A pipeline can be run only if it has a RunToFinish step connected (end-of-pipeline marker).
If RunToFinish is missing or not connected, you won’t be able to run the full pipeline — but you can still run parts of it using Preview on individual steps/boxes.
Fix
RunToFinish.You can run pipelines instantly without moving them to Production.
Production is not mandatory.
When you version a project, it is automatically moved to Production.
This is important when you want a stable “published” state.
“Run with cache” reuses cached intermediate results to speed up repeated runs during testing.
Important behavior
Because cached runs can reuse earlier intermediate data. If your input file, parameters, or upstream logic changed, running with cache may not reflect the latest changes.
When in doubt, run once without cache to confirm the real end-to-end behavior.
No. Assets are for reading input files only.
You cannot set an output path that writes into Assets.
Where should I write outputs instead?
Outputs written to Records are stored (not treated as “Assets”).
Use Records when you need run results to remain available after the execution.
Many actions rely on values stored in Global parameters (and often come from Sources / Services / Definitions).
Common causes:
Fix options
Because pipeline-level parameters override source-level parameters if they share the same name.
This is a frequent source of confusion when troubleshooting credentials.
Best practice
This is one of the most common issues.
Fast checklist
Some actions require input data to be sorted by one or more columns before they run.
Fix
Add a Sort box between the input step and the action that requires sorted data.
Configure sorting to match the required criteria:
Tip
If you sort as “string” but the data is numeric or date-like, results may be wrong. Always match the true data type.
platform supports both:
When something fails, the fastest resolution happens when you share:
“When I look at the DataTable view or inside my exported text file, I see ‘NaN’ in the columns where there should actually be numbers!”
“NaN” means “Not a Number”. It usually means that you tried to compute a sum, a mean, or any mathematical expression using some columns that did not contain any number in the very first place. All Numbers in ETL are represented in “English Notation” (standard or scientific notation are also accepted): The decimal separator is the dot (‘.’) and there are no “thousands” separators. In particular, the “French notation” for numbers is not (directly) considered by ETL as correct numbers and will lead to “NaN” results.
Here are some examples of Correct&Wrong number formatting:
| GOOD | BAD |
|---|---|
| 1528.7 | 1.528,7 (usage of “French” notation) |
| 1.5287E3 | 1,528.7 (usage of a “thousand” separator) |
You can use the ChangeDataType action to “correct” the numbers that are in the wrong format.
More precisely, the ChangeDataType action will not only “correct” the numbers, it will also change the meta-data of the column to reflect that the column contains floating-point values.