Conference decwet::winnt-clusters

761.0. "Any way to clean things up?" by MPOS01::naiad.mpo.dec.com::mpos01::cerling (I'[email protected]) Tue Apr 22 1997 16:39

	Is there any way to clean out all references to certain databases from
	the master database without reinstalling?  I am having some problems
	getting a database enrolled on V1.1.

	I had the cluster up and running.  I had created databases and enrolled
	them.  Everything appeared to be working fine.  I started having some
	weird hardware problems on the disk that the shared database was on.
	I ended up with a system where the cluster thought I had an enrolled
	database, but SQL did not recognize any database.  There was no device
	and no database.

	So I uninstalled the cluster software, deleted what was left after the
	uninstall, and removed all reference from the Registry.  Reinstalled 
	the cluster software.  It recognized that SQL was installed.  Now,
	however, when I try to create a shared database, it never shows up to
	be enrolled.  I used the sp_fallback_help to see what SQL thought was
	going on and it listed one database that I had tried to enroll, and it
	never showed.  It did not list another database that I had tried to 
	enroll, and never showed.  So I did an sp_fallback_enroll_svr_db on
	the database I was trying to enroll.  It shows up in the list from
	sp_fallback_help, but it does not show in the cluster administrator.

	Therefore, I would like to see if I can clean the master db up so that
	I can start over.  If not, I guess I will go back, deinstall clusters
	and SQL, then reinstall SQL and clusters.  Even though it is a demo
	system, I would rather not do this.

tgc

    I really struggled trying to install a SQL database on seperate disks
    servered by seperate servers.  I had duplicate enrolled databases, the
    database enrolled on the primary server, and not on the failover
    server. The database enrolled on both systems but would not failover.
    No amount of running any of the fallback stored procedures helped at
    any time apart from telling me what I already knew.
    
    I eventually resolved the problems by uninstalling both Clusters and SQL 
    from both nodes. Reinstalling both and recreated the databases on their 
    respective parent node. I then failed over the failover node group onto 
    the primary server enrolled the database and then failed back.  I have had
    no problems since.  I do several demos a week and failover by the 
    three finger salute and power failure.
    
    Hth,
    
    
    /Shaun
                                                 

Here's the sequence we used at Wizard's to clean up 
between lab sessions:

1) uninstall Clusters 1.1
2) rebuild SQL Master Database
   (available thru Start -> Programs -> Microsoft SQL Server 6.5
    -> Setup)
3) reapply Microsoft SQL 6.5 Service Pack
4) install Clusters 1.1

Somewhere in there you should delete the database-related files 
created on the shared storage.  I typically do that before I
deinstall Clusters, but I don't think the order of this particular
step is important.