Train and Deploy on GCP from a Local Notebook

    This guide introduces you to using Kubeflow Fairing to train and deploy amodel to Kubeflow on Google Kubernetes Engine (GKE), and Google Cloud ML Engine.As an example, this guide uses a local notebook to demonstrate how to:

    • Train an XGBoost model in a local notebook,
    • Use Kubeflow Fairing to train an XGBoost model remotely on Kubeflow,
    • Use Kubeflow Fairing to train an XGBoost model remotely on Cloud ML Engine,
    • Use Kubeflow Fairing to deploy a trained model to Kubeflow, and
    • Call the deployed endpoint for predictions.

    This guide has been tested on Linux and Mac OS X. Currently, this guide has not beentested on Windows.

    Clone the Kubeflow Fairing repository to download the files used in this example.

    Set up Python, Jupyter Notebook, and Kubeflow Fairing

    • You need Python 3.6 or later to use Kubeflow Fairing. To check ifyou have Python 3.6 or later installed, run the following command:

    The response should be something like this:

    1. Python 3.6.5

    If you do not have Python 3.6 or later, you can from the Python SoftwareFoundation.

    • Use virtualenv to create a virtual environment to install KubeflowFairing in. To check if you have virtualenv installed, run thefollowing command:
    1. which virtualenv

    The response should be something like this.

      If you do not have virtualenv, use pip3 to install virtualenv.

      1. pip3 install --upgrade virtualenv
      1. virtualenv venv --python=python3
      2. source venv/bin/activate
      • Install Jupyter Notebook.
      • Install Kubeflow Fairing from the cloned repository.
      1. pip3 install --upgrade .
      • Install the Python dependencies for the XGBoost demo notebook.
      1. pip3 install -r examples/prediction/requirements.txt

      In order to use Kubeflow Fairing to train or deploy to Kubeflow on GKE,or Cloud Machine Learning Engine, you must configureyour development environment with access to GCP.

      • If you do not have the Cloud SDK installed, install theCloud SDK.

      • Use gcloud to set a default project.

      1. export PROJECT_ID=<your-project-id>
      2. gcloud config set project ${PROJECT_ID}
      • Kubeflow Fairing needs a service account to make API calls to GCP. Therecommended way to provide Fairing with access to thisservice account is to set the environmentvariable. To check for the GOOGLE_APPLICATION_CREDENTIALS environmentvariable, run the following command:
      1. ls "${GOOGLE_APPLICATION_CREDENTIALS}"

      The response should be something like this:

      1. /.../.../key.json

      If you do not have a service account, then create one and grant itaccess to the required roles.

      1. export SA_NAME=<your-sa-name>
      2. gcloud iam service-accounts create ${SA_NAME}
      3.     --member serviceAccount:${SA_NAME}@${PROJECT_ID}.iam.gserviceaccount.com \
      4.     --role 'roles/editor'

      Create a key for your service account.

      Create the GOOGLE_APPLICATION_CREDENTIALS environment variable.

      1. export GOOGLE_APPLICATION_CREDENTIALS=~/key.json

      Set up Docker

      1. docker ps
      • If you get a message like , then installDocker.
      • If you get an error like Error response from daemon: Bad response fromDocker engine, then .
      • If you are using Linux and you use sudo to access Docker, follow thesesteps to add your user to the docker group. Note, thedocker group grants privileges equivalent to the root user. To learn moreabout how this affects security in your system, see the guide to the.

      Authorize Docker to access your GCP Container Registry.

      1. gcloud auth configure-docker

      Use the following instructions to set up and configure your Kubeflow anddevelopment environments for training and prediction from Kubeflow Fairing.

      • If you do not have a Kubeflow environment, follow the guide to to set up your Kubeflow environment.The guide provides two options for setting up your environment:

      • Update your kubeconfig with appropriate credentials and endpointinformation for your Kubeflow cluster. To find yourcluster’s name, run the following command to list the clusters in yourproject:
      1. gcloud container clusters list

      Update the following command with your cluster’s name and GCP zone, thenrun the command to update your kubeconfig to provide it with credentialsto access this Kubeflow cluster.

      1. export CLUSTER_NAME=kubeflow
      2. export ZONE=us-central1-a
      3. gcloud container clusters get-credentials ${CLUSTER_NAME} --region ${ZONE}

      Use Kubeflow Fairing to train a model locally and on GCP

      • Launch the XGBoost quickstart in a local Jupyter notebook.
        • Follow the instructions in the notebook to train a model locally, onKubeflow, and on Cloud ML Engine. Then deploy the trained modelto Kubeflow for predictions and send requests to the prediction endpoint.

        Was this page helpful?

        Glad to hear it! Please .

        Sorry to hear that. Please tell us how we can improve.