GitHub - MiguelRipoll23/gameserver: A game server for multiplayer games

A game server for multiplayer peer-to-peer games.

Used by these games:

• Hood Ball - 2D Rocket League inspired-game
• Air Combat - 3D flight combat simulator

Features

• User registration and authentication using device authenticator
• Secure cloud-based game configuration to update game settings remotely
• Server messages and notifications for connected players
• Matchmaking using token-based pairing with tunneling support
• Player and host identity verification
• Chat messages integrity using digital signatures
• Secure player score management
• Discord bot for management and moderation
• Anti-cheat rules with automatic violation reporting

Configuration

The application runs on Cloudflare Workers, with wrangler.jsonc as the source of truth for the Worker's bindings.

1. Configure the Worker variables and secrets listed in .env.example via the Cloudflare dashboard or wrangler secret put.
2. Create the resources declared in wrangler.jsonc and fill in their IDs for each environment (top-level, staging, and production):
   • GAMESERVER_KV — one KV namespace per environment.
   • HYPERDRIVE — one Hyperdrive configuration per environment (same binding name, different id).
   • WEBSOCKET_DURABLE_OBJECT — declared via durable_objects; no ID required.
3. Connect the Worker to the repository in the Cloudflare dashboard (Settings → Builds), set the build command to pnpm run predeploy, and keep the deploy command as npx wrangler deploy. Commit and push — Workers Builds runs predeploy (migrations + Discord registration) and deploys on every push.

Database configuration

Provision a PostgreSQL database and create the authenticated_user role before running migrations. The deployed Worker connects through Cloudflare Hyperdrive, so the Worker itself never needs a DATABASE_URL secret — only the migration/CI commands (which run outside the Worker) connect to PostgreSQL directly.

Local configuration

Copy .env.example to .env and set:

• DATABASE_URL — used by the local migration commands (pnpm run db:migrate and pnpm run predeploy).
• CLOUDFLARE_HYPERDRIVE_LOCAL_CONNECTION_STRING_HYPERDRIVE — used by pnpm run dev to emulate the HYPERDRIVE binding.

The dev script loads .env into the process environment, which is how Wrangler reads the local Hyperdrive connection string. The same variable works whether you run plain pnpm run dev or --env staging/--env production, because the binding name stays HYPERDRIVE.

CI/CD (staging and production)

Deployment runs through Cloudflare Workers Builds, which authenticates through the dashboard connection — no CLOUDFLARE_API_TOKEN is needed. Set the build command to pnpm run predeploy to apply migrations and register the Discord slash commands before each deploy, and keep the deploy command as npx wrangler deploy. For separate staging and production Workers, connect each environment's Worker and add --env staging/--env production to the deploy command.

Add DATABASE_URL, DISCORD_APPLICATION_ID, and DISCORD_BOT_TOKEN as build variables/secrets in the Workers Builds settings (Settings → Build). The build command runs on every push to the production branch (and on preview branches if enabled), so migrations run before each deploy.

Contributing

I welcome contributions of all kinds! Whether you're fixing bugs, adding new features, improving documentation, or suggesting enhancements, your efforts are appreciated.

Play, Create & Share