ApexSQL Log中的Redo Script跟原始SQL不一致问题

最近遇到一个误更新数据的问题，使用apexsql log做挖掘事务日志的时候，发现apexsql log生成的redo script跟原始sql是有区别的。它们并不是完全一致的。只是逻辑上等价而已。如下所示，我们做一个测试，如下所示，创建一个表后，并模拟忘记添加条件，出现误删除数据的情况

 

select * into kerry_test from sys.objects;

delete from kerry_test

 

 

然后我们用apexsql log挖掘事务日志，你会看到有很多对应的delete记录，而且对应的redo script跟原始sql是不一样的。如下所示，

 

 

 

 

下面是其中两个redo  script（为了方便查看，对脚本进行了格式化）

 

 

--delete (0003f73c:000065fc:000d) done at 2019-06-13 10:35:56.876 by xxx\xxxx in transaction 0000:0adf73ea (committed)

begin transaction;

delete  from [dbo].[kerry_test]

where   /*** warning: where clause for this statement was generated for a table with no primary key and no clustered index ***/

        [name] = n'sysfgfrag' collate chinese_prc_ci_as

        and [object_id] = 19

        and [principal_id] is null

        and [schema_id] = 4

        and [parent_object_id] = 0

        and [type] = n's ' collate latin1_general_ci_as_ks_ws

        and [type_desc] = n'system_table' collate latin1_general_ci_as_ks_ws

        and [create_date] = '20160101 10:46:55.060'

        and [modify_date] = '20160101 10:46:55.100'

        and [is_ms_shipped] = 1

        and [is_published] = 0

        and [is_schema_published] = 0;

if @@rowcount <= 1

    commit transaction;

else

    begin

        rollback transaction;

            print 'error: statement affected more than one row. all the changes were rolled back.';

    end;

--delete (0003f73c:000065fc:000a) done at 2019-06-13 10:35:56.876 by xxx\xxx in transaction 0000:0adf73ea (committed)

begin transaction;

delete  from [dbo].[kerry_test]

where   /*** warning: where clause for this statement was generated for a table with no primary key and no clustered index ***/

        [name] = n'sysseobjvalues' collate chinese_prc_ci_as

        and [object_id] = 9

        and [principal_id] is null

        and [schema_id] = 4

        and [parent_object_id] = 0

        and [type] = n's ' collate latin1_general_ci_as_ks_ws

        and [type_desc] = n'system_table' collate latin1_general_ci_as_ks_ws

        and [create_date] = '20160101 10:47:02.050'

        and [modify_date] = '20160101 10:47:02.057'

        and [is_ms_shipped] = 1

        and [is_published] = 0

        and [is_schema_published] = 0;

if @@rowcount <= 1

    commit transaction;

else

    begin

        rollback transaction;

            print 'error: statement affected more than one row. all the changes were rolled back.';

    end;

 

 

如果有聚集索引，你会看到没有where条件的update、delete操作都会生成很多个对应的redo script，对应的where条件则是聚集索引字段。如下测试所示：

select  *

into    kerry_test2

from    sys.objects;

create clustered index pk_kerry_test2 on kerry_test2(object_id);

update  kerry_test2

set     name = 'kerry';

delete  from kerry_test2;

 

 

 

下面是对应的update、delete操作的redo script

--update (0003f73c:0000700a:001f) done at 2019-06-13 11:27:26.036 by xxx\xxx in transaction 0000:0adf7404 (committed)

begin transaction; 

update  [dbo].[kerry_test2]

set     [name] = n'kerry' collate chinese_prc_ci_as

where   [object_id] = 3;

if @@rowcount <= 1

    commit transaction;

else

    begin

        rollback transaction;

            print 'error: statement affected more than one row. all the changes were rolled back.';

    end;

--    delete (0003f73c:00007482:00d9) done at 2019-06-13 11:27:35.406 by xxx\xxx in transaction 0000:0adf7405 (committed)

begin transaction;

delete  from [dbo].[kerry_test2]

where   [object_id] = 836302139;

if @@rowcount <= 1

    commit transaction;

else

    begin

        rollback transaction;

            print 'error: statement affected more than one row. all the changes were rolled back.';

    end;

 

这里只是简单记录一下这个事情，千万不要以为redo script跟原始sql是完全一致的。如果遇到这种问题，如何判断呢？可以查看transaction id，以及对应的begin time , end time,如下所示，你会发现它们是一致的