2 Replies Latest reply on Aug 23, 2018 2:13 AM by Crystal D'souza

    Limiting Extract Refresh jobs per site

    Crystal D'souza

      Hi all,

       

      So, we have a single node , 8 core server running on Tableau 10.5. We have 20+ sites on the server, each for a different team/department.

       

      The way we have configured our server is to ensure all extract jobs run overnight or early morning while ensuring minimal "adhoc" manually triggered jobs during the day time. Normally, not more than 2-3 manual job triggers per site. We do however have some scheduled jobs during the day time due to its refresh criticality. (as in, the data needs to be updated say every 4 hours).

       

      Recently however, we have noticed that users tend to manually trigger a lot of jobs which take up time and backgrounder processes and ends up denying the schedule jobs (Especially the day time jobs) to run on time.

      I do know that we can close the task manually by logging on to the server. But that requires constant monitoring of the server and then doing a guess work over how long the job would run to decide whether to manually terminate the task.

       

      What I would like to know right now is,

      1. If there is a possibility to restrict the number of extract jobs per site. (doesnt matter if its manual or scheduled. Its a first step for us to bring some order to the refresh jobs.) If yes, how can this be done?

      2. If there is a possibility to restrict the timeout only for manual jobs. backgrounder.querylimit tends to affect all jobs, scheduled and manually triggered.

       

       

      Thanks

      Crystal

        • 1. Re: Limiting Extract Refresh jobs per site
          Ritesh Bisht

          Hi Crystal,

           

          1) Schedules are not site specific but are server-wide.  Really not specific to a particular site. Also please note that Site Administrators cannot create schedules.

           

          2) Restrict the timeout only for manual jobs , I have not seen anyone doing this and I don't think there is an easy way to do the same.

           

           

          OR (If you can )

           

          Don't give them access instead let someone else do on behalf of them (Owner / Administrator )

          refreshextracts workbook-name or datasource-name

          Performs a full or incremental refresh of extracts belonging to the specified workbook or data source.

          This command takes the name of the workbook or data source as it appears on the server, not the file name when it was published. Only an administrator or the owner of the workbook or data source is allowed to perform this operation.

          Examples

          tabcmd refreshextracts --datasource sales_ds

          tabcmd refreshextracts --project "Sales External" --datasource sales_ds

          tabcmd refreshextracts --workbook "My Workbook"

          tabcmd refreshextracts --url SalesAnalysis

           

          tabcmd Commands

           

          Please mark the answer as HELPFUL/CORRECT if it really helped you so that it can help others as well .

          Thanks,

          Ritesh

          • 2. Re: Limiting Extract Refresh jobs per site
            Crystal D'souza

            Hey Ritesh,

             

            Im aware that the schedules are Server Wide. Which is why I mentioned, that most of the extracts on the server (across sites) run overnight.

            The site administrators are informed about the schedules.

             

            What we're trying to achieve here is to first limit the number of manually triggered extract jobs using the "refresh now" option.

            We do inform that the site administrators trigger the refresh extract jobs and also to make use of the Schedules, but the challenge here is

            1. Since the sites are for a department. Some site admins may not actually be involved in a project and grants the refresh rights to the publisher. Now, multiply this by the number of sites and publishers, you get an idea of how many people can actually trigger manual jobs.
            2. Prioritization of the jobs by a site administrator would work best when we have a limit of the jobs run per site. This way, we can actually ensure that not more x number of jobs will run in a day. Right now, there isn't anyway to see this.
            3. Users often refresh data as part of development. Problem is, we notice extracts run to the full timeout and notice
              • Joins weren't properly configured
              • Filters weren't set properly. 

             

            What I have explained above are just some of the common issues we notice. Hence my second point if we have a time out limit set specifically for the manual jobs.

             

            We have a self service model established where the users/site admins are expected to handle all things pertaining to the site. But this freedom manifests itself as a server wide problem which is where we need some stability to ensure everyone gets access. So, as a server admin, we'd like to get some control and let the users get exactly what they want.

             

            Thanks

            Crystal