Eddy AutoML Helm Charts

Helm Charts for deploying all the components of Eddy AutoML.

Deploying without Ingresses

Build dependent charts

helm dependency build

Create a namespace

kubectl create ns eddy-automl

Install all the charts including a Kafka cluster

helm install -name demo --namespace eddy-automl  ./

or

Install only the Eddy AutoML charts wihout the Kafka cluster

helm install -name demo --namespace eddy-automl --set kafka.enabled=False ./

Deploying with NodePorts

Deploy all the components:

helm install -name demo --namespace eddy-automl -f values.minikube.yaml ./

If you are using minikube, it exposes all services of type NodePort to the host.

Run the following command to get the ports and update the eddy-automl-ui-config ConfigMap accordingly.

minikube service list

Example:

|-------------|--------------------------------|--------------|----------------------------|
|  NAMESPACE  |              NAME              | TARGET PORT  |            URL             |
|-------------|--------------------------------|--------------|----------------------------|
| default     | kubernetes                     | No node port |
| eddy-automl | demo-eddy-automl-backend       | http/8000    | http://192.168.64.10:30887 |
| eddy-automl | demo-eddy-automl-ui            | http/80      | http://192.168.64.10:31951 |
| eddy-automl | demo-eddy-kafka-graphql-bridge | http/8090    | http://192.168.64.10:32306 |

kubectl edit configmap demo-eddy-automl-ui-config

    {
      "debug": false,
      "dataGraphQLClient": {
        "httpEndpoint": "http://192.168.64.10:30887/api/graphql",
        "wsEndpoint": "http://192.168.64.10:32306/subscriptions"
      }
    }

And rollout restart the deployment:

kubectl rollout restart deployment demo-eddy-automl-ui 

And open the UI at: http://192.168.64.10:31951

Deploying with Ingresses (minikube)

Enable the ingress and ingress-dns addons by running:

minikube addons enable ingress

Run minikube ip and add edit values.minikube-ingress.yaml by replacing all the occurence of the IP.

Deploy the components:

helm install -name demo --namespace eddy-automl -f values.minikube.yaml -f values.minikube-ingress.yaml ./

Open the UI at: http://automl-ui.<minikube-ip>.nip.io, for example: http://automl-ui.192.168.64.10.nip.io

Deploying with Ingresses (ngnix-ingress)

Requires proper DNS Zone configuration, cert-manager and ngnix-ingress.

helm install -name demo --namespace eddy-automl -f values.production.yaml ./

Runing instances of these charts can be found at:

https://automl.k.eddy-analytics.org

https://eddy-automl-backend.k.eddy-analytics.org

https://eddy-automl-subscriptions.k.eddy-analytics.org

Uninstalling

helm uninstall demo