2004年11月01日

情况说明:
系统:SUN Solaris8
数据库版本:9203
问题描述:工程人员报告,数据库在重新启动时无法正常启动.检查发现UNDO表空间丢失.
问题诊断及解决过程如下:

1. 登陆系统检查alert.log文件

检查alert.log文件是通常是我们诊断数据库问题的第一步

SunOS 5.8

login: root
Password:
Last login: Thu Apr 1 11:39:16 from 10.123.7.162
Sun Microsystems Inc. SunOS 5.8 Generic Patch October 2001
You have new mail.
# su – oracle
bash-2.03$ cd $ORACLE_BASE/admin/*/bdump
bash-2.03$ vi *.log

“alert_gzhs.log” 7438 lines, 283262 characters
Sat Feb 7 20:30:06 2004
Starting ORACLE instance (normal)
LICENSE_MAX_SESSION = 0
LICENSE_SESSIONS_WARNING = 0
SCN scheme 3
Using log_archive_dest parameter default value
LICENSE_MAX_USERS = 0
SYS auditing is disabled
Starting up ORACLE RDBMS Version: 9.2.0.3.0.
System parameters with non-default values:
processes = 150
timed_statistics = TRUE
shared_pool_size = 1157627904
large_pool_size = 16777216
java_pool_size = 637534208
control_files = /u01/oradata/gzhs/control01.ctl,
/u02/oradata/gzhs/control02.ctl,
/u03/oradata/gzhs/control03.ctl
db_block_size = 8192
db_cache_size = 2516582400
compatible = 9.2.0.0.0
log_archive_start = TRUE
log_archive_dest_1 = LOCATION=/u06/oradata/gzhs/arch
log_archive_format = %t_%s.dbf
db_file_multiblock_read_count= 16
fast_start_mttr_target = 300
undo_management = AUTO
undo_tablespace = UNDOTBS1
undo_retention = 10800
remote_login_passwordfile= EXCLUSIVE
db_domain =
instance_name = gzhs
dispatchers = (PROTOCOL=TCP) (SERVICE=gzhsXDB)
job_queue_processes = 10
hash_join_enabled = TRUE
background_dump_dest = /oracle/admin/gzhs/bdump
user_dump_dest = /oracle/admin/gzhs/udump
core_dump_dest = /oracle/admin/gzhs/cdump
sort_area_size = 524288
db_name = gzhs
open_cursors = 300
star_transformation_enabled= FALSE
query_rewrite_enabled = FALSE
pga_aggregate_target = 838860800
aq_tm_processes = 1
PMON started with pid=2
DBW0 started with pid=3
LGWR started with pid=4
CKPT started with pid=5
SMON started with pid=6
“alert_gzhs.log” 7438 lines, 283262 characters
USER: terminating instance due to error 30012
Instance terminated by USER, pid = 26433
ORA-1092 signalled during: ALTER DATABASE OPEN…
Thu Apr 1 11:11:08 2004
Starting ORACLE instance (normal)
LICENSE_MAX_SESSION = 0
LICENSE_SESSIONS_WARNING = 0
SCN scheme 3
Using log_archive_dest parameter default value
LICENSE_MAX_USERS = 0
SYS auditing is disabled
Starting up ORACLE RDBMS Version: 9.2.0.3.0.
System parameters with non-default values:
processes = 150
timed_statistics = TRUE
shared_pool_size = 1157627904
large_pool_size = 16777216
java_pool_size = 637534208
control_files = /u01/oradata/gzhs/control01.ctl, /u02/oradata/gzhs/control02.ctl, /u03/oradata/gzhs/control03.ctl
db_block_size = 8192
db_cache_size = 2516582400
compatible = 9.2.0.0.0
log_archive_start = TRUE
log_archive_dest_1 = LOCATION=/u06/oradata/gzhs/arch
log_archive_format = %t_%s.dbf
db_file_multiblock_read_count= 16
fast_start_mttr_target = 300
undo_management = AUTO
undo_tablespace = UNDOTBS1
undo_retention = 10800
remote_login_passwordfile= EXCLUSIVE
db_domain =
instance_name = gzhs
