2 Replies Latest reply on May 2, 2018 12:58 AM by Sanket Kelkar

    Steps for Multi node Server Upgradation 10.1 to 10.5.2

    Sanket Kelkar

      Hi All,

       

      I am working on up-gradation of Server 10.1 to 10.5.2 - It is a 3 node server with one primary and 2 workers installed on VM. My upgrade plan is as below -

       

      1. Stop the primary server. We don't want users to change anything during the upgrade.

      2. Take the full manual backup of primary server. This will include the backup of working nodes as well.

      3. Run the 10.5.2 server setup on primary VM.

      4. The setup will automatically detect that 10.1 is already installed and ask for backup. We can skip this as it is already done in #2.

      5. The new setup will uninstall 10.1.

      6. If the upgrade requires workers to be upgraded manually, hold on to the up-gradation of primary and -

        6.a. Sign in to each worker node and manually run 10.5.2 worker setup program.

        6.b. This will also automatically uninstall previous version and install a new version.

      7. Return to primary node installation and follow prompts to finish the upgrade.

      8. Roll-out new server to users and validate.

       

      We are aware of the fact that extracts will be converted to .hyper after the refresh and this is acceptable.

       

      Can someone please validate these steps and suggest any changes if needed ?

       

      Thanks,

      Sanket K.

        • 2. Re: Steps for Multi node Server Upgradation 10.1 to 10.5.2
          Sanket Kelkar

          Sharing my observations on the upgrade for the community -

           

          We upgraded from 10.1 to 2018.1 last week -

           

          1. Stopped the primary server and took full manual backup as we did not want users to make any changes while the upgrade was in progress. Our server is of 100 GB size, and the backup took 52 mins to complete and generated a .tsbak file with 37 GB size. The server has 64 GB RAM with 8 processors.

          2. Started the upgrade by running 2018.1 server installable file. It automatically detected that we already have 10.1 and uninstalled it.

          3. VERY IMP - we have a non-default server location, and hence needed to be very careful while selecting the path. If the path goes wrong, then it installs a completely new version of 2018.1 and does not copy any configuration.

          Path in our case -

           

          Original Path – D:\Programs\Tableau\Tableau Server\10.1\bin

          Browse to - D:\Programs\Tableau (please note that it automatically appends Tableau server folder with the browsed path and creates a new folder with a name 2018.1 )

          New path - D:\Programs\Tableau\Tableau Server\2018.1\bin

           

          4. While the upgrade was running, it asked to manually upgrade workers. At that moment, we signed-in to each worker and ran 2018.1 server worker program. It also automatically detected the previous version and upgraded to 2018.1

          5. The entire upgrade took 1 and a half hour and it successfully copied each and every configuration including - all workbooks, sites, extracts, schedules, subscriptions, SSL configuration, windows AD config, license information etc from 10.1 to 2018.1

           

          Performance stats -

          A full refresh pointing to Oracle database -

           

          On 10.1

          Time to refresh- 49 minutes

          Size of the .tde file- 1.11 GB

          Data - 2 million rows and 30 columns

           

          On 2018.1

          Time to refresh- 22 minutes

          Size of the .hyper file - 800 MB

          Data - 2 million rows and 30 columns

           

          Overall, we are getting more than 50% performance improvement in the extract refresh and 20-30% performance improvement in dashboard interactivity.

           

          Please let me know if I have missed any point and I would be happy to help!

           

          Thanks,

          Sanket K.