This document describes the deployment process to a new Google Cloud environment. If you just want to learn how to get started or what tools you need to install, read the contribution guidelines.
Log in to Google Cloud console and create a
new project. We'll refer to the project ID you chose as $PROJECT
.
You may have to activate billing and upgrade your account to access all features required for this project.
Run gcloud init
to initialise your project in your terminal, then authenticate
via gcloud auth login
. In case you're having trouble with Python 2 vs 3, try:
CLOUDSDK_PYTHON=python3 gcloud auth login
Open the App Engine dashboard and create a new app in the region of your choice with the flexible environment.
Open the Storage dashboard and
create the buckets $PROJECT-data
, $PROJECT-logs
and $PROJECT-responses
in
the same region as the App Engine app above. Leave the default options
otherwise.
Open the PubSub dashboard and
create the topic users
, then two subscriptions attached to that topic:
crawl
with "Pull" delivery type, "Never expire", 600 seconds acknowledgement deadline, and 1 day retention duration,logs
with "Pull" delivery type, "Never expire", 600 seconds acknowledgement deadline, and 7 day retention duration.
Also create another topic responses
and one subscription attached to that topic:
response_logs
with "Pull" delivery type, "Never expire", 600 seconds acknowledgement deadline, and 7 day retention duration.
Then make sure to update the PubSub project, topic, and subscription:
crawl
in the scraper,logs
andresponse_logs
in.env
anddocker-compose.yaml
.
Go to the APIs & Services dashboard, find the Google Container Registry API, and enable it.
Go to the IAM & admin dashboard,
section Service accounts,
and find the App Engine default service account. Select "Create key" from the
actions, and download the key in JSON format. Move that file to the root of this
project as gs.json
. This is a private key, do not check it into version
control!
Make sure gcloud
uses these credentials by editing the following lines in
your local ~/.boto
:
# Google OAuth2 service account credentials (for "gs://" URIs):
gs_service_key_file = /path/to/gs.json
Now you should be able to log in to Container Registry:
cat gs.json | docker login -u _json_key --password-stdin https://gcr.io
Read more about using JSON credentials to access GCR.
Edit your .env
file to use the correct project:
GC_PROJECT=$PROJECT
Similarly, edit app.yaml
to use the correct environment variables:
env_variables:
GC_PROJECT: $PROJECT
GC_DATA_BUCKET: $PROJECT-data
PUBSUB_QUEUE_PROJECT: $PROJECT
PUBSUB_QUEUE_TOPIC_USERS: users
PUBSUB_QUEUE_TOPIC_RESPONSES: responses
The App Engine domain should be automatically added to ALLOWED_HOSTS
in
settings.py if $GC_PROJECT
is configured correctly.
Should you experience problems with your domain not being whitelisted, check
there first.
You should now be able to deploy the service. For a full release, simply run
./release.sh
If you don't need to build a new recommender and datebase version, it should suffice to run
pipenv run pynt syncdata releaseserver
Either way, after successful deployment the service should be available at https://$PROJECT.appspot.com/.
If you're using custom domains, navigate to the corresponding settings and follow the instructions. If the domain was previously used in a different project, you will need to unassign it first.
Everything should be done! Sit back and relax...