Given what Microsoft told me, and based on my own experience, here’s what I did before performing the in-place upgrade in order to reduce the risk of problems to the lowest I could:
- I double-checked that all of the hardware (and their drivers) were on Microsoft’s Hardware Compatibility list for SQL Server 2000 clustering.
- I double-checked all log files, looking for potential problems of any kind.
- I took a copies of my most recent database backups on the SQL Server 7.0 cluster to be upgraded and restored them on another SQL Server 7.0 test server. I then ran all the appropriate DBCC commands on this restore. I did this for two reasons. First, to test my backups to be sure they worked successfully. In a worst case scenario I wanted to ensure that I could successfully recover my databases. And second, I didn’t want to run the DBCC commands on the production server as it would present too much of a load on that server, negatively affecting performance.
- While I didn’t have a test SQL Server 7.0 cluster available, I did have a test SQL Server 7.0 server (non-clustered) available. So after performing the previous step, I did an in-place upgrade of the SQL Server 7.0 test server databases to SQL Server 2000. While this wasn’t a complete test, it would be the closest I could get to the real conversion. Everything worked fine, which was reassuring.
After completing all of the above, I was ready for the actual upgrade.
Performing the Conversion
Here are the steps I followed to perform the actual in-place upgrade from a SQL Server 7.0 cluster to a SQL Server 2000 cluster.
- Prevent access by users to the SQL Server 7.0 cluster.
- Perform a full backup of each of the databases on the server to be upgraded. For best protection, these backups should be made to a server over the network, or to tape.
- Perform a test failover and failback of the nodes one last time before the upgrade to see if there are any cluster-related issues. Check the error logs to be sure there are no unexpected issues. Be sure that the node used to originally install SQL Server 7.0 clustering is the primary node after you have successfully performed your testing.
Once these preliminary steps were completed, the first major step of the in-place upgrade is to remove SQL Server 7.0 clustering from the current cluster. Here are the steps.
- Log into the primary node of the SQL Server 7.0 cluster using the account and password used for the mssqlserver service. (Recommended by Microsoft.)
- Start the SQL Server 7.0 Failover Cluster Wizard on the primary node of the cluster.
- Select the option to remove clustering.
- Once the SQL Server 7.0 Failover Cluster Wizard has competed removing the cluster, exit the Wizard.
- Reboot the primary node of the cluster. Then log into the primary node of the SQL Server 7.0 cluster using the account and password used for the mssqlserver service.
- Reboot the secondary node of the cluster. Wait until the secondary node fully restarts.
- Once the second node of the cluster has successfully rebooted, use Cluster Administrator on the primary node to verify that the primary node controls the resources of the Windows 2000 cluster. If all of the above steps have been successful, this will be the case.
Now that SQL Server 7.0 clustering has been removed from both nodes of the cluster, it is time to upgrade the SQL Server 7.0 files (they still remain on the primary node because only the clustering-related files were removed in the previous step).
In essence, in this step, you will be performing a stand-alone upgrade of SQL Server 7.0 to SQL Server 2000. Once this step is done, then you will convert the SQL Server 2000 stand-along version to a SQL Server 2000 clustered version. This section assumes that you are familiar with the general steps required to install SQL Server 2000.
- If you haven’t done so already, log into the primary node of the SQL Server 7.0 cluster using the account and password used for the mssqlserver service.
- From the SQL Server 2000 Enterprise CD, or from the local hard disk if the setup files are installed locally on drive C:, begin the SQL Server 2000 setup program.
- In the Welcome dialog box, click Next.
- When asked to select the SQL Server 2000 components to install, select all of the appropriate options for your unique situation.
- In the Computer Name dialog box, leave the defaults.
- In the Installation Selection dialog box, select “Upgrade, remove, or add components to an existing instance of SQL Server.”
- In the Instance Name dialog box, leave the defaults.
- In the Existing Installation dialog box, select “Upgrade your existing installation.”
- In the Upgrade dialog box, leave the defaults.
- In the Connect to Server dialog box, select the appropriate authentication mode.
- In the Start Copying Files dialog box, click Next. This starts the upgrade process. When done, you will be prompted to restart the computer. Go ahead and do so. As you might expect, this will cause the Windows 2000 clustering to fail over to the second node.
- Once the primary node of the cluster has successfully rebooted, log on using the mssqlserver service account and password, and then use Cluster Administrator to move the cluster resources from the secondary node back to the primary node.
-
Continues…