Manual RAC DB creation on Windows.
Assumption:
1. The clusterware is installed and is up and running on both the nodes.
2. The Oracle Home is installed on both the nodes.
Steps:
1. First we need a shared storage to store the Oracle DB files, Redo Logs, Control Files etc. So we need to create ASM diskgroups on which the mentioned files can be accomodated. CREATE ASM DISKGROUPS
2. Create the pwd(orapwd)file and service(oradim) utility. Please note the service to be created for the instance names so, SID will change between the 1st and 2nd node.
3. Create the init.ora parameters as we do for normal single instance and create the DB using CREATE DATABASE statement. Please exercise caution while mentioning
Update the value in init.ora
At this point you ll have one instance fully functional in open mode.
4. Now make the necessary parameter changes to the init.ora.
Assumption:
1. The clusterware is installed and is up and running on both the nodes.
2. The Oracle Home is installed on both the nodes.
Steps:
1. First we need a shared storage to store the Oracle DB files, Redo Logs, Control Files etc. So we need to create ASM diskgroups on which the mentioned files can be accomodated. CREATE ASM DISKGROUPS
2. Create the pwd(orapwd)file and service(oradim) utility. Please note the service to be created for the instance names so, SID will change between the 1st and 2nd node.
3. Create the init.ora parameters as we do for normal single instance and create the DB using CREATE DATABASE statement. Please exercise caution while mentioning
- MAXINSTANCES
 - MAXLOGFILES
 - MAXLOGMEMBERS
 
3. Make sure to point the control file to the right file name. Just pointing the control_files parameter to a disk group will not help. After DB creation find the exact file name of the control_files from 
select * from v$controlfile;
Update the value in init.ora
At this point you ll have one instance fully functional in open mode.
4. Now make the necessary parameter changes to the init.ora.
- Add cluster_database=true
 - Add cluster_database_instance=n where n is the number of instances
 - Then add instance_name, instance_number, Undo_tablespace & Thread specific to instances
 
5. Create respective Undo tablespace & Threads for the second instance.
6. Copy the init.ora & pwd file to node 2 and start the instance. The instance should come up fine w/o issues.
7. Once started, stop the instances and add them to the OCR throught the srvctl command so that the instances can be controlled as a clustered resource.
8. Create a centric spfile and update the same in OCR so that the agent of the CRS knows which file to refer to start the DB.
Hope this helps!!!
No comments:
Post a Comment