This repository contains the infrastructure manifests for deploying Feature Flags App.

Feature Flags App running on Amazon EKS, managed with Argo CD and Helm, following modern GitOps principles - “App-of-Apps” pattern.

Components overview

• Core App: Feature Flags API (with MongoDB)
• Monitoring: Kube-Prometheus-Stack (Prometheus, Grafana)
• Logging: EFK Stack (Elasticsearch, Fluent Bit, Kibana)

• Argocd Deployment Flow
• Sync waves
• Argocd Applications Structure
• Grafana Dashboards
• Kibana Dashboards
• Deploy Locally

graph TD
  A[**Root Application**] --> B[Bootstrap Layer]
  B --> B2[ECK Operator]
  B --> B3[MongoDB Operator]
  B --> B4[External Secrets Operator]

  A --> C[Infrastructure Layer]
  C --> C1[Kube Prometheus Stack]
  C --> C2[EFK Stack]
  C --> C3[MongoDB]
  C --> C4[Cert Manager]
  C --> C5[Ingress NGINX]
  C --> C6[External Secrets Setup]

  A --> D[Application Layer]
  D --> D1[Feature-Flags App]

1. Bootstrap

   • Install ECK operator + crds
   • Install MongoDB operator + crds
   • Install External Secrets operator + crds

2. Infrastructure

   • Deploy kube-prometheus-stack
   • Deploy MongoDB
   • Deploy EFK
     • ECK (Elasic Cloud Kubernetes - for Elasticsearch & kibana)
     • Fluent Bit
   • Deploy Ingress NGINX
   • Deploy external-secrets-setup (ClusterSecretStore configuration)
   • Deploy cert-manager

3. Applications

   • Deploy Feature-Flags API

This repository uses a centralized configuration pattern. A single values file overrides settings across all applications and infrastructure components (like domains, regions, and secrets) to ensure consistency.

Source of Truth: argocd/applications/values.yaml

Argo CD Applications reference the global values file using valueFiles.

Example:

source:
  helm:
    valueFiles:
    - ../../argocd/applications/values.yaml  # References the global config

This repository includes several Helm charts located in the infrastructure/ directory, each managing a specific component of the observability and database stack:

Manages TLS certificates for Kubernetes applications. Automatically provisions and renews certificates from Let's Encrypt and other certificate authorities. Essential for securing ingress traffic and enabling HTTPS.

Complete logging stack combining Elasticsearch, Fluent Bit, and Kibana. Fluent Bit collects logs from all pods, forwards them to Elasticsearch for indexing, and Kibana provides a web UI for log analysis and visualization.

Configures the ClusterSecretStore and required ServiceAccount resources for External Secrets integration. Enables secure integration with external secret management systems like AWS Secrets Manager. The External Secrets Operator is deployed separately in the bootstrap layer.

Comprehensive monitoring solution including Prometheus for metrics collection, Grafana for visualization, and Alertmanager for alerting. Pre-configured with custom dashboards for the Feature Flags API.

Deploys MongoDB Community Edition custom resources managed by the MongoDB Kubernetes Operator. Provides the persistent data store for the Feature Flags application.

This Grafana dashboard monitors the Feature Flags API on Kubernetes. Provisioned via the kube-prometheus-stack Helm chart (ConfigMap: feature-flags-api-grafana-dashboard), it combines Flask app metrics (prometheus_flask_exporter) with Kubernetes metrics from Prometheus to visualize traffic, performance, and resource usage.

The dashboard, titled "Feature Flags Dashboard", focuses on high-level log volume, service activity, and error rates, particularly targeting feature flags application common log fields like HTTP status codes (status/code) and Python/standard logging levels (levelname).

The dashboard is structured into several panels for immediate observability into log health and volume.

The dashboard applies a Global Filter requiring the existence of the field kubernetes.container_name.keyword.

• Implication: Only logs originating from Kubernetes containers will be visible. Logs shipped from external sources (e.g., VMs, bare metal) without this specific metadata field will be automatically filtered out.

The dashboard relies on the specific mappings defined in the index_template.json to function correctly.

• Numeric Fields (short/integer): Fields like status, code, and latency_ms are explicitly mapped as numeric types. This allows the dashboard to perform range queries (e.g., status >= 400) and aggregations. If these were mapped as default strings, the error rate logic would fail.
• Keyword Fields: Fields like levelname and kubernetes.container_name use the keyword type, which is required for the "Top 10" bucket aggregations used in the visualization splits.

# Install Argo CD (if not already installed)
kubectl create namespace argocd

kubectl apply -n argocd -f https://raw.githubusercontent.com/argoproj/argo-cd/stable/manifests/install.yaml
t
# Apply the root app
kubectl apply -f argocd/root-app.yaml -n argocd

# Verify
kubectl get ns
kubectl get pods -A
kubectl get ingress -A

# Get the password
kubectl -n argocd get secret argocd-initial-admin-secret \
  -o jsonpath="{.data.password}" | base64 -d; echo

# Login with the password from above
argocd login localhost:8080 --username admin --password <PASSWORD> --insecure

# Port forward 
 kubectl port-forward svc/argocd-server -n argocd 8080:443 

# Get the password 
kubectl -n argocd get secret argocd-initial-admin-secret \
  -o jsonpath="{.data.password}" | base64 -d; echo

# Open argocd UI
# Login with admin user and the retrieved password
http://localhost:8080