"Select data to load" in Data Load Editor fetches entire table

[gandalfgray]

When we use the "Select data to load" in the Data Load Editor using a data connection via Amazon Redshift ODBC Sense sends a

SELECT * FROM <tablename> 

without any limit it seems.

One of our developers clicked on a large table and that query was running for 22 hours. 

Is there a way to make the "Select data to load" add a limit on what it fetches?

[Gysbert_Wassenaar]

When developing you can use the Debug mode and in that mode choose to load a limited number of records.
Another way is to add a WHERE clause to the generated load statement:
SELECT * FROM <tablename> WHERE <... condition(s) to limit the data retrieved ...>

[gandalfgray]

Thanks Gysbert, but you are missing my point.

This happens when you pick a table from the "Select data to load" window (see below),

that is, when you click on any of the tables in the table list.

I would expect Sense to be smart enough to just fetch a few 1000 rows or something, to be able to present a preview, but it seems it fetches the full table.

[Gysbert_Wassenaar]

In my experience the data preview does not load all the data from the source table.

[gandalfgray]

We usually don't have problems with SQLServer or PostgreSql data, but this is Amazon Redshift, and there it performs a full table load according to the logging

[Gysbert_Wassenaar · Accepted solution]

I've found what QS supposedly does with the data preview: https://support.qlik.com/articles/000060939

A select count(*) is executed first. Perhaps Amazon Redshift doesn't do this very well and executes a full table scan.
Or QS tries to load only 100 records and QS does something wrong there or Amazon Redshift doesn't get it and tries to return everything.  Though I don't see really how that could happen in this age of cursors and pagination and whatnot.