This is an excerpt from the bestselling book Oracle Grid & Real Application Clusters.  To get immediate access to the code depot of working RAC scripts, buy it directly from the publisher and save more than 30%.

Also see how to add an instance to an existing Oracle server

The following steps can be followed to add instances to RAC nodes:

1. From any of the newly added and configured nodes, the additional nodes listener processes are configured using the Network Configuration Assistant (NETCA) utility. Choose the same port number and protocol used on the existing nodes. If the NETCA utility shows the existing node on the node list page, there is no need to select this node because it already has a listener configured on it.

2. Once the listeners are up, convert the database to RAC starting up the instances on the other nodes using one of the following procedures:

* Automated Conversion Procedure

* Manual Conversion Procedure

Automated Conversion Procedure

If a duplicate of the single instance database was created as described in the section Duplication of the Original Single Instance Database, the DBCA is used to complete the conversion to a multi-instance RAC database.

1. Start by executing the DBCA utility from the primary or initial node. On the node selection screen, select the names of the nodes that to be included as part of the cluster database. On the Template Selection screen, select the preconfigured template, which is the single instance duplicate, that was created earlier. Enter the database name for the RAC database and respond to the remaining DBCA prompts.

2. Enter the name of the raw device for the SPFILE on the Initialization Parameters screen to use raw devices for the cluster database files. On the Storage page, if a RAW file mapping file has not been specified, the default database file names should be replaced with the raw devices for the control files, redo logs, and datafiles to create the cluster database. Click the Finish button and create the database.

3. Once DBCA completes the conversion to the RAC database, the Password Management screen is displayed.  It is a requirement that the passwords for database users who have SYSDBA and SYSOPER roles be changed. Once the conversion process is complete, DBCA will exit.

Manual Conversion Procedure

If DBCA was not used to create a duplicate image of the single instance database, the following steps are performed complete the manual conversion to RAC:

1. For each new node, create an Optimal Flexible Architecture (OFA) directory structure to support the Oracle system. This structure should be the same on all of the nodes.

2. If there is a need to convert the single instance database files from a regular cooked file system to RAW devices, on UNIX/Linux copy the database datafiles, control files, redo logs, and server parameter file into corresponding raw devices using the DD command, or use the OCOPY command on Windows-based systems. If no conversion is required, continue on to the next step.

3. Use the ALTER DATABASE BACKUP CONTROLFILE TO TRACE command to create a control file recreation script in the location specified by the user_dump_dest initialization parameter.

4. Edit the control file recreation script with the REUSE keyword, a specification for maxinstances with the Oracle suggested setting being 32, maxlogfiles and such for a RAC database.

5. Recreate the control files by executing the script as explained in the script header.

6. If the single instance database is using an SPFILE parameter file, a temporary PFILE must be created from the SPFILE using the following SQL statement:

CREATE PFILE='pfile_name' from spfile='spfile_name'

(the file names should be full path names)

7. Shut down the database instance.

8. Edit the temporary parameter file and set the cluster_database parameter to TRUE, and set the instance_number parameter to a unique value for each instance, using the sid.parameter=value syntax.

9. At this point, the size of the SGA must be adjusted to avoid swapping and paging when converting to RAC. RAC requires around 350 bytes for each buffer to accommodate the Global Cache Service (GCS). Therefore, if there are 15,250 buffers, RAC requires about 350*15,250 bytes or about 5 MB more memory. Change the size of the SGA by changing the db_cache_size and db_nk_cache_size parameters accordingly.

10. Once the PFILE has been edited, use it to start up the database instance.

11. Once the instance is started, create an undo tablespace for each additional instance using the CREATE UNDO TABLESPACE SQL statement. If RAW devices are in used, then ensure that the datafile for the undo tablespace is on a RAW device.

12. Create a new redo thread that has at least two redo logs for each additional instance. If RAW devices are being used, ensure that the new redo log files are also on RAW devices. Enable the new redo threads by using the ALTER DATABASE SQL statement.

13. Shutdown the database instance using either a shutdown normal or shutdown immediate command.

14. Copy the Oracle password file from the source node to the corresponding location on the additional RAC cluster nodes. Make sure that the oracle_sid name is replaced in each password file name appropriately for each additional instance.

15. Edit the PFILE and add the remote_listener=listeners_db_name and sid.local_listener=listener_sid parameters.

16. Configure the net service entries for the database and instances and address entries for the local_listener for each instance and remote_listener in the tnsnames.ora file and copy it to all nodes.

17. Create an SPFILE from the PFILE using the command:

CREATE SPFILE='Spfile_name' from pfile='pfile_name'

(the file names should be full path names)

18. If a cluster file system is not in use, ensure that the SPFILE is on a RAW device.

19. Create a $oracle_home/dbs/initsid.ora file on UNIX-based systems or a %oracle_home%\database\initsid.ora file on Windows-based systems that contains the following entry:

spfile='spfile_path_name'
where spfile_path_name is the complete path name of the SPFILE.

20. Add the configuration for the RAC database and its instance-to-node mapping using SRVCTL

21. Start the RAC database using SRVCTL

Once the database has been started with SRVCTL, the conversion process is complete, and the following SQL statement can be used to see the status of all the instances in the converted RAC database:

select * from v$active_instances