#Tableau Server127 discussing

Thank you @Emeric LE SAULNIER DE SAINT JOUAN

 for your response. 

I agree on your points, regarding 2TB file.  

We have small extracts incremental & full refresh hyper , approx 15 to 20 GB file, As checked in logs , when these jobs are running in Node1 , the copy activity taking longer duration, tabsvc/dataengine/extracts to tabsvc/temp.  Even peak and non-peak hours, there is no CPU /RAM high usage during that time, but still copy is taking longer time. We are using Premium SSD (P80) - with 900MBs throughput 20000 IOPS.

 com.tableausoftware.sos.Access - event=sos.access.copyfolder 

Extracts folder: 

/data/tableau-data/data/tabsvc/dataengine/extract 

Temp folder :  

 /data/tableau-data/data/tabsvc/temp/backgrounder_1.20242.25.0405.0338/ 

Hey Mike, I don't have a Tableau Server environment open right now to grab screenshots, but here's how to get there: 

1. Your Tableau Server needs to be version 2019.3 or later
2. Enable the Metadata API, go to your Tableau Server, open a terminal and run: tsm maintenance metadata-services enable
3. Once enabled, access the GraphQL explorer
4. Paste the query I shared above into the left panel and hit the play button
5. Results show up on the right side

The GraphiQL interface is self-documenting, there's a Docs panel on the right side where you can explore all available fields. 

If you get stuck on any specific step, let me know what you're seeing and I can help troubleshoot.

Hi @Vinh Ton

a content owner is a privileged user, so something is up.  

Has somebody disabled this option ?

This usually comes down to caching. When you browse a workbook on Server, you are very often hitting the query cache or the native presentation model cache, so you never pay the full query cost. Desktop does not have those same layers, so every time you open a view you are running the queries fresh against the data source. 

A few other things can widen the gap. Desktop holds a single VizQL session for your machine, while Server has a tuned backgrounder and query pipeline with concurrency handling. If the source is on a remote database, the network path from Server to the database is often faster than from your laptop. Extracts also behave differently: on Server they live next to VizQL, on Desktop they are read from your local disk which can be slower if the file is big. 

If you want to narrow it down, check Help then Settings and Performance then Start Performance Recording on Desktop and see which queries dominate. In my experience it is often one or two expensive calculated fields or a Custom SQL block doing more work than it needs to. If it helps to get a readout of every calculated field and Custom SQL in the workbook without opening them one by one, I recommend using ExportCalculatedFields (you can google it) which spits out that inventory from the .twbx directly. Very useful and quick. Cheers!

I use ExportCalculatedFields (you can google it) that does a lot of this automatically if you upload a workbook file. It extracts all calculated fields with their formulas and shows which fields and worksheets are actually in use. Could save you a ton of manual XML digging if you just need the field inventory.

Hi @Emeric LE SAULNIER DE SAINT JOUAN

we are on Tableau Server v 2024.2.9. We got in touch with Tableau/Salesforce today and they let us know that we need the REST API v.3.24 for better CustomView functionality. That API version is only available for Tableau Server 2024.3+. So that explains the the issue. 

Once we upgrade, I'll report back. It sounds like it may still be a challenge to satisfy our requirements for this project as the CustomView endpoints of the API may not be as robust as we need. 

Thanks for all your follow through. We're grateful. 

Best 

Mike