Aspir8 from Scratch

Let's deploy Aspire-flavoured apps to a Kubernetes cluster, through Aspir8! Are you new to Kubernetes? Don't worry. Let's start from scratch.

Prerequisites

• for Aspire
  • .NET 8 SDK 8.0.200 or higher with the Aspire workload
  • Visual Studio Code with the C# Dev Kit extension
• for local Kubernetes cluster
  • Docker Desktop
• for Azure
  • Azure subscription
  • Azure CLI
• for AWS
  • AWS subscription
  • AWS CLI
• for GKE
  • TBD
• for NHN Cloud
  • NHN Cloud subscription

Local Kubernetes Cluster Setup through Docker Desktop

1. Install Docker Desktop on you local machine.
2. Enable Kubernetes in Docker Desktop.
3. Deploy sample app to a Kubernetes cluster.

Kubernetes Dashboard Setup

Note: This is only applicable for Kubernetes Dashboard v2.x. From v3.x, use Helm Charts approach.

References

• Deploy and Access the Kubernetes Dashboard
• Creating sample user

1. Get dashboard version.

   # Bash
   dashboard_version=$(curl 'https://api.github.com/repos/kubernetes/dashboard/releases' | \
       jq -r '[.[] | select(.name | contains("-") | not)] | .[0].name')
   
   # PowerShell
   $dashboard_version = $($(Invoke-RestMethod https://api.github.com/repos/kubernetes/dashboard/releases) | `
       Where-Object { $_.name -notlike "*-*" } | Select-Object -First 1).name

2. Install dashboard.

   # Bash
   kubectl apply -f \
     https://raw.githubusercontent.com/kubernetes/dashboard/$dashboard_version/aio/deploy/recommended.yaml
   
   # PowerShell
   kubectl apply -f `
     https://raw.githubusercontent.com/kubernetes/dashboard/$dashboard_version/aio/deploy/recommended.yaml

1. Create admin user.

   kubectl apply -f ./admin-user.yaml

2. Get the access token. Take note the access token to access the dashboard.

   # Bash
   kubectl get secret admin-user \
       -n kubernetes-dashboard \
       -o jsonpath={".data.token"} | base64 -d
   
   # PowerShell
   kubectl get secret admin-user `
       -n kubernetes-dashboard `
       -o jsonpath='{ .data.token }' | `
       % { [Text.Encoding]::UTF8.GetString([Convert]::FromBase64String($_)) }

3. Run the proxy server.

   kubectl proxy

4. Access the dashboard using the following URL:

   http://localhost:8001/api/v1/namespaces/kubernetes-dashboard/services/https:kubernetes-dashboard:/proxy/
   

Aspire-flavoured App Build

1. Install .NET Aspire workload.

   # Bash
   sudo dotnet workload update && sudo dotnet workload install aspire
   
   # PowerShell
   dotnet workload update && dotnet workload install aspire

2. Create a new Aspire starter app.

   dotnet new aspire-starter -n Aspir8

3. Build the app.

   dotnet restore && dotnet build

4. Run the app locally.

   dotnet run --project Aspir8.AppHost

5. Open the app in a browser, and go to the weather page to see whether the API is working or not. The port number might be different from the example below.

   http://localhost:18888
   

Aspire-flavoured App Deployment to Kubernetes Cluster through Aspir8

Use local container registry

1. Install Distribution (formerly known as Registry) as a local Docker Hub (Container Registry).

   docker run -d -p 6000:5000 --name registry registry:latest

   Note: The port number of 6000 is just an arbitrary number. You can choose your own one.

2. Install Aspir8.

   dotnet tool install -g aspirate --prerelease

3. Initialise Aspir8.

   cd Aspir8.AppHost
   aspirate init

   • Set the fall-back container registry to localhost:6000.
   • Skip the repository prefix.
   • Skip the fall-back container tag.
   • Skip the custom directory for the kustomize manifest template.

4. Build and publish the app to the local container registry.

   aspirate generate

   • Choose all components while generating.
   • Select the impage pull policy to IfNotPresent.
   • Generate the top level kustomize manifest for the Kubernetes cluster.

