Global ingress in practice on Google Container Engine — Part 2: Demo

This article is a follow up to a couple previous ones, Global Kubernetes in 3 Steps on GCP which walks through setting up a global cluster, and Global ingress in practice on Google Container Engine — Part 1: Discussion which discusses how you would use a Global cluster with a Google LoadBalancer for Ingress.

Source Code
The source code for this walkthrough can be found on GitHub here

In there you’ll find:
- /app — the code for the apps used in the demo
- /cluster — a single script to provision a federated cluster
- /deploy — the k8s yaml files we’ll be using in the demo

Setup the cluster

I’ve covered this in detail in the Global Kubernetes in 3 Steps post and have provided it again in the README. This article is more about how to use it so I won’t go into the setup process here.

Deploy the Applications and Ingress

First off, grab the repo and clone it locally

git clone https://github.com/cgrant/global-k8s-ingress-with-gce-controller
cd global-k8s-ingress-with-gce-controller

For this example I’ve provided sample python applications in the /apps directory. I’ve prebuilt them and posted them to docker hub as well. You should be able to customize the code or swap in your own images.

Deploy the apps

kubectl apply -f deploy/app-with-context.yaml
kubectl apply -f deploy/simple-echo.yaml

As mentioned in Part 1, you’ll need to explicitly set the NodePort values on the Deployment. I’ve already set these for you to `30048` and `30050`

Review the workloads page in Google Cloud Console

Create a global IP
Once the application pods are deployed we can look at sticking them together with an ingress.

First thing we need to do is create a global IP for the Global LoadBalancer.

This is critical for ingress to work on multiple clusters. the default ephemeral IPs are only regional and won’t allow proper federation

Create a global ip named ingress-ip

gcloud compute addresses create ingress-ip — global

Deploy the Ingress

Now deploy the Ingress itself

kubectl apply -f deploy/ingress.yaml

This will take awhile, 5 minutes or so, for the loadbalancers to be created worldwide and for all the health checks to pass.

You can review the progress on the Discovery & load balancing page in the Cloud Console

Once the load balancer is up you can access the following urls:

/
/foo
/foo/bar
/echo
/echo/anyString

From the Discovery & load balancing page in the Cloud Console, click on the Load Balancer name for details. Then scroll down to the backends.

There will be a backend for each service in your ingress yaml. Go ahead and click on one of the backends

As traffic flows to your new service, you can see the break down of location and other metrics on this page.

Thats it, you’re good to go!

On Your Own

Thats the short and sweet of the demo. Be sure to explore the code to understand the mechanics. Also try updates on your own

Go ahead and deploy your own service, modify the ingress.yaml and run

kubectl apply -f deploy/ingress.yaml

New services on the load balancer take a few minutes to come up but changes to existing services is pretty quick

Try it by updating the image in the app Deployment to something new, for example maybe i update `simple-echo.yaml `

spec:
 containers:
 — name: simple-echo
 image: cgrant/simple-echo-server
```

to

spec:
 containers:
 — name: simple-echo
 image: cgrant/simple-echo-server:version1

Then apply

kubectl apply -f deploy/simple-echo.yaml

The containers will update promptly and you’re all set!

I really like the ability of using native Google Cloud products with Kubernetes. It takes much of the hassle out of the infrastructure while providing me with robust managed tools for my apps.

Hope you enjoyed. Please leave any comments or questions