Local Mode does not work on Sagemaker Studio

[bloodywolf3]

Describe the bug
I am trying to do training in a local container by running the example python notebook: mxnet_gluon_mnist/mxnet_mnist_with_gluon_local_mode.ipynb

To reproduce
A clear, step-by-step set of instructions to reproduce the bug.
In the sagemaker studio window

Select a) MXNET GPU optimized kernel and b) ml.g4dn.xlarge instance

1. Running the script: !/bin/bash ./setup.sh
   The output is ./setup.sh: line 3: sudo: command not found
   The user does not have root access. Everything required to run the notebook is already installed and setup. We are good to go!
2. Run m.fit(inputs)
   The output is ImportError: 'docker-compose' is not installed. Local Mode features will not work without docker-compose. For more information on how to install 'docker-compose', please, see https://docs.docker.com/compose/install/

System information
SageMaker Python SDK version:
Framework name (eg. PyTorch) or algorithm (eg. KMeans): MXNET GPU optimized kernel
Framework version:
Python version: 3
CPU or GPU: ml.g4dn.xlarge
Custom Docker image (Y/N): N

[chuyang-deng]

Hi @bloodywolf3, docker itself will not work on Studio notebooks as Studio Notebooks run inside docker containers. You could try to run this example on SageMaker Notebook instance instead.

[seanpmorgan]

Closing. As mentioned Studio runs inside of docker so in order to build images you can either use this utility:
https://github.com/aws-samples/sagemaker-studio-image-build-cli

Or build them outside of studio (in SM notebooks, locally with aws cli) and continue with the example after it's been pushed to ECR.

[samadwar]

I have written an article describing a solution that makes it possible to use 'Local Mode' in SageMaker Studio. Apologies for opening a closed issue, but adding this for completeness.