project-ktb/side-project-gitops

Purpose and Delivery Model

This is a GitOps repository for deploying a Kubernetes platform and a single side-project-backend workload. Argo CD application manifests define the reconciliation entry points; Helm charts and pinned dependencies define desired cluster state. There is no application source code or README guidance in the repository.

GitOps Control Plane

Argo CD continuously deploys both platform services and the backend from this repository. Each application is independently declared, allowing lifecycle separation between workload, observability, storage, and logging.

• argocd/applications/side-project-backend.yaml

• argocd/applications/monitoring-stack.yaml

• argocd/applications/loki.yaml

Backend Workload Chart

The custom Helm chart packages the backend’s Kubernetes-facing contract: deployment, service discovery, ingress exposure, runtime configuration, and Prometheus scraping. Production settings are separated from defaults, making the chart the primary workload deployment boundary.

• charts/side-project-backend/templates/deployment.yaml

• charts/side-project-backend/templates/configmap.yaml

• charts/side-project-backend/templates/ingress.yaml

• charts/side-project-backend/values.yaml

• charts/side-project-backend/values-prod.yaml

Traffic and Observability Boundary

The backend is exposed through a Kubernetes Service and Ingress, while a ServiceMonitor connects it to the Prometheus operator’s discovery model. Metrics therefore flow from backend pods through the ServiceMonitor into the monitoring stack.

• charts/side-project-backend/templates/service.yaml

• charts/side-project-backend/templates/ingress.yaml

• charts/side-project-backend/templates/servicemonitor.yaml

• platform/monitoring/kube-prometheus-stack/values.yaml

Monitoring Platform

The repository deploys a kube-prometheus-stack for metrics collection, alerting, and visualization components supplied by the upstream Helm dependency. Metrics Server provides Kubernetes resource metrics. A MySQL exporter exposes database metrics to the same Prometheus-based monitoring plane.

• argocd/applications/monitoring-stack.yaml

• platform/monitoring/kube-prometheus-stack/Chart.yaml

• argocd/applications/metrics-server.yaml

• platform/monitoring/mysql-exporter/values.yaml

Logging Platform

Grafana Alloy is deployed as the log collection/forwarding agent, and Loki is the centralized log backend. Their charts are locally vendored with locked versions, so log ingestion and retention behavior is controlled through repository values rather than floating remote chart versions.

• argocd/applications/alloy.yaml

• platform/logging/alloy/values.yaml

• argocd/applications/loki.yaml

• platform/logging/loki/values.yaml

Persistent Storage and AWS Dependency

Persistent-volume provisioning is delegated to the AWS EBS CSI driver. The repository also defines a StorageClass, establishing AWS EBS as the expected durable block-storage provider for stateful platform components such as Loki or monitoring storage when configured to use persistence.

• argocd/applications/aws-ebs-csi-driver.yaml

• platform/storage/aws-ebs-csi-driver/templates/storageclass.yaml

• platform/storage/aws-ebs-csi-driver/values.yaml

Runtime and Dependency Boundaries

The runtime is Kubernetes with Helm rendering and Argo CD reconciliation. External infrastructure assumptions are AWS EBS storage and a MySQL instance reachable by the MySQL exporter; the database itself is not provisioned here. Upstream platform charts are vendored as .tgz dependencies and version-locked, while the backend chart is maintained directly in-repo.

• platform/monitoring/mysql-exporter/Chart.lock

• platform/logging/loki/Chart.lock

• platform/storage/aws-ebs-csi-driver/Chart.lock