5. Deploy the app to the Kubernetes cluster.

   aspirate apply

   • Choose the Kubernetes cluster to docker-desktop.

6. Check the services in the Kubernetes cluster.

   kubectl get services

7. Update the service type of webfrontent to NodePort.

   kubectl patch svc webfrontend -n default -p '{"spec": {"type": "NodePort"}}'

8. Confirm the webfrontend service type updated, and note the port number of the webfrontend service, 32689 for example.

   kubectl get services

9. Open the app in a browser, and go to the weather page to see whether the API is working or not.

   http://localhost:32689
   

Use Azure Kubernetes Services (AKS)

Note: It uses Azure CLI, which is the imperative approach. The declarative approach using Bicep is TBD.

1. Set environment variables. Make sure that you use the closest or preferred location for provisioning resources (eg. koreacentral).

   # Bash
   export AZURE_ENV_NAME="aspir8$RANDOM"
   export AZ_RESOURCE_GROUP=rg-$AZURE_ENV_NAME
   export AZ_NODE_RESOURCE_GROUP=rg-$AZURE_ENV_NAME-mc
   export AZ_LOCATION=koreacentral
   export ACR_NAME=acr$AZURE_ENV_NAME
   export AKS_CLUSTER_NAME=aks-$AZURE_ENV_NAME
   
   # PowerShell
   $AZURE_ENV_NAME = "aspir8$(Get-Random -Minimum 1000 -Maximum 9999)"
   $AZ_RESOURCE_GROUP = "rg-$AZURE_ENV_NAME"
   $AZ_NODE_RESOURCE_GROUP = "rg-$AZURE_ENV_NAME-mc"
   $AZ_LOCATION = "koreacentral"
   $ACR_NAME = "acr$AZURE_ENV_NAME"
   $AKS_CLUSTER_NAME = "aks-$AZURE_ENV_NAME"

2. Create a resource group.

   az group create -n $AZ_RESOURCE_GROUP -l $AZ_LOCATION

3. Create an Azure Container Registry (ACR).

   # Bash
   az acr create \
       -g $AZ_RESOURCE_GROUP \
       -n $ACR_NAME \
       -l $AZ_LOCATION \
       --sku Basic \
       --admin-enabled true
   
   # PowerShell
   az acr create `
       -g $AZ_RESOURCE_GROUP `
       -n $ACR_NAME `
       -l $AZ_LOCATION `
       --sku Basic `
       --admin-enabled true

4. Get ACR credentials.

   # Bash
   export ACR_LOGIN_SERVER=$(az acr show \
       -g $AZ_RESOURCE_GROUP \
       -n $ACR_NAME \
       --query "loginServer" -o tsv)
   export ACR_USERNAME=$(az acr credential show \
       -g $AZ_RESOURCE_GROUP \
       -n $ACR_NAME \
       --query "username" -o tsv)
   export ACR_PASSWORD=$(az acr credential show \
       -g $AZ_RESOURCE_GROUP \
       -n $ACR_NAME \
       --query "passwords[0].value" -o tsv)
   
   # PowerShell
   $ACR_LOGIN_SERVER = $(az acr show `
       -g $AZ_RESOURCE_GROUP `
       -n $ACR_NAME `
       --query "loginServer" -o tsv)
   $ACR_USERNAME = $(az acr credential show `
       -g $AZ_RESOURCE_GROUP `
       -n $ACR_NAME `
       --query "username" -o tsv)
   $ACR_PASSWORD = $(az acr credential show `
       -g $AZ_RESOURCE_GROUP `
       -n $ACR_NAME `
       --query "passwords[0].value" -o tsv)

5. Create an AKS cluster.

   Note: Depending on the location you create the cluster, the VM size might vary.

