- Replace non-persistent emptyDir with PersistentVolumeClaim for database storage
- Add imagePullSecrets to both API and frontend deployments for private registry access
- Implement database initialization Job that creates schema and populates fake data
- Fix incomplete frontend-deployment.yaml YAML structure
- Add database initialization ServiceAccount with minimal privileges
- Ensure idempotent initialization (checks if DB exists before creating)
- Update kustomization.yaml to include all new resources in correct order
These changes ensure the deployment:
1. Persists database across pod restarts
2. Authenticates with private container registry
3. Automatically initializes the database with schema and sample KPI data
4. Follows DRY and KISS principles with single reusable init job
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
ARCHITECTURE CHANGES:
- API service: Node.js server on port 3001 (Dockerfile.api)
- Frontend service: Nginx serving React on port 80 (Dockerfile.frontend)
- Each service has its own deployment, service, and replicas
- Ingress routes / to frontend and /api/ to API
KUBERNETES MANIFESTS:
- api-deployment.yaml: 2 replicas of Node.js API server
- api-service.yaml: ClusterIP service for API
- frontend-deployment.yaml: 2 replicas of Nginx frontend
- frontend-service.yaml: ClusterIP service for frontend
- Updated ingress.yaml: Routes traffic based on paths
- Updated kustomization.yaml: References new deployments
DOCKER IMAGES:
- Dockerfile.api: Minimal Node.js image for API (~200MB)
- Dockerfile.frontend: Nginx + React build (~50MB)
- Separate builds in workflow for independent versioning
NGINX CONFIGURATION:
- Removed API proxy (separate service now)
- Simplified config for static file serving only
BENEFITS:
- Independent scaling (can scale frontend/API separately)
- Smaller images with minimal base images
- API errors don't affect frontend availability
- Easier to update one service without affecting the other
🤖 Generated with Claude Code
Co-Authored-By: Claude <noreply@anthropic.com>
- Remove PersistentVolumeClaim definition from deployment.yaml
- Replace persistentVolumeClaim volume with emptyDir for temporary storage
- Database files will be stored in pod memory/disk (ephemeral)
- Pods can now schedule without waiting for storage provisioning
Note: Data is not persistent across pod restarts. For production, implement proper storage class or migrate to persistent storage solution.
🤖 Generated with Claude Code
Co-Authored-By: Claude <noreply@anthropic.com>
- Update kustomization.yaml with newName field for proper image replacement
- Replace individual kubectl apply commands with kustomize build
- Use kustomize edit to dynamically set image tag from CI/CD pipeline
- Simplify deployment to single 'kubectl apply -k' command
- Consolidate verification steps
Benefits:
- Single source of truth for manifest configuration
- Declarative image tag management
- Easier to manage overlays for different environments
- Reduced template duplication
🤖 Generated with Claude Code
Co-Authored-By: Claude <noreply@anthropic.com>
Major Changes:
- Migrate UI to Shadcn components with Tailwind CSS v3
- Implement dark theme as default with improved color scheme
- Optimize homepage layout to fit single screen without scrolling
- Fix chart visibility with explicit colors for dark mode
Deployment Infrastructure:
- Add Docker multi-stage build with Nginx + Node.js
- Create Kubernetes manifests (deployment, service, ingress, PVC)
- Configure Gitea CI/CD workflow with registry integration
- Add deployment scripts with registry support
CI/CD Configuration:
- Registry: gitea.vidoks.fr/sortifal/pfee
- Automatic build and push on commits
- Kubernetes deployment with image pull secrets
- Three-stage pipeline: build, deploy, notify
Documentation:
- Add DEPLOYMENT.md with comprehensive deployment guide
- Add SETUP-REGISTRY.md with step-by-step registry setup
- Add workflow README with troubleshooting guide
- Include configuration examples and best practices
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
