magda-function-template

An Openfass Serverless Function template for Magda. You can also use faas-cli to create a Openfaas function. However, this template leverage our own toolset and works with our own deployment / CI system better.

You can click the Use this template Github function button above to create a new repository from this template repository instead of forking it. More details see Github Help Document: Creating a repository from a template

Supply Your Own Function Code

You can supply your own function code in src/index.ts. e.g.:

export default async function myFunction(input: any) {
    return "hello world!\n";
}

Invoke you function with Content-Type: application/json header will make your function receive unserialised data as input parameter.

Install Project Dependencies

yarn install

Build & Run Function in Minikube

• Deploy Magda v0.0.57-0 or later
• Build the function
  • Run yarn build
• Push docker image to minikube
  • Run eval $(minikube docker-env)
  • Run yarn docker-build-local
• Deploy function to Minikube
  • Make sure namespacePrefix field in deploy/minikube-dev.yaml contains correct magda-core deploy namespace. By default, it's default and it works if you've deployed Magda to default namespace.
  • Run yarn deploy-local
• Invoke your Function:
  • Install faas-cli
  • Run kubectl --namespace=[openfaas gateway namespace] port-forward svc/gateway 8080 to port-forward openfaas gateway
    • Here, [openfaas gateway namespace] is [magda-core namespace]-openfaas. e.g. if magda is deployed to default namespace, [openfaas gateway namespace] would be default-openfaas
  • Invoke by Run echo "" | faas-cli faas-cli invoke magda-function-template
  • Alternatively, you can use Postman to send a HTTP Request (HTTP method doesn't matter here) to Magda gateway /api/v0/openfaas/function/magda-function-template

Deploy with Magda

• Add as Magda dependencies:

- name: magda-function-template
  version: 0.0.57-0
  repository: https://charts.magda.io
  tags:
      - all
      - magda-function-template

• Run helm dep build to pull the dependency
• Deploy Magda

Verify the function is deployed

• Method One:
  • Access Magda Gateway: /api/v0/openfaas/system/function with your web browser
    • You might need Admin access to access this endpoint. However, you can disable the admin auth in Magda config.
• Method Two:
  • Run kubectl --namespace=[openfaas function namespace] get functions
    • Here, [openfaas function namespace] is [magda-core namespace]-openfaas-fn. e.g. if magda is deployed to default namespace, [openfaas function namespace] would be default-openfaas-fn

If the Scale to Zero option is set for the function (it's set to true by default), you won't see function pod in openfaas function namespace until you invoke the function

CI Setup

This repo comes with script to build, test & release script to release docker image & helm chart to Magda repo. You need to setup the following Github action secrets to make it work:

• AWS_ACCESS_KEY_ID: Magda helm chart repo S3 bucket access key
• AWS_SECRET_ACCESS_KEY: Magda helm chart repo S3 bucket access key secret
• DOCKER_HUB_PASSWORD: Magda docker hub bot password
• GITHUB_ACCESS_TOKEN: Magda github bot access token

Requirements

Kubernetes: >= 1.14.0-0

Repository	Name	Version
https://charts.magda.io	magda-common	1.0.0-alpha.4

Values

Key	Type	Default	Description
defaultImage.imagePullSecret	bool	false
defaultImage.pullPolicy	string	"IfNotPresent"
defaultImage.repository	string	"docker.io/data61"
global.image	object	{}
global.openfaas	object	{}
image.name	string	"magda-function-template"
resources.limits.cpu	string	"100m"
resources.requests.cpu	string	"50m"
resources.requests.memory	string	"30Mi"