ETERNUS SF Recovery Manager for Oracle EE 13.0.1 User's Guide - Solaris(TM) Operating System -

Contents

Chapter 4 RECOVERING DATABASES

This chapter explains the recovery function of Recovery Manager for Oracle.

4.1 Overview 

This section gives an overview of the recovery procedure.

When a database is damaged, it can be recovered as follows:

1. Confirming the status of the damage
   

   When a database is damaged, it is necessary to assess which file has been damaged.
   

   1. Checking the alert log file
      

      When a database is damaged, information is output to the alert log (alert_ORACLE_SID.log). Refer to the alert log and check which file has been damaged.
      

   2. Checking the database

      Execute the DB check from the Recovery Main Menu screen to identify which file has been damaged.

2. Recovering the database

   After identifying the damage, execute the recovery operation from the recovery screen.

   Recovery Manager for Oracle provides the following recovery functions:

   Recovery function	Explanation
   Auto recovery	Determines the type of database damage and executes the appropriate restore and recover function.
   Full recovery	Executes the restore of control files and all data files, and executes the recovery function.
   Custom recovery	Executes the recovery of files selected on-screen. This database recovery function is suited to advanced users.

   

   Recovery Manager for Oracle supports disk-to-disk database restore and recovery operations. Restoring and recovery from a tape is not supported. If a backup volume on a disk is damaged, restore the volume from a tape using backup software prior to activating the recovery screen on the Recovery Manager for Oracle.

3. Confirming database recovery
   

   After the recovery operation has ended normally, check the transaction data.

4.2 Recovery Main Menu screen

This section explains the Recovery Main Menu screen.

1. Starting the Recovery Main Menu screen
   
   
   1. Log in as root on the X Window System.
      

      Log in to the database server*1 as root.
      

      *1: Log in to the primary node in the cluster operation (HA and RAC)

   2. Start the Recovery Main Menu screen
      

      Execute the "/opt/FJSVrmfo/bin/rmforecovery" command from the terminal emulator or the file manager.

      An example of command execution from the terminal emulator is shown as follows:

      [Starting the terminal emulator]

      

      [Activating the rmforecovery command from the terminal emulator]

      

   3. The Recovery Main Menu screen displays.

      [Recovery Main Menu screen]

      

2. Explanation of the Recovery Main Menu screen
   
   
   1. Database list

      The database list displays the databases already configured for backup and recovery operations.
      

      Select a database on which "DB check" or "recovery" is to be performed.

   2. DB check

      This operation confirms whether the database selected from the "Database list" is normal or damaged.

   3. Recovery Database

      This operation recovers the database selected from the "Database list".

4.3 DB Check

This section explains the DB check.

The DB check operation checks the status of files in the database (control files, online redo log files and all data files).

• If the database is not open or the control file is damaged, the database cannot be checked correctly.
• When the online redo log file comprises multiple members, and some of those members are damaged, the damage may not be detected correctly.
• Damage to the temporary tablespace (TEMPFILE) may not be detected correctly.
• Any tablespace found to be in offline status will be turned online forcibly.

4.3.1 DB Check Procedure 

This section explains the procedure for DB check as follows:

1. Starting the Recovery Main Menu screen

   Start the Recovery Main Menu screen as directed in the section, "Starting the Recovery Main Menu screen".

2. Confirming activation of the listener for target database connection

   Confirm that the listener for target database connection has been activated.
   

   Log in as the Oracle Install user of the target database, and confirm, using the "lsnrctl status" command, that the listener has been activated. Check if the target database can be connected to by specifying the connect strings with the "tnsping" command.

   Example: Checking the listener (when the listener name is "listener")

   % id uid=1001(ora920) gid=1001(dba) % lsnrctl status listener LSNRCTL for Solaris: Version 9.2.0.5.0 - Production on 13-APR-2005 11:46:28 Copyright (c) 1991, 2002, Oracle Corporation. All rights reserved. Connecting to (DESCRIPTION=(ADDRESS=(PROTOCOL=TCP)(HOST=rmfo200)(PORT=1521))) : <snip> : Listening Endpoints Summary... (DESCRIPTION=(ADDRESS=(PROTOCOL=tcp)(HOST=rmfo200)(PORT=1521))) Services Summary... Service "ora920" has 1 instance(s). Instance "ora920", status UNKNOWN, has 1 handler(s) for this service... The command completed successfully

   
   

   Confirm that the listener service has been activated for the target database as shown above.

