):  03 c2 64 31 ff 06 02 c0 1d a5 00 00
  mask: (2048):
--//通过bbed观察看看.
--//03 c2 64 31 ,03表示长度.后面3位表示oracle数字.

scott@test01p> @ conv_n c26431
       n20
----------
      9948

bbed> set dba 11,2446
        dba             0x02c0098e (46139790 11,2446)
--//注:windows下bbed,无法识别10g以上版本的os头,block存在+1的偏移.

bbed> map
 file: d:\app\oracle\oradata\test\test01p\users01.dbf (11)
 block: 2446                                  dba:0x02c0098e
------------------------------------------------------------
 ktb data block (index leaf)
 struct kcbh, 20 bytes                      @0
 struct ktbbh, 72 bytes                     @20
 struct kdxle, 32 bytes                     @100
 b2 kd_off[399]                             @132
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
 ub1 freespace[822]                         @930
 ub1 rowdata[6380]                          @1752
 ub4 tailchk                                @8188

bbed> x /rnnx *kd_off[3]
rowdata[6352]                               @8104
-------------
flag@8104:     0x00 (none)
lock@8105:     0x00
data key:
col    0[3] @8107: 9583
col    1[2] @8111: 10
col    2[6] @8114:  0x02  0xc0  0x1d  0x9f  0x00  0x19

--//9948-9583+3 = 368

bbed> x /rnnx *kd_off[368]
rowdata[516]                                @2268
------------
flag@2268:     0x00 (none)
lock@2269:     0x00
data key:
col    0[3] @2271: 9948
col    1[2] @2275: 10
col    2[6] @2278:  0x02  0xc0  0x1d  0xa5  0x00  0x00

bbed> x /rxxx *kd_off[368]
rowdata[516]                                @2268
------------
flag@2268:     0x00 (none)
lock@2269:     0x00
data key:
col    0[3] @2271:  0xc2  0x64  0x31
col    1[2] @2275:  0xc1  0x0b
col    2[6] @2278:  0x02  0xc0  0x1d  0xa5  0x00  0x00
--//可以看出原来的key是 03 c2  64 31 02 c1  0b 06 02 c0 1d a5 00 00
--//而shrink space后,索引的键值发生了变化,变为如下:
--//key: (12):  03 c2 64 31 ff 06 02 c0 1d a5 00 00
--//0xff表示null,参考链接:http://blog.itpub.net/267265/viewspace-2120439/=>[20160619]null在数据库的存储.txt
--//也就是索引的第2字段oracle认为是null,也就是遇到这样的情况shrink space时.oracle错误的认为y=null,
--//因为这样的情况y=10的值并没有保存在数据块中,而是放在sys.ecol$中.

scott@test01p> select *  from sys.ecol$ where tabobj# in (select data_object_id from dba_objects where owner = user and object_name = 't');
   tabobj#     colnum binarydefval                     guard_id
---------- ---------- ------------------------------ ----------
     27978          3 c10b
--//c10b对应number类型是数字10.
--//对于这样的情况如果要降低hwm,仅仅ctas建立表以及索引.
--//如果增加字段时写入数据块中,应该不会出现这样的情况.看了一下隐含参数,应该是_add_col_optim_enabled.
sys@test> @ hide _add_col_optim_enabled
name                   description                        default_value session_value system_value isses issys_mod
---------------------- ---------------------------------- ------------- ------------- ------------ ----- ---------
_add_col_optim_enabled allows new add column optimization true          true          true         true  immediate

scott@test01p> alter session set "_add_col_optim_enabled"=false;
session altered.

create table t1(x int, pad varchar2(100)) enable row movement;
insert /*+ append*/  into t1 select level, lpad('x', 100, 'x') from dual connect by level<=1e4;
alter table t1 add y int default 10 not null;
create index i_t1_xy on t1(x,y);
delete t1 where x<=5000;
commit ;
alter table t1 shrink space;

scott@test01p> alter table t1 shrink space;
table altered.
--//当然这样增加字段就很慢!!

总结:
如果要做shrink space,最好先检查看看是否曾经这样增加过新字段.