RMAN: RAC Backup and Recovery using RMAN [ID 243760.1]

RMAN: RAC Backup and Recovery using RMAN [ID 243760.1]
		修改时间 30-JUN-2011     类型 BULLETIN     状态 PUBLISHED

Applies to:

Oracle Server - Standard Edition
Oracle Server - Personal Edition
Oracle Server - Enterprise Edition - Version: 9.0.1.0 and later    [Release: 9.0.1 and later]
Information in this document applies to any platform.

Purpose

The purpose of this document is to give a quick guide about how to perform. Backup and Recovery on RAC databases. We will follow this points:

1. Verify the database mode and archive destination.
2. Verify connectivity using sqlnet for target and catalog.
3. Determine your backup Device.
4. Understand how to create an RMAN persistent configuration for a RAC env.
5. Create backups to disk using the new persistent configuration parameters.
6. Backupset Maintenance using the configured retention policy.
7. Restore and Recover
    a. Complete
    b. Incomplete
8. Review and understand the impact of resetlogs on the catalog.
9. RMAN Sample Commands.

Scope and Application

• This discussion is for a 2-node Oracle RAC Cluster.
• The logs are being archived to their respective node.
• We are allocating channels to each node to enable the autolocate feature of RMAN in a RAC env.

RMAN: RAC Backup and Recovery using RMAN

1. Verify the databases are in archivelog mode and archive destination.

 a. NODE 1: thread 1

SQL> archive log list;
Database log mode              Archive Mode
Automatic archival             Enabled
Archive destination            /u02/app/oracle/product/9.2.0/dbs/arch
Oldest online log sequence     20
Next log sequence to archive   21
Current log sequence           21

 b. NODE 2: thread 2

SQL> archive log list;
Database log mode              Archive Mode
Automatic archival             Enabled
Archive destination            /u02/app/oracle/product/9.2.0/dbs/arch
Oldest online log sequence     8
Next log sequence to archive   9
Current log sequence           9

2. Verify connectivity to the target nodes and catalog if used.

  $ setenv TNS_ADMIN $ORACLE_HOME/network/admin
  $ sqlplus /nolog
     SQL> connect sys/pwd@node1 as sysdba
     SQL> connect sys/pwd@node2 as sysdba
     SQL> connect rman/rman@rcat

3. Set your testing areas.

Testing HOME for logs:  /u02/home/usupport/rman

Backups HOME Location:  /rman/V920

4. Connect using RMAN to verify and set the controlfile persistent configuration. 

The controlfiles are shared between the instances so configuring the controlfile on node 1 also sets it for all nodes in the RAC cluster.

 * Always note the target DBID
    connected to target database: V920 (DBID=228033884)

 *  Default Configuration

    RMAN> SHOW ALL;
    CONFIGURE RETENTION POLICY TO REDUNDANCY 1; # default
    CONFIGURE BACKUP OPTIMIZATION OFF; # default
    CONFIGURE DEFAULT DEVICE TYPE TO DISK; # default
    CONFIGURE CONTROLFILE AUTOBACKUP ON;
    CONFIGURE CONTROLFILE AUTOBACKUP FORMAT FOR DEVICE TYPE DISK TO '%F'; # default
    CONFIGURE DEVICE TYPE DISK PARALLELISM 1; # default
    CONFIGURE DATAFILE BACKUP COPIES FOR DEVICE TYPE DISK TO 1; # default
    CONFIGURE ARCHIVELOG BACKUP COPIES FOR DEVICE TYPE DISK TO 1; # default
    CONFIGURE MAXSETSIZE TO UNLIMITED; # default
    CONFIGURE SNAPSHOT CONTROLFILE NAME TO '/u02/app/oracle/product/9.2.0/dbs/snapcf_V9201.f'; # default

 * Make changes to the default that fit your business requirements.
     Note the retention policy can be set "TO REDUNDANCY x" or it can  be set "TO RECOVERY WINDOW OF x DAYS", this is new in Oracle9i.

