这篇主要介绍的是RAC 环境下的RMAN 备份。 关于Oracle 备份与恢复的一些理论知识参考我的Blog：
Oracle 备份 与 恢复 概述
http://blog.****.net/tianlesoftware/archive/2010/04/16/5490733.aspx
一． RAC 归档的设置
1.1 相关理论知识
RAC 在运行的时候，每个实例都会产生归档日志，所有实例的归档日志集中在一起，才能完整地代表数据库的操作历史，此外，只有进行介质恢复（Media Recovery）时，才会用到归档日志。 进行介质恢复时，才要求在执行恢复操作的那个节点访问所有实例的归档日志。
正是因为归档日志有这些特点，所以归档位置的设计也有两种方案：
（1）各节点生成的归档放到共享存储上，这样自然可以确保每个节点都能够访问到，比如将归档存放到ORACLE的ASM，或者是第三方提供的集群文件系统中。 对于这种方法，一些集群的配置比较麻烦，而且也增加了ASM的维护，出现问题也不好处理。
（2）各节点除在本地生成归档文件外，另外向其它节点或者说执行备份的节点发送归档日志，以确保执行备份的那台节点能够访问到所有的归档文件。在这种方法中，因为ORACLE 的重做日志发送机制非常灵活，在10g版本中可以同时向10个目标地写入归档(11g增加到了30个)，所以利用这种特性，将各节点生成的归档发送到执行备份的节点中，来实现该节点能够访问所需的归档文件。
在第二种方案中，我们可以在每个节点建2个归档目录,分别存放本地和其他节点节点的归档日志，这里假设是2个节点的RAC.
|
归档位置 |
实例1 |
实例2 |
|
本地磁盘 |
Mkdir /rac1_arch Mkdir /rac2_arch |
Mkdir /rac1_arch Mkdir /rac2_arch |
|
Log_archive_dest_1 |
Location=’/rac1_arch’ |
Location=’/rac2_arch’ |
|
Log_archive_dest_2 |
Service=’rac2’ |
Service=’rac1’ |
|
Standby_archive_dest |
‘/rac2_arch’ |
‘/rac1_arch’ |
在每个节点上建2个目录: /u02/rac1_arch, /u02/rac2_arch， 并赋予读写的权限：
[root@rac2 /]# mkdir /u02/rac1_arch
[root@rac2 /]# mkdir /u02/rac2_arch
[root@rac2 /]# chown oracle:oinstall /u02/*
[root@rac2 /]# chmod 777 /u02/*
1.2 RAC 设置成归档模式
RAC的归档设置和单实例归档设置差不多。 先将所有实例设置为非OPEN状态，然后在任意一个处于MOUNT状态的实例执行ALTER DATABASE命令，操作成功后，再正常启动其它实例即可。
之前整理的一篇RAC 归档切换的文档，不过和今天这个实验不太匹配，就重新在整理下。
Oracle RAC 归档 与 非归档 切换
http://blog.****.net/tianlesoftware/archive/2010/04/20/5501184.aspx
1.2.1 设置归档参数
1.2.1.1 设置实例orcl1的参数：
SQL> alter system set log_archive_dest_1 = 'LOCATION=/u02/rac1_arch' scope=both sid='orcl1';
System altered.
SQL> alter system set log_archive_dest_2 = 'service=orcl2' scope=both sid='orcl1';
System altered.
SQL> alter system set standby_archive_dest = '/u02/rac2_arch' scope=both sid='orcl1';
System altered.
1.2.1.2 设置实例orcl2的参数：
SQL> alter system set log_archive_dest_1 = 'LOCATION=/u02/rac2_arch' scope=both sid='orcl2';
System altered.
SQL> alter system set log_archive_dest_2= 'SERVICE=orcl1' scope=both sid='orcl2';
System altered.
SQL> alter system set standby_archive_dest = '/u02/rac1_arch' scope=both sid='orcl2';
System altered.
