flabunsky / azure-container-storage-samples

Azure Container Storage code samples

Introduction

Getting Started with Azure Container Storage

This repo contains the code and instructions to deploy Azure Container Storage using CLI and deploy Jupyter & Kafka workloads.

Pre-requisites

• Install Azure CLI
• Install Kubectl

Installation

# Upgrade to the latest version of the aks-preview cli extension by running the following command.
az extension add --upgrade --name aks-preview

# Add or upgrade to the latest version of k8s-extension by running the following command.
az extension add --upgrade --name k8s-extension

# Set subscription context
az account set --subscription <subscription-id>

# Register resoure providers
az provider register --namespace Microsoft.ContainerService --wait 
az provider register --namespace Microsoft.KubernetesConfiguration --wait

# Create a resource group
az group create --name <resource-group-name> --location <location>

# Create an AKS cluster with Azure Container Storage extension enabled
az aks create -n <cluster-name> -g <resource-group-name> --node-vm-size Standard_D4s_v3 --node-count 3 --enable-azure-container-storage azureDisk

# Connect to the AKS cluster
az aks get-credentials --resource-group <resource-group-name> --name <cluster-name>

# Display available storage pools, one was created when Azure Container Storage was enabled
kubectl get sp -n acstor

# Display storage classes - there should be a storage class created that corresponds to the storage pool
kubectl get sc

Jupyterhub

Pre-requisites

• Install Python
• Install requests Python library
• Install Chocolatey to install helm

Deployment

1. Create config.yaml file to specify that each single user that will be created will get provisioned 1 Gi of storage, from a StorageClass created via Azure Container Storage.

code config.yaml

singleuser:
  storage:
    capacity: 1Gi
    dynamic:
      storageClass: acstor-azuredisk
hub:
  config:
    JupyterHub:
      admin_access: false
    Authenticator:
      admin_users:
        - admin

2. Install jupyterhub

(if not running on Cloud Shell) choco install kubernetes-helm

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

This is installing JupyterHub using the information provided on config.yaml, and creating a namespace (jhub1) where the demo resources will sit.

helm upgrade --cleanup-on-fail --install jhub1 jupyterhub/jupyterhub --namespace jhub1 --create-namespace --values config.yaml

kubectl get service --namespace jhub1

3. Port forward to connect to the service

kubectl --namespace=jhub1 port-forward service/proxy-public 8080:http

Note: If the port-forward doesn't work from CloudShell, you can try using your computer's local terminal

Create additional user sessions

1. From the browser - log on to: http://localhost:8080/ using the credentials from the config.yaml file (username: admin).

2. Generate the token from http://localhost:8080/hub/token. Tokens are sent to the Hub for verification. The Hub replies with a JSON model describing the authenticated user.

3. Update the value of the token in the 'api_token' parameter in the Python script (user_creation.py)

4. Run python script

py user_creation.py

5. Run these commands in your cluster to get the pods and PVCs

kubectl get pvc -n jhub1
kubectl get pods -n jhub1