Oracle：使用RMAN恢复数据库（一）：单机到单机

tlin82大约 4 分钟

Caveat

Notes to myself only . Use this instrucution at your own risk in production.

Scenario

Oracle Single instance (Source) to Single instance(target)
Netbackup Data Backup Environment

Environment

mynbumaster.example.comopen in new window (Netbackup master)
target-db.example.comopen in new window (Database server to receive the backup for full restore)
source-db.example.comopen in new window (Database server where the backup was taken)

Recovery Procedure

Check System Enviroment Variables

target database SID must be same as source database. If not modify the one on the target.

[oracle@targetServer ~]$ cat ~/.bash_profile | tr -s '\n'
# .bash_profile
# Get the aliases and functions
if [ -f ~/.bashrc ]; then
        . ~/.bashrc
fi
 
# User specific environment and startup programs
PATH=$PATH:$HOME/bin
export PATH
export TMP=/tmp
export TMPDIR=$TMP
export ORACLE_SID=orcl
export ORACLE_UNQNAME=orcl
export ORACLE_BASE=/data/u01/app/oracle
export ORACLE_HOME=$ORACLE_BASE/product/11.2.0/dbhome_1/
export TNS_ADMIN=$ORACLE_HOME/network/admin
export PATH=/usr/sbin:$PATH
export PATH=$ORACLE_HOME/bin:$PATH
export LD_LIBRARY_PATH=$ORACLE_HOME/lib:/lib:/usr/lib

Create new init spfile

example spfile:

[oracle@xmdevoaapp02 ~]$ cat /u01/oradata/init_orcl.ora
orcl.__db_cache_size=3271557120
orcl.__java_pool_size=33554432
orcl.__large_pool_size=234881024
orcl.__oracle_base='/data/u01/app/oracle'#ORACLE_BASE set from environment
orcl.__pga_aggregate_target=2332033024
orcl.__sga_target=4345298944
orcl.__shared_io_pool_size=0
orcl.__shared_pool_size=754974720
orcl.__streams_pool_size=0
*._allow_level_without_connect_by=TRUE
*.audit_file_dest='/data/u01/app/oracle/admin/orcl/adump'
*.audit_trail='db'
*.compatible='11.2.0.4.0'
*.control_file_record_keep_time=31
*.control_files='/u01/oradata/control01.ctl','/data/u01/app/oracle/fast_recovery_area/orcl/control02.ctl'
*.db_block_size=8192
*.db_domain=''
*.db_name='orcl'
*.db_recovery_file_dest='/u01/oradata/fast_recovery_area'
*.db_recovery_file_dest_size=21474836480
*.diagnostic_dest='/data/u01/app/oracle'
*.dispatchers='(PROTOCOL=TCP) (SERVICE=orcl)'
*.memory_target=6677331968
*.open_cursors=1000
*.processes=500
*.remote_login_passwordfile='EXCLUSIVE'
*.session_cached_cursors=300
*.sessions=555
*.undo_tablespace='UNDOTBS1'

Start the instance with the spfile

SQL> startup nomount pfile='/u01/oradata/init_orcl.ora';

Build the spfile

SQL> create spfile from pfile='/u01/oradata/init_orcl.ora';
file created.

Start the instance with a `nomount` option

SQL> shutdown immediate;
SQL> startup nomount;

Restore the control file

restore the control file from backup server,in this case, Netbackup mynbumaster is the NBU master and the backup was taken on the source database which is source-db.example.com in this casae.

the control file also act as RMAN backup catalog where contains backup info we need in the subsequent steps

Pick up the controle file and note down the file name in backup images

RMAN>
run
{
ALLOCATE CHANNEL CH00 TYPE 'SBT_TAPE';
SEND 'NB_ORA_SERV=mynbumaster,NB_ORA_CLIENT=source-db.example.com';
restore controlfile from 'cntrl_17662_1_1139688297';
release CHANNEL CH00;
}

start the datbase instance and mount the database but keep it unopen

SQL> alter database mount

to keep the files consistent (files include datafiles,control files , online logs), the database recovery is required before openning the database)

