Deployment

How to run a bot 24/7:

VPS (DigitalOcean, Linode, AWS EC2):

• Standard $5-10/month box.
• Run process in tmux/screen or systemd.
• ~99.9% uptime.

PaaS (Heroku, Railway, Fly.io):

• Push code, they run it.
• Easy.
• Free tiers limited.

Serverless (AWS Lambda, Cloudflare Workers):

• Webhook-triggered.
• Pay per invocation.
• Cold start latency.
• Good for low-traffic bots.

Container (Docker on K8s):

• Reproducible deployment.
• Scale horizontally.
• Heavyweight for single bot.

systemd service example:

[Unit]
Description=My Bot
After=network.target

[Service]
ExecStart=/usr/bin/python /opt/bot/main.py
Restart=always
User=botuser
Environment=BOT_TOKEN=secret

[Install]
WantedBy=multi-user.target

Environment variables:

• Tokens, API keys, DB URLs.
• Never commit to git.
• Use .env in dev; secrets manager in prod.

Logging:

• stdout to systemd journal / Docker logs.
• File for persistence.
• Structured (JSON) for parsing.

Monitoring:

• Uptime monitor (UptimeRobot, BetterUptime).
• Heartbeat: bot sends "I'm alive" to monitor every 5 min.
• Alerts on missed heartbeats.

Auto-restart:

• systemd Restart=always.
• pm2 in Node.js.
• Watchdogs for stuck bots.

Updates:

• Hot reload (limited).
• Rolling restart: stop, deploy, start.
• Brief downtime acceptable for most bots.

Backup:

• Database snapshots.
• Token rotation if compromised.

Privacy:

• Don't log message contents (if avoidable).
• Honor data deletion requests.
• GDPR for EU users.

Cost:

• Small bot: $5/month.
• Mid: $20-50/month.
• Large (many servers): $200+/month + DB + Redis.