RMAN恢复单实例至RAC 并升级Oracle 版本
一、描述：
- 本文主要讲解单实例RMAN异机恢复至RAC集群并升级数据库的相关事宜。
- 本文主要记录的是使用RMAN恢复ORACLE 11.2.0.3 至 11.2.0.4的详细步骤。
- 本文源端环境为Redhat 6.8 + Oracle 11.2.0.3 目标端环境为 Redhat 7.4 RAC + Oracle
11.2.0.4。 - 话不多说，进入实战。
二、准备环境
- 安装RAC的步骤就不在这里描述。
三、恢复
3.1 恢复spfile/pfile
-
首先从源端环境获取pfile，然后进行修改，以下为修改好的参数文件，可用于恢复目标库。
-
然后在节点1，编辑好参数文件：
-
下述代码加粗的表示需要修改或者注意的地方。
[oracle@golddb1 rman]$ cat /soft/rman/pfile_gold_20190918.ora
*.audit_file_dest=’/u01/app/oracle/admin/gold/adump’
*.audit_trail=‘NONE’
*.compatible='11.2.0.4.0’
*.control_files=’+DATA/GOLD/CONTROLFILE/control01.ctl’,’+DATA/GOLD/CONTROLFILE/control02.ctl’#Restore Controlfile
*.db_block_size=8192
*.db_domain=’’
*.db_name=‘gold’
*.db_unique_name=‘GOLD’
*.diagnostic_dest=’/u01/app/oracle’
*.dispatchers=’(PROTOCOL=TCP) (SERVICE=goldXDB)’
*.log_archive_dest_1='LOCATION=+ARCH/'
*.log_archive_dest_state_1=‘ENABLE’
*.log_archive_format=‘arch_%t_%s_%r.dbf’
*.open_cursors=300
*.processes=1000
*.remote_login_passwordfile=‘EXCLUSIVE’
*.sessions=1105
*.standby_file_management=‘AUTO’
*.undo_tablespace=‘UNDOTBS1’
[oracle@golddb1 rman]$
-
新建下列目录（用于审计文件/节点1、2均做）
mkdir -p /u01/app/oracle/admin/gold/adump
-
节点1启动实例进程
sqlplus / as sysdba
create spfile from pfile=’/soft/rman/pfile_gold_20190918.ora’;
startup nomount;
exit;
3.2 Restore控制文件
rman target /
run
{
allocate channel c1 type disk;
restore controlfile from '/soft/rman/20190712/cf_opt_c-3028359837-20190712-01';
release channel c1;
}
-
加载数据库
alter database mount;
-
注册备份恢复目录
rman target /
CATALOG START WITH ‘/soft/rman/20190712/’;
exit;
3.3 Restore 数据库
- 节点1编辑恢复脚本
more /home/oracle/scripts/restore.sh
source /home/oracle/.bash_profile
log_dir=/home/oracle/scripts/
dt=`date+%y%m%d%H%M%S`
export log_dir
export dt
$ORACLE_HOME/bin/rman target=/ log=$log_dir/log_restore_$dt.log <- 启动后台任务
nohup sh restore.sh &
-
等待执行结果
-
或者使用查看进程
ps -ef |grep rman
waitting…
…
…
…
[oracle@golddb1 scripts]$ tail -f log_restore_.log
renamed tempfile 1 to +DATA/ in control file
renamed tempfile 2 to +DATA/ in control file
released channel: c1
released channel: c2
released channel: c3
released channel: c4
RMAN>
Recovery Manager complete.
until…
- 因为是异机恢复，需要考虑文件的路径问题，切记一定要在recover之前做datafile的rename，否则需要重新restore。
sqlplus / as sysdba
alter database rename file '/u02/redo/gold/redo04.log' to '+ARCH/GOLD/ONLINELOG/redo04.log';
alter database rename file '/u02/redo/gold/redo05.log' to '+ARCH/GOLD/ONLINELOG/redo05.log';
alter database rename file '/u02/redo/gold/redo06.log' to '+ARCH/GOLD/ONLINELOG/redo06.log';
3.4 recover database
- 制定recover的脚本跑后台，避免会话中断等问题。
more /home/oracle/scripts/recover.sh
source /home/oracle/.bash_profile
log_dir=/home/oracle/scripts/
dt=`date+%y%m%d%H%M%S`
export log_dir
export dt
$ORACLE_HOME/bin/rman target=/ log=$log_dir/log_recover_$dt.log <[oracle@golddb1 scripts]$ nohup sh /home/oracle/scripts/recover.sh &
waitting…
…
…
…
- 查看日志
[oracle@golddb1 scripts]$ tail -f log_recover_.log
released channel: c4
RMAN-00571: ===========================================================
RMAN-00569: =============== ERROR MESSAGE STACK FOLLOWS ===============
RMAN-00571: ===========================================================
RMAN-03002: failure of recover command at 09/18/2019 13:31:15
RMAN-06054: media recovery requesting unknown archived log for thread 1 with sequence 1014 and starting SCN of 8098654444
RMAN>
Recovery Manager complete.
until…
- 使用不完全恢复
- 因为rman备份时最后一个归档未备份，导致需要不完全恢复，但是不影响数据（只要确认在备份时没有数据进入）
rman target /
recover database until scn 8098654444;
exit;
-
修改数据库内存大小
-
避免后续升级报share pool 的错误
sqlplus / as sysdba alter system set sga_target=70G scope=spfile; alter system set sga_max_size=70G scope=spfile; alter system set pga_aggregate_target=20G scope=spfile; shutdown immediate; startup mount; -
在升级之前先查看组件状态
-
如果存在组件失效，建议先编译生效
set linesize 300
col comp_name for a60
select comp_id,comp_name,version,status from dba_registry;
—check valid
3.5 准备升级
cd /u01/app/oracle/product/11.2.0.4/db_1/rdbms/admin/
sqlplus / as sysdba
alter database open upgrade;
spool /home/oracle/catupgrd_20190918.txt
@catupgrd.sql
- 启动数据库并检查组件
sqlplus / as sysdba
startup;
set linesize 300
col comp_name for a60
select comp_id,comp_name,version,status from dba_registry;
COMP_ID COMP_NAME VERSION STATUS
------------------------------ ------------------------------------------------------------ ------------------------------ ----------------------
OWB OWB 11.2.0.3.0 VALID
APEX Oracle Application Express 3.2.1.00.12 VALID
EM Oracle Enterprise Manager 11.2.0.4.0 VALID
AMD OLAP Catalog 11.2.0.4.0 VALID
SDO Spatial 11.2.0.4.0 VALID
ORDIM Oracle Multimedia 11.2.0.4.0 VALID
XDB Oracle XML Database 11.2.0.4.0 VALID
CONTEXT Oracle Text 11.2.0.4.0 VALID
EXF Oracle Expression Filter 11.2.0.4.0 VALID
RUL Oracle Rules Manager 11.2.0.4.0 INVALID
OWM Oracle Workspace Manager 11.2.0.4.0 VALID
COMP_ID COMP_NAME VERSION STATUS
------------------------------ ------------------------------------------------------------ ------------------------------ ----------------------
CATALOG Oracle Database Catalog Views 11.2.0.4.0 VALID
CATPROC Oracle Database Packages and Types 11.2.0.4.0 VALID
JAVAVM JServer JAVA Virtual Machine 11.2.0.4.0 VALID
XML Oracle XDK 11.2.0.4.0 VALID
CATJAVA Oracle Database Java Packages 11.2.0.4.0 VALID
APS OLAP Analytic Workspace 11.2.0.4.0 VALID
XOQ Oracle OLAP API 11.2.0.4.0 VALID
18 rows selected.
SQL>
- 编译生效
cd /u01/app/oracle/product/11.2.0.4/db_1/rdbms/admin/
sqlplus / as sysdba
@utlrp.sql
- 再次检查组件
SQL> set linesize 300
SQL> col comp_name for a60
SQL> select comp_id,comp_name,version,status from dba_registry;
COMP_ID COMP_NAME VERSION STATUS
------------------------------ ------------------------------------------------------------ ------------------------------ ----------------------
OWB OWB 11.2.0.3.0 VALID
APEX Oracle Application Express 3.2.1.00.12 VALID
EM Oracle Enterprise Manager 11.2.0.4.0 VALID
AMD OLAP Catalog 11.2.0.4.0 VALID
SDO Spatial 11.2.0.4.0 VALID
ORDIM Oracle Multimedia 11.2.0.4.0 VALID
XDB Oracle XML Database 11.2.0.4.0 VALID
CONTEXT Oracle Text 11.2.0.4.0 VALID
EXF Oracle Expression Filter 11.2.0.4.0 VALID
RUL Oracle Rules Manager 11.2.0.4.0 VALID
OWM Oracle Workspace Manager 11.2.0.4.0 VALID
COMP_ID COMP_NAME VERSION STATUS
------------------------------ ------------------------------------------------------------ ------------------------------ ----------------------
CATALOG Oracle Database Catalog Views 11.2.0.4.0 VALID
CATPROC Oracle Database Packages and Types 11.2.0.4.0 VALID
JAVAVM JServer JAVA Virtual Machine 11.2.0.4.0 VALID
XML Oracle XDK 11.2.0.4.0 VALID
CATJAVA Oracle Database Java Packages 11.2.0.4.0 VALID
APS OLAP Analytic Workspace 11.2.0.4.0 VALID
XOQ Oracle OLAP API 11.2.0.4.0 VALID
18 rows selected.
SQL>
- @catupgrd.sql 可以重复运行,所以如果出现报错,应用解决方案之后,重新运行升级脚本.并反复检查输出log.
