MSTICPy and other packages do not install properly when switching between the Python 3.6 or 3.8 Kernels

We've seen this happen intermittently. Using !pip install pkg will install correctly in one environment but then does not install in the other. The package cannot be imported and used in the second environment.

Avoid using !pip install... to install packages in AML notebooks. Instead, use one of the following options:

Use the %pip line magic within a notebook.

%pip install --upgrade msticpy

Install from a terminal. Open a terminal in Azure ML notebooks and run the following commands:

conda activate azureml_py38
pip install --upgrade msticpy

Close the terminal and restart the kernel.