Loading...

FAQ - Oracle RAC iDataAgent

How do we protect Oracle data on a local zone?

In order to protect Oracle data on a local/non-global zone, you need to install the Oracle iDataAgent on the non-global zone.

Can we protect data on non-global zones if the File System iDataAgent is installed only on the global zone?

Yes. We can protect file system data on non-global zones if the File System iDataAgent is installed only on the global zone. However, in order to enable consistent backups of application specific data on the non-global zones, you will need to install the corresponding application specific iDataAgent on the non-global zone.

Can we perform command line backups of Oracle RAC instances with different Oracle SID and database names?

Yes. Different SID and database names for Oracle RAC instances are supported for Oracle command line backups.

Can we perform command line backups if Oracle RAC database name and instance name (as defined in the CommCell Console) are different?

No. Whenever you run backups from the RMAN command line, ensure that the Oracle database name and the Instance Name as defined in the CommCell Console are the same, otherwise you will need to pass the name of the instance in the RMAN script as follows:

allocate channel ch1 type 'sbt_tape'
PARMS="SBT_LIBRARY=<software_install_path>/Base/libobk.so,ENV=(CvClientName=<rac_client_name>,CVOraRacDBName=<rac_database_name>,CvInstanceName=<instance_name>)"

where <rac_client_name> and <instance_name> are the names of the RAC Client and Instance (e.g., Instance001) on which the iDataAgent is installed.

On Windows clients for RAC, can we run multiple RMAN jobs simultaneously for the same instance from the CommCell Console?

No. You should always limit the RMAN jobs for the same instance to run one at a time in the CommCell Console and Scheduler.

Can we run third-party command line backup operations when a selective online backup job is running?

No. Third-party command line backup operations cannot be run when a selective online full backup job is running.

When does a non-full backup automatically get converted to a full backup?

  • When you select backup type as Full in CommCell Console and enter the backup type as Incremental on customize script, the job will run as Full.
  • When you select backup type as Incremental in CommCell Console and enter the backup type as Full on customize script, the job will run as Full. However, the Job Manager will treat this job as Incremental (on job controller and job history and job report). This job will be pruned as it reaches the data retention date.
  • When a storage policy is changed and if backup type is switched to full from job manager, the increment level will be changed to 0.
  • If you do not enter the backup type on customize script, then the system picks up the backup type selected in CommCell Console.

When does incorrect data transfer count displayed in job details?

When you resume a backup job, if the subclient content includes the SP file, the backup job gets restarted from the beginning and all the data that was backed up previously gets backed up again. This causes an incorrect display of data transferred count in job details.

How is the data retrieved during browse and restore operations?

When you browse or restore data, the browse function by default returns the requested data based on the latest backups available. This is usually the information that users are interested in. The browse function does this by using the current date and time as the effective date.

What are the tables that can be restored in Oracle 11g2?

You can restore tables under non-system tablespaces in Oracle 11g2.

How are table indexes handled during restore?

When you include the indexes during a table restore, note that only the index definition is restored and not the indexed data.

Can we manually import tables from the auxiliary instance?

Once the tables are restored to the auxiliary instance, you can either use the CommCell Console to export the data from the auxiliary instance and import it to a destination that you choose, or you can leave the data in the staging path and import the data manually by using Oracle import utilities.

How are stored procedures restored?

Stored procedures are restored from the Schema level. Schema is the collection of data objects created by the user to contain or reference their data. Hence, if one of the table within the schema is selected for restore, all the stored procedures for that schema will also get restored.

Can we restore tables when there is a schema change in the database?

No. Table restores are not supported when there are schema changes in the database.

What happens if one of the streams fail when running third-party command line operations with multiple streams?

Oracle third-party command line operations running on multiple streams will share the same Job ID in the Job Manager. If all the streams return failure, then the job is marked as failed. However, if one of the streams fail, it is submitted to the other stream for completion.

How are third-party command line jobs handled when Oracle services use a different service account instead of local system?

