sql data reader step is not working in procedure whereas it is running manually

Question

There are bunch of data readers which i have ported along with some cubes and procedures , so everything is fine except the calling sql data reader step in procedure , it is throwing an error and going to error group whereas manually it is running perfectly, can any body help with this. please find the error from the logs.

Screenshot 2025-12-15 170446.png

Robert Barkaway · Answer

Hi Robert-Jan, I'd be surprised if it was the complexity of the SQL causing a problem, purely because I've had to use some fairly complex SQL statements myself in the past and never had any issues - although I would say that best practice is typically to create larger SQL statements as Views where possible and query those instead.

What actual error message are you getting? When you run the query in the data reader it only retrieves a sample data set and not the full thing, so it's possible it isn't encountering the rows which are causing a problem.

Also have you tried disabling table sampling and row counts? Sometimes depending on the structure of the SQL queries those can pose a problem.

Robert-Jan van Kuppeveld · Answer

I encounter the same behaviour, when I run the SQL queries in the data reader, they work fine, Board is not able to return the correct number of imported / rejected records.

If I run the same SQL reader via a procedure, it will go to the error group.

I have no clue how to fix them, the sql statement is fine, it is also working in the data pipeline, but it seems that the reader can not handle it if the manual query is to complex.

Bale Goutham · Answer

Hi @Bart Scott

When I went through the logs, I found that disable table sampling and count records were throwing errors. When the DataReader was run manually, these errors appeared but did not stop the execution of the DataReader. However, when we ran the same steps through the procedure, the new on Error Go to Group feature introduced by the board caused the group to stop running. Previously, this feature did not exist via group it was through selection, which is why—even though errors occurred—we were still able to run the DataReader. Now, that is no longer the case.