四、添加集群资源
- 修改参数文件
- 指定参数文件到ASM
[oracle@golddb1 ~]$ vi pfile_gold1_new_20190918.ora
gold1.__db_cache_size=60397977600
gold2.__db_cache_size=60397977600
gold1.__java_pool_size=1879048192
gold2.__java_pool_size=1879048192
gold1.__large_pool_size=805306368
gold2.__large_pool_size=805306368
gold1.__pga_aggregate_target=21206401024
gold2.__pga_aggregate_target=21206401024
gold1.__sga_target=73551314944
gold2.__sga_target=73551314944
gold1.__shared_io_pool_size=0
gold2.__shared_io_pool_size=0
gold1.__shared_pool_size=9932111872
gold2.__shared_pool_size=9932111872
gold1.__streams_pool_size=0
gold2.__streams_pool_size=0
gold1.__oracle_base=’/u01/app/oracle’#ORACLE_BASE set from environment
gold2.__oracle_base=’/u01/app/oracle’#ORACLE_BASE set from environment
*.audit_file_dest=’/u01/app/oracle/admin/gold/adump’
*.audit_trail=‘NONE’
*.compatible='11.2.0.4.0’
*.control_files=’+DATA/gold/controlfile/control01.ctl’,’+DATA/gold/controlfile/control02.ctl’#Restore Controlfile
*.db_block_size=8192
*.db_domain=’’
*.db_name=‘gold’
*.db_unique_name=‘GOLD’
*.diagnostic_dest=’/u01/app/oracle’
*.dispatchers=’(PROTOCOL=TCP) (SERVICE=goldXDB)’
*.log_archive_dest_1='LOCATION=+ARCH/'
*.log_archive_dest_state_1=‘ENABLE’
*.log_archive_format=‘arch_%t_%s_%r.dbf’
*.open_cursors=300
*.pga_aggregate_target=20971520000
*.processes=1000
*.remote_login_passwordfile=‘EXCLUSIVE’
*.sessions=1105
*.sga_max_size=73400320000
*.sga_target=73400320000
gold1.undo_tablespace=‘UNDOTBS1’
gold2.undo_tablespace='UNDOTBS2’
*.cluster_database=true
*.db_create_file_dest=’+DATA’
gold1.instance_number=1
gold2.instance_number=2
*gold2.thread=2
gold1.thread=1
.undo_management=‘AUTO’
*.remote_listener='gold-scan:1521’
按照上述修改。
- 重定向参数文件
- 如果是目标端数据库没有在集群资源中则需要创建本地的spfile
- 注意上述提到的节点2的undotbs2 ，在节点1使用该参数文件启动后，会自动创建，亦可手动创建
- 节点2的online redo 会自动创建，但是检查会发现并没有实际的文件，故后续需要手动drop 然后重建redo。
create spfile='+DATA' from pfile='/home/oracle/pfile_gold1_new_20190918.ora';
create pfile='/home/oracle/pfile_gold1_new_20190918.ora' from spfile='/u01/app/oracle/product/11.2.0.4/db_1/dbs/spfilegold1.ora';
su - grid
asmcmd
ASMCMD> cd DATA/GOLD/PARAMETERFILE
ASMCMD> ls
spfile.272.1019315119
ASMCMD> pwd
+data/gold/parameterfile
ASMCMD>
ASMCMD>
ASMCMD>
ASMCMD> mkalias '+DATA/GOLD/PARAMETERFILE/spfile.297.1019320369' '+DATA/GOLD/spfilegold.ora'
su - oracle
sqlplus / as sysdba
SQL> show parameter spfile
NAME TYPE VALUE
------------------------------------ ----------- ------------------------------
spfile string /u01/app/oracle/product/11.2.0
.4/db_1/dbs/spfilegold1.ora
shutdown immediate;
create spfile from pfile='/home/oracle/pfile_gold1_new_20190918.ora';
- 传输参数文件至节点2
scp /home/oracle/pfile_gold1_new_20190918.ora golddb2:/home/oracle/pfile_gold2_new_20190918.ora
- 节点2创建本地spfile
- 本地spfile目的是为了所恢复库不在集群资源的情况
su - oracle
sqlplus / as sysdba
create spfile from pfile='/home/oracle/pfile_gold2_new_20190918.ora';
- 节点1 &&节点2分别起库
节点1
su - oracle
vi $ORACLE_HOME/dbs/initgold1.ora
spfile='+DATA/GOLD/spfilegold.ora'
sqlplus / as sysdba
startup;
节点2
su - oracle
vi $ORACLE_HOME/dbs/initgold2.ora
spfile='+DATA/GOLD/spfilegold.ora'
sqlplus / as sysdba
startup;
- 添加集群资源
- 使用节点1 oracle用户
su - oracle
srvctl add database -d GOLD -o /u01/app/oracle/product/11.2.0.4/db_1 -p '+DATA/GOLD/spfilegold.ora'
srvctl add instance -d gold -i gold1 -n golddb1
srvctl add instance -d gold -i gold2 -n golddb2
- 两节点
- 做重启尝试验证集群命令
sqlplus / as sysdba
shutdown immediate;
节点1
su - grid
srvctl start database -d GOLD
srvctl stop database -d GOLD
- 同时观察alert日志。
- 至此集群升级恢复完成
邮箱：[email protected]
昵称：A这货不是DBA
人生短暂，何须惨淡！
欢迎指正，共同学习！