This workshop walks through setting up Kubeless on an Azure Kubernetes Service cluster and deploying some example functions to get familiar with Kubeless.

• Azure account with the ability to create Kubernetes clusters
• Azure CLI
• Git
• kubectl

Follow this walkthrough to setup your Azure Kubernetes Service cluster and connect to it from your local machine using kubectl.

Head over to the Kubeless releases page to download the latest binary for your operating system. For example, for macOS:

curl -LO https://github.com/kubeless/kubeless/releases/download/v1.0.0-alpha.8/kubeless_darwin-amd64.zip
unzip kubeless_darwin-amd64.zip
chmod +x bundles/kubeless_darwin-amd64/kubeless
mv bundles/kubeless_darwin-amd64/kubeless /usr/local/bin/kubeless

One installed correctly, you should be able to run kubeless version, for example:

$ kubeless version
Kubeless version: v1.0.0-alpha.8

To install the server-side components of Kubeless on your Kubernetes cluster, copy and run the instructions in the release notes for the version you are installing. For example, for version v1.0.0-alpha.8 in an RBAC-enabled cluster:

kubectl create ns kubeless
kubectl create -f https://github.com/kubeless/kubeless/releases/download/v1.0.0-alpha.8/kubeless-v1.0.0-alpha.8.yaml 

Once the server-side components have successfully been installed, you should be able to run the kubeless get-server-config command:

$ kubeless get-server-config
INFO[0000] Current Server Config:
INFO[0000] Supported Runtimes are: python2.7, python3.4, python3.6, nodejs6, nodejs8, nodejs_distroless8, ruby2.4, php7.2, go1.10, dotnetcore2.0, java1.8, ballerina0.981.0, jvm1.8

With Kubeless now installed in your cluster, you'll notice that there is a new functions API resource available to you:

$ kubectl get functions
No resources found.

We haven't created any functions yet, so this will come back empty. As we create functions later in this walkthrough, we'll see this list will become populated.

The easiest way to use the examples in this repository is to clone it:

git clone https://github.com/prydonius/kubeless-on-azure-walkthrough-ossvc
cd kubeless-on-azure-walkthrough-ossvc

The first function we will deploy is a simple one-line function that will return the text "hello world!" in Node.js:

module.exports = {
  foo: function (event, context) {
    return 'hello world!';
  }
}

Run the following command to deploy the function:

kubeless function deploy helloget --runtime nodejs6 --handler helloget.foo --from-file helloget.js

We can see that creating this function has resulted in the creation of a Pod to run our small bit of code:

$ kubectl get functions
NAME       AGE
helloget   1m

$ kubectl get pods
NAME                        READY     STATUS            RESTARTS   AGE
helloget-675b75d66d-nt68m   0/1       PodInitializing   0          5s

To run the function, we use the kubeless function call command:

$ kubeless function call helloget
hello world!

One way to manage resources in your Kubernetes cluster is by writing declarative definitions in YAML or JSON that describe the desired state of your cluster. Kubeless fits into this workflow as it extends the Kubernetes API. To demonstrate this, we can deploy the same function in the previous step using a Kubernetes manifest:

apiVersion: kubeless.io/v1beta1
kind: Function
metadata:
  name: hello
spec:
  handler: handler.hello
  runtime: nodejs6
  function: |
      module.exports = {
        hello: function(event, context) {
          return 'Hello World directly using Kubernetes!'
        }
      }

This function is mostly the same as the previous one, but changes the message returned to "Hello World directly using Kubernetes!".

To deploy this function using kubectl:

kubectl apply -f function.yaml

Once this function is running, we can try calling it:

$ kubeless function call hello
Hello World directly using Kubernetes!

Typically, you would want to pass around some data along with your function call to do some processing. With Kubeless, this is pretty easy. Your functions accept and event and context argument that give it more information about the incoming request.

To build a basic echo service using Kubeless functions, we can deploy the following function:

module.exports = {
  handler: (event, context) => {
    console.log(event);
    return event.data;
  },
};

Deploy hellowithdata.js using kubeless:

kubeless function deploy hellowithdata --runtime nodejs6 --handler hellowithdata.handler --from-file hellowithdata.js

Once this function is up and running, we can try calling it with some data:

$ kubeless function call hellowithdata --data '{"hello": "world"}'
{"hello": "world"}

When writing real-world functions, you might find you need to import some dependencies from the language's package manager. Kubeless supports installing dependencies for the core runtimes it supports.

As an example, we can create a function that uses the Node.js lodash package:

'use strict';

const _ = require('lodash');

module.exports = {
    handler: (event, context) => {
        _.assign(event.data, {date: new Date().toTimeString()})
        return JSON.stringify(event.data);
    },
};

This, accompanied with the following package.json file to define the dependency and version, will instruct Kubeless to fetch this package before running the function:

{
    "name": "hellowithdeps",
    "version": "0.0.1",
    "dependencies": {
        "lodash": "3.10.1"
    }
}

To deploy this, we'll need to specify the --dependencies flag and point it to our package.json file:

kubeless function deploy hellowithdeps --runtime nodejs6 --handler hellowithdeps.handler --from-file hellowithdeps.js --dependencies package.json

Once the dependencies have been installed and the function is running, we can test it by calling it:

$ kubeless function call hellowithdeps --data '{"hello": "world"}'
{"hello": "world","date":"00:43:31 GMT+0000 (UTC)"}

To remove the example functions from your cluster:

kubectl delete functions --all

To remove Kubeless from your cluster:

kubectl delete -f https://github.com/kubeless/kubeless/releases/download/v1.0.0-alpha.8/kubeless-v1.0.0-alpha.8.yaml 

Checkout the Kubeless documentation to learn more about Kubeless and what you can do with it!