How to change the Exchange Server database paths on a cluster

This article describes how Microsoft Cluster Server processes registry changes, and describes how to ensure that values that are created by Performance Optimizer are retained in a clustered environment.

The information in this article does not apply to Microsoft Windows 2000 and Windows Server 2003 environments. If you delete the Mscs folder on the quorum drive in those environments the cluster does not function.

When Exchange Server 5.5 is running on a Cluster Server, the Cluster service keeps track of registry changes that are made to the Exchange Server services. If a registry change is made on one node, the cluster service copies these changes to a checkpoint file on the quorum disk. These changes are then copied to the other node. This only occurs when the Exchange Server service is running. If the Exchange Server service is not running and changes are made to the registry, these changes are overwritten with the values that are stored in the checkpoint files when the Exchange Server service is started. For more information, click the following article number to view the article in the Microsoft Knowledge Base: Because Exchange Server 5.5 Performance Optimizer shuts down all of the Exchange Server services, all of the changes that Performance Optimizer makes (for example, database paths) are overwritten with the old values that are stored in the checkpoint files when Performance Optimizer starts these services.

To ensure that the new values that Performance Optimizer creates are kept:

1. Shut down node 2.
2. Use the Services icon in Control Panel to stop the Cluster service on node 1.
3. Use the Services icon in Control Panel to start the Cluster service on node 1 by using the /fixquorum switch.
4. Use Cluster Administrator to verify that all of the resources are online in all groups. If not, bring the resources online.
5. Delete the MSCS folder on the quorum drive.
6. Run Performance Optimizer.
7. After Performance Optimizer is finished running and the Exchange Server resources are back online, verify there are no .cpt files in the MSCS folder on the quorum drive.
8. Use the Services icon in Control Panel to stop the Cluster service on node 1.
9. Use the Services icon in Control Panel to start the Cluster service normally (with no switches).
10. You can use Cluster Administrator to change dependencies on the affected services so that they depend on the correct disk. To do so:
    1. In Cluster Administrator, expand the Server container, and then click Active Resources. (You must do this for each node.)
    2. Double-click each of the Exchange instances, click the Dependencies tab, and verify that the correct resource dependencies are listed. For example, in the Exchange Information Store instance, the System Attendant is the correct resource dependency. Click Modify to make changes.
11. Use Cluster Administrator to verify that all of the resources are online in all groups. If not, bring the resources online.
12. Bring node 2 back online.
    
    NOTE: Sometimes the registry will not update. To force an update of the registry, you can Fail-over and Fall-back. To Fail-over, in Active Resources of node 1, right-click the System Attendant, and then click Initiate Failure. Wait for all "pending" actions to complete. To Fall-back, expand Groups, right-click the Exchange Virtual Server, and then click Move Group.

For more information on changing the Exchange 5.5 database or log file paths on an Windows 2000 cluster, click the following article number to view the article in the Microsoft Knowledge Base: