Infrastructure

[WIP] Project is still in progress

Simple infrastructure build with Terraform to create EKS cluster which supporting Istio.

VPC:

• Public and private subnets
• Internet gateway
• NAT gateway in each AZs

EKS:

• EKS managed node groups based on SPOT instances
• Fargate profiles configured for namespace without sidecar injection
• Using public and private subnets

Limitations You cannot use AWS fargate for all pods - aws/containers-roadmap#682.

Terraform

Input

Name	Default value	Description
environment	stg	Environment e.g. staging, production
dns_zone_name	kk.dev
kubernetes_cluster_name	k8s	Name for the Kubernetes cluster
tags	{}	Default tags assigned to all AWS resources
aws_region	eu-west-1
aws_profile	playground	Profile used to deploy changes to AWS
nodes_instances_sizes	["t3.medium"]
auto_scale_options	{ min = 1 max = 2 desired = 1 }	Scale configuration for EKS managed node groups
helm_istio_version	1.15.3
helm_prometheus_version	15.9.0
helm_kiali_version	1.55.1
helm_grafana_version	6.31.1
helm_flagger_version	1.24.1
helm_telepresence_version	2.8.3
release_version	1.0.0	Release version of your infrastructure
telepresence_enabled	true	Determine if telepresence should be in cluster

Output

Name	Description

CD

For progressive delivery operator it's used Flagger. You will find an example of usage canary pattern in Canary example.

Development and debugging

For better DX, there is installed Telepresence.

To start use it on your local computer, follow the instructions.

Maintenance of your apps

Prefered structure of your app

Project should be placed in separate repository. The name of repository should follow the pattern service-{language}-{project_name} e.g. service-go-ecommerce-order-orchestrator.

Structure of catalogs:

service-go-project_name
│   README.md
│
└───source
│   │   your-application-file1.ext
│   │   your-application-file2.ext
|   |   ...
│   
└───charts
│    │   Chart.yaml
│    │   values.yaml
│    |   ...
│    

@TODO Add terraform resource which check your organisation repositories and based on repository name pattern upload all charts. This feature should have configuration option.

Dasboards

To add new dashboard to grafana, add your json to tools/dasboards catalog and run make compress-grafana-dasboards command. Command will minify jsons and put it in tools/dashboards_compressed catalog. Next step is creating kubernetes_config_map resource, you will find examples in istio.tf.

• https://github.com/istio/istio/tree/master/manifests/addons/dashboards
• https://github.com/fluxcd/flagger/tree/main/charts/grafana/dashboards

Makefile

➜ make
run-dry                        Run terraform without deployment
run                            Run terraform with deployment
init                           Terraform init
plan                           Terraform plan
apply                          Terraform apply
destroy                        Terraform destroy
test                           Terraform test
fix                            Fix style
update-kubeconfig              Create or update kubeconfig e.g. make update-kubeconfig profile=playground cluster=my-cluster
apply-podinfo                  Apply podinfo to EKS
generate-cerfificate           Generate certificate e.g. make generate-cerfificate domain=yourdomain.com
compress-grafana-dasboards     Compress grafana dasboards
help                           Display this help message

How to use

Required tools:

• Terraform
• AWS cli
• Kubectl
• Telepresence
• JQ
• GH

Run in your cluster:

1. Configure your AWS profile
2. To run with deployment, use command:

make run

To run without deployment, use command:

make run-dry

3. Optionally you can deploy podinfo app for testing purposes (kubectl is required on your machine):

make apply-podinfo

4. Run make update-kubeconfig profile=playground cluster=stg-k8s-kk-dev-eks to store actual kubeconfig file, you can test if connection working properly by kubectl get svc -n app command.

---

@TODO:

• Route53
• mTLS
• State in dynamodb
• infracost.io
• Optional logging stack
• Components settings adjustment
• Podinfo as helm + variable which decides if should be deployed or not
• Github projects detetion and auto deployment (section: Prefered structure of your app)
• SSM agent
• JWT auth