how to "de-cluster" SQL 2000 and SQL 2005?

Question

We have a Windows 2003 cluster with one each of SQL 2000 and SQL 2005 instances, and the hardware is totally overwhelmed with the current workload.  We have new hardware, but want to start out with just an SQL 2008 installation; that has to be postponed because some client applications need to be upgraded as well.  In order to give our existing 2000/2005 applications a shot in the arm, we want to virtualize the SQL cluster with Hyper-V.  We've had considerable success with this transition for other platforms.

The only supported method for virtualizing a cluster is with the clustered volume resources on an iSCSI storage array, which we don't have.  I ran one of the cluster nodes through the System Center-VMM wizard, producing VHD -based virtual disks, and none of the clustered resources came up.  This was to be expected, as the disk signatures and "hardware" was different.  No amount of hackery would allow me to bring up the cluster and thus the SQL instances.

I think the best chance we have at making this work is to de-cluster the SQL instances, i.e. to make the SQL instances independent of the cluster service.  What is the most expedient way to do this?  After P2V, I have all the disks with the same drive letters, so all the paths would be the same.  I know that both SQL 2000 and 2005 are cluster aware, and there are dependencies such as databases being on clustered volumes.  Can I start and run the SQL services if the cluster service isn't running?  What if I uninstall the cluster service?  Will there be any stability issues?  Can I uninstall and reinstall SQL Server, and re-use the master database?  Any other ideas?

Answer 1

Without the Cluster Service, a Clustered SQL Server does not start. You need to uninstall your clustered instances and install stand-alone instances if you do not want your SQL Server instances to be non-clustered.

Regarding to the system databases, I'm not really sure; but you can move your user databases with no problem. You can also transfer your Logins and Jobs using SSIS or some other scripts (to move the Logins with their passwords etc.) which can be found by Googling.

---

Ekrem Önsoy - MCDBA, MCITP:DBA & DBD, MCSD.Net, SQL Server MVP

Answer 2

Hi

If you wanted to avoid recreating the system objects (logins etc) you could install the binaries on the new VM (same SQL version and SP, with the same drive-letters and install path) and then just copy the data - and log - files from the cluster (after you take the SQL resource offline). Bring up the SQL Service on the VM and it should all work. You'll need to change the servername though internally (sp_dropserver etc) if the name changes. Otherwise, it should just work.

HTH

Ewan

---

If you have found this post helpful, please click the 'Vote as Helpful' link (the green triangle and number on the top-left).

If this post answers your question, click the 'Mark As Answered' link below. It helps others who experience the same issue in future to find the solution.

Answer 3

I can re-use master/model/msdb from the clustered instances?  I'll try that, then.