undo 损坏案列

文章由LinuxBoy分享于2019-03-30 02:03:38热评（83）

undo 损坏案列

公司一台测试环境的基于linux 平台下 Oracle 11.2.0.3 的数据库，为开归档，未备份。 21号晚上，因/目录下空间使用%100，oracle HOME目录在系统 / 目录下：

因硬盘资源占尽，不能连接操作，oracle 数据库挂起。

某人的操作,查看undotbs1 占用最大，通过mv 移动到另一目录，同时系统被重启，使得undotbs1 数据文件损坏，不能使用，最后又做了一个rm 操作，重启库，导致故障出现！

报错一：

Wed Jan 22 09:42:50 2014
ALTER DATABASE OPEN
Errors in file /u01/app/oracle/diag/rdbms/gtadata13/gtadata13/trace/gtadata13_dbw0_4245.trc:
ORA-01157: cannot identify/lock data file 3 - see DBWR trace file
ORA-01110: data file 3: '/u01/app/oracle/oradata/gtadata13/undotbs01.dbf'
ORA-27047: unable to read the header block of file
Linux-x86_64 Error: 25: Inappropriate ioctl for device
Additional information: 1
Wed Jan 22 09:42:52 2014
Checker run found 1 new persistent data failures
Errors in file /u01/app/oracle/diag/rdbms/gtadata13/gtadata13/trace/gtadata13_ora_4361.trc:
ORA-01157: cannot identify/lock data file 3 - see DBWR trace file
ORA-01110: data file 3: '/u01/app/oracle/oradata/gtadata13/undotbs01.dbf'
ORA-1157 signalled during: ALTER DATABASE OPEN...

--- 就是oracle 在mount后，不能加载到open 状态。

2 接下来操作：因为undo tablespace 数据文件undotbs1 没有了，想通过重建一个undo 表空间 undotbs2 把数据库启动到open 状态

操作：

SQL> show parameter undo
NAME TYPE VALUE
------------------------------------ ----------- ------------------------------
undo_management string AUTO
undo_retention integer 900
undo_tablespace string UNDOTBS1

SQL > CREATE UNDO TABLESPACE UNDOTBS2 DATAFILE '/XXXX.DBF' SIZE 32M AUTOEXTEND ON NEXT 32M MAXSIZE 10G; --重创建表空间
SQL > SELECT *　FROM V$TABLESAPCE SELECT NAME,STATUS FROM V$DATAFILE -- 查询其状态值
SQL > ALTER SYSTEM SET UNDO_TABLESPACE=UNDOTBS2 SCOPE=BOTH -- 通过show parameter undo 查看是否使用。

3 此时，数据库可以open起来，但是通过client ,或者其他用户连接时，报错：

报错二
SQL> conn input/INPUT
ERROR:
ORA-00604: error occurred at recursive SQL level 1
ORA-00376: file 3 cannot be read at this time
ORA-01110: data file 3: '/u01/app/oracle/oradata/gtadata13/undotbs01.dbf'
ORA-02002: error while writing to audit trail
ORA-00604: error occurred at recursive SQL level 1
ORA-00376: file 3 cannot be read at this time
ORA-01110: data file 3: '/u01/app/oracle/oradata/gtadata13/undotbs01.dbf'

4 根据报错，发现不仅仅是 undotbs1数据文件有问题，还有开启了审计 audit: 如是

先关闭审计
SQL > SHOW PARAMETER AUDIT
NAME TYPE VALUE
------------------------------------ ----------- ------------------------------
audit_file_dest string /u01/app/oracle/admin/gtadata1
3/adump
audit_sys_operations boolean FALSE
audit_syslog_level string
audit_trail string DB

SQL > alter system set audit_trail=none scope=spfile -- 设置后需要重启库。 --具体见审计

5 再通过对undotbs1数据文件操作，使其offline 处理（看行否）

SQL > alter database datafile 3 offline drop ;

6 通过 v$logfile,dba_tablespaces, dba_data_files 查看数据表空间，数据文件的状态：

SQL> select tablespace_name,file_id,file_name from dba_data_files;
TABLESPACE_NAME FILE_ID FILE_NAME
------- ---------- -------------------------------------------------------------
USERS 4 /u01/app/oracle/oradata/gtadata13/users01.dbf
UNDOTBS1 3 /u01/app/oracle/oradata/gtadata13/undotbs01.dbf
SQL> select status,tablespace_name from dba_tablespaces;

STATUS TABLESPACE_NAME
--------- ------------------------------
ONLINE SYSTEM
ONLINE SYSAUX
ONLINE UNDOTBS1

7 此时发现undotbs1 数据文件还在，同时undotbs1 表空online

如是操作：

报错三
SQL> alter tablespace UNDOTBS1 offline;
alter tablespace UNDOTBS1 offline
*
ERROR at line 1:
ORA-01191: file 3 is already offline - cannot do a normal offline
ORA-01110: data file 3: '/u01/app/oracle/oradata/gtadata13/undotbs01.dbf'
--- 此时心想，怎么不能offline了，看能否风能 temporary offline

查询数据文件头，select FILE#,checkpoint_change#,recover, fuzzy from v$datafile_header;

最后通过 SQL> alter system checkpoint; --做一个检查点，再试试：
System altered.
SQL> alter tablespace undotbs1 offline temporary;
Tablespace altered.

再次通过 dba_tablespaces 查看 undotbs1 的状态，发现是否offline。 offline 状态。

8 测试再看看能否通过其他用户连接或client 连接：
-- 发现ok，可以通过其他用户连接了，但是一些程序涉及到报错：

报错四：

执行存储过程失败 ORA-00376: 此时无法读取文件 3
ORA-01110: 数据文件 3: '/u01/app/oracle/oradata/gtadata13/undotbs01.dbf'
ORA-06512: 在 "GTA_DATA.SP_QA_TIMELINESS", line 54
ORA-06512: 在 line 1

如是想了想，确实，因为undotbs1 是通过物理删除的，那么oracle 一致性会是这些需要recovery恢复：

9 既然offline，可否删除掉，（估计比较麻烦，这回退给干掉了，怎么回退了？）

通过dba_rollback_segs 发现还有很多 recovery 的undotbs1 段需要回滚恢复，是数据一致性。

SQL> select segment_name,tablespace_name,status from dba_rollback_segs;
SEGMENT_NAME TABLESPACE_NAME STATUS
------------------------------ ------------------------------ ----------------
SYSTEM SYSTEM ONLINE
_SYSSMU122_928896348$ UNDOTBS1 OFFLINE
_SYSSMU121_4101333926$ UNDOTBS1 OFFLINE
_SYSSMU120_471964226$ UNDOTBS1 OFFLINE
_SYSSMU119_3645569891$ UNDOTBS1 OFFLINE
_SYSSMU118_1816999230$ UNDOTBS1 OFFLINE
_SYSSMU117_3513527861$ UNDOTBS1 OFFLINE
_SYSSMU116_2167311593$ UNDOTBS1 OFFLINE
_SYSSMU90_1969094056$ UNDOTBS1 NEEDS RECOVERY
_SYSSMU89_2804401042$ UNDOTBS1 NEEDS RECOVERY
_SYSSMU88_3446396459$ UNDOTBS1 NEEDS RECOVERY
_SYSSMU87_268667266$ UNDOTBS1 NEEDS RECOVERY
_SYSSMU86_1912503840$ UNDOTBS1 NEEDS RECOVERY
_SYSSMU85_2732352333$ UNDOTBS1 NEEDS RECOVERY
_SYSSMU84_1805825668$ UNDOTBS1 NEEDS RECOVERY
_SYSSMU83_1984855352$ UNDOTBS1 NEEDS RECOVERY
_SYSSMU212_1777710046$ UNDOTBS2 ONLINE
_SYSSMU211_3260590093$ UNDOTBS2 ONLINE
_SYSSMU210_1915944113$ UNDOTBS2 ONLINE
_SYSSMU209_2868303011$ UNDOTBS2 ONLINE
_SYSSMU208_3687438092$ UNDOTBS2 ONLINE
_SYSSMU207_752508113$ UNDOTBS2 ONLINE