dispatchers = (PROTOCOL=TCP) (SERVICE=gzhsXDB)
job_queue_processes = 10
hash_join_enabled = TRUE
background_dump_dest = /oracle/admin/gzhs/bdump
user_dump_dest = /oracle/admin/gzhs/udump
core_dump_dest = /oracle/admin/gzhs/cdump
sort_area_size = 524288
db_name = gzhs
open_cursors = 300
star_transformation_enabled= FALSE
query_rewrite_enabled = FALSE
pga_aggregate_target = 838860800
aq_tm_processes = 1
PMON started with pid=2
DBW0 started with pid=3
LGWR started with pid=4
CKPT started with pid=5
SMON started with pid=6
RECO started with pid=7
CJQ0 started with pid=8
Thu Apr 1 11:11:13 2004
starting up 1 shared server(s) …
QMN0 started with pid=9
Thu Apr 1 11:11:13 2004
starting up 1 dispatcher(s) for network address ‘(ADDRESS=(PARTIAL=YES)(PROTOCOL=TCP))’…
ARCH: STARTING ARCH PROCESSES
ARC0 started with pid=12
ARC0: Archival started
ARC1 started with pid=13
Thu Apr 1 11:11:13 2004
ARCH: STARTING ARCH PROCESSES COMPLETE
Thu Apr 1 11:11:13 2004
ARC0: Thread not mounted
Thu Apr 1 11:11:13 2004
ARC1: Archival started
ARC1: Thread not mounted
Thu Apr 1 11:11:14 2004
ALTER DATABASE MOUNT
Thu Apr 1 11:11:18 2004
Successful mount of redo thread 1, with mount id 1088380178.
Thu Apr 1 11:11:18 2004
Database mounted in Exclusive Mode.
Completed: ALTER DATABASE MOUNT
Thu Apr 1 11:11:27 2004
alter database open
Thu Apr 1 11:11:27 2004
Beginning crash recovery of 1 threads
Thu Apr 1 11:11:27 2004
Started first pass scan
Thu Apr 1 11:11:28 2004
Completed first pass scan
1 redo blocks read, 0 data blocks need recovery
Thu Apr 1 11:11:28 2004
Started recovery at
Thread 1: logseq 177, block 2, scn 0.33104793
Recovery of Online Redo Log: Thread 1 Group 3 Seq 177 Reading mem 0
Mem# 0 errs 0: /u01/oradata/gzhs/redo03.log
Thu Apr 1 11:11:28 2004
Completed redo application
Thu Apr 1 11:11:28 2004
Ended recovery at
Thread 1: logseq 177, block 3, scn 0.33124794
0 data blocks read, 0 data blocks written, 1 redo blocks read
Crash recovery completed successfully
Thu Apr 1 11:11:28 2004
LGWR: Primary database is in CLUSTER CONSISTENT mode
Thread 1 advanced to log sequence 178
Thread 1 opened at log sequence 178
Current log# 1 seq# 178 mem# 0: /u01/oradata/gzhs/redo01.log
Successful open of redo thread 1.
Thu Apr 1 11:11:28 2004
ARC0: Evaluating archive log 3 thread 1 sequence 177
Thu Apr 1 11:11:28 2004
ARC0: Beginning to archive log 3 thread 1 sequence 177
Creating archive destination LOG_ARCHIVE_DEST_1: ‘/u06/oradata/gzhs/arch/1_177.dbf’
Thu Apr 1 11:11:28 2004
SMON: enabling cache recovery
ARC0: Completed archiving log 3 thread 1 sequence 177
Thu Apr 1 11:11:28 2004
Errors in file /oracle/admin/gzhs/udump/gzhs_ora_27781.trc:
ORA-30012: \263\267\317\373\261\355\277\325\274\344 ‘UNDOTBS1′ \262\273\264\346\324\332\273\362\300\340\320\315\262\273\325\375\310\
267
Thu Apr 1 11:11:28 2004
Error 30012 happened during db open, shutting down database
USER: terminating instance due to error 30012
Instance terminated by USER, pid = 27781
ORA-1092 signalled during: alter database open…

:q

在警报日志末尾显示了数据库在Open状态因为错误而异常终止.

