Cloud-Native DevOps Platform
A self-managed Kubernetes platform on AWS — provisioned with Terraform, configured with Ansible, delivered through Jenkins and Argo CD.
- terraform
- ansible
- kubernetes
- aws
- jenkins
- argocd
- prometheus
- grafana
View sourceMIT
Author
Why it exists
Managed Kubernetes hides the parts worth understanding. This cluster was built without one: three control plane nodes, an etcd quorum, containerd and Calico, provisioned and configured from scratch. Two small applications run on it, but they are not the point — they exist to exercise the delivery path end to end. The platform is the artifact.
What's inside
- 12 Terraform modules across dev, staging and prod environments
- 13 Ansible roles, from baseline hardening to the full cluster
- kubeadm HA control plane with an etcd quorum across three AZs
- Jenkins pipeline with an enforced SonarQube quality gate and Trivy image scan
- Argo CD app-of-apps with self-heal and drift detection
- kube-prometheus-stack with alert routing
Decisions worth explaining
- Why kubeadm instead of EKS?
- Because the goal was to understand what EKS abstracts away. EKS is the right production answer for most teams; it is the wrong learning answer if you have never seen an etcd quorum lose a member.