   # Bash
   az aks create \
       -g $AZ_RESOURCE_GROUP \
       -n $AKS_CLUSTER_NAME \
       -l $AZ_LOCATION \
       --node-resource-group $AZ_NODE_RESOURCE_GROUP \
       --node-vm-size Standard_B2s \
       --network-plugin azure \
       --generate-ssh-keys \
       --attach-acr $ACR_NAME
   
   # PowerShell
   az aks create `
       -g $AZ_RESOURCE_GROUP `
       -n $AKS_CLUSTER_NAME `
       -l $AZ_LOCATION `
       --node-resource-group $AZ_NODE_RESOURCE_GROUP `
       --node-vm-size Standard_B2s `
       --network-plugin azure `
       --generate-ssh-keys `
       --attach-acr $ACR_NAME

6. Connect to the AKS cluster.

   # Bash
   az aks get-credentials \
       -g $AZ_RESOURCE_GROUP \
       -n $AKS_CLUSTER_NAME \
   
   # PowerShell
   az aks get-credentials `
       -g $AZ_RESOURCE_GROUP `
       -n $AKS_CLUSTER_NAME `

7. Connect to ACR.

   Note: This is the demo purpose only. You should manually enter username and password from your input.

   docker login $ACR_LOGIN_SERVER -u $ACR_USERNAME -p $ACR_PASSWORD

8. Install Aspir8.

   dotnet tool install -g aspirate --prerelease

9. Initialise Aspir8.

   cd Aspir8.AppHost
   aspirate init -cr $ACR_LOGIN_SERVER -ct latest --non-interactive

   Note: If you are asked to enter or skip the repository prefix, enter n to skip it.

10. Build and publish the app to ACR.

    aspirate generate --image-pull-policy IfNotPresent --non-interactive

11. Deploy the app to the AKS cluster.

    aspirate apply -k $AKS_CLUSTER_NAME --non-interactive

12. Install a load balancer to the AKS cluster.

    kubectl apply -f ../load-balancer.yaml

13. Confirm the webfrontend-lb service type is LoadBalancer, and note the external IP address of the webfrontend-lb service.

    kubectl get services

14. Open the app in a browser, and go to the weather page to see whether the API is working or not.

    http://<EXTERNAL_IP_ADDRESS>
    

15. Once you are done, delete the entire resources from Azure.

    az group delete -n $AZ_RESOURCE_GROUP -f Microsoft.Compute/virtualMachineScaleSets -y --no-wait

Use Amazon Elastic Kubernetes Service (EKS)

Note:

• It uses both AWS Console and AWS CLI to provision resources to AWS.
• It uses the Root account for this demo purpose only. You should use the IAM user with the least privilege.

1. Set environment variables. Make sure that you use the closest or preferred location for provisioning resources (eg. ap-northeast-2).

   # Bash
   export AWS_ENV_NAME="aspir8$RANDOM"
   export AWS_ACCOUNT_ID=$(aws sts get-caller-identity --query "Account" --output text)
   export AWS_LOCATION=ap-northeast-2 # Seoul
   export ECR_LOGIN_SERVER=$AWS_ACCOUNT_ID.dkr.ecr.AWS_LOCATION.amazonaws.com
   export EKS_STACK_NAME=aspir8-stack
   export EKS_CLUSTER_NAME=eks-$AWS_ENV_NAME
   export EKS_NODE_GROUP_NAME=aspir8-nodegroup
   
   # PowerShell
   $AWS_ENV_NAME = "aspir8$(Get-Random -Minimum 1000 -Maximum 9999)"
   $AWS_ACCOUNT_ID = $(aws sts get-caller-identity --query "Account" --output text)
   $AWS_LOCATION = "ap-northeast-2" # Seoul
   $ECR_LOGIN_SERVER = "$($AWS_ACCOUNT_ID).dkr.ecr.$($AWS_LOCATION).amazonaws.com"
   $EKS_STACK_NAME = "aspir8-stack"
   $EKS_CLUSTER_NAME = "eks-$AWS_ENV_NAME"
   $EKS_NODE_GROUP_NAME = "aspir8-nodegroup"