1.2.1.3 在2个节点上分别验证参数的状态：
SQL> set wrap off
SQL> col dest_name format a20
SQL> select dest_name,status,error from v$archive_dest;
DEST_NAME STATUS ERROR
-------------------- --------- -------------------------------------------------
LOG_ARCHIVE_DEST_1 VALID
LOG_ARCHIVE_DEST_2 VALID
LOG_ARCHIVE_DEST_3 INACTIVE
LOG_ARCHIVE_DEST_4 INACTIVE
LOG_ARCHIVE_DEST_5 INACTIVE
LOG_ARCHIVE_DEST_6 INACTIVE
LOG_ARCHIVE_DEST_7 INACTIVE
LOG_ARCHIVE_DEST_8 INACTIVE
LOG_ARCHIVE_DEST_9 INACTIVE
LOG_ARCHIVE_DEST_10 INACTIVE
10 rows selected.
1.2.2 将RAC 切换成归档模式
1.2.2.1 修改数据库的归档模式
SQL> alter system set cluster_database=false scope=spfile sid='*';
System altered.
1.2.2.2 关闭所有实例
SQL> shutdown immediate
1.2.2.3 在任意一个实例上将数据库启动到mount状态，修改数据库归档模式
SQL> startup mount
ORACLE instance started.
Total System Global Area 167772160 bytes
Fixed Size 1266392 bytes
Variable Size 117443880 bytes
Database Buffers 46137344 bytes
Redo Buffers 2924544 bytes
Database mounted.
SQL> alter database archivelog;
Database altered.
SQL> alter system set cluster_database=true scope=spfile sid='*';
System altered.
SQL> shutdown immediate
1.2.2.4 重启数据库，确定归档生效
SQL> archive log list;
Database log mode Archive Mode
Automatic archival Enabled
Archive destination /u02/rac1_arch
Oldest online log sequence 54
Next log sequence to archive 55
Current log sequence 55
SQL> archive log list;
Database log mode Archive Mode
Automatic archival Enabled
Archive destination /u02/rac2_arch
Oldest online log sequence 26
Next log sequence to archive 27
Current log sequence 27
1.2.2.5 在两个节点分别切换归档日志，并检查对应的目录是否产生归档日志
SQL> alter system switch logfile;
System altered.
SQL> select inst_id,recid,dest_id,name from gv$archived_log ;
INST_ID RECID DEST_ID NAME
---------- ---------- ---------- -----------------------------------------------
2 12 1 /u02/rac1_arch/1_5_730181171.dbf
2 13 2 /u01/rac1_arch1_5_730181171.dbf
2 14 2 /u02/rac2_arch/2_3_730181171.dbf
2 15 1 /u02/rac1_arch/1_6_730181171.dbf
2 16 2 /u02/rac1_arch/1_6_730181171.dbf
2 17 1 /u02/rac1_arch/1_7_730181171.dbf
……
1 12 1 /u02/rac1_arch/1_5_730181171.dbf
1 13 2 /u01/rac1_arch1_5_730181171.dbf
1 14 2 /u02/rac2_arch/2_3_730181171.dbf
1 15 1 /u02/rac1_arch/1_6_730181171.dbf
1 16 2 /u02/rac1_arch/1_6_730181171.dbf
提示：RAC 数据库各实例拥有各自的REDO线程，归档文件名的生成规则由LOG_ARCHIVE_FORMAT初始化参数控制，默认情况下是 %t_%s_%r.dbf ，所以不会导致重复的发生。
注意一个参数：LOG_ARCHIVE_LOCAL_FIRST，用来设置是否首先归档文件到本地，默认为true.
LOG_ARCHIVE_LOCAL_FIRST 这个参数是Oracle 10g 新增的，它主要针对Standby环境退出，在Oracle 10g 以前的Standby中，本地和远程的归档都完成后，联机日志文件才可以被重用，在网络速度慢的环境中，远程归档的配置会很大程度的影响节点的处理能力。而设置LOG_ARCHIVE_LOCAL_FIRST=true，Oracle 会先进行本地归档，本地归档结束后在进行远程传递，同时使联机日志可以重用，从而减少了网络环境对本地的影响。 如果把这个参数设置为FALSE， 则相当于Oracle 10g 之前的方式，这个参数默认是True，如果在应用中遇到找不到归档日志的问题，就可以把这个参数改成FALSE.
