Preview Timed Out During Dataset Creation with Multiple Joins and Some Columns Missing in Topic

Question & Answer
, ,
1

Question:

I’m facing an issue while working on Amazon QuickSight related to dataset preview and topic creation.

Issue Description:

While creating a dataset, QuickSight is unable to preview the data and also seems to ignore a few columns when the dataset is used to create a Q Topic.

Join Configuration:

Here’s how I’ve set up my joins:

  • Dataset 1: Created using a direct SQL query (from tables A and B).
  • This dataset is then used as a data source in another dataset, where additional joins are added:
    • tableA <--> table C
    • tableA <--> table D
    • Plus, a previously created dataset: tableA ↔ table E <--> table F which is previously created dataset is also joined.

This entire structure was tried out with SPICE and DIRECT QUERY but even with that, the issue persists.

Error Message:

Can’t show preview  
Preview timed out. You can save and visualize data, or...

Also, when I try to create a Topic, all columns from the dataset are present

Suggestions Needed:

  • What could be causing QuickSight to fail the preview and omit columns in the topic creation process?
  • Could it be due to:
    • Complex join logic?
    • Dataset size or performance limits?
    • SPICE caching issues?
  • Will this issue affect the performance or accuracy of Q Topics or downstream dashboards?
  • Are there any best practices to avoid timeouts or data preview failures in complex dataset configurations?

Any insights or solutions would be really helpful. Thank you!

2

Personal experience, I have used complex sql with multiple joins. You don’t need to worry, if the preview does not show up.

I had faced this issue when there are I believe ~30+ columns fetched in this data source.
It did not hamper performance for me, we were using Spice capacity for such most of our joins.
About the “Excluded columns”, I have usually faced this scenario when I have replaced some joined tables/sql or tried replacing data sources but not during data source creation.

3

But How do I know if my data is being retrieved from data source or not, I’ve been using direct query

4

Hey @Francis_Joel

The way that I have tested the data coming through, especially for Q Topics, is to create a control analysis where I can compare Q topic responses against my data.