操作系统环境：WIndows 2003 Server （32Bit）
数据库版本：9.2.0.1 ARCHIVED mode
错误现象：写入数据文件11时报告I/O错误，尝试进行recover时报告下列错误：
        ORA-00283：恢复会话因错误而取消
        ORA-01115：从文件11读取块时出现IO错误（块#1046535）
        ORA-01110：数据文件11:‘E:\ORACLE\ORA92\DQSBNET\NETREP_DATA.DBF’
        ORA-27069：skgfdisp:尝试在文件范围外执行I/O
        OSD-04026：传递的参数无效（OS 1046535）

检索Metalink发现此现象为Oracle Bug，
        DB <9.2.0.3 Cannot AUTOEXTEND to 4Gb boundary in NT [ID 317174.1]

Symptoms

 When a datafile has been configured such that it will autoextend onto a 4GB boundary, any attempt to autoextend the file causes  the file the database to crash (when in noarchive log mode). If the database was in archive log mode,will be marked offline recover and access to the database continues.

Solution

This bug is fixed in  9203. Upgrading the database to the latest relase followed by a recovery will solve this problem.

Here are the steps:

1.Do a clean shutdown of the database

2. Upgrade the database to the latest version ( greater than 9.2.0.3) as this bug is fixed in 9203.

If the database version is below 9i Rel 2(9.2.0.1) then first upgrade to 9iRel2 and then apply the latest available patchset.
Kindly check metalink for the latest patchset available

3.Mount the database

4.Do a COMPLTE recovery of the dataabse

5.Then start the database in migrate mode and upgrade the database. Refer Note 159657.1: Complete Upgrade Checklist for Manual Upgrades from 8.X / 9.0.1 to Oracle9i

我的操作：

1. 下载9.2.0.4补丁p3095277_9204_WINNT.zip
2. alter database datafile 11 offline;
3. shutdown immediate
4. 升级Oracle软件至9.2.0.4
5. startup nomount
6. alter system set shared_pool_size=150m scope=spfile;
7. alter system set java_pool_size=150m scope=spfile;
8. shutdown immediate
9. startup migrate
10. spool patch.log
11. @?\rdbms\admin\catpatch.sql
12. spool off
13. shutdown immediate
14. startup
15. recover datafile 11;      -- 选择auto方式
    
16. alter database datafile 11 online;
17. shutdown immedaite
18. cold backup