2 Replies Latest reply on Mar 25, 2012 4:39 AM by dmshimself

    Upgrade to 7.5

    lianne.redding Apprentice

      Hi

       

      Would anyone have a complete list of the ALTER TABLE xxxx DROP COLUMN xxxx statements that are executed by running the MDM to upgrade to 7.5?

       

      There are some legacy jobs implemented by the DBAs a long time ago that were running against the database which created statistics.  Therefore when I run the upgrade and one of the drop column statements is executed against a column that is referenced by a statistic the MDM errors.  I then have to restore the database, drop the statistic causing the issue and re-run the MDM.  I have done this around 10 times and it's getting a bit tedious!!!

       

      I appreciate this is an issue that has been caused here and is nothing to do with the MDM.  I asked for these jobs to be stopped as part of our last upgrade when I hit the same issue - unfortunatley the damage has already been done!  Last time i stuck with it and managed to compile a list myself

       

      Any help greatly appreciated

       

      Thanks

      Lianne

        • 1. Re: Upgrade to 7.5
          dgaughran Apprentice

          hi Lianne,

          you may already be aware of this but the mdm log will give you a blow by blow account include tables/columns changed/dropped see C:\WINDOWS\Temp\MetaDataManagerLog.xml. it will however only have the data up until the failure of the mdm changes to the database.

          rgds,

          dave

          • 2. Re: Upgrade to 7.5
            dmshimself ITSMMVPGroup

            It's pretty hard to do this as every site tends to have different objects/indexes.  The best I've managed is to turn database trace on before running MDM.  The trace produced has a lot of stuff in it that isn't in the metadata files, but even so each sites contents will tend to change..

             

            How about dropping all the stats?  There is usually a naming convention that would allow you to spot them.  As MDM recreates all the indexes, I suspect the value of any remaining stats would be quite low.

             

            I also thoroughly recommend a VMWARE DB server with a snap shot making it a bit quicker to get back to where you were?  (But you're probably doing that already)