Initial Setup

Setup a Google Cloud Platform project and your local environment.

Method A:

Method B: Alternately, you can skip down to to the Command line section and try completing the full process from the command line.

Note that neither method has been fully tested, end-to-end, on a new GCP account and project.

Create a GCP Project

  1. Create a new Google Cloud Platform (GCP) project.

  • A. Go to the Cloud Resource Manager and click “Create Project”.

    1. Write down your Project ID for the following code.

    1. Click “Create”.

See Creating and managing projects for more information, including instructions for creating a project programatically.

  1. Enable billing and APIs for required services.

# enable APIs
gcloud services list --available  # check what's available
gcloud services enable  # enable pubsub
# will need others like bigquery and storage

Setup Local Environment

Broker instances run 100% in the Google Cloud, as determined by the code in the broker package. You can develop the code and/or deploy an instance to the Cloud from your local machine. Setup your environment as follows:

  1. Install Google Cloud SDK command-line tools using one of the following options. Included tools: gcloud, gsutil, and bq. Version 323.0.0 or above is recommended.

1b. You may need to authenticate gcloud using the code below. You may be asked to re-authenticate occasionally in the future.

PROJECT_ID=my-pgb-project  # replace with your GCP Project ID

gcloud auth login  # follow the instructions to login to GCP
gcloud config set project $PROJECT_ID  # set your project ID
  1. Install Python libraries for GCP services and other tools such as Pandas, Astropy, etc. as needed. You can install packages individually, or use the requirements.txt file at the top level of the repo. The following code creates a new Conda environment and installs the requirements.

# create a Conda environment (optional)
conda create -n pgb python=3.7
conda activate pgb

# install the requirements. assumes txt file is in current directory
pip3 install -r requirements.txt

Note: On an M1 Mac, first use Conda to install Astropy (conda install astropy=3.2.1), then comment the related line out of the requirements file before doing pip install.

  1. Create a GCP service account and download a key file for authentication using the following code. See Getting started with authentication for more information.

PROJECT_ID=my-pgb-project  # replace with your GCP Project ID
NAME=my-service-account  # replace with desired account name
KEY_PATH=local/path/GCP_auth_key.json  # replace with desired path (ending in .json)

gcloud iam service-accounts create $NAME
gcloud projects add-iam-policy-binding $PROJECT_ID --member="serviceAccount:$NAME@$" --role="roles/owner"
gcloud iam service-accounts keys create $KEY_PATH --iam-account=$NAME@$
  1. Set environment variables

PROJECT_ID=my-pgb-project  # replace with your GCP Project ID
KEY_PATH=local/path/GCP_auth_key.json  # same path as in step 3


If you are using a Conda environment, you can configure the environment variables as follows:

PROJECT_ID=my-pgb-project  # replace with your GCP Project ID
KEY_PATH=local/path/for/key/file.json  # same path as in step 3

# log into the environment and create de/activate files
conda activate pgb
mkdir -p ./etc/conda/activate.d
mkdir -p ./etc/conda/deactivate.d
touch ./etc/conda/activate.d/
touch ./etc/conda/deactivate.d/

# add environment variables
echo "export GOOGLE_CLOUD_PROJECT='$PROJECT_ID'" >> ./etc/conda/activate.d/
echo "export GOOGLE_APPLICATION_CREDENTIALS='$KEY_PATH'" >> ./etc/conda/activate.d/
echo 'unset GOOGLE_CLOUD_PROJECT' >> ./etc/conda/deactivate.d/
echo 'unset GOOGLE_APPLICATION_CREDENTIALS' >> ./etc/conda/deactivate.d/
  1. Check that your authentication works by making an API request. Here we request a list of Cloud Storage buckets (in Python):

from import storage

storage_client = storage.Client()
# Make an authenticated API request
buckets = list(storage_client.list_buckets())
# If the request succeeded, your authentication works
print(buckets)  # this list will be empty if you haven't created any buckets yet

Command line

# choose your GCP Project ID (it must be unique, so at least add a number here)
# choose a name for your service account
# choose a location for your key file

# install the SDK for the command line
    # Linux and MacOS:
curl | bash
    # Windows:
    # see
# follow the directions

# open a new terminal or restart your shell
# exec -l $SHELL

# connect gcloud to the Google account you want to use (assumes you have one already)
gcloud init
gcloud auth login
# this will open a browser and prompt you for authorization. follow the instructions

# create the project, set it as the gcloud default, and enable the Pub/Sub API
gcloud projects create $PROJECT_ID
gcloud config set project $PROJECT_ID
gcloud services enable

# create an owner service account and download a key file
gcloud iam service-accounts create $NAME
gcloud projects add-iam-policy-binding $PROJECT_ID --member="serviceAccount:$NAME@$" --role="roles/owner"
gcloud iam service-accounts keys create $KEY_PATH --iam-account=$NAME@$

# set environment variables

# install Pub/Sub Python API
pip install google-cloud-bigquery
pip install google-cloud-pubsub
pip install google-cloud-storage

# if you would like to delete the project with you are done, use:
# gcloud projects delete $PROJECT_ID