delivery

Backend MVP scaffold for multi-vendor, single-rider orchestration.

Prerequisites

• Bun 1.2+
• PowerShell (for smoke test script)
• Redis (only if running BullMQ mode)
• PostgreSQL (only if running Prisma repository mode)

Environment

Copy .env.example to .env and update values.

cp .env.example .env

Important values:

• ORDER_API_PORT=3000
• QUEUE_DRIVER=in-memory or QUEUE_DRIVER=bullmq
• REPOSITORY_DRIVER=memory or REPOSITORY_DRIVER=prisma
• DATABASE_URL=postgresql://postgres:postgres@localhost:5432/delivery
• OSRM_ENABLED=true
• OSRM_BASE_URL=https://router.project-osrm.org
• OSRM_TIMEOUT_MS=10000 (0 disables timeout)
• OSRM_DISABLE_TIMEOUT=false
• OSRM_HEALTH_RETRIES=3
• OSRM_HEALTH_RETRY_DELAY_MS=600
• OSRM_FAILURE_THRESHOLD=3
• OSRM_COOLDOWN_MS=60000
• OSRM_CACHE_TTL_MS=300000
• OSRM_MAX_CONCURRENT_REQUESTS=4
• OSRM_LOG_FAILED_ROUTE_URL=true
• OSRM_LOG_SUCCESS_ROUTE_URL=true
• REDIS_URL=redis://localhost:6379 (required for BullMQ mode)
• VENDOR_MIN_CONFIDENCE=0.65

Run

bun install
bun run dev

Startup logs

• Queue driver (bullmq or in-memory)
• Repository driver (prisma or memory)
• Redis connection status (connected/ping/reconnect/error/closed)
• OSRM API availability
• OSRM route request logs (success/failure URL, coords, timeout, minutes/reason)
• Route planner logs (final stop sequence + rider map URL)
• Dispatch logs (final rider guidance payload)

Endpoints

• GET /health
• POST /api/v1/orders
• GET /api/v1/orders/:orderId
• GET /api/v1/orders/:orderId/route
• POST /api/v1/orders/:orderId/cancel
• GET /api/v1/users/:userId/routes

Notes on OSRM timeouts

If you run many concurrent orders against the public endpoint (router.project-osrm.org), you can hit timeouts/rate limits. This project throttles and caches OSRM estimates, logs route URLs, and falls back to local estimation during cooldown. For heavy load tests, prefer your own OSRM server and set OSRM_BASE_URL to that instance.

Smoke test

bun run smoke:test