Skip to content

deployment.md

Latest commit

 

History

History
146 lines (100 loc) · 4.79 KB

deployment.md

File metadata and controls

146 lines (100 loc) · 4.79 KB
Raw

Deployment

Docker Compose (recommended for local and demo environments)

The repository includes a full stack in docker-compose.yml:

  • Application services: api, coordinator, worker
  • Infrastructure: redis, rabbitmq, consul
  • Observability: jaeger, prometheus, grafana

Quick start

From the repository root:

./setup.sh
make up-detached
make wait-api
make ui

Or use a single command to start backend and web UI:

make dev

Manual equivalent:

docker compose up -d --build
curl http://localhost:5281/health/ready
cd frontend && npm install && npm run dev

Service endpoints

Service URL
API http://localhost:5281
Coordinator http://localhost:5200
Worker health http://localhost:5101/health/ready
RabbitMQ management http://localhost:15672
Consul UI http://localhost:8500
Jaeger UI http://localhost:16686
Prometheus http://localhost:9090
Grafana http://localhost:3000

Swagger UI is available at http://localhost:5281/swagger when ASPNETCORE_ENVIRONMENT=Development.

Demo accounts

The API seeds two development accounts on startup when running in Development (Docker Compose sets this by default):

Role Email Password
Admin admin@example.com ChangeMe-Admin-12
Standard user user@example.com ChangeMe-User-12

Sign in at http://localhost:5173/login/. See configuration.md for seed settings.

Health checks

Compose uses service readiness endpoints:

  • API: GET /health/ready
  • Coordinator: GET /health/ready
  • Worker: GET /health/ready on port 5101

The worker service sets stop_grace_period: 35s to allow drain and Consul deregistration before shutdown.

Container images

Each host has a multi-stage Dockerfile:

  • src/DistributedQuery.Api/Dockerfile
  • src/DistributedQuery.Coordinator/Dockerfile
  • src/DistributedQuery.Worker/Dockerfile

Images publish Release builds, run as the non-root app user, and expose only required ports.

Running services individually

For local development without Compose, start infrastructure first:

docker compose up -d redis rabbitmq consul jaeger

Then run each host:

dotnet run --project src/DistributedQuery.Coordinator
dotnet run --project src/DistributedQuery.Worker
dotnet run --project src/DistributedQuery.Api

Ensure CoordinatorClient:BaseUrl, Redis, Consul, and RabbitMQ settings match your environment.

Web UI

The React frontend in frontend/ is not included in docker-compose.yml. Use make dev or make ui from the repository root (see README).

For production, build static assets and serve them from a CDN or web server:

cd frontend
npm ci
npm run build

Deploy the frontend/dist/ output as a static SPA with fallback routing to index.html. Place the UI on the same origin as the API or behind a reverse proxy that forwards API paths to the backend. The backend does not configure CORS for browser cross-origin requests.

See frontend.md for environment variables and authentication setup.

Troubleshooting

Problem What to check
docker compose fails Docker Desktop is running
API not ready Run docker compose ps and wait for api, coordinator, and worker to be healthy
Frontend loads but login fails Backend is up at http://localhost:5281/health/live; frontend/.env has VITE_API_PROXY_TARGET=http://localhost:5281
Redirect to login on every page Sign in at /login/ with a seeded account (see Demo accounts above) or register at /signup/
Port already in use Stop other services on ports 5281 (API) or 5173 (frontend), or change the frontend port: npm run dev -- --port 5174

Production considerations

  • Replace Compose defaults with environment-specific secrets and connection strings.
  • Set stable RSA JWT keys for the API (Authentication__JwtSigning__*).
  • Disable development account seeding (Authentication__Seed__Enabled=false) and use real identity management.
  • Connect shard databases as described in connecting-databases.md.
  • Run Redis, RabbitMQ, and Consul as managed or clustered services for high availability.
  • Enable OTLP export and Prometheus scraping in your observability platform.
  • Terminate TLS at a load balancer or ingress in front of the API.

Kubernetes manifests are not included in this repository. Add them when you need cluster orchestration, autoscaling, and managed secret distribution.

Observability stack

Compose provisions:

  • OTLP trace export to Jaeger (OpenTelemetry:OtlpEndpoint=http://jaeger:4317)
  • Prometheus scraping via infra/prometheus.yml
  • Grafana datasource and dashboard provisioning under infra/grafana/

Do not log SQL parameter values, credentials, or raw result rows.