   Example: Checking "tnsping" (when the connect string for the target database is "ora920")

   % id uid=1001(ora920) gid=1001(dba) % tnsping ora920 TNS Ping Utility for Solaris: Version 9.2.0.5.0 - Production on 13-APR-2005 11:49:15 : <snip> : Used TNSNAMES adapter to resolve the alias Attempting to contact (DESCRIPTION = (ADDRESS_LIST = (ADDRESS = (PROTOCOL = TCP)(HOST = rmfo200)(PORT = 1521))) (CONNECT_DATA = (SERVICE_NAME = ora920))) OK (10 msec)

   Confirm, using '"tnsping", that the target database can be connected to as shown above.

3. Performing the DB check
   

   Select the "Database name" subject to the check from the database list and "DB check" on the Recovery Main Menu screen, and click the [Start] button.

   [ DB Check Executing screen]

   

4. Confirming the DB check result

   When the DB check has completed the screen will display whether or not the database is normal or abnormal.
   

   1. When the DB check result is normal

      When the DB check result is normal, the following DB Check Result screen will be displayed.

      

      To terminate the processing, click [Exit].

   2. When the DB check result is abnormal

      When the DB check result is abnormal, the following DB Check Result screen will be displayed.

      

      When the tablespace check result is [NG], the damaged tablespace can be checked by clicking [Show Detail].

      

      To recover the damage immediately, click [Recovery]. To terminate the processing, click [Exit].

      

      Depending on the extent of the database damage, executing the DB check may cause the instance to terminate.
      

      Prior to executing the recovery operation, follow the procedures in the sections, "Confirming activation of the listener for target database connection" and "Pre-check in the cluster operation (HA and RAC)". If an error is found, remove the cause of the error then execute the recovery operation.

5. Confirming completion

   From the DB Check Result screen, click [Exit] and the following screen will be displayed:

   

   To terminate the processing, click [Yes]. To redisplay the DB Check Result screen without terminating the processing, click [No].

4.3.2 Removing DB Check Execution Error  

This section explains the action to take in the event of common errors generated during the DB check.

1. Cannot connect to database

   The listener may not have been activated if the following screen is displayed. Follow the procedure to confirm in the section, "Confirming activation of the listener for target database connection", activate the listener, and re-execute the DB check.

   

2. Cannot perform DB check as DB has not been activated
   

   The database may not have been activated or damaged if the following screen is displayed.
   

   Activate the database and re-execute the DB check if the database has not been activated.
   

   Execute the recovery operation if the database has been damaged.

   

4.4 Recovering Databases 

This section explains how to recover databases.

The procedure for recovering databases is explained as follows:

1. Starting the Recovery Main Menu screen

   Start the Recovery Main Menu screen as directed in the section, "Starting the Recovery Main Menu screen".

2. Confirming activation of the listener for target database connection

   Confirm by following the same procedure as described in the section, "Confirming activation of the listener for target database connection" in "DB check".

3. Pre-checking for the cluster operation (HA and RAC)
   

   Confirm the status of each Reliant(R) Monitor Software (hereafter referred to as "RMS") resource through the Global Cluster Services of the PRIMECLUSTER Web-Based Admin View.

   [Items to confirm]
   

   1. Are the disks with the target database volume and the backup volume visible from the node executing the recovery operation?
      
   2. Has the listener for target database connection been activated?
      
   3. When the OPC/EC is in use, is the ACM logical host (inheriting host) the node that executes the recovery operation?
      

   If any of the above items is abnormal, recover the cluster resources.
   

   For details on how to recover the cluster resources, refer to the "PRIMECLUSTER Web-Based Admin View Operation Guide".

4. Recovery

   Select the "Database name" to be recovered from the database list and "Recovery Database" on the Recovery Main Menu screen, and click the [Start] button.

   The following recovery screen will be displayed:

   

   [Recovery type]

   Auto (auto recovery)	Determines the type of database damage and executes the appropriate restore and recovery function.
   Full (full recovery)	Executes the restore of control files and all data files, and executes the recovery function.
   Custom (custom recovery)	Executes the recovery of files selected on-screen. This database recovery function is suited to advanced users.

   [Button]

   Recovery point	Displays the Recovery Point screen.
   Start	Executes the recovery processing.
   Exit	Exits the application.
   Show Detail	Displays the details of the DB check result.

4.4.1 Auto Recovery

The auto recovery function assesses the status of the database damage, and executes the restore and recovery operation according to the status.

The recovery operation may not be performed normally under the following circumstances:

• when the listener is down
• when the online redo log file comprises multiple members and only some of the members are damaged
• when the temporary tablespace (TEMPFILE) is damaged
• Oracle RAC 10g

1. Executing auto recovery

   Select [Auto] as the recovery type on the Recovery Main screen, and click [OK].

   

2. Confirming continuation of recovery operation

   A screen confirming the continuation of the recovery operation will be displayed.

   

   To execute the recovery operation, click [Run]. To cancel, click [Cancel].

   

   The data cannot be recovered to the status immediately before the database damage if the current online redo log file has been damaged, or any of the archived redo log files generated after backup are missing.

3. Recovery Executing screen

   The Recovery Executing screen will be displayed.

   

   Wait until the recovery processing has completed.

4. Recovery Completion screen

   The following screen will be displayed when the recovery processing has completed.

   

   To display the execution log, click [Show Log]. To terminate, click [Exit].

   

Depending on the status of database damage, the redo log may sometimes be reset (resetlogs). If the redo log is reset, back up the database immediately because resetting invalidates the existing backup of the database.

4.4.2 Full Recovery

The full recovery function restores all the control files and data files, and executes the recovery operation.

1. Executing full recovery

   Select [Full] as the recovery type on the Recovery Main screen, and click [OK].

   

2. Confirming continuation of recovery operation

   A screen confirming the continuation of the recovery operation will be displayed.

   

   To execute the recovery operation, click [Run]. To cancel, click [Cancel].

   

   The data cannot be recovered to the status immediately before the database damage if the current online redo log file has been damaged, or if any of the archived redo log files generated after backup are missing.

3. Recovery Executing screen

   The Recovery Executing screen will be displayed.

   

   Wait until the recovery processing is completed.

4. Recovery Completion screen

   The following screen will be displayed when the recovery processing has completed.
   

   To display the execution log, click [Show Log]. To terminate, click [Exit].

   

In the full recovery operation, the redo log will always be reset (resetlogs). Back up the database immediately because resetting invalidates the existing backup of the database.

4.4.3 Custom Recovery

The custom recovery function executes the recovery operation for the files and the method both selected on-screen. This database recovery function is suited to advanced users. If the files and the method of recovery are not specified correctly, the recovery operation will not be performed normally.

Those who are not familiar with the database backup and recovery function should select "Auto recovery" or "Full recovery".

1. Executing custom recovery
   

   Select [Custom] as the recovery type on the Recovery Main screen.

   

2. Selecting the database file to be recovered
   

   Click the tabs [Control File], [Redo Log File], and [Tablespace], and select files to recover.

   1. Control file

      

      [Function]

      None	Does not recover the control files.
      Restore	Recovers the control files.

      [Button]

      Reset

      Cancels the function selection.

      

      When the control files are recovered, the redo log will always be reset (resetlogs). Back up the database immediately because resetting invalidates the existing backup of the database.

   2. Redo log file

      

      [Function]

      None	Does not recover the redo log files.
      Restore	Recovers the redo log files.

      [Button]

      Reset

      Cancels the function selection.

      

      • When the redo log files are recovered, the redo log will always be reset (resetlogs). Back up the database immediately because resetting invalidates the existing backup of the database.
      • When [Restore] is selected for the redo log files, select ether the temporary tablespace (TEMPFILE) from "All Tablespaces" or "Tablespace Name" too.

      
      

   3. Tablespaces

      All Tablespaces, Tablespace Specification, or Volume Name can be selected.

      

      [Function]

      All Tablespaces	Recovers all tablespaces.
      Tablespace Name	Recovers the selected tablespaces.
      Volume Name	Restores the selected volume by the OPC and recovers it Note: Can be selected only when the Data file backup method (BK_TYPE) is OPC/EC.

      [Button]

      Reset

      Cancels the function selection.

      

      • In the recovery operation where Tablespace name has been selected, when the Data file volume type (DBF_TYPE) is a file system, data files will be restored by RMAN (not by OPC) even if the Data file backup method (BK_TYPE) is OPC/EC.
        
