Deployment

Production deployment guide for AttendiBot. The system consists of three components: Rust backend (bot + API), PostgreSQL, and Next.js web (BFF + dashboard).

Architecture overview

Key principle: The Rust API on port 8080 should be reachable from the web BFF only — not exposed publicly. Auth.js routes (/api/auth/*) must stay on the Next.js app.

Prerequisites

• Docker (recommended) or bare-metal with Rust + Node.js
• PostgreSQL 18+
• Discord application configured — see Discord Setup
• Domain with HTTPS for the web app

Backend deployment

Docker (recommended)

Build and run from backend/:

cd backend
docker compose up --build

The Dockerfile uses cargo-chef with BuildKit cache mounts for faster rebuilds.

Production environment variables

Production rejects default placeholders for WEBHOOK_SECRET and API_INTERNAL_SECRET.

Dokploy notes

• Keep Clean Cache disabled for normal deploys (preserves Docker layer and BuildKit caches)
• Enable Clean Cache only when debugging stale build artifacts
• Typical build times: 1–2 min (source change), 3–4 min (cold or dependency change)

Health check

curl http://localhost:8080/health

Expected: {"status":"ok"} (when database is connected)

Web deployment

Docker / Dokploy

Build from the attendibot-web repo root (Fumadocs MDX in content/docs/ is compiled at build time):

docker build -t attendibot-web .

When working from the AttendiBot monorepo checkout, regenerate MDX from markdown before building:

cd web && npm run sync-docs

Dokploy should use the web repository with build context . (repo root) and Dockerfile at the repository root.

Runtime environment variables:

See billing setup for Stripe Dashboard configuration.

Build args (Docker):

Web → backend connectivity

Set on the web service:

ATTENDIBOT_API_BASE_URL=http://attendibot-backend-xxx:8080
ATTENDIBOT_API_INTERNAL_SECRET=<same as backend>

Use the backend container's internal Docker hostname, not the public site URL.

Verify from the web container:

node -e "console.log(process.env.ATTENDIBOT_API_BASE_URL)"
node -e "fetch(process.env.ATTENDIBOT_API_BASE_URL+'/health').then(r=>r.text()).then(console.log)"

Vercel

Deploy web/ to Vercel. Set the same environment variables. Point ATTENDIBOT_API_BASE_URL to your backend's reachable URL (ensure network access from Vercel to your backend).

Invite URL after deploy

For "Add to Discord" to work, set at least one of:

• Runtime: AUTH_DISCORD_ID or DISCORD_INVITE_URL
• Build time: NEXT_PUBLIC_DISCORD_INVITE_URL or NEXT_PUBLIC_DISCORD_CLIENT_ID

Verify: view page source, search for oauth2/authorize, confirm client_id= is present.

Database

PostgreSQL stores all session records, leaderboard data, signing keys, and configuration.

Migrations

Migrations run automatically when the backend starts (sqlx::migrate!).

Migration files in backend/migrations/:

Backups

Back up PostgreSQL regularly. Signing keys (encrypted) and all session records live in the database.

Post-deploy checklist

After deployment:

1. Backend health check passes (/health)
2. Web can reach backend (verify from web container)
3. Discord OAuth redirect URI matches production URL
4. "Add to Discord" button works
5. Sign in to dashboard succeeds
6. Generate and unlock signing keys
7. Configure tracking and test a voice session
8. Verify a session on the public verifier

See also docs/marketing/post-deploy-checklist.md for SEO and bot directory submission.

Network security

Related docs

• Discord Setup
• Architecture
• Troubleshooting
• Developer Guide