Deploying with Docker, off the laptop for good

ngrok dies with your laptop. Production means a small server (any VPS with 1 GB RAM is plenty for this bot), your code in a container, and a real HTTPS domain that Meta and Midtrans can call day and night.

Step 1: the Dockerfile

FROM node:20-slim
WORKDIR /app
COPY package*.json ./
RUN npm ci --omit=dev
COPY . .
ENV NODE_ENV=production
CMD ["node", "server.js"]

Step 2: build and run on the server

# .dockerignore first: node_modules, data, .env
docker build -t wa-bot .

docker run -d --name wa-bot \
  --restart unless-stopped \
  --env-file .env \
  -p 127.0.0.1:3000:3000 \
  -v "$(pwd)/data:/app/data" \
  wa-bot

docker logs -f wa-bot   # watch it boot

Step 3: HTTPS with Caddy

# /etc/caddy/Caddyfile  (point bot.yourdomain.com's DNS at the server)
bot.yourdomain.com {
  reverse_proxy localhost:3000
}

Step 4: repoint the webhooks

In the Meta console change the callback URL to https://bot.yourdomain.com/webhook and re-verify; in Midtrans set the notification URL to /payment/notify on the same domain. Send a test message and a sandbox payment to prove both paths reach the server.

The going-live checklist

• Replace the temporary Meta token with a permanent system-user token, and store it only in .env on the server.
• To message any customer (beyond 5 test recipients), complete business verification and register a real number in the Meta console.
• Switch Midtrans from sandbox to production keys and domain when you are ready for real money.
• Back up data/bot.db daily (a cron job copying the file is enough at this scale).
• Point an uptime monitor at the server so you learn about downtime before customers do.

Final checkpoint: turn your laptop off, message the bot, complete a sandbox order and payment. If the confirmation arrives, you have built and shipped a production WhatsApp commerce system, the same architecture our team operates for clients, just smaller.