Well, this is not a promising start. I went to uninstall the UnusedSQLCLuster. I got errors (apparently there was an application installed (IBM CDC) that was holding the master.dbf open, and also hanging on to a clustered disk resource. After the uninstall SQL instance failed, I finally figured out it was probably the CDC app, and uninstalled it. And now the uninstall SQL instance shows no instance.
Of course, Failover Manager *does* show the instance, and the AD account is not disabled. <sigh> Now, in this particular case, I can (probably) live with this, since I am getting ready to nuke the box entirely. I can - I think - just delete the AD account of the SQL instance, with no ill effects, especially since I am not planning on reusing that particular name for a SQL instance in future (HOPEFULLY). Sound about right? I'm about to evict the node from the Windows cluster (just to be complete). Then gonna nuke the box, and make a new Win2012 R2 cluster, and then a SQL 2012 R2 cluster, with a completely different name. On Thu, Nov 2, 2017 at 10:41 AM, Michael Leone <oozerd...@gmail.com> wrote: > Looking for some advice. I've just been told that in a couple weeks, > I've got some cluster re-arranging to do. (Mind you, I've had a week's > vacation scheduled for the last 6 months, so I really only have 1 week > to do this ...) > > Here's what I have now: > > Windows 2008 R2 / SQL 2008 R2 cluster (call it "UnusedSQLCluster) > - This is a 1 node cluster, at the moment. > Windows 2008 R2 / SQL 2008 R2 cluster with 3 named SQL instances > - This is a 2 node cluster > > So here's what they want. > > Get rid of UnusedSQLCluster; reformat it as a Win2012 R2/SQL 2012 R2 cluster. > Move one of the named SQL instances to what used to be > UnusedSQLCluster but *keeping* the current instance name and IP > address. > > Got all that? > > So here's what I think I need to do. > > Gracefully uninstall the old UnususedSQLCluster (both SQL and Win clusters); > (This should be relatively straightforward - uninstall SQL node; evict > Windows node) > reformat it and install Win 2012 R2/SQL 2012 R2, using a temporary SQL > cluster name - "TempSQLCluster". > > Uninstall SQLInstance01 from the cluster it is in. > (this cluster remains in production, servicing the 2 other SQL instances) > > Rename TempSQLCluster to SQLInstance01, using same IP address, > > Have I missed a step? > > BIG QUESTIONS / CONCERNS: > > 1. It is possible to rename a SQL cluster instance name to something > else, that looks straightforward enough. And there is no log shipping, > event forwarding, etc, that I know of. I know I may have to do some > DNS cleanup. > > 2. I know that SQL cluster instances register themselves into AD, > hence why I want to gracefully uninstall everything. But re-using the > same SQL instance name should be OK? Or do I need to clean that object > out of AD after uninstalling it? > > I did this once, months ago, and I know I didn't do all the steps > properly at that time, as my boss deleted the AD account of a SQL > instance without uninstalling it first. So it's still there in SQL, in > a zombie state. > > What have I missed? What do I need to look out for? (DNS, etc) > > I'm sure I didn't explain everything clearly enough, so feel free to > interrogate further.