      • When the temporary tablespace (TEMPFILE) has been selected for the recovery operation by Tablespace Specification, TEMPFILE will be recreated.

      When "Recovery by selecting the tablespace name" is selected and the tablespace is other than the "system tablespace, undo tablespace, or temporary tablespace (TEMPFILE)", the following inquiry will be made:

      

      To recover online without terminating the database, select [Online Recovery] and click [OK]. To recover offline by terminating the database, select [Offline Recovery] and click [OK].

3. Confirming continuation of recovery operation

   A screen confirming the continuation of the recovery operation will be displayed.

   

   To execute the recovery operation, click [Run]. To cancel, click [Cancel].

4. Recovery Executing screen

   The Recovery Executing screen will be displayed.
   

   Wait until the recovery processing has completed.

5. Recovery Completion screen

   The following screen will be displayed when the recovery processing has completed.

   

   To display the execution log, click [Show Log]. To terminate, click [Exit].

4.4.4 Recovery Point 

The following screen will be displayed when [Recovery Point] is clicked on the Recovery Main screen.

[Function]

*1: The range of selection varies depending on the value of the number of backup generation managements (BK_GEN)

For 1, only the Latest Generation can be selected.

For 2, the Latest Generation or the Previous Generation (the Latest Generation only for EC) can be selected.

For 3, all (the Latest Generation and the Previous Generation only for EC) can be selected.

Specification must be within the specification range.

*2: Always specify "1" as the Thread for the single server or cluster operation (HA).

*3: Logs up to the specified log sequence number - 1 are applicable.

[Button]

*1: Displays the number of lines (from the bottom) specified in ALT_ROWS of the backup basic information file (rmfo_bkcom.ini).

4.5 Checking Recovery Result 

Confirm that the recovery operation has ended normally from the Recovery Completion screen and the recovery execution log files.

4.5.1 Recovery Completion screen 

1. Recovery Completion screen at normal ending

   The following screen will be displayed when recovery processing ends normally:

   

2. Recovery Completion screen at error ending

   The following screen will be displayed when recovery processing ends in error:

   

   Refer to the execution log and remove the cause of the error. Then re-execute the recovery operation.

4.5.2 Execution Log file 

Detailed information is output to the log file at the time of recovery operation.

Log files are created under the "/opt/FJSVrmfo/log/recovery/DB_NAME" directory.

The log file names are as follows:

YYYYMMDDHHMISS: Execution time (year/month/day/hour/minute/second)

When the recovery operation has ended in error, refer to the log file and remove the cause of the error. Then re-execute the recovery operation.

• The message "The recovery operation ended abnormally" may be output even when the database has been recovered successfully because the DB check performed after the recovery generated an error. Ensure the contents of the execution log are checked.
• The DB check error may be caused by "the listener is down" or "The cluster resource became abnormal.". Reactivate the listener or recover the cluster resource, re-execute the DB check only, and confirm that the database has been recovered normally.
• Recovery processing invalidates the backup data collected so far. Back up the database immediately.
  

Example: Execution log showing the recovery operation to be normal but the DB check to be in error.

4.6 Recovery Post-Processing 

1. Checking transaction data

   When the recovery operation has ended normally, check the transaction data

2. Checking for the cluster operation (HA and RAC)
   

   Confirm the status of each RMS resource through the Global Cluster Services of the PRIMECLUSTER Web-Based Admin View for the cluster operation (HA and RAC).
   

   Recover the cluster resources if the resources are abnormal.
   

   Oracle resources may be in the warning status. In this case, ether execute the "hvoraenable" command for the each resources concerned and re-open the resource monitor function, or restart (OFFLINE/ONLINE) all the Oracle user applications from the Global Cluster Services screen of the PRIMECLUSTER Web-Based Admin View.
   

   For details on how to recover the cluster resources, refer to the "PRIMECLUSTER Web-Based Admin View Operation Guide".
   

   For Oracle RAC 10g, confirm using the server control utility (SRVCTL) that the status of the databases, instances, and cluster database services are normal.
   

   For details on and how to use the server control utility (SRVCTL), refer to the manual "Oracle Real Application Clusters Management"

3. Executing backup

   Back up the database immediately.

Contents