Awesome

IllumiDesk Helm Chart

Overview

Use this helm chart to install IllumiDesk into your Cluster. This chart depends on the jupyterhub.

This setup pulls images defined in the illumidesk/values.yaml file from DockerHub. To push new versions of these images or to change the image's tag(s) (useful for testing), then follow the instructions in the build images section.

TL;DR

  helm repo add illumidesk https://illumidesk.github.io/helm-chart/
  helm repo update
  helm upgrade --install $RELEASE illumidesk/illumidesk --version $SEMVER --namespace $NAMESPACE --values example-config/values.yaml --debug

Prerequsites

• helm >= v3
• Kubectl >= 1.17

Testing Locally

1. Install Kind
2. Create a directory called illumidesk-nb-exchange
3. Inside that directory create a sub directory called illumidesk-courses
4. Create a cluster with the following YAML config

      kind: Cluster
      apiVersion: kind.x-k8s.io/v1alpha4
      nodes:
      - role: control-plane
        extraMounts:
        - hostPath: ./illumidesk-nb-exchange
          containerPath: "/illumidesk-nb-exchange"
        - hostPath: ./illumidesk-nb-exchange/illumidesk-courses
          containerPath: "/illumidesk-courses"

5. Run the following command to create the cluster
   • kind create cluster --name {cluster name} --config test-cluster.yaml
6. Load images as needed into local Kind cluster
   • kind load docker-image illumidesk/grader-notebook:712 --name {cluster name}
7. Create namespace in kubernetes
   • kubectl create namespace {}
8. If testing locally, fetch pull request and cd into root directory, helm-chart
9. To Deploy:
   • helm upgrade --install {name} charts/illumidesk/ -n {namespace} -f bar-us-west-2-config_go_grader.yaml --debug --dry-run
   • NOTE: name and namespace should match
   • NOTE: remove --dry-run flag to run deploy illumidesk stack
10. Port forward with * kubectl port-forward svc/proxy-public -n {namespace} 8000:80

Troubleshooting

* Get logs
  * `kubectl logs {pod} -n {namespace}`
* Exec into pod
  * `kubectl exec -it {pod} -n {namespace} -- /bin/bash`
* If helm chart is hanging, it is likely beacuse you need to load the docker image locally
  * `kubectl get pods -n {namespace}`
    * NOTE: if the `hook-image-puller` is in `init` status, log the pod and see if there are images that the cluster cannot pull

Installing the chart

Create config.yaml file and update it with your setup.

NOTE: to get a token use openssl rand -hex 32:

• Add Illumidesk repository to HELM:

    helm repo add illumidesk https://illumidesk.github.io/helm-chart/
    helm repo update

• Install a release of the illumidesk helm chart:

  RELEASE=illumidesk
  NAMESPACE=illumidesk
  helm upgrade \
    --install $RELEASE \
    illumidesk/illumidesk \
    --version 3.2.0
    --namespace $NAMESPACE \
    --values my-custom-config.yaml

• Install a release wtihout config.yaml file

kubernetes create namespace test
helm upgrade --install test --set proxy.secretToken=XXXXXXXXXX illumidesk/illumidesk --version 3.2.0 -n test

Steps to setup argo on a new cluster

helm install argo argo/argo-workflows -n argo --create-namespace

helm install argo-events argo/argo-events --set singleNamespace=false --set namespace=''  -n argo-events --create-namespace

helm install argo-events-nats nats/nats --set cluster.enabled=true -n argo-events

helm install argo-events-stan nats/stan --set stan.nats.url=nats://argo-events-nats.argo-events.svc.cluster.local:4222 --set stan.clusterID=argo-events-stan --set cluster.enabled=true -n argo-events

Uninstall the Chart

    helm uninstall $RELEASE -n $NAMESPACE

Configuration

Note: Please follow instructions to install the Cert Manager if you are using the ALB or Nginx Ingress Controller

Note: Please follow instructions to setup external dns, if you plan to use this resource

Note: Please follow reference guides in the values.yaml in order to properly configure the resource during a deployment

NOTE: The following envars must be set:

• Jupyterhub.hub.extraEnvar.POSTGRES_NBGRADER_PASSWORD
• Jupyterhub.hub.extraEnvar.POSTGRES_JUPYTERHUB_PASSWORD
• Jupyterhub.hub.extraEnvar.JUPYTERHUB_API_TOKEN
• Jupyterhub.hub.extraEnvar.JUPYTERHUB_CRYPT_KEY

NOTE: The following envars must be set depending on autentication type

• Jupyterhub.hub.extraEnvar.LTI_SHARED_SECRET
• Jupyterhub.hub.extraEnvar.OIDC_CLIENT_SECRET

Cluster Helm Chart

EFS CSI Driver

1. Navigate to the `policy folder' and create a policy for EFS
2. Create a policy for EFS CSI driver using the policy document

      aws iam create-policy \
    --policy-name AmazonEKS_EFS_CSI_Driver_Policy \
    --policy-document file://policy/iam-policy-efs-csi-driver.json

3. Get the region-code and oidc-id to pass into trust policy

    aws eks describe-cluster --name {cluster} --query "cluster.identity.oidc.issuer" --output text

4. Use the example policy/trust-efs-csi-driver-policy-example.json to create the trust policy for efs csi driver

    aws iam create-role \
  --role-name AmazonEKS_EFS_CSI_DriverRole \
  --assume-role-policy-document file://"policy/trust-efs-csi-driver-policy-example.json"

5. Attach efs csi driver IAM policy to the role created in the previous step

aws iam attach-role-policy \
  --policy-arn arn:aws:iam::{account_id}:policy/AmazonEKS_EFS_CSI_Driver_Policy \
  --role-name AmazonEKS_EFS_CSI_DriverRole

6. Deploy Cluster level resources

  helm upgrade --install {release} illumidesk/cluster --namespace kube-system -f {cluster-stage-custom-config}.yaml --debug --dry-run

Configuration

Validate the Helm Chart

• For nodeport you will need to use your one of your node ips and also the port you defined in your values file.
  • Open up your browser and use the NODE_IP:NODE_PORT
  • Use the following command to list out your nodes:

   kubectl get nodes -o wide

• For load balancer you will need to get the external IP for proxy-public
  • Use this command to view your services and then paste the loadbalancer dns that is is the external ip of proxy-public

  kubectl get svc -n $NAMESPACE

• For Application Load Balancer, you must have specified the host in your values file
  • Verify the dns has propgates your domain

    dig $HOST 

• Open up your browser and paste the value for your host

Cleanup

  helm delete <release name> --purge