Sharing a backgrounder manager workbook I built.
+ Each backgrounder is single threaded; therefore can only process one task a time.
+ Most tasks processed by backgrounders are system tasks.
+ Exceptions to this rule: Active Directory Sync tasks, Subscription tasks, and Extract Refresh tasks which are scheduled by end users/admins.
Since Extract Refreshes tend to take the most amount of computing power - most deployments have a least a few long running extract refreshes - workbook does focus on Extract Refresh tasks, but also have views for all backgrounder tasks.
The Dashboard ‘Resource Utilization per Extract’ is my take on how we should prioritize which extract refresh is taking the most amount of Tableau Server resources.
For each Extract Refresh, the dashboard outlines the number of minutes it took to complete the refresh and the number of time the refresh was scheduled for given time range filter.
For example, if an extract takes ~10 minutes to run and is scheduled to run every hour, this extract will take ~240 minutes a day of system resources (represented in the dashboard as 24 rectangles of 10 minutes width)
In the other hand if an extract takes ~1.5 hours to run and is scheduled to run once a day, this extract will take ~90 minutes a day of system resource (represented as one rectangle of ~90 minutes width)
Tooltip contains error message. If extracts is failing with timeout error (default 7200 seconds or 2 hours), I would prioritize those extracts over other failures, as those tasks are taking the system resources (~2hours of a backgrounder process) and are not bringing new data into the system.
Similar to LogShark's Backgrounder workbook, but LIVE connection to Tableau Server internal Postgres database, the dashboards ‘Backgrounder Utilization’ and ‘Extract Task Details’ can help get up-to-date insights on Extracts running at Tableau Server.
The following workbook has 6 dashboards designed to help understand and manage backgrounder load.
I hope it is helpful. Please do like/rate this post so that I know your feedback.