2. 尝试重新启动数据库

bash-2.03$ sqlplus “/ as sysdba”

SQL*Plus: Release 9.2.0.3.0 – Production on 星期四 4月 1 11:43:52 2004

Copyright (c) 1982, 2002, Oracle Corporation. All rights reserved.

已连接到空闲例程。

SQL> startup
ORACLE 例程已经启动。

Total System Global Area 4364148184 bytes
Fixed Size 736728 bytes
Variable Size 1845493760 bytes
Database Buffers 2516582400 bytes
Redo Buffers 1335296 bytes
数据库装载完毕。
ORA-01092: ORACLE 例程终止。强行断开连接

工程人员报告的问题重现.

3. 检查数据文件

bash-2.03$ cd /u01/ oradata/gzhs
bash-2.03$ ls -l
total 55702458
-rw-r—– 1 oracle dba 1073750016 Apr 1 11:44 UNDOTBS2.dbf
-rw-r—– 1 oracle dba 1073750016 Apr 1 11:44 WAP12_BILLINGDETAIL.dbf
-rw-r—– 1 oracle dba 1073750016 Apr 1 11:44 WAP12_MAIN.dbf
-rw-r—– 1 oracle dba 2097160192 Apr 1 11:44 WAP12_MAIN10.dbf
-rw-r—– 1 oracle dba 2097160192 Apr 1 11:44 WAP12_MAIN11.dbf
-rw-r—– 1 oracle dba 2097160192 Apr 1 11:44 WAP12_MAIN2.dbf
-rw-r—– 1 oracle dba 2097160192 Apr 1 11:44 WAP12_MAIN3.dbf
-rw-r—– 1 oracle dba 2097160192 Apr 1 11:44 WAP12_MAIN4.dbf
-rw-r—– 1 oracle dba 2097160192 Apr 1 11:44 WAP12_MAIN5.dbf
-rw-r—– 1 oracle dba 2097160192 Apr 1 11:44 WAP12_MAIN6.dbf
-rw-r—– 1 oracle dba 2097160192 Apr 1 11:44 WAP12_MAIN7.dbf
-rw-r—– 1 oracle dba 2097160192 Apr 1 11:44 WAP12_MAIN8.dbf
-rw-r—– 1 oracle dba 2097160192 Apr 1 11:44 WAP12_MAIN9.dbf
-rw-r—– 1 oracle dba 1073750016 Apr 1 11:44 WAP12_MVIEW.dbf
-rw-r—– 1 oracle dba 1073750016 Mar 24 17:15 WAP12_TEMP1.dbf
…………………….

发现存在文件UNDOTBS2.dbf

4. mount数据库,检查系统参数

bash-2.03$ sqlplus “/ as sysdba”

SQL*Plus: Release 9.2.0.3.0 – Production on 星期四 4月 1 11:46:20 2004

Copyright (c) 1982, 2002, Oracle Corporation. All rights reserved.

已连接到空闲例程。

SQL>
SQL>
SQL> startup mount;
ORACLE 例程已经启动。

Total System Global Area 4364148184 bytes
Fixed Size 736728 bytes
Variable Size 1845493760 bytes
Database Buffers 2516582400 bytes
Redo Buffers 1335296 bytes
数据库装载完毕。
SQL> select name from v$datafile;

NAME
——————————————————————————–
/u01/oradata/gzhs/system01.dbf
/u01/oradata/gzhs/cwmlite01.dbf
/u01/oradata/gzhs/drsys01.dbf
/u01/oradata/gzhs/example01.dbf
/u01/oradata/gzhs/indx01.dbf
/u01/oradata/gzhs/odm01.dbf
/u01/oradata/gzhs/tools01.dbf
/u01/oradata/gzhs/users01.dbf
/u01/oradata/gzhs/xdb01.dbf
…………………….
/u01/oradata/gzhs/UNDOTBS2.dbf

