notebookutils.fs.cp() cannot copy from Lakehouse shortcut to Files in Python notebook

Hi svencken​,

I think there is one useful detail to add to the workaround already provided.

In Fabric Python notebooks, Microsoft's current NotebookUtils file-system documentation states that notebookutils.fs.cp() internally uses the same AzCopy-based mechanism as fastcp().

So in a Python 3.12 notebook, switching between cp() and fastcp() is not really testing two separate copy implementations. That also fits with both methods producing the same AzCopyFailedException in your case.

What I find interesting about your reproduction is:

• SharePoint shortcut -> notebook-local storage works
• notebook-local storage -> Lakehouse Files works
• SharePoint shortcut -> Lakehouse Files directly fails

That suggests the shortcut can be read and the Lakehouse destination can be written independently. The failure appears to be specific to the direct AzCopy-backed source/destination path.

Microsoft does document some specific cp() / fastcp() limitations for OneLake shortcuts, particularly S3/GCS shortcuts, but I do not see a current documented limitation stating that a SharePoint shortcut cannot be copied directly into Lakehouse Files.

SharePoint/OneDrive shortcuts themselves are now GA and are documented as accessible from Fabric workloads.

Before treating the two-step local copy as the permanent solution, I would test a minimal case and capture:

• the exact shortcut source path passed to cp()
• the exact Lakehouse Files destination path
• the SharePoint shortcut authentication type
• whether notebookutils.fs.exists() and notebookutils.fs.ls() succeed against that same source path

If those checks work and a single small file still fails only on the direct shortcut -> Files copy, I would raise it with Microsoft Support as a reproducible NotebookUtils/AzCopy interoperability issue.

For now, your shortcut -> local -> Lakehouse Files sequence is a reasonable workaround, but I would not consider the direct SharePoint-shortcut failure a documented product limitation based on the current Learn documentation.

AI-assisted drafting: AI was used to help structure and phrase this response. I reviewed and validated the technical content before posting.