2 Replies Latest reply on Jan 26, 2018 2:41 AM by Andrew Harper

    tabcmd export parallel processes seem to 'get stuck'

    Andrew Harper

      We are running four tabcmd export --pdf processes in parallel (using python's multiprocessing module and --no-cookie) from a remote machine against Tableau Server (TS) 10.3. TS is configured with two processes for each of the VizQL, Cache and Data Servers.

       

      Against TS 9.1 this configuration worked. Against TS 10.3 the tabcmd processes run for some minutes and then all seem to get stuck together. For example, the four processes will start and finish almost at the same time (rendering the view correctly), having run for 15m. The view that each is rendering takes about 15s on its own! Interestingly, each process very often takes almost exactly 900s...

       

      Tableau Support has recommended reviewing firewall/antivirus settings which we are doing. Does anyone have any other ideas? Thank you!

        • 1. Re: tabcmd export parallel processes seem to 'get stuck'
          Jeff Strauss

          Are you running live against a database or against an extract?  I am wondering if somehow parallel queries are coming into action with your upgrade and somehow it's getting contention within the database.

          • 2. Re: tabcmd export parallel processes seem to 'get stuck'
            Andrew Harper

            Thanks for your reply. We are running against extracts. At the moment we are investigating the antivirus setup. I fired up four Command Prompt windows on the Tableau Server machine and in each ran a series of tabcmd export commands. Before too long, the commands running in two of the windows hang. Consistently, fifteen minutes later, the command processing continues. Trend Micro Office Scan is installed on the machine. Although this was installed after Tableau Server, we have since upgraded Tableau Server, and I wonder whether the new installation path is not being ignored by Trend Micro. This is being checked... When I find out more I'll post again... Thanks again.