Issue with copy data from csv

[Anonymous]

I'm copying data from a csv file that is uploaded to a lakehouse to a table in the same lakehouse.  I've set it up three different ways.

1. Pipeline copy task with the csv as the source and new table as destination

2. Using the "Load to Tables" festure on the file from the Lakehouse explorer

3. A simple data flow with the csv file as a source and a new table as a destination

Option 2 and 3 work as expected.  Option 1, the pipeline, produces the following.

1. The table seems to be missing the data for the last column when viewing in the Lakehouse explorer.

2. An error occurs when accessing the SQL Endpoint with a table in this state.

3. Any further use of the table in other ways, like a dataflow, has the last column's data missing.

4. However, if querying it using Spark SQL in a Notebook shows the data.

Here are a series of screen shots depicted the issue.  NOTE: Conducted a similar experiment with a JSON file and all was good so it would seem to ba an issue with handling CSV's in the Pipeline Copy activitiy.

Load via Pipeline (missing last column)

Load via "Load to Tables" (last column OK)

Load via Dataflow (last column OK)

Error in SQL Endpoint when table loaded via Pipeline exists.

Spark SQL Query of the table with the last column's data missing everywhere else.

[ajarora]

Yes that was my assumption. In that case reset any existing mappings and your issue should get resolved.

[sudhav]

Hi, in pipeline case, you have to take care of import scheme option, there you need to map the column names and data types, then you will get succeeded output.

If I answer your question, please mark it as solution.

[ajarora]

Make sure you have no explicit mapping set in the copy activity.

Please log a support ticket with your activity run id as well as pipeline json. We will resolve it ASAP.

[Anonymous]

Thanks Ajarora.  Upon inspecting the JSON of the Copy Task, it contained a schema from a different CSV file than the opne selected for the source.  What I'm thinking happened is that when creating a new pipeline, I used the same name as one that I previously deleted and somehow got content from the old one mixed in with the new one I'm not sure.  I will try to reproduce.