The core component of my Sagemaker lifecycle template for DS

lifecycle_core.sh

#!/bin/bash


# --- script documentation
# recieves two arrays as input using the length prefix convention. See:
# https://stackoverflow.com/questions/43686878/pass-multiple-arrays-as-arguments-to-a-bash-script

# add mandatory packages here
# packages+=("pandas")
echo ">>> conda_envs: ${conda_envs[@]}"
echo ">>> packages: ${packages[@]}"
# echo ">>> vars: ${vars[@]}"

# ==== Part 1: Clone Repository ====

# Clones the the right branch of the right repository
# using the credentials of the right user
# (so commits from SageMaker are legible)

echo ">>> === Part 1 - Repo Cloning ==="

exc

if [ -d "./SageMaker/$REPO/" ]; then
    echo ">>> $REPO repository already cloned!"
else
    cd SageMaker
    echo ">>> Cloning $REPO and checking out branch $BRANCH ..."
    git clone --branch "$BRANCH" "https://$GITHUB_UNAME:$GITHUB_PAT@github.com/$GITHUB_ORG/$REPO.git"
    echo ">>> Done cloning!"
    cd "$REPO"
    echo ">>> See git status output:"
    git status
    cd ..
    cd ..
    echo ">>> Back to working dir!"
fi


# ==== Part 2: MLflow-on-DataBricks intergration ====

echo ">>> === Part 1 - Databricks Integration  ==="

echo "Injecting env vars into all desired conda environments..."
# Iterating over all conda environments
# (Note that "base" is special environment name, include it there as well)
for env in base ${conda_envs[*]}; do

    # 2.1. databricks-required env var setup
    echo ">>> Injecting env vars into $env conda environment..."
    mkdir -p "/home/ec2-user/anaconda3/envs/$env/etc/conda/activate.d/"
    touch "/home/ec2-user/anaconda3/envs/$env/etc/conda/activate.d/env_vars.sh"
    for var in ${vars[*]}; do 
        conda env config vars set "$var"
        echo "$var" >> "/home/ec2-user/anaconda3/envs/$env/etc/conda/activate.d/env_vars.sh"
    done
done
echo "Done injecting env vars into all desired conda environments..."


# ==== Part 3: Installing Python Packages ====

echo ">>> === Part 3 - Python Packages Installs ==="

# Iterate over SOME conda environments, and
# 1) Install all required Python packages from pip
# 2) Install the local repo's components as packages

for env in ${conda_envs[*]}; do
    # installing required python packages on all conda environments:
    echo ">>> Installing pip packages in $env ..."
    
    # 3.1. Activate the current conda environment
    conda activate "$env"
    # source /home/ec2-user/anaconda3/bin/activate $(basename "$env")
    echo ">>> Check active conda env:"
    conda env list

    pip install --upgrade pip
    
    # 3.2.0. prevent installation errors when installing mlflow:
    # pip uninstall -y enum34

    # 3.2.1 Installing packages in the Jupyter system environment can affect stability
    # of your SageMaker Notebook Instance. You can remove this check if you'd like to
    # install Jupyter extensions, etc.
    if [ $env = 'JupyterSystemEnv' ]; then
        echo ">>> Skipping installation inside the Jupyer system environment."
        continue
    fi
    
    # 3.3. Running pip installs of packages

    for package in ${packages[*]}; do
        echo ">>> - Installing $package package..."
        pip install "$package"
    done
    # (You can also perform "conda install" here as well)
    echo ">>> Finished installing packages from PyPI. Check list:"
    pip list
    
    # 3.4. pip install local research repo components
    echo ">>> Installing cloned repo as a package"
    pip install -e "./SageMaker/$REPO[test]"
    
    # 3.5. Deactivate the current conda environment
    # source /home/ec2-user/anaconda3/bin/deactivate
    conda deactivate
done
echo "Done installing packages!"


# ===== Part 4: Logging Utilization Metrics to CloudWatch ====

echo ">>> === Part 4 - Utilization Logging  ==="

# Installs the cloud watch agent on the notebook instance in order to collect load and utilization metrics.

# OVERVIEW
# This script publishes the system-level metrics from the Notebook instance to Cloudwatch.
#
# Note that this script will fail if either condition is not met
#   1. Ensure the Notebook Instance has internet connectivity to fetch the example config
#   2. Ensure the Notebook Instance execution role permissions to cloudwatch:PutMetricData to publish the system-level metrics
#
# https://aws.amazon.com/cloudwatch/pricing/

# PARAMETERS
NOTEBOOK_INSTANCE_NAME=$(jq '.ResourceName' \
                      /opt/ml/metadata/resource-metadata.json --raw-output)

echo "Fetching the CloudWatch agent configuration file."
wget https://raw.githubusercontent.com/aws-samples/amazon-sagemaker-notebook-instance-lifecycle-config-samples/master/scripts/publish-instance-metrics/amazon-cloudwatch-agent.json

sed -i -- "s/MyNotebookInstance/$NOTEBOOK_INSTANCE_NAME/g" amazon-cloudwatch-agent.json

echo "Starting the CloudWatch agent on the Notebook Instance."
sudo /opt/aws/amazon-cloudwatch-agent/bin/amazon-cloudwatch-agent-ctl -a \
    fetch-config -m ec2 -c file://$(pwd)/amazon-cloudwatch-agent.json -s

rm amazon-cloudwatch-agent.json