二、RAC数据库的RMAN备份
RAC 和 单实例数据库备份机制是一样的，有两点需要注意：
（1） RMAN 要连接到集群中的某个实例，而不是连接到整个集群
（2） 备份归档日志时，必须保证在备份实例上能够访问所有实例的归档日志，否则就会报错。
2.1 先看一个归档日志不一致的问题
在这种情况下备份是会报错的。 之前启动归档之后，2个节点的归档目录文件是相同的，现在我们模拟归档日志不一致的情况。
先关闭两个节点的归档位置2。 此时归档日志都不能传递到对方的归档的目录下。
SQL> alter system set log_archive_dest_state_2 =defer scope=both sid='*';
System altered.
在手动档产生归档日志：
SQL> alter system switch logfile;
System altered.
此时两个节点归档目录下文件不一致。我们连接到rac1节点，然后用rman 备份一下，看报什么错
RMAN> backup database plus archivelog;
Starting backup at 20-SEP-10
current log archived
using channel ORA_DISK_1
RMAN-00571: ===========================================================
RMAN-00569: =============== ERROR MESSAGE STACK FOLLOWS ===============
RMAN-00571: ===========================================================
RMAN-03002: failure of backup plus archivelog command at 09/20/2010 14:13:32
RMAN-06059: expected archived log not found, lost of archived log compromises recoverability
ORA-19625: error identifying file /u02/rac2_arch/2_11_730181171.dbf
ORA-27037: unable to obtain file status
Linux Error: 2: No such file or directory
Additional information: 3
这里报错了。 现在我们手动把缺失的日志从rac2节点copy到节点1. 然后启用归档位置2.
SQL> alter system set log_archive_dest_state_2 =enable scope=both sid='*';
System altered.
然后在备份一下看看。
RMAN> backup database plus archivelog;
Starting backup at 20-SEP-10
current log archived
using channel ORA_DISK_1
channel ORA_DISK_1: starting archive log backupset
channel ORA_DISK_1: specifying archive log(s) in backup set
input archive log thread=1 sequence=11 recid=29 stamp=730213616
input archive log thread=1 sequence=15 recid=36 stamp=730217603
input archive log thread=1 sequence=16 recid=40 stamp=730221061
input archive log thread=1 sequence=17 recid=45 stamp=730222093
input archive log thread=2 sequence=1 recid=3 stamp=730194853
input archive log thread=2 sequence=2 recid=4 stamp=730194856
input archive log thread=2 sequence=3 recid=14 stamp=730195685
input archive log thread=2 sequence=4 recid=20 stamp=730196124
input archive log thread=2 sequence=5 recid=23 stamp=730204019
input archive log thread=2 sequence=6 recid=25 stamp=730205106
input archive log thread=2 sequence=7 recid=27 stamp=730206847
input archive log thread=2 sequence=8 recid=28 stamp=730213614
input archive log thread=2 sequence=11 recid=37 stamp=730217603
input archive log thread=2 sequence=12 recid=41 stamp=730221072
input archive log thread=2 sequence=13 recid=44 stamp=730222092
channel ORA_DISK_1: starting piece 1 at 20-SEP-10
channel ORA_DISK_1: finished piece 1 at 20-SEP-10
piece handle=+FLASH_RECOVERY_AREA/orcl/backupset/2010_09_20/annnf0_tag20100920t152820_0.265.730222103 tag=TAG20100920T152820 comment=NONE
channel ORA_DISK_1: backup set complete, elapsed time: 00:00:37
Finished backup at 20-SEP-10
Starting backup at 20-SEP-10
using channel ORA_DISK_1
channel ORA_DISK_1: starting full datafile backupset
channel ORA_DISK_1: specifying datafile(s) in backupset
input datafile fno=00001 name=+DATA/orcl/datafile/system.276.730181051
input datafile fno=00003 name=+DATA/orcl/datafile/sysaux.277.730181053
input datafile fno=00002 name=+DATA/orcl/datafile/undotbs1.278.730181053
input datafile fno=00005 name=+DATA/orcl/datafile/undotbs2.284.730181347
input datafile fno=00004 name=+DATA/orcl/datafile/users.279.730181053
channel ORA_DISK_1: starting piece 1 at 20-SEP-10
channel ORA_DISK_1: finished piece 1 at 20-SEP-10
piece handle=+FLASH_RECOVERY_AREA/orcl/backupset/2010_09_20/nnndf0_tag20100920t152900_0.266.730222143 tag=TAG20100920T152900 comment=NONE
channel ORA_DISK_1: backup set complete, elapsed time: 00:02:28
channel ORA_DISK_1: starting full datafile backupset
