This project is a demo of how we could write a dockerised micro-service using nodejs to run on Google Cloud Platform (GCP).
This application receives messages via subscriptions to Google pub/sub topics rather than HTTP requests. The build and deployment model should also work for an HTTP-based micro-service using express or koa, but that is untested.
- You will need to download a key for connecting to the pub/sub topic. This can be done in the api management console for GCP.
- Put the key in this project's root folder and name it
keyfile.json. This could potentially be tidied up to use environmental variables instead. - The application can be run as a normal node application using
npm start. - Once the application starts up, it will make a connection to pub/sub to setup a subscription. Messages are passed to a stub rest client that merely prints out the incoming message.
Make sure you have the latest version of the command line tools available:
brew install Caskroom/cask/google-cloud-sdk
gcloud components install kubectl
Alias kubectl for your sanity:
alias kubectl="/usr/local/Caskroom/google-cloud-sdk/latest/google-cloud-sdk/bin/kubectl"
Google don't currently support running arbitrary nodejs applications using App Engine (their platform-as-a-service offering). Instead, we produce a Docker image which can run on Google Container Engine (referred to as GKE to distinguish it from Google Compute Engine). The following commands can either be run on a local laptop, CI environment or a developer's laptop.
First, login to your Google account with the newly install SDK. This will launch a web browser for a normal OAuth 2 flow:
gcloud auth application-default login
Build the docker image with a prefix of the GCP region, followed by the project name. Label it with a version string:
docker build -t eu.gcr.io/integrations-154709/example-image:1.3 .
Then push that labelled image to the Google-hosted docker image registry:
gcloud docker -- push eu.gcr.io/integrations-154709/example-image:1.3
Connect to the currently running kubernetes cluster (named cluster-1) in our test project:
gcloud container clusters get-credentials cluster-1 --zone europe-west1-b --project integrations-154709
Rollout the new image for our deployment named deployment/example-image:
kubectl set image deployment/example-image example-image=eu.gcr.io/integrations-154709/example-image:1.3
List our pods in the deployment to check everything is running nicely:
kubectl get pods
You should see something like:
NAME READY STATUS RESTARTS AGE
example-image-2448862826-1n9aj 1/1 Running 0 19m
Try looking at the console of this running container with:
kubectl attach example-image-2448862826-1n9aj -c example-image
When you publish a message on the pub/sub topic, you should see it printed to the console of the container running on Google's infrastructure, displayed on your laptop.
See how many pods are in your deployment currently:
kubectl get deployments
Scale the deployment:
kubectl scale deployment/example-image --replicas=2
See the status of the individual pods:
kubectl get pods