已选择23行。

SQL>
SQL> show parameter undo

NAME TYPE VALUE
———————————— ———– ——————————
undo_management string AUTO
undo_retention integer 10800
undo_suppress_errors boolean FALSE
undo_tablespace string UNDOTBS1
SQL> show parameter spfile

NAME TYPE VALUE
———————————— ———– ——————————
spfile string

发现系统没有使用spfile,而初始化参数设置的undo表空间为UNDOTBS1

5. 检查参数文件

bash-2.03$ cd $ORACLE_HOME/dbs
bash-2.03$ ls
init.ora initgzhs.ora initgzhs.ora.old orapwgzhs
initdw.ora initgzhs.ora.hurray lkGZHS snapcf_gzhs.f
bash-2.03$ vi initgzhs.ora
“initgzhs.ora” [Incomplete last line] 105 lines, 3087 characters
##############################################################################
# Copyright (c) 1991, 2001, 2002 by Oracle Corporation
##############################################################################

###########################################
# Archive
###########################################
log_archive_dest_1=’LOCATION=/u06/oradata/gzhs/arch’
log_archive_format=%t_%s.dbf
log_archive_start=true

###########################################
# Cache and I/O
###########################################
db_block_size=8192
db_cache_size=2516582400
db_file_multiblock_read_count=16

###########################################
# Cursors and Library Cache
###########################################
open_cursors=300

………………….

###########################################
# System Managed Undo and Rollback Segments
###########################################
undo_management=AUTO
undo_retention=10800
undo_tablespace=UNDOTBS1

:q!

这个设置是极其可疑的.
怀疑参数文件和实际数据库设置不符.

6. 再次检查alert文件
查找对于UNDO表空间的操作

第一部分,创建数据库时的信息:

Sat Feb 7 20:30:12 2004
CREATE DATABASE gzhs
MAXINSTANCES 1
MAXLOGHISTORY 1
MAXLOGFILES 5
MAXLOGMEMBERS 3
MAXDATAFILES 100
DATAFILE ‘/u01/oradata/gzhs/system01.dbf’ SIZE 500M REUSE AUTOEXTEND ON NEXT 10240K MAXSIZE UNLIMITED
EXTENT MANAGEMENT LOCAL
DEFAULT TEMPORARY TABLESPACE TEMP TEMPFILE ‘/u01/oradata/gzhs/temp01.dbf’ SIZE 1000M REUSE AUTOEXTEND ON NEXT 250M MAXSIZE UNLIMITED
UNDO TABLESPACE “UNDOTBS1″ DATAFILE ‘/u01/oradata/gzhs/undotbs01.dbf’ SIZE 1000M REUSE AUTOEXTEND ON NEXT 100M MAXSIZE UNLIMITED
CHARACTER SET ZHS16GBK
NATIONAL CHARACTER SET AL16UTF16
LOGFILE GROUP 1 (‘/u01/oradata/gzhs/redo01.log’) SIZE 256M,
GROUP 2 (‘/u01/oradata/gzhs/redo02.log’) SIZE 256M,
GROUP 3 (‘/u01/oradata/gzhs/redo03.log’) SIZE 256M

注意,这也是OCP教材上提到的两种创建UNDO表空间的方式之一

第二部分,发现创建UNDOTBS2的记录信息:

Wed Mar 24 20:20:58 2004
/* OracleOEM */ CREATE UNDO TABLESPACE “UNDOTBS2″ DATAFILE ‘/u01/oradata/gzhs/UNDOTBS2.dbf’ SIZE 1024M AUTOEXTEND ON NEXT 100M MAXSIZE UNLIMITED
Wed Mar 24 20:22:37 2004
Created Undo Segment _SYSSMU11$
Created Undo Segment _SYSSMU12$
Created Undo Segment _SYSSMU13$
Created Undo Segment _SYSSMU14$
Created Undo Segment _SYSSMU15$
Created Undo Segment _SYSSMU16$
Created Undo Segment _SYSSMU17$
Created Undo Segment _SYSSMU18$
Created Undo Segment _SYSSMU19$
Created Undo Segment _SYSSMU20$
Completed: /* OracleOEM */ CREATE UNDO TABLESPACE “UNDOTBS2″
Wed Mar 24 20:24:25 2004
Undo Segment 11 Onlined
Undo Segment 12 Onlined
Undo Segment 13 Onlined
Undo Segment 14 Onlined
Undo Segment 15 Onlined
Undo Segment 16 Onlined
Undo Segment 17 Onlined
Undo Segment 18 Onlined
Undo Segment 19 Onlined
Undo Segment 20 Onlined
Successfully onlined Undo Tablespace 15.
Undo Segment 1 Offlined
Undo Segment 2 Offlined
Undo Segment 3 Offlined
Undo Segment 4 Offlined
Undo Segment 5 Offlined
Undo Segment 6 Offlined
Undo Segment 7 Offlined
Undo Segment 8 Offlined
Undo Segment 9 Offlined
Undo Segment 10 Offlined
Undo Tablespace 1 successfully switched out.

第三部分,新的UNDO表空间被应用

Wed Mar 24 20:24:25 2004
ALTER SYSTEM SET undo_tablespace=’UNDOTBS2′ SCOPE=MEMORY;

我们发现问题就在这里,创建了新的UNDO表空间以后,因为使用的是pfile文件,修改的只对当前实例生效,操作人员忘记了修改pfile文件.

如果使用spfile,缺省的修改范围是both,会同时修改spfile文件,就可以避免以上问题的出现.

第四部分,删除了UNDOTBS1的信息

Wed Mar 24 20:25:01 2004
/* OracleOEM */ DROP TABLESPACE “UNDOTBS1″ INCLUDING CONTENTS AND DATAFILES CASCADE CONSTRAINTS
Wed Mar 24 20:25:03 2004
Deleted file /u01/oradata/gzhs/undotbs01.dbf
Completed: /* OracleOEM */ DROP TABLESPACE “UNDOTBS1″ INCLUDI

这样再次重新启动数据库的时候,问题出现了,pfile中定义的UNDOTBS1找不到了,而且操作实在很久以前,没人能回忆起来,甚至无法得知是什么人的操作。

7. 更改pfile,启动数据库

修改undo表空间

###########################################
# System Managed Undo and Rollback Segments
###########################################
undo_management=AUTO
undo_retention=10800
undo_tablespace=UNDOTBS2

….

bash-2.03$ sqlplus “/ as sysdba”

SQL*Plus: Release 9.2.0.3.0 – Production on 星期四 4月 1 11:55:11 2004

Copyright (c) 1982, 2002, Oracle Corporation. All rights reserved.

连接到:
Oracle9i Enterprise Edition Release 9.2.0.3.0 – 64bit Production
With the Partitioning, OLAP and Oracle Data Mining options
JServer Release 9.2.0.3.0 – Production

SQL> select * from v$version;

BANNER
—————————————————————-
Oracle9i Enterprise Edition Release 9.2.0.3.0 – 64bit Production
PL/SQL Release 9.2.0.3.0 – Production
CORE 9.2.0.3.0 Production
TNS for Solaris: Version 9.2.0.3.0 – Production
NLSRTL Version 9.2.0.3.0 – Production

SQL> exit
从Oracle9i Enterprise Edition Release 9.2.0.3.0 – 64bit Production
With the Partitioning, OLAP and Oracle Data Mining options
JServer Release 9.2.0.3.0 – Production中断开
bash-2.03$

在这里我们可以看到,使用spfile可以免去手工修改pfile文件的麻烦,减少了犯错的可能。

既然Oracle9i给我们提供了这个新特性,就值得我们学习使用它.