channel ORA_DISK_1: specifying datafile(s) in backupset
including current control file in backupset
including current SPFILE in backupset
channel ORA_DISK_1: starting piece 1 at 20-SEP-10
channel ORA_DISK_1: finished piece 1 at 20-SEP-10
piece handle=+FLASH_RECOVERY_AREA/orcl/backupset/2010_09_20/ncsnf0_tag20100920t152900_0.267.730222301 tag=TAG20100920T152900 comment=NONE
channel ORA_DISK_1: backup set complete, elapsed time: 00:00:15
Finished backup at 20-SEP-10
Starting backup at 20-SEP-10
current log archived
using channel ORA_DISK_1
channel ORA_DISK_1: starting archive log backupset
channel ORA_DISK_1: specifying archive log(s) in backup set
input archive log thread=1 sequence=18 recid=49 stamp=730222306
input archive log thread=2 sequence=14 recid=48 stamp=730222306
channel ORA_DISK_1: starting piece 1 at 20-SEP-10
channel ORA_DISK_1: finished piece 1 at 20-SEP-10
piece handle=+FLASH_RECOVERY_AREA/orcl/backupset/2010_09_20/annnf0_tag20100920t153150_0.268.730222313 tag=TAG20100920T153150 comment=NONE
channel ORA_DISK_1: backup set complete, elapsed time: 00:00:04
Finished backup at 20-SEP-10
RMAN>
2.2 归档文件的删除问题
在上面备份成功后，这里有一个新问题，就是对归档文件的删除。通常我们习惯于在备份完归档之后删除原归档文件，以释放相应的磁盘空间，不过在RAC环境下，只能删除节点1上的归档文件，节点2上的无法删除。因为在节点1上操作不了节点2上的归档文件。
这个问题我们可以通过配置通道来解决。一般有几个实例，我们就配置几个通道。 配置通道命令如下：
RMAN> configure channel 1 device type disk connect 'sys/oracle@orcl1';
new RMAN configuration parameters:
CONFIGURE CHANNEL 1 DEVICE TYPE DISK CONNECT 'sys/oracle@orcl1';
new RMAN configuration parameters are successfully stored
RMAN> configure channel 2 device type disk connect 'sys/oracle@orcl2';
new RMAN configuration parameters:
CONFIGURE CHANNEL 2 DEVICE TYPE DISK CONNECT 'sys/oracle@orcl2';
new RMAN configuration parameters are successfully stored
RMAN>
查看通道情况：
RMAN> show channel;
RMAN configuration parameters are:
CONFIGURE CHANNEL 1 DEVICE TYPE DISK CONNECT 'sys/oracle@orcl1';
CONFIGURE CHANNEL 2 DEVICE TYPE DISK CONNECT 'sys/oracle@orcl2';
关于RMAN 通道，在我blog：
RMAN 系列（二） ---- RMAN 设置和配置
http://blog.****.net/tianlesoftware/archive/2010/06/16/5674309.aspx
中的3.4节： 配置RMAN默认设置 中有介绍。
这样设置通道除了能删除其他节点的归档日志外， RMAN还能够智能的对备份中出现"问题"的任务进行再分配，比如说通道1和通道2同时执行备份，其中通道1备份数据文件1，3，4，通道2备份数据文件2和5，假如通道2执行过程中由于某些原因崩溃，RMAN可以自动将任务放回到通道1上执行。
刚才已经配置过2个不同的通道，现在我们来删除归档文件试试：
RMAN> crosscheck archivelog all;
released channel: ORA_DISK_1
allocated channel: ORA_DISK_1
channel ORA_DISK_1: sid=130 instance=orcl1 devtype=DISK
validation failed for archived log
archive log filename=/u02/rac1_arch/1_16_730181171.dbf recid=38 stamp=730221051
validation failed for archived log
archive log filename=/u02/rac1_arch/1_17_730181171.dbf recid=42 stamp=730222091
validation failed for archived log
archive log filename=/u02/rac1_arch/1_18_730181171.dbf recid=46 stamp=730222305
validation failed for archived log
archive log filename=/u02/rac2_arch/2_1_730181171.dbf recid=1 stamp=730194852
validation failed for archived log
archive log filename=/u02/rac2_arch/2_2_730181171.dbf recid=4 stamp=730194856
validation failed for archived log
archive log filename=/u02/rac2_arch/2_3_730181171.dbf recid=11 stamp=730195680
validation failed for archived log
archive log filename=/u02/rac2_arch/2_4_730181171.dbf recid=19 stamp=730196123
validation failed for archived log
archive log filename=/u02/rac2_arch/2_12_730181171.dbf recid=39 stamp=730221056
validation failed for archived log
archive log filename=/u02/rac2_arch/2_13_730181171.dbf recid=43 stamp=730222091
validation failed for archived log
archive log filename=/u02/rac2_arch/2_14_730181171.dbf recid=47 stamp=730222304
Crosschecked 10 objects
RMAN> delete archivelog all;