2. Create a VPC stack for EKS.

   # Bash
   aws cloudformation create-stack \
       --region $AWS_LOCATION \
       --stack-name $EKS_STACK_NAME \
       --template-url https://s3.us-west-2.amazonaws.com/amazon-eks/cloudformation/2020-10-29/amazon-eks-vpc-private-subnets.yaml
   
   # PowerShell
   aws cloudformation create-stack `
       --region $AWS_LOCATION `
       --stack-name $EKS_STACK_NAME `
       --template-url https://s3.us-west-2.amazonaws.com/amazon-eks/cloudformation/2020-10-29/amazon-eks-vpc-private-subnets.yaml

3. Create an EKS cluster role and attach it to the policy.

   # Bash
   aws iam create-role \
       --role-name Aspir8AmazonEKSClusterRole \
       --assume-role-policy-document file://"eks-cluster-role-trust-policy.json"
   aws iam attach-role-policy \
       --policy-arn arn:aws:iam::aws:policy/AmazonEKSClusterPolicy \
       --role-name Aspir8AmazonEKSClusterRole
   
   # PowerShell
   aws iam create-role `
       --role-name Aspir8AmazonEKSClusterRole `
       --assume-role-policy-document file://"eks-cluster-role-trust-policy.json"
   aws iam attach-role-policy `
       --policy-arn arn:aws:iam::aws:policy/AmazonEKSClusterPolicy `
       --role-name Aspir8AmazonEKSClusterRole

4. Create an EKS cluster node role and attach it to the policies.

   # Bash
   aws iam create-role \
       --role-name Aspir8AmazonEKSNodeRole \
       --assume-role-policy-document file://"eks-node-role-trust-policy.json"
   aws iam attach-role-policy \
       --policy-arn arn:aws:iam::aws:policy/AmazonEKSWorkerNodePolicy \
       --role-name Aspir8AmazonEKSNodeRole
   aws iam attach-role-policy \
       --policy-arn arn:aws:iam::aws:policy/AmazonEC2ContainerRegistryReadOnly \
       --role-name Aspir8AmazonEKSNodeRole
   aws iam attach-role-policy \
       --policy-arn arn:aws:iam::aws:policy/AmazonEKS_CNI_Policy \
       --role-name Aspir8AmazonEKSNodeRole
   
   # PowerShell
   aws iam create-role `
       --role-name Aspir8AmazonEKSNodeRole `
       --assume-role-policy-document file://"eks-node-role-trust-policy.json"
   aws iam attach-role-policy `
       --policy-arn arn:aws:iam::aws:policy/AmazonEKSWorkerNodePolicy `
       --role-name Aspir8AmazonEKSNodeRole
   aws iam attach-role-policy `
       --policy-arn arn:aws:iam::aws:policy/AmazonEC2ContainerRegistryReadOnly `
       --role-name Aspir8AmazonEKSNodeRole
   aws iam attach-role-policy `
       --policy-arn arn:aws:iam::aws:policy/AmazonEKS_CNI_Policy `
       --role-name Aspir8AmazonEKSNodeRole

5. Create an EKS cluster by following this document.

6. Create an EKS cluster nodes by following this document.

7. Connect to the EKS cluster.

   aws eks update-kubeconfig --name $EKS_CLUSTER_NAME --region $AWS_LOCATION

8. Connect to ECR.

   aws ecr get-login-password --region $AWS_LOCATION | docker login --username AWS --password-stdin $ECR_LOGIN_SERVER

9. Create repositories in ECR.

   aws ecr create-repository --repository-name apiservice --region $AWS_LOCATION
   aws ecr create-repository --repository-name webfrontend --region $AWS_LOCATION

10. Install Aspir8.

    dotnet tool install -g aspirate --prerelease

11. Initialise Aspir8.

    cd Aspir8.AppHost
    aspirate init -cr $ECR_LOGIN_SERVER -ct latest --non-interactive

    Note: If you are asked to enter or skip the repository prefix, enter n to skip it.