When Oracle Services use a different service account other than Local System, third party command line jobs are run with multiple job IDs. In such cases, use the following steps to enable third party command line jobs to run with a single job ID:

  1. From the Start menu, point to Administrative Tools, and then click Local Security Policy.
  2. Double-click Local Policies, and then double-click User Rights Assignment.
  3. In the details pane, double-click Adjust memory quotas for a process.
  4. Click Add User or Group.
  5. In the Enter the object names to select box, type the user or group name to which you want to assign the user right, and then click OK.
  6. In the details pane, double-click Replace a process level token.
  7. Click Add User or Group.
  8. In the Enter the object names to select box, type the user or group name to which you want to assign the user right, and then click OK.

How do we delete archive logs from a Flash Recovery Area that is protected?

When the Oracle database is configured to save archive logs in the Flash recovery area, backup jobs will fail if the Oracle RAC subclients have both the protect backup recovery area and the Archive Delete options enabled at the same time.

To avoid such backup failures, configure two different subclients, one to protect the backup recovery area and the other for enabling Archive Deletion for Archive logs.

How do we configure backups on Oracle Exadata Database machines?

In order to backup Oracle Exadata Machines, do the following:
  1. Install the Oracle iDataAgent on all the Exadata nodes.
  2. Create a RAC pseudo client that includes all the Exadata nodes.
  3. Run the backups on the pseudo client. The backups are performed using RMAN scripts.

    Since MediaAgents cannot be installed on Exadata database machines, they communicate with the Exadata nodes using infiniband network during the backup.

Can you restore from a RAC One Node in a multiple node configuration?

Yes. You can restore from a RAC One Node in a multiple node configuration.

In the case of emergency failover, the instance (same ORACLE_SID) is relocated to another physical node. The emergency failover is supported during a backup. Make sure that the active ORACLE instance and client combination are the same as configured in the CommCell Console for a control file or sp file restore.

For RAC One node configurations, use the Switch Database Mode for Restore check box (Options tab),  for restores.

How do I view the list of RMAN backup pieces from command line?

For information on how to view the RMAN backup pieces, see Viewing Oracle RMAN Backup Pieces

How do I restore multiplexed data if the number of drives is less than the allocated channels?

If the number of drives is less than the allocated channels, you can enable multiplexing and restore the multiplexed data to disk (application free restore).

For detailed instructions, see Restoring to a Disk (App Free Restore).

Why is an auxiliary instance being created on the source client when the destination client is using a table level restore?

If a destination client is selected without a user defined auxiliary instance, then an auxiliary instance is automatically created on the source client.

Can I delete the FRA content using the Archive Delete option when the backup does not include archive logs?

No. The flash recovery area (FRA) content is deleted based on the RMAN retention rules for FRA content, therefore it is not required to delete the FRA content manually.

Set the retention rules for the FRA content using the following command from the RMAN prompt with "n" representing the number of days

CONFIGURE ARCHIVELOG DELETION POLICY TO RECOVERY WINDOW of n days

When does the SnapProtect software use OS authentication (/) in place of the connect string?

The SnapProtect software uses OS authentication (/) and not the connect string for:

  • Control file restores
  • SP file restores
  •  Table level restores
  • Instance refreshes

You can use the connect string in place of OS authentication. For more information see Using the Connect String Instead of OS Authentication to Connect to the Oracle RAC Database

Are there additional steps to take when the Agent is installed on an existing client?

If you install the agent on on an existing client, you must run the cvpkgchg command to update the group permission after you install the agent. You can use this command to set the group and permission to the correct level which prevents the backup job from failing. For information on how to change the group permissions, see Changing UNIX Groups and Permissions on UNIX Clients.

Why does the CommCell Console show the instance status as UNKNOWN when the instance is in the NOMOUNT state and the instance is configured to use SYS or a user with DBA privileges?

 The SnapProtect software uses the connect string to connect to the instance and when a static listener is not configured and the instance is in the NOMOUNT state, the CommCell Console shows the instance status as UNKNOWN.