     In this example, using PARALLELISM 2 as 2 nodes are used.
     The PARALLELISM will than automatically start 2 channels and will   use the related CONFIGURE CHANNEL for additional clauses.

    CONFIGURE RETENTION POLICY TO REDUNDANCY 3;
    CONFIGURE BACKUP OPTIMIZATION OFF;
    CONFIGURE DEFAULT DEVICE TYPE TO DISK;
    CONFIGURE CONTROLFILE AUTOBACKUP ON;
    CONFIGURE CONTROLFILE AUTOBACKUP FORMAT FOR DEVICE TYPE DISK TO '/rman/V920/%F';
    CONFIGURE DEVICE TYPE DISK PARALLELISM 2;
    CONFIGURE DATAFILE BACKUP COPIES FOR DEVICE TYPE DISK TO 1;
    CONFIGURE ARCHIVELOG BACKUP COPIES FOR DEVICE TYPE DISK TO 1;
    CONFIGURE MAXSETSIZE TO UNLIMITED;
    CONFIGURE SNAPSHOT CONTROLFILE NAME TO '/rman/V920/snapcf_V92321.f';
    CONFIGURE CHANNEL 1 DEVICE TYPE DISK connect 'SYS/rac@node1';
    CONFIGURE CHANNEL 2 DEVICE TYPE DISK connect 'SYS/rac@node2';

 *  Review/Verify your new configuration.
 
    RMAN> SHOW ALL;

    CONFIGURE RETENTION POLICY TO REDUNDANCY 3;
    CONFIGURE BACKUP OPTIMIZATION OFF;
    CONFIGURE DEFAULT DEVICE TYPE TO DISK;
    CONFIGURE CONTROLFILE AUTOBACKUP ON;
    CONFIGURE CONTROLFILE AUTOBACKUP FORMAT FOR DEVICE TYPE DISK TO '/rman/V920/%F';
    CONFIGURE DEVICE TYPE DISK PARALLELISM 2;
    CONFIGURE DATAFILE BACKUP COPIES FOR DEVICE TYPE DISK TO 1;
    CONFIGURE ARCHIVELOG BACKUP COPIES FOR DEVICE TYPE DISK TO 1;
    CONFIGURE CHANNEL 1 DEVICE TYPE DISK CONNECT  'SYS/rac@node1';
    CONFIGURE CHANNEL 2 DEVICE TYPE DISK CONNECT  'SYS/rac@node2';
    CONFIGURE MAXSETSIZE TO UNLIMITED;
    CONFIGURE SNAPSHOT CONTROLFILE NAME TO '/rman/V920/snapcf_V92321.f';

 

5. Make a backup using the new persistent configuration parameters.

*  Backup database with differential incremental 0 and then archived logs   using the delete input option.

     BACKUP INCREMENTAL LEVEL 0 FORMAT '/rman/V920/%d_LVL0_%T_%u_s%s_p%p' DATABASE;
     BACKUP ARCHIVELOG ALL FORMAT '/rman/V920/%d_AL_%T_%u_s%s_p%p'DELETE INPUT;

 * Backup again using differential incremental level 1

     BACKUP INCREMENTAL LEVEL 1 FORMAT '/rman/V920/%d_LVL1_%T_%u_s%s_p%p' DATABASE;
     BACKUP ARCHIVELOG ALL FORMAT '/rman/V920/%d_AL_%T_%u_s%s_p%p' DELETE INPUT;

 * To simplify this from Oracle9i we can also use PLUS ARCHIVELOG 

    BACKUP INCREMENTAL LEVEL 0 FORMAT '/rman/V920/%d_LVL0_%T_%u_s%s_p%p'
        DATABASE PLUS ARCHIVELOG FORMAT '/rman/V920/%d_AL_%T_%u_s%s_p%p' DELETE INPUT;

 This uses a different algorithm than backup database and backup archivelog in separate commands, the algorithm for PLUS ARCHIVELOG is:

     1. Archive log current
     2. Backup archived logs
     3. Backup database level 0
     4. Archive log current
     5. Backup any remaining archived log created during backup

 
6. Backup set Maintenance using the configured retention policy

    RMAN> LIST BACKUP SUMMARY;
    RMAN> LIST BACKUP BY DATAFILE;
    RMAN> LIST BACKUP OF DATABASE;
    RMAN> LIST BACKUP OF ARCHIVELOG ALL;
    RMAN> LIST BACKUP OF CONTROLFILE;

   These above can be enhanced with the "until time" clause as well as the archivelog backups using "not backed up x times" to cut down on  many copies of a log in several backup sets.
   Then continuing with SMR  Server Managed Recovery use the change archivelog from...until...delete  to remove old logs no longer needed on disk.

  To check/delete obsolete backups  or archivelogs we use:

    RMAN> REPORT OBSOLETE;

    RMAN> DELETE OBSOLETE;
           or
    RMAN> DELETE NOPROMPT OBSOLETE;

  To check the database files:

    RMAN> REPORT SCHEMA; 

7. Restore and Recover

        Complete Recovery

  With the database mounted on the node1 and no-mount on node2 connect to the target and catalog using RMAN.

      rman target / catalog rman/rman@rcat

    This script. will restore and recover the database completely and open.
    All previous backup will still be available for use because there was not RESETLOGS command given.

     run {
             RESTORE DATABASE;
             RECOVER DATABASE;
             ALTER DATABASE OPEN;
          }

         Incomplete Recovery

If you are using instance registration the database must be mounted to register with the listener. This means you must use the current controlfile for restore and recovery or setup a dedicated listener if not  already done. RMAN requires a dedicated server connection and does not work with using instance registration before mounting the controlfile.
Using the autobackup controlfile feature requires the DBID of the  TARGET database. It must be set when the database is not mounted and only the controlfile and spfile (from 9.2) can be restored this way.

    1. Shutdown node1 and node2

    2. Startup no-mount node2 and node1

    3. Start rman

     rman trace reco1.log

     RMAN> CONNECT CATALOG rman/rman@rcat

     RMAN> SET DBID=228033884;

     RMAN> CONNECT TARGET

   4. Restore the controlfile from autobackup

     % rman trace recocf.log

     RMAN> SET DBID=228033884;

     RMAN> CONNECT TARGET

     RMAN> RUN
           {
             SET CONTROLFILE AUTOBACKUP FORMAT FOR DEVICE TYPE disk TO '/rman/V920/%F';
             ALLOCATE CHANNEL d1 DEVICE TYPE disk;
             RESTORE CONTROLFILE FROM AUTOBACKUP
                 MAXSEQ 5           # start at sequence 5 and count down (optional)
                 MAXDAYS 5;         # start at UNTIL TIME and search back 5 days (optional)
             MOUNT DATABASE;
            }

--此處應該有問題，用catalog後就不用設DBID了

  5. Verify what is available for incomplete recovery.

    We will recover to the highest SCN log sequence and thread. We will use the log sequence in this  case.

     The options are "until time", "until scn", or "until sequence".

   *  First we need to fine the highest sequence of each thread:

    SQL> select max(sequence#) from v$archived_log  where thread#=1;

       MAX(SEQUENCE#)
       --------------
               25

    SQL> select max(sequence#) from v$archived_log  where thread#=2;

       MAX(SEQUENCE#)
       --------------
             13

  * Next is to find the thread with highest  SCN ( NEXT_CHANGE# )

In this case the SCN is greater in thread 2 sequence 13 than in sequence 25  thread 1.   So we will use  thread 2.  We will set sequence 14  thread 2 for RMAN  'until sequence'  recovery,  because RMAN stops the recovery  before applying the indicated sequence. Log sequence for  recovery  needs always  be sequence+1 to end  at +1 after applying the prior sequence.

