When running %pip install
in my Synapse notebook manually, it works as intended, without any problems.
However, when I try to run that same notebook in a pipeline, I get:
{
"errorCode": "6002",
"message": "MagicUsageError: %pip magic command is disabled.",
"failureType": "UserError",
"target": <my_notebook_name>,
"details": []
}
I have looked at these two posts:
Why magic command doesn't work when running pipeline?
Synapse : Execute the Magic Command (%run) Notebook from Pipeline
But these are both only concerned with the %run
command, not %pip
The solution to both of those posts is to add an absolute path, which makes sense in the context of a %run
command, but I'm not sure how to translate adding an absolute path to %pip
? Or if there is something else that can make this work?
I have received same ERROR when using the %pip
command and running pipeline.
I have tried the below approach:
I have created a requirements.txt
file for the dependencies that need to be installed.
For example I have created a dependency for splink==1.0.3
I have uploaded the requirements.txt
Go to Manage tab > Spark Pool
Now I ran the notebook to verify the if the version of the dependency is correct?
Next, I ran the pipeline.
Results: