Simplify Amazon EKS multi-tenant application deployment by using Flux

In the bootstrap folder, open the bootstrap.sh file and update the variable values for the S3 bucket name, DynamoDB table name, and AWS Region:

S3_BUCKET_NAME="<S3_BUCKET_NAME>" 
DYNAMODB_TABLE_NAME="<DYNAMODB_NAME>" 
REGION="<AWS_REGION>"

Run the bootstrap.sh script. The script requires the AWS CLI, which you installed as part of prerequisites.

cd bootstrap
./bootstrap.sh

After the bootstrap process completes successfully, copy the S3 bucket and DynamoDB table name from the variables section of the bootstrap.sh script:

# Variables
S3_BUCKET_NAME="<S3_BUCKET_NAME>"
DYNAMODB_TABLE_NAME="<DYNAMODB_NAME"

Paste those values to the run.sh script, which is in the root directory of the project:

BACKEND_BUCKET_ID="<SAME_NAME_AS_S3_BUCKET_NAME>"
DYNAMODB_ID="<SAME_NAME_AS_DYNAMODB_NAME>"

Upload the project code to a CodeCommit repository. You can automatically create this repository through Terraform by setting the following variable to true in the demo/pipeline/locals.tf file:

create_new_repo = true

Update the locals.tf file according to your requirements to create pipeline resources.

In the templates folder, make sure that the buildspec files have the following variable set to network:

TF_MODULE_TO_BUILD: "network"

On the CodePipeline console, on the pipeline details page, start the pipeline by choosing Release change.

In the demo/pipeline folder, open the locals.tf file, and set the vpc_enabled variable to true.

Run the run.sh script to apply the changes to the pipeline module:

demo/pipeline/locals.tf
./run.sh -m pipeline -env demo -region <AWS_REGION> -tfcmd init
./run.sh -m pipeline -env demo -region <AWS_REGION> -tfcmd plan 
./run.sh -m pipeline -env demo -region <AWS_REGION> -tfcmd apply

Open the values.yaml file in the following location:

cd cfg-terraform/demo/compute/cfg-tenant-mgmt

The file looks like this:

---
global:
  clusterRoles:
    operator: platform-tenant
    flux: flux-tenant-applier
  flux:
    tenantCloneBaseUrl: ${TEANT_BASE_URL}
    repoSecret: ${TENANT_REPO_SECRET}
tenants:
  tenant-1:
    quotas:
      limits:
        cpu: 1
        memory: 1Gi
    flux:
      path: overlays/tenant-1
  tenant-2:
    quotas:
      limits:
        cpu: 1
        memory: 2Gi
    flux:
      path: overlays/tenant-2

In the global and tenants sections, update the configuration based on your requirements:

Tenant namespaces were created, as specified in values.yaml:

kubectl get ns -A

Quotas are assigned to each tenant namespace, as specified in values.yaml:

kubectl get quota --namespace=<tenant_namespace>

Details of quotas are correct for each tenant namespace:

kubectl describe quota cpu-memory-resource-quota-limit -n <tenant_namespace>

Cilium Network Policies were applied to each tenant namespace:

kubectl get CiliumNetworkPolicy -A

Flux is able to connect to the CodeCommit repository that's specified in the GitOps module:

kubectl get gitrepositories -A

The Flux kustomization controller has deployed the YAML files in the CodeCommit repository:

kubectl get kustomizations -A

All application resources are deployed in their tenant namespaces:

kubectl get all -n <tenant_namespace>

An ingress has been created for each tenant:

kubectl get ingress -n <tenant_namespace>