microsoft / helm-web-api

Enabling applications in your Kubernetes cluster to programmatically install helm charts and expose them through a single public facing IP

On-demand micro-services deployment for Kubernetes via HTTP endpoints

Enabling applications in your Kubernetes cluster to programmatically install helm charts and expose them through a single public facing IP

The solution exposes web endpoint which provides the Helm functionalities of Install, Delete and Upgrade charts, as well as automatically configure the ingress controller to expose them to the internet all within a single public IP. The solution we propose consists of two parts, both as web servers:

Helm charts deployer: let developers install, delete and upgrade helm charts from inside the cluster, using a simple REST API.

Ingress rule setter: expose installed helm charts to the internet, via a single IP.

When to use this solution

• Automating deployments in the cluster.
• Programmatically managing the cluster from the code.

Requirements

• Kubernetes 1.9+ with RBAC enabled.
• Helm

Installation

1. Install Helm with RBAC

   Make sure to grant tiller sufficient permissions to run helm inside the cluster, and install the app Chart. The example below will configure helm and tiller to work with the chart's default values. Execute the following lines if using the default chart values:

   kubectl apply -f https://raw.githubusercontent.com/Microsoft/k8s-On-demand-micro-services-deployment/master/rbac-example/tiller.yaml
   
   helm init --service-account tiller

2. Clone the repository

   • Build and push the image into a docker repository

   docker build -t registryname/on-demand-micro-services-deployment-k8s .
   docker push registryname/on-demand-micro-services-deployment-k8s

   • Edit the values.yaml file to point to the newly published image and registry

3. Install the app chart

   helm install chart/on-demand-micro-services-deployment --name on-demand-micro-services-deployment --set rbac.create=true

4. Call on-demand-micro-services-deployment-k8s and install nginx-ingress-controller, to expose other helm charts via a single public IP:

   curl -d '{"chartName":"stable/nginx-ingress", "releaseName":"myingress"}' -H "Content-Type: application/json" -X POST http://<on-demand-micro-services-deployment-k8s-ip>:4000/install

5. Label each ingress controller. This is required, since this is our way of telling the system, which IPs to use:

   kubectl label service myingress-nginx-ingress-controller appingress=ingress

Now you have a working API awaiting HTTP requests.

Using the API

If you used the default settings, the API will be accessible internally at: http://on-demand-micro-services-deployment-k8s.default.svc.cluster.local:4000

Here is a quick node.js snippet that makes use of the API to install RabbitMQ with default settings.

Note: any chart will be suitable, RabbitMQ is just a specific example:

let chart = { name: "stable/rabbitmq", servicePort: 5672 };

// perform helm install
var installResponse = await requestPostAsync(Paths.HelmInstall, { form: { chartName: "" } });

// create a rule to expose the new service expternally
var ingressResponse = await requestGetAsync(Paths.SetIngressRule, { serviceName: installResponse.serviceName, servicePort: chart.servicePort });

return `Your new service: ${ingressResponse.releaseName}, is publicly accessible on ${ingressResponse.ip}:${ingressResponse.port}`;

For example, installig a chart from a private repository is done by sending the following json to the exposed endpoint:

{
  "chartName":"sampleApp",
  "releaseName":"sampleApp1",
  "privateChartsRepo": "https://raw.githubusercontent.com/username/helm_repo/master/index.yaml"
}

In most cases the deployed chart will be a private one, which makes sense to be deployed on demand, for example a service which adds support for a new format of requests. However, public charts are fully supported (remove the 'privateChartsRepo' key-value).

API Documentation

Check out the API documentation here

How it works

Feel free to read the extended summary here

Contributing

This project welcomes contributions and suggestions. Most contributions require you to agree to a Contributor License Agreement (CLA) declaring that you have the right to, and actually do, grant us the rights to use your contribution. For details, visit https://cla.microsoft.com.

When you submit a pull request, a CLA-bot will automatically determine whether you need to provide a CLA and decorate the PR appropriately (e.g., label, comment). Simply follow the instructions provided by the bot. You will only need to do this once across all repos using our CLA.

This project has adopted the Microsoft Open Source Code of Conduct. For more information see the Code of Conduct FAQ or contact [email protected] with any additional questions or comments.

helm-web-api's Issues

Helm 3 Support

Does this work with helm 3? I noticed in the instructions it lists tiller, which is no longer used with helm 3.

Custom values not working

Custom values provided are ignored, because the check on whether values are set is inverted.

helm-web-api/on-demand-micro-services-deployment-k8s/helm.js

Line 116 in b8d9452

if (this.deployObject) {

I think the this keyword should be omitted, because it is in a static method and I am not sure what it refers to here.

A follow up issue is that the sample code uses a url-encoded form to send the POST data, and the body parser does not seem to work well as it is. Using JSON encoding in the POST solves the issue.

"Error: Not Found" | when trying to post

Getting the below error when trying to post any releases or when trying to access the endpoint through the browser (helm 2.14.3 & kubectl 1.16.0):

Error: Not Found
at /usr/src/index.js:140:15
at Layer.handle [as handle_request] (/usr/src/node_modules/express/lib/router/layer.js:95:5)
at trim_prefix (/usr/src/node_modules/express/lib/router/index.js:317:13)
at /usr/src/node_modules/express/lib/router/index.js:284:7
at Function.process_params (/usr/src/node_modules/express/lib/router/index.js:335:12)
at next (/usr/src/node_modules/express/lib/router/index.js:275:10)
at urlencodedParser (/usr/src/node_modules/body-parser/lib/types/urlencoded.js:91:7)
at Layer.handle [as handle_request] (/usr/src/node_modules/express/lib/router/layer.js:95:5)
at trim_prefix (/usr/src/node_modules/express/lib/router/index.js:317:13)
at /usr/src/node_modules/express/lib/router/index.js:284:7

It is possible to use my own charts (local or from a repo server)?

Hi,
good project, but I've a question. In your examples you use the charts from k8s repo like nginx-ingress.
Can I use my charts, from local drive(mounted in pod) or from a web server?

privateChartsRepo functionality if not working

Steps to Replicate:

curl -d '{"chartName":"chart-name", "releaseName":"chart-name-1", "privateChartsRepo":"https://<gs-bucket-name-helm-repo>.storage.googleapis.com/index.yaml"}' -H "Content-Type: application/json" -X POST http://<service-endpoint>:4000/install

Expected behavior:

should return success

Actual result:

{"status":"failed","reason":"Installation failed."}