Skip to content

deployment.md

Latest commit

 

History

History
113 lines (76 loc) · 3.43 KB

deployment.md

File metadata and controls

113 lines (76 loc) · 3.43 KB
Raw

Deployment

Two supported ways to run CarePlan: local Docker Compose, and a short-lived AWS practice stack via Terraform.

Prerequisites

Environment Needs
Local / Docker Docker Desktop (or Compose), optional Python 3.x for non-Docker runs
Next.js product UI Node.js 22 and npm
LLM (Gemini / Vertex) GCP project, ADC via gcloud auth application-default login, .env with GCP_PROJECT / GCP_LOCATION
AWS practice AWS CLI credentials, Terraform ≥ 1.5, password for RDS

Local: Docker Compose

From the repository root:

docker compose up --build

Services:

Service Role Host ports
web Django 8000
worker Celery —
postgres Database 5432
redis Celery broker 6379

Open http://127.0.0.1:8000/.

The same Compose command also starts the frontend service. Open http://127.0.0.1:3000/ for the Next.js product interface. Inside Compose, DJANGO_API_BASE_URL is http://web:8000.

Product UI without Docker

Start Django on port 8000, then run:

cd web
cp .env.example .env.local
npm ci
npm run dev

The UI runs at http://127.0.0.1:3000/. Its server-only DJANGO_API_BASE_URL defaults to http://127.0.0.1:8000; it is not exposed to browser JavaScript.

Optional monitoring profile (Prometheus / Grafana) is documented in Compose; enable only when needed.

Local without Docker

See the Local Run section in README.md: install requirements.txt, set POSTGRES_*, migrate, runserver, and start a Celery worker in a second terminal. PostgreSQL and Redis must already be available.

Tear down (local)

docker compose down

Add -v only if you intend to wipe local database volumes.

Cloud: Terraform practice stack

Infrastructure lives in infra/practice/ (API Gateway, three Lambdas, SQS + DLQ, RDS). This is a learning / validation stack, not a production account layout.

Apply

cd infra/practice
$env:TF_VAR_db_password = "YourStrongPassword1"
terraform init   # first time only
terraform plan
terraform apply

Useful outputs after apply:

  • api_gateway_endpoint / api_post_orders_url
  • queue_url, rds_endpoint
  • Lambda function names

Example smoke calls (PowerShell: use curl.exe, not the curl alias):

curl.exe -X POST https://<api-id>.execute-api.us-east-1.amazonaws.com/orders
curl.exe https://<api-id>.execute-api.us-east-1.amazonaws.com/orders/123

Destroy (required after testing)

cd infra/practice
$env:TF_VAR_db_password = "YourStrongPassword1"
terraform destroy

Cloud resources are created only for testing and destroyed afterward to minimize AWS cost. Leaving RDS and related resources running will incur charges.

Notes

  • Terraform state under infra/practice/ is local and gitignored; do not commit *.tfstate.
  • Lambda packages in this practice module currently use stub handlers under runtimes/aws/stubs/; wiring (IAM, event source, routes, env vars) is what Terraform validates end to end.
  • RDS password is supplied only via TF_VAR_db_password — never commit secrets.

CI

Pull requests run tests via GitHub Actions (see .github/workflows/). CI does not apply Terraform or create AWS resources.

Related docs