released channel: ORA_DISK_1
allocated channel: ORA_DISK_1
channel ORA_DISK_1: sid=130 instance=orcl1 devtype=DISK
List of Archived Log Copies
Key Thrd Seq S Low Time Name
------- ---- ------- - --------- ----
38 1 16 X 20-SEP-10 /u02/rac1_arch/1_16_730181171.dbf
42 1 17 X 20-SEP-10 /u02/rac1_arch/1_17_730181171.dbf
46 1 18 X 20-SEP-10 /u02/rac1_arch/1_18_730181171.dbf
1 2 1 X 20-SEP-10 /u02/rac2_arch/2_1_730181171.dbf
4 2 2 X 20-SEP-10 /u02/rac2_arch/2_2_730181171.dbf
11 2 3 X 20-SEP-10 /u02/rac2_arch/2_3_730181171.dbf
19 2 4 X 20-SEP-10 /u02/rac2_arch/2_4_730181171.dbf
39 2 12 X 20-SEP-10 /u02/rac2_arch/2_12_730181171.dbf
43 2 13 X 20-SEP-10 /u02/rac2_arch/2_13_730181171.dbf
47 2 14 X 20-SEP-10 /u02/rac2_arch/2_14_730181171.dbf
Do you really want to delete the above objects (enter YES or NO)? yes
deleted archive log
archive log filename=/u02/rac1_arch/1_16_730181171.dbf recid=38 stamp=730221051
deleted archive log
archive log filename=/u02/rac1_arch/1_17_730181171.dbf recid=42 stamp=730222091
deleted archive log
archive log filename=/u02/rac1_arch/1_18_730181171.dbf recid=46 stamp=730222305
deleted archive log
archive log filename=/u02/rac2_arch/2_1_730181171.dbf recid=1 stamp=730194852
deleted archive log
archive log filename=/u02/rac2_arch/2_2_730181171.dbf recid=4 stamp=730194856
deleted archive log
archive log filename=/u02/rac2_arch/2_3_730181171.dbf recid=11 stamp=730195680
deleted archive log
archive log filename=/u02/rac2_arch/2_4_730181171.dbf recid=19 stamp=730196123
deleted archive log
archive log filename=/u02/rac2_arch/2_12_730181171.dbf recid=39 stamp=730221056
deleted archive log
archive log filename=/u02/rac2_arch/2_13_730181171.dbf recid=43 stamp=730222091
deleted archive log
archive log filename=/u02/rac2_arch/2_14_730181171.dbf recid=47 stamp=730222304
Deleted 10 objects
2.3 RMAN 备份的一个完整实例
关于搭建RMAN 备份平台的其他相关操作参考Blog：
如何 搭建 RMAN 备份平台
http://blog.****.net/tianlesoftware/archive/2010/07/16/5740896.aspx
RMAN 备份的脚本参考Blog：
Linux 平台下 RMAN 全备 和 增量备份 shell 脚本
http://blog.****.net/tianlesoftware/archive/2010/07/16/5740630.aspx
这里我们用nocatalog 模式下的全备备份为例，在运行这个脚本之前需要配置好归档。 具体参考前面的说明。
Shell 脚本如下：
########################################################################
## RAC_hot_database_backup.sh ##
## created by Tianlesoftware ##
## 2010-9-21 ##
#########################################################################
#!/bin/sh
# ---------------------------------------------------------------------------
# Determine the user which is executing this script.
# ---------------------------------------------------------------------------
CUSER=`id |cut -d"(" -f2 | cut -d ")" -f1`
# ---------------------------------------------------------------------------
# Put output in <this file name>.out. Change as desired.
# Note: output directory requires write permission.
# ---------------------------------------------------------------------------
RMAN_LOG_FILE=${0}.out
# ---------------------------------------------------------------------------
# You may want to delete the output file so that backup information does
# not accumulate. If not, delete the following lines.
# ---------------------------------------------------------------------------
if [ -f "$RMAN_LOG_FILE" ]
then
rm -f "$RMAN_LOG_FILE"