此时，百度，及询问了一些高手，说最好做个备份：如是想通过expdp 导入导出：

报错五：

[oracle@gtadata13 dump_dir]$ impdp dcsys/DCSYS directory=dump_dir dumpfile=TBL_CHN_FN_ForecFin.dmp
Import: Release 11.2.0.3.0 - Production on Wed Jan 22 14:40:30 2014
Copyright (c) 1982, 2011, Oracle and/or its affiliates. All rights reserved.
Connected to: Oracle Database 11g Enterprise Edition Release 11.2.0.3.0 - 64bit Production
With the Partitioning, OLAP, Data Mining and Real Application Testing options
ORA-31626: job does not exist
ORA-06512: at "SYS.DBMS_SYS_ERROR", line 79
ORA-06512: at "SYS.KUPV$FT", line 1042
ORA-31637: cannot create job SYS_IMPORT_FULL_01 for user DCSYS
ORA-31632: master table "DCSYS.SYS_IMPORT_FULL_01" not found, invalid, or inaccessible
ORA-31635: unable to establish job resource synchronization
ORA-06512: at "SYS.DBMS_SYS_ERROR", line 79
ORA-06512: at "SYS.KUPV$FT_INT", line 2401
ORA-00376: file 3 cannot be read at this time
ORA-01110: data file 3: '/u01/app/oracle/oradata/gtadata13/undotbs01.dbf'
-- 这也不行，看来，只能老实的弄了

10 ,打算删除这offline undotbs1表空间，看是否跳过：

报错六：

SQL> drop tablespace undotbs1;
drop tablespace undotbs1
*
ERROR at line 1:
ORA-01548: active rollback segment '_SYSSMU1_1240252155$' found, terminate dropping tablespace
SQL> DROP ROLLBACK SEGMENT "_SYSSMU1_1240252155$";
DROP ROLLBACK SEGMENT "_SYSSMU1_1240252155$"
*
ERROR at line 1:
ORA-30025: DROP segment '_SYSSMU1_1240252155$' (in undo tablespace) not allowed

再次通过百度，高手请教：发现需要在pfile 上添加隐藏参数文件_offline_rollback_segments （‘xx’）和 _corrupted_rollback_segments ('xx') 后再删除，看否跳过

在pfile中加入参数
_offline_rollback_segments=(‘’)
_corrupted_rollback_segments=(‘’) ---括号参数为dba_rollback_segs中 undotbs1 status 为need recovery 状态的这种值“_SYSSMU122_928896348$”

10 ：于是通过 pfile添加影藏参数或者
alter system set _offline_rollback_segments = " 值 " socpe=spfile
alter system set _corrupted_rollback_segments = " 值 " socpe=spfile 进行操作。

当时我通过重建pfile参数文件 *._offline_rollback_segments=('_SYSSMU90_1969094056$',。。。。）
*._corrupted_rollback_segments=('_SYSSMU90_1969094056$', 来操作

然后通过删除所有 dba_rollback_segs 下的所有值后，在drop undotbs1 表空间：

SQL> drop rollback segment "_SYSSMU1_1240252155$"; ---注意双引号不能有空格
Rollback segment dropped. ---对应的值，一个一个删除。

11 ：最后删除 undotbs1 表空间

---ok,可以删除了，再通过dba_rollback_segs发现，没有了undtotbs1 的表空间了。

12 ：扫尾： a: 恢复原来好审计功能设置，
b: 多切换几次，查看业务数据
c: 这样操作，虽然可以了，但是有部分业务数据丢失
d: 做好备份
e: 就像大师说的，遇事，莫急躁

推荐文章：

undo 损坏案列