1 Reply Latest reply on Oct 18, 2018 8:02 PM by Ritesh Bisht

    How to find which workbook or data extract job consuming most Memory (RAM)


      Hi experts,

      recently our Vizql process started consuming much higher then usual RAM - just to put in context - out of 256GB - about 30 to 40% consumption was usual - nothing changed on server side yet vizql processes started consuming most of RAM i-e about upto 80% which causes view load times to degrade extremely.


      nothing changed on server so wondering what objects/jobs might be causing this consumption?


      Is there any tool through which we could map RAM used with object at point in time?


      looking for some help please.


      We're using Server 10.5.3.



        • 1. Re: How to find which workbook or data extract job consuming most Memory (RAM)
          Ritesh Bisht

          Hi Syed,


          Below post is from a Tableau employee


          Yes there is a tool called Logshark. ...Find more --->Introducing Logshark: Analyze your Tableau Server log files with Tableau | Tableau Software



          Full post ---->Why is my VizQLServer memory consumption so high?


          Hey Adrian --


          VizQLs *do* consume lots of RAM, and the behavior is completely normal. there is no appreciable difference between the way they work in 9.1 and 9.3 so I'm not sure why you're seeing something different.


          When you watch the memory consumption on one (or more) machines over time, you'll typically see more and more RAM being used, and then it levels off. Here's a time compressed version of this behavior as a load test with 60 concurrent users execute constant load against Tableau:



          Screen Shot 2016-11-08 at 4.24.19 PM.png


          See how available RAM goes down sharply between seconds 0 and 500 and then stays about the same? This is normal, but you're seeing it over the course of an hour vs. a couple of days.


          You may ALSO see available RAM go "way back up" if Tableau Server chooses to recycle a vizqlserver process. The Server Resource Manager will do that from time to time. This behavior could explain why earlier on in the thread Toby saw one of his VizQLs using lots of RAM, and another "not so much". The process using less RAM was probably "newer" than the other, having recently been recycled.


          Now that Logshark is available to the public, one could use the SRM report which it generates to actually see this happening. It's kind of cool!






          Please mark the answer as CORRECT & HELPFUL if it really helps you so that it can help others as well

          1 of 1 people found this helpful