fi
# -----------------------------------------------------------------
# Initialize the log file.
# -----------------------------------------------------------------
echo >> $RMAN_LOG_FILE
chmod 666 $RMAN_LOG_FILE
# ---------------------------------------------------------------------------
# Log the start of this script.
# ---------------------------------------------------------------------------
echo Script $0 >> $RMAN_LOG_FILE
echo ==== started on `date` ==== >> $RMAN_LOG_FILE
echo >> $RMAN_LOG_FILE
# ---------------------------------------------------------------------------
# Oracle home path.
# ---------------------------------------------------------------------------
ORACLE_HOME=/u01/app/oracle/product/10.2.0/db_1
export ORACLE_HOME
# ---------------------------------------------------------------------------
# the Oracle SID of the target database.
# ---------------------------------------------------------------------------
ORACLE_SID=orcl1 -- 注意，这里需要连接任意节点就可以
export ORACLE_SID
# ---------------------------------------------------------------------------
# The Oracle DBA user id (account).
# ---------------------------------------------------------------------------
ORACLE_USER=oracle
export ORACLE_USER
# ---------------------------------------------------------------------------
# Set the Oracle Recovery Manager name.
# ---------------------------------------------------------------------------
RMAN=$ORACLE_HOME/bin/rman
# ---------------------------------------------------------------------------
# Print out the value of the variables set by this script.
# ---------------------------------------------------------------------------
echo >> $RMAN_LOG_FILE
echo "RMAN: $RMAN" >> $RMAN_LOG_FILE
echo "ORACLE_SID: $ORACLE_SID" >> $RMAN_LOG_FILE
echo "ORACLE_USER: $ORACLE_USER" >> $RMAN_LOG_FILE
echo "ORACLE_HOME: $ORACLE_HOME" >> $RMAN_LOG_FILE
# ---------------------------------------------------------------------------
# Print out the value of the variables set by bphdb.
# ---------------------------------------------------------------------------
#echo >> $RMAN_LOG_FILE
#echo "NB_ORA_FULL: $NB_ORA_FULL" >> $RMAN_LOG_FILE
#echo "NB_ORA_INCR: $NB_ORA_INCR" >> $RMAN_LOG_FILE
#echo "NB_ORA_CINC: $NB_ORA_CINC" >> $RMAN_LOG_FILE
#echo "NB_ORA_SERV: $NB_ORA_SERV" >> $RMAN_LOG_FILE
#echo "NB_ORA_POLICY: $NB_ORA_POLICY" >> $RMAN_LOG_FILE
# ---------------------------------------------------------------------------
# NOTE: This script assumes that the database is properly opened. If desired,
# this would be the place to verify that.
# ---------------------------------------------------------------------------
echo >> $RMAN_LOG_FILE