     SQL> select sequence#, thread#, first_change#, next_change#
                from v$archived_log
               where sequence# in (13,25);

      SEQUENCE#    THREAD# FIRST_CHANGE# NEXT_CHANGE#
     -------------------- -------------- ------------------------- -------------------------
                            25            1                           1744432                    1744802
                            13            2                           1744429                    1744805

      SQL> select sequence#, thread#, first_change#, next_change#
           from v$backup_redolog
           where sequence# in (13,25);

      SEQUENCE#    THREAD# FIRST_CHANGE# NEXT_CHANGE#
      -------------------- -------------- ------------------------- -------------------------
                            25            1                           1744432                    1744802
                            13            2                           1744429                    1744805

  ---        SET UNTIL SEQUENCE 14 THREAD 2  ----

  6. Get the sentences to add TEMPFILES after opening DB.

Locally Managed Temporary Tablespaces are not restored by  RESTORE command, we need to create them manually after recovery is complete.

If using LMT Temporary tablespace the controlfile will have the syntax  to add the tempfile after recovery is complete. The following command will give us the create controlfile sencence:

               SQL> alter database backup controlfile to trace;

                Example:

      # Commands to add tempfiles to temporary tablespaces.
      # Online tempfiles have complete space information.
      # Other tempfiles may require adjustment.
        ALTER TABLESPACE TEMP ADD TEMPFILE '/dev/db/rV92B_temp_01.dbf'  SIZE 41943040  REUSE AUTOEXTEND OFF;
      # End of tempfile additions.
      #

 7. Run the rman script

Since log sequence 13 thread 2 next_change# is 3 changes ahead of thread 1 sequence 25 we are using sequence 14 to stop recovery. This will restore  the data files and recover them completely using the online logs.

     run {
          SET UNTIL SEQUENCE 14 THREAD 2;
          RESTORE DATABASE;
          RECOVER DATABASE;
          ALTER DATABASE OPEN RESETLOGS;
          }

Finally add the tempfiles with sentences from step 6.

 8. Review and understand the impact of resetlogs on the catalog.

After resetlogs there are 2 incarnations in the recovery catalog. Only  one incarnation can be current at one time for a given DBID. The Inc Key  keeps track of the database incarnations.

RMAN> LIST INCARNATION OF DATABASE V920;

List of Database Incarnations
DB Key  Inc Key  DB Name  DB ID            CUR  Reset SCN  Reset Time
-----------  ----------- ------------ ----------------  ------- --------------  ---------------
2656          2657       V920         228033884      NO   1                   29-MAY-03
2656          3132       V920         228033884      YES  1744806       13-JUN-03

9. RMAN Sample Commands

* With a dedicated listener (not using instance registration)  restoring the controlfile.

 run {
 ALLOCATE CHANNEL D1 TYPE DISK CONNECT 'SYS/RAC@NODE1';
 ALLOCATE CHANNEL D2 TYPE DISK CONNECT 'SYS/RAC@NODE2';
  SET UNTIL SEQUENCE 14 THREAD 2;
   RESTORE CONTROLFILE;
  ALTER DATABASE MOUNT;
 RELEASE CHANNEL D1;
 RELEASE CHANNEL D2;
 }

 * Backup Archivelog

 BACKUP ARCHIVELOG ALL NOT BACKED UP 3 TIMES;
 BACKUP ARCHIVELOG UNTIL TIME 'SYSDATE-2' NOT BACKED UP 2 TIMES;

相关的 产品 Oracle Database Products > Oracle Database > Oracle Database > Oracle Server - Standard Edition Oracle Database Products > Oracle Database > Oracle Database > Oracle Server - Personal Edition Oracle Database Products > Oracle Database > Oracle Database > Oracle Server - Enterprise Edition

返回页首

来自 “ ITPUB博客 ” ，链接：http://blog.itpub.net/38267/viewspace-704870/，如需转载，请注明出处，否则将追究法律责任。

转载于:http://blog.itpub.net/38267/viewspace-704870/