If your VPS is down for two hours and you didn't know, monitoring failed. This guide walks through the free monitoring stack that catches real problems — uptime, resource exhaustion, disk fill, cron failure — without paying for Datadog.

What you're actually monitoring for

• Site down. External check that hits your URL every 1-5 minutes.
• Server unreachable. Ping / SSH availability.
• Disk filling. Log growth, database growth, forgotten /tmp files.
• RAM exhaustion. OOM killer running on your services.
• CPU pinned. Sustained 100% CPU means something's stuck or being abused.
• Cron didn't run. Silent backup failure is the classic disaster.
• SSL certificate about to expire. Automatically renew or die.
• Unusual traffic. DDoS, spam, spike.

The free stack

• UptimeRobot — external URL monitoring. Free tier: 50 monitors, 5-minute checks.
• Netdata — local resource dashboards. Or Prometheus + node_exporter + Grafana if you want more control.
• Healthchecks.io — cron ping monitoring. Free tier: 20 checks.
• Fail2ban — blocks brute-force SSH attempts.
• logwatch — nightly emailed log summary.

Total cost: $0. Total setup time: 1 hour.

Layer 1: UptimeRobot

• Sign up at uptimerobot.com. Free.
• Add HTTP monitor for https://yourdomain.com.
• 5-minute check interval on free tier (paid: 1-minute).
• Alerts via email, Discord webhook, Slack webhook, Telegram bot.

Set up a Discord webhook alert — Discord's mobile notifications are instant and free:

• Discord server → channel → Settings → Integrations → Webhooks → New Webhook. Copy URL.
• UptimeRobot → Alert Contacts → Add → Discord → paste webhook URL.

Now when your site is down, your phone buzzes.

Add ping monitors for critical URLs — the login page, the API health endpoint, checkout.

Layer 2: Netdata (local dashboards)

One-command install:

```bash

bash <(curl -Ss https://my-netdata.io/kickstart.sh)

```

Netdata now runs on localhost:19999 — every metric (CPU, RAM, disk I/O, network, per-service resource) visible in real-time.

Firewall Netdata — don't expose it to the internet:

```bash

ufw allow from 127.0.0.1 to any port 19999

```

Access via SSH tunnel: ssh -L 19999:localhost:19999 you@vps → open http://localhost:19999 locally.

For remote access, use Netdata Cloud's free tier (aggregates your nodes). Or use Tailscale to reach the port over your private tailnet.

Layer 3: Alerting from Netdata

Netdata has built-in health monitoring. Configure /etc/netdata/health_alarm_notify.conf:

```

SEND_EMAIL="YES"

DEFAULT_RECIPIENT_EMAIL="[email protected]"

SEND_DISCORD="YES"

DISCORD_WEBHOOK_URL[sysadmin]="https://discord.com/api/webhooks/..."

DEFAULT_RECIPIENT_DISCORD="sysadmin"

```

Now high CPU, low disk, unresponsive services all page you.

Layer 4: Healthchecks.io for cron

For every cron job that matters (backups, log rotation, cache purges):

• Sign up at healthchecks.io. Free.
• Create a check. Copy the ping URL.
• Add to your cron script: curl -fsS -m 10 --retry 5 -o /dev/null https://hc-ping.com/YOUR-UUID after successful run.

If your cron doesn't ping in the expected window, Healthchecks alerts you.

Ideal for daily backup verification: no ping in 25 hours → email alert → you know backup broke last night.

Layer 5: Fail2ban

Install and configure fail2ban to block brute-force attempts:

```bash

apt install fail2ban

```

Default config bans IPs after 5 failed SSH attempts for 10 minutes. Turn it up:

```

/etc/fail2ban/jail.local

[sshd]

enabled = true

findtime = 10m

maxretry = 3

bantime = 1h

```

Restart: systemctl restart fail2ban.

Check active bans: fail2ban-client status sshd.

Layer 6: logwatch

Nightly emailed summary of your logs:

```bash

apt install logwatch

```

Config: /etc/logwatch/conf/logwatch.conf — set Detail = Med and MailTo = [email protected].

Runs via /etc/cron.daily/00logwatch. Every morning you get an email summarising the previous 24 hours of logs. You'll see brute-force patterns, unusual traffic, and errors before they become incidents.

Layer 7 (optional): Prometheus + node_exporter + Grafana

For teams: run Prometheus + node_exporter on the VPS, Grafana on a separate management box.

• Prometheus scrapes metrics from node_exporter.
• Grafana visualises them.
• AlertManager routes alerts.

Overkill for a single-VPS setup; correct for a fleet of 5+ boxes.

SSL certificate expiry

If you use Certbot with --nginx, auto-renewal is already set up via /etc/cron.d/certbot.

Verify:

```bash

certbot renew --dry-run

```

Add a Healthchecks.io check that pings when renewal runs — silent Certbot failure means expired certs at 3am on a Saturday.

What NOT to do

• Enable Netdata's public dashboard on port 19999 over the open internet. Exposes system information; also occasionally used as an attack vector.
• Ignore log emails. logwatch is only useful if you actually read it.
• Alert on everything. Fatigue makes you ignore alerts. Alert on what actually needs immediate action.
• Rely only on external monitoring. External uptime says nothing about database corruption or slow queries.

Cost view

• UptimeRobot free tier: $0.
• Netdata (self-hosted): $0.
• Healthchecks.io free tier: $0.
• Fail2ban: $0.
• logwatch: $0.

For a paid upgrade path:

• UptimeRobot Pro (1-minute checks): $8/mo.
• Netdata Cloud: free for individuals, ~$10/user/month for teams.
• Healthchecks.io paid: $20/mo for 100+ checks.
• Total under $40/mo for the paid stack. Under $0 for the free stack.

Related guides

• How to set up automated backups on any VPS — pairs with monitoring.
• Shared hosting to plain VPS migration — includes basic monitoring setup.
• Best hosting for a small SaaS MVP — MVP-scale monitoring recommendations.