Set up uptime monitoring and alerting

[michaeldistel]

Problem

No monitoring means:

• Downtime invisible until users report
• No SLA tracking
• Can't measure availability
• Slow incident response

Solution

Use UptimeRobot (free tier) for basic monitoring:

UptimeRobot Setup

1. Create account: https://uptimerobot.com

2. Add HTTP(s) monitor:

   • URL: https://controlforge.dev
   • Type: HTTPS
   • Interval: 5 minutes (free tier)

3. Add ping endpoints:

   • Homepage: https://controlforge.dev
   • Docs: https://controlforge.dev/docs
   • Health: https://controlforge.dev/api/health (if created)

Alert Channels

• Email
• Slack webhook
• Discord webhook
• SMS (paid tier)

Health Check Endpoint

// src/routes/api/health/+server.ts
export async function GET() {
  const health = {
    status: 'healthy',
    timestamp: new Date().toISOString(),
    uptime: process.uptime(),
    version: import.meta.env.VITE_BUILD_COMMIT
  };
  
  return new Response(JSON.stringify(health), {
    headers: { 'Content-Type': 'application/json' }
  });
}

Alternative: BetterUptime

More features, better UX:

• Status page (public)
• Incident management
• On-call scheduling
• Free tier: 10 monitors

Monitoring Checklist

• Homepage (https://controlforge.dev)
• Documentation (https://controlforge.dev/docs)
• API health endpoint
• SSL certificate expiry
• DNS resolution
• Response time (<500ms)

Status Page

Create public status page showing:

• Uptime % (30/60/90 day)
• Current status
• Incident history
• Scheduled maintenance

<!-- src/routes/status/+page.svelte -->
<script>
  let status = { operational: true, uptime: 99.9 };
</script>

<h1>System Status</h1>

<div class="status {status.operational ? 'operational' : 'outage'}">
  {status.operational ? '✓ All Systems Operational' : '⚠️ Experiencing Issues'}
</div>

<div class="uptime">
  <h2>Uptime</h2>
  <p>30 days: {status.uptime}%</p>
</div>

Alert Configuration

Send alerts when:

• Site down for >2 minutes
• Response time >3 seconds
• SSL certificate expires <30 days
• 4xx/5xx error rate >5%

Success Criteria

• Monitoring configured for all endpoints
• Alerts sent to email/Slack
• Public status page available
• Historical uptime data tracked
• Response time monitored

[andrianvaleanu]

BetterUptime is a solid alternative for the status page alone, but I've found Pulsetic a bit more reliable for managing regional latency checks without hitting a paywall immediately. Also, make sure that health endpoint returns a 503 instead of a 500 if the database connection drops, otherwise the monitor might miss partial outages.

[michaeldistel]

Thanks for the suggestions!

I've been chatting with Anton about this, and we’re leaning towards Gatus. It’s a great middle ground very lightweight, self-hostable, and uses a simple YAML config for health checks.

It gives us a clean status page and easy alerting (Telegram/Discord) without the complexity of a full-blown monitoring suite. Seems like the most efficient fit for our current stack.