Creating a Docker image for GCP

In this final part of the topic guide, we will customize our own Docker image to contain additional libraries and dependencies. A working knowledge of Docker is useful to understand this section, but if you have never used Docker before, you can still follow the instructions below.

The instructions below show how the public Tumult Analytics demo image was created, and can be used as a template for your own image.

First, we will need to create a new Docker repository in the Artifact Registry.

Next, we will create the image that will be placed in the repository. We will need to create two files for this locally. The first, Dockerfile, contains Docker instructions to build the image.

FROM python:3.9-bullseye

# Install the dependencies needed for GCP
RUN apt-get update && apt-get install -y procps tini

# Install Tumult Analytics
RUN pip install --upgrade pip && \
    pip install tmlt.analytics

# Add additional dependancies here as needed
# RUN pip install <package>

# Set up the Spark user for GCP integration
RUN useradd -ms /bin/bash spark -u 1099
USER 1099
WORKDIR /home/spark
ENV PYSPARK_PYTHON="/usr/local/bin/python"

The second file, cloudbuild.yaml, contains the instructions for the Google command line tool to build the image and place it in the repository. In our example, we named our repository analytics, and our image tutorial. You will need to replace REPOSITORY NAME and IMAGE NAME with your repository and image names you set earlier. We do not need to set the $PROJECT_ID variable as it is automatically set by the Google command line.

steps:
- name: 'gcr.io/cloud-builders/docker'
  args: ['build', '-t', 'us-docker.pkg.dev/$PROJECT_ID/[REPOSITORY NAME]/[IMAGE NAME]', '.']
images:
- 'us-docker.pkg.dev/$PROJECT_ID/[REPOSITORY NAME]/[IMAGE NAME]'

Then, to build the image, we need to install the Google Cloud CLI tool.

Finally, we can build the image by running the following command.

gcloud builds submit --region=global --config cloudbuild.yaml --project=[PROJECT NAME]

If it ran successfully, you should see a new completed image in the Artifact Registry.

With the guide complete, you should be able to run Tumult Analytics programs on Google Cloud Platform. If you have any questions, please feel free to join our Slack, or contact us at our support email.