Dashboard calculation queue grooming
In a larger organization, with a lot of dashboard usage, you can benefit from having a policy of removing older or unused dashboards. These charts can otherwise add to the calculation queue and take up room for more frequently used and viewed dashboards.
Below you will find some instructions on how to find these charts and how to remove them.
In the Dashboards section of the Admin page you can see all the charts, their names, the ID of the Dashboard page they are on, the page name and the chart owner.
Clicking the "Calculation statistics" button will bring up information about the current calculation queue and the recently processed charts.
The statistics page has two sections. The upper part is the current queue with a number of columns explained below.
TimP: Time priority. Increases over time to avoid that a chart stays at the bottom of the queue forever.
Boost: Boost priority. Temporary priority added when a chart is refreshed.
Prio: Priority based on the number of chart viewers.
SubP: Sub priority set to number of chart viewers.
Time: Timestamp when chart was added to queue. Number of seconds since server started.
ID: Chart ID
Pend: If the chart is currently processed, this is the running time in seconds.
NumS: Number of users using the chart.
GenT: Generation time in seconds of last chart processing.
WaitT: Waiting time in queue in seconds before chart was processed.
Page: The page in which the chart resides and the owner of it.
The lower part is the list of processed charts with a subset of the same columns as above.
If your users are experiencing long refresh times you can check the dashboard statistics for charts with long generation times that are clogging the queue. Using the chart ID you can then find the owner of the chart in question and ask them if the chart is still in use. If not you can ask them to delete it or you can delete it from the admin view yourself. If it is still relevant you can ask the user if it's possible to reconstruct it to make it simpler of maybe break it up into a number of smaller charts instead in order to help move through the queue.