Loading
  • Hey Laim,

     

    Are you seeing gradual growth or has it suddenly started to grow and not shrink (when the garbage collector kicks in) ?

     

    I ask this because I recently had a huge issue with this particular table.

    On one platform we noticed that, over the space of a couple of weeks, it had blown from 65 million rows to over 550 million and the underlying DB more than doubled.

    The underlying issue was that the garbage collector had stopped working and the whole system was slowly grinding to a halt.

    I have some scripts we ran to get things back working but I don't really want to side-track the discussion if its not a garbage collector issue.

     

    Tom

     

     

     

     

     

     

    Expand Post
    • Hey Tom,

       

      We've seen a huge increase in the last few weeks that blew up out of nowhere. We do have a support ticket open for one of the platforms that I believe they've mentioned the Garbage Collector so I may go back and request those scripts. We're planning to do some shifting around of platforms eventually but even if we had it working for a period of time while we do that, it means it lowers the urgency ever so slightly to get the migration(s) done!

       

      Out of curiosity, is this a standalone platform or an SPE platform?

       

      Thanks,

      Laim

      Expand Post
      • Hi Laim,

         

        It is an SPE platform, which has been around for 5+ years - .

         

        Here are some areas for you to look at.

        If you decide to cleanup using scripts you need to ensure you have adequate backups and time to do this etc.

         

        Cleanup of inv.datasoftware

        As far as I understand it each computer software item will be logged daily with a new sequence number.

        The garbage collector is supposed to effectively clear out software entries for computers with old sequence numbers.

        Snow Support helped with the creation of a script to manually delete huge swathes of computer/sequence numbers. We started with small batches and gradually increased the batch size. I ran this script at off peak times only when the DUJ and other processor hog tasks weren't running.

         

        Check/Cleanup of the inv.UsersForDeletion table

        I secretly suspect that this was the root cause of my garbage collector issue.

        Under normal circumstances this table grows and and shrinks every day when the GC runs.

        It should normally have 0 rows at some point during the day.

        When I started looking tbl.software issue I had 182000 unprocessed users for deletion.

        These are effectively user accounts that the GC shoud have processed - but couldn't.

        Snow Support helped with the creation of a script to instantly delete huge swathes of users from this list.

        We performed this while garbage collection was disabled in SnowInventory.

        This took days - but once we had cleaned out all references to these users the GC started working again.

         

        Check your inv.ClientsForDeletion table

        Under normal circumstances this table also grows and shrinks every day when the GC runs.

        It should normally have 0 rows at some point during the day.

        Snow Support helped with the creation of a script to run the deletion process for computers that should be deleted. We ran this script off peak, while garbage collection was disabled in SnowInventory.

        This took many days - but once we had cleaned out all references to these users the GC started working again.

         

        Cleanout non-current user accounts

        This is on a long running SPE tenant. I was surprised that there were hundreds of thousands of user accounts listed in inv.DataUserAccount which were for non-current customers.

        In fairness some of these were for customers from Inventory 3 days - but they were all still hiding there.

        I guess I'd always thought that the system was self cleaning - but I was wrong.

        I adapted the user deletion script mentioned above to delete all the non-current users (from old sites and customers) from this table.

        We ran this script off peak, while garbage collection was disabled in SnowInventory.

        This also took days.

         

        We also ran scripts to cleanout inv.DataDeviceDriver and inv.DataNTService which had also grown alarmingly when the GC was not working.

        I feel your current pain !. This is very fresh for me, I feel like an ex-addict when I say that I am about 16 days clean.I hope that some of this pain I went through, combined with the these steps may provide some tips on how you might tackle your issue.

         

        I'm not going to post the cleanup scripts in this public forum but I'm happy to share offline, next week if you are interested :)

         

        Good luck

        Tom

         

         

         

         

         

         

         

        Expand Post
      • Hey Tom,

         

        Thank you for the very detailed explanation, are you able to send me the scripts please? Specifically the one for inv.UsersForDeletion, I've just had a look and that's what our Garbage Collector is complaining about the most - unable to delete users marked for removal!

         

        E: laimmc (at) softcat dot com

        Expand Post

Related  Product Forums


                     → Flexera One



                      → Snow Atlas



                      → FlexNet Manager



                      → Snow License Manager



                       → App Broker


       Need help finding an answer?


        Ask a Question →


Loading
inv.DataSoftware