# ---------------------------------------------------------------------------
# ---------------------------------------------------------------------------
# Call Recovery Manager to initiate the backup.
# ---------------------------------------------------------------------------
CMD_STR="
ORACLE_HOME=$ORACLE_HOME
export ORACLE_HOME
ORACLE_SID=$ORACLE_SID
export ORACLE_SID
$RMAN nocatalog target sys/oracle msglog $RMAN_LOG_FILE append << EOF
RUN {
allocate channel c1 device type disk connect 'sys/oracle@orcl1';
allocate channel c2 device type disk connect 'sys/oracle@orcl2';
BACKUP FORMAT '/u01/backup/orcl_%U_%T' skip inaccessible filesperset 5 DATABASE TAG orcl_hot_db_bk;
sql 'alter system archive log current';
BACKUP FORMAT '/u01/backup/arch_%U_%T' skip inaccessible filesperset 5 ARCHIVELOG ALL DELETE INPUT;
backup current controlfile tag='bak_ctlfile' format='/u01/backup/ctl_file_%U_%T';
backup spfile tag='spfile' format='/u01/backup/ORCL_spfile_%U_%T';
release channel c1;
release channel c2;
}
allocate channel for maintenance device type disk connect 'sys/oracle@orcl1'; allocate channel for maintenance device type disk connect 'sys/oracle@orcl2';
report obsolete;
delete noprompt obsolete;
crosscheck backup;
delete noprompt expired backup;
list backup summary;
release channel;
#EOF
"
# Initiate the command string
if [ "$CUSER" = "root" ]
then
echo "Root Command String: $CMD_STR" >> $RMAN_LOG_FILE
su - $ORACLE_USER -c "$CMD_STR" >> $RMAN_LOG_FILE
RSTAT=$?
else
echo "User Command String: $CMD_STR" >> $RMAN_LOG_FILE
/bin/sh -c "$CMD_STR" >> $RMAN_LOG_FILE
RSTAT=$?
fi
# ---------------------------------------------------------------------------
# Log the completion of this script.
# ---------------------------------------------------------------------------
if [ "$RSTAT" = "0" ]
then
LOGMSG="ended successfully"
else
LOGMSG="ended in error"
fi
echo >> $RMAN_LOG_FILE
echo Script $0 >> $RMAN_LOG_FILE
echo ==== $LOGMSG on `date` ==== >> $RMAN_LOG_FILE
echo >> $RMAN_LOG_FILE
/bin/mailx -s "RMAN Backup SID " tianlesoftware@vip.qq.com < $RMAN_LOG_FILE
exit $RSTAT
脚本中有指定备份文件的位置。 这里要注意，如果是直接备份到硬盘上，那么在两个节点上都要创建对应的备份目录。 因为节点1通道的备份文件会放在节点1上。节点2的备份文件会放在节点2上。 这种把备份文件分开存放，当然不方便我们恢复。
开始在这个问题上，也是碰了钉子，后来才弄明白：
RAC RMAN 备份 RMAN-03009 ORA-19504 channel c3 not allocated 错误分析
http://blog.****.net/tianlesoftware/archive/2010/09/22/5901031.aspx
所以，我们可以通过NFS mount 把备份文件放在一个节点。 或者直接把备份文件直接存放在存储上。
关于NFS mount的配置，参考我的Blog：
Linux NFS 和 Samba 共享配置
http://blog.****.net/tianlesoftware/archive/2010/07/22/5752092.aspx
这里只是测试，所以采用的是NFS方式。 当然在实际RAC环境下，是放在存储上的。
--本篇文章转自：RAC之RMAN备份