Restore the datafiles

rman target / cmdfile "rman_restore_dbfiles"

rman_restore_dbfiles

run {
allocate channel ch00 type 'sbt_tape';
allocate channel ch01 type 'sbt_tape';
allocate channel ch02 type 'sbt_tape';
SEND 'NB_ORA_SERV=nbumaster,NB_ORA_CLIENT=source-db.example.com';
set newname for datafile 1 to '/u01/oradata/system.259.965837843';
set newname for datafile 2 to '/u01/oradata/sysaux.260.965837853';
set newname for datafile 5 to '/u01/oradata/users.264.965837853';
set newname for datafile 6 to '/u01/oradata/ecology01.dbf';
set newname for datafile 7 to '/u01/oradata/ecology02.dbf';

SET NEWNAME FOR TEMPFILE 1 to '/u01/oradata/temp.262.965837849';
SET NEWNAME FOR TEMPFILE 2 to '/u01/oradata/t_ecology01.dbf';
restore database;
switch datafile all;
switch tempfile all;
release channel ch00;
release channel ch01;
release channel ch02;

}

Recover the database

Recovery is the last step to be done. The recovery will requires archive log in order to be able to open the database. My understanding of the primary operation is to apply the archive log to the database

method 1: restore from tape backup or backcup server

run {
allocate channel ch00 type 'sbt_tape';
allocate channel ch01 type 'sbt_tape';
SEND 'NB_ORA_SERV=mynbumaster,NB_ORA_CLIENT=source-db.example.com';
RESTORE ARCHIVELOG SEQUENCE BETWEEN 24324 AND 24338;
release channel ch00;
release channel ch01;
}

Once this step. Try to open it:

recover database open resetlogs

if you're lucky enough you should be able to open the database. if you failed . don't panic you're not alone .Acutally in many cases, the recovery at this point complaints file inconsistency. the error looks similar to this :

RMAN-06556: datafile 1 must be restored from backup older than SCN 52427660576

meaning data file 1 restored from backup is newer than the others or database it should be. SCN is the database flag number. It must be the same and one accross the data files.

option1: restore the data file 1 to old version
option2: apply the archive log to database

** how to check SCN**

SYS@ora11g> col name format a60

SYS@ora11g> select checkpoint_change# from v$database;

CHECKPOINT_CHANGE#


           2220979

SYS@ora11g> SELECT a.FILE#,a.NAME,a.RECOVER,a.CHECKPOINT_CHANGE#,status  FROM v$datafile_header a;

    FILE# NAME                                                        REC CHECKPOINT_CHANGE# STATUS


         1/u01/app/oracle/oradata/ora11g/system01.dbf                  NO             2220979ONLINE

         2/u01/app/oracle/oradata/ora11g/sysaux01.dbf                  NO             2220979ONLINE

         3/u01/app/oracle/oradata/ora11g/undotbs01.dbf                 NO             2220979ONLINE

solution####

the solution is to do more restores until that no compliants

Method 1: continue running rman restore to restore any archive logs from backup as needed
Method 2: if the backup of needed archive log no longer available (if deleted), try to get the archved logs either from the source database (very useful when you get a corrupted the backup or missing)

physical file copying and paste of the archive logs file to the target database

use RMAN catalog to register the archive logs so that RMAN recover will be able to locate thes archves during the recover process

catalog archivelog '/u01/oradata/fast_recovery_area/ORCL/archivelog/2023_06_17/o1_mf_1_24339_l8t2v396_.arc';
catalog archivelog '/u01/oradata/fast_recovery_area/ORCL/archivelog/2023_06_17/o1_mf_1_24340_l8t30xd4_.arc';

when archive los get ready . re-try the recover operation by executing the recover. if the inconsistency persists, repeat the restore of archvie log until it succeeds

Finally,

SQL> recover database open resetlogs

database is altered

Congratulations!

Side note

why backup size sometimes is larger than actual database size. To verify;

SQL> select sum(bytes)/1024/1024/1024  from dba_segments;

SUM(BYTES)/1024/1024/1024

86.4510498

SQL> select sum(bytes)/1024/1024/1024  from dba_data_files;

SUM(BYTES)/1024/1024/1024

295.922852

Oracle：使用RMAN恢复数据库（一）：单机到单机

# Scenario

# Environment

# Recovery Procedure

# Check System Enviroment Variables

# Create new init spfile

# Start the instance with the spfile

# Build the spfile

# Start the instance with a nomount option

# Restore the control file

# start the datbase instance and mount the database but keep it unopen

# Restore the datafiles

# Recover the database

#

# solution####

# Side note