12. Build and publish the app to ECR.

    aspirate generate --image-pull-policy IfNotPresent --non-interactive

13. Deploy the app to the EKS cluster.

    aspirate apply -k $EKS_CLUSTER_NAME --non-interactive

14. Install a load balancer to the EKS cluster.

    kubectl apply -f ../load-balancer.yaml

15. Confirm the webfrontend-lb service type is LoadBalancer, and note the URL under the external IP address column of the webfrontend-lb service.

    kubectl get services

16. Open the app in a browser, and go to the weather page to see whether the API is working or not.

    http://<xxxx.ap-northeast-2.elb.amazonaws.com>
    

17. Once you are done, delete the entire resources from AWS.

    # Delete EKS node group
    aws eks delete-nodegroup --nodegroup-name $EKS_NODE_GROUP_NAME --cluster-name $EKS_CLUSTER_NAME
    
    # Delete EKS cluster
    aws eks delete-cluster --name $EKS_CLUSTER_NAME
    
    # Delete ECR repositories
    aws ecr delete-repository --repository-name apiservice --force --region $AWS_LOCATION
    aws ecr delete-repository --repository-name webfrontend --force --region $AWS_LOCATION
    
    # Delete CloudFormation stack
    aws cloudformation delete-stack --stack-name $EKS_STACK_NAME

    Note:

    • Deleting the EKS node group takes 5-10 mins.
    • Only after the EKS node group is deleted, the EKS cluster can be deleted.
    • While deleting the CloudFormation stack, you might be failing the deletion process. It's highly likely because of Elastic Load Balancer. Go to EC2 Dashboard, and delete the existing load balancer instance first.

Use Google Kubernetes Engine (GKE)

TBD

Use NHN Kubernetes Services (NKS)

Note:

• It uses NHN Cloud Console to manage NHN Kubernetes Service (NKS).
• It uses NHN Container Registry (NCR) as the container registry.

1. Add the following Docker Hub repository details to Aspir8.ApiService/Aspir8.ApiService.csproj.

   <PropertyGroup>
     <ContainerRepository>{{DOCKER_USERNAME}}/apiservice</ContainerRepository>
   </PropertyGroup>

2. Add the following Docker Hub repository details to Aspir8.Web/Aspir8.Web.csproj.

   <PropertyGroup>
     <ContainerRepository>{{DOCKER_USERNAME}}/webfrontend</ContainerRepository>
   </PropertyGroup>

3. Set environment variables.

   export NHN_ENV_NAME="aspir8$RANDOM"
   export NKS_CLUSTER_NAME=nks-$NHN_ENV_NAME

4. Create an NKS cluster from the console.

5. Get the kubeconfig of the NKS cluster from the console.

6. Connect to the NKS cluster using the kubeconfig.

   export KUBECONFIG=~/.kube/config:~/path/to/downloaded/kubeconfig
   kubectl config view --merge --flatten > ~/.kube/merged_kubeconfig
   mv ~/.kube/config ~/.kube/config.bak
   mv ~/.kube/merged_kubeconfig ~/.kube/config

7. Change the context to the NKS cluster.

   kubectl config use-context default

8. Connect to Docker Hub.

   Note: This is the demo purpose only. You should manually enter username and password from your input.

   docker login registry.hub.docker.com -u <DOCKER_USERNAME> -p <DOCKER_PASSWORD>

9. Initialise Aspir8.

   cd Aspir8.AppHost
   aspirate init -cr registry.hub.docker.com -ct latest --non-interactive

10. Build and publish the app to Docker Hub.

    aspirate generate --image-pull-policy IfNotPresent --non-interactive

11. Deploy the app to the NKS cluster.

    aspirate apply -k toast-$NKS_CLUSTER_NAME --non-interactive

12. Install a load balancer to the NKS cluster.

    kubectl apply -f ./load-balancer.yaml

13. Confirm the webfrontend service type is LoadBalancer, and note the external IP address of the webfrontend service.

    kubectl get services

14. Open the app in a browser, and go to the weather page to see whether the API is working or not.

    http://<EXTERNAL_IP_ADDRESS>
    

15. Once you are done, delete the entire resources from the console and container images from Docker Hub.