< Previous  | TOC | Index | Next  >

Creating Oracle 10g database on CFS

Creating a starter database on CFS involves the following tasks:

• Creating a Cluster File System for the Oracle 10g database
• Running the dbca utility for cluster file system

Creating a Cluster File System for the Oracle 10g database

If you plan to use a cluster file system to store the Oracle database, you can use the following procedure to create the file system.

 To create Cluster File System for database tablespaces

1. Create a disk group (for example: rac_dg):

   # vxdg -s init oradatadg c4t2d3

2. Create a single shared volume (for example: oradatavol), large enough to contain a file system for all the tablespaces (see Oracle documentation the tablespace sizes). Assuming 6.8 GB are required for the tablespaces:

   # vxassist -g oradatadg make oradatavol 6800M

3. Deport and import the group in the shared mode to enable I/O fencing:

   # vxdg deport oradatadg

   # vxdg -s import oradatadg

4. Set the activation mode (sw) to allow shared access to the disk group:

   # vxdg -g oradatadg set activation=sw

5. Start the volume in the disk group:

   # vxvol -g oradatadg startall

6. Create a VxFS file system in this volume. From one node, enter:

   # mkfs -F vxfs -o largefiles /dev/vx/rdsk/oradatadg/oradatavol

7. Create a mount point for the shared file system:

   # mkdir /oradata

8. From the same system, mount the file system:

   # mount -F vxfs -o cluster /dev/vx/dsk/oradatadg/oradatavol \

   /oradata

9. Set "oracle" to be the owner of the file system, and set "755" as the permissions:

   # chown -R oracle:oinstall /oradata

   # chmod 755 /oradata

10. On the other system(s), do step 4 and step 7 through step 9.

    You can now create the database; refer to Oracle documentation.

Running the dbca utility for cluster file system

As oracle user, use the dbca utility on the master node to create a general purpose database on a cluster file system. This utility is a graphical user interface and requires setting the DISPLAY environment variable.

 To create a database on cluster file system

1. Make sure an oracle account is created on all nodes.
2. Verify that remsh works among all the nodes under the oracle account.
3. From one node, log in as oracle user.
4. Create and start the Oracle listener using NETCA utility.

   Refer to the Oracle9i Installation Guide for more information.

   $ netca

5. Run the dbca utility. When starting the utility with a cluster file system, use the -datafileDestination option to specify the mount point.

   For example:

   $ dbca -datafileDestination /rac_ts

   Refer to the Oracle documentation for details on the dbca utility.