Synapse Pipeline Notebook 无法从引用的 Notebook 中解析方法

Question

I have a Synapse Pipeline which runs a notebook containing unit tests before executing the business job (another notebook). The unit test notebook references the functions using the mssparkutils.notebook.run() command, and works fine when I run the notebook on its own in Synapse Studio. However, when I trigger the notebook in a pipeline, it fails with the error:

{
    "errorCode": "6002",
    "message": "NameError: name 'get_latest_report_per_user' is not defined",
    "failureType": "UserError",
    "target": "Run Tests",
    "details": []
}

get_latest_report_per_user is defined in the referenced notebook. The reference works fine when run on its own outside of the pipeline.

The above snippet is from the unit test notebook I am running, and get_latest_report_per_user is defined in the "dependency" notebook.

I tried using the magic %run command instead of mssparkutils.notebook.run() to no avail.

Why is this only failing when executed as part of a pipeline?

Answer 1

I've been having similar issues and some of them were resolved by "Publishing" the Notebooks. You could also try the "Enable unpublished notebook reference" option under Notebook properties, but I can't recall if this option is respected by Notebooks which are invoked by a Synapse Pipeline.