🛡 24/7 MONITORING · YOUR VPS

Your site is down at 3 AM. Guardian
restarts it before you wake up.

A lightweight Python daemon that lives on your VPS. It checks your site every 60 seconds, auto-restarts on a crash, and sends a Telegram alert when an issue needs your attention. Installed in 20 minutes. From $70 one-time. The source code is yours—no subscriptions, no per-check pricing.

See Pricing →
60s
Check Interval
< 2min
Down to Restored
256MB
RAM Footprint
$0/mo
After Setup

What actually happens when your site goes down

Most "monitoring" services just send a panic email 10 minutes after your visitors have already left.

Without Guardian: Your site crashes at 03:17 AM (OOM, hung DB, cron deadlock—whatever). UptimeRobot notices at 03:22. You see the email at 09:00 when you wake up. 5 hours and 43 minutes of downtime. Google noticed. Customers left. Your Telegram bot missed webhooks.

With Guardian: Crash at 03:17. Guardian detects it at 03:18, restarts your service at 03:18, and confirms recovery at 03:19. You wake up to a ✅ http_root recovered · 0.01s message in Telegram. Total downtime: 1-2 minutes.

17 Health Checks Out of the Box

Each site gets its own tailored checklist. These are the checks we use on our own AskOracle stack right now.

HTTP 200 + Speed
Main page + key API endpoints
systemd Active
Service state check per unit
PostgreSQL Connect
Live query + timeout
SQLite Health
File readable, queries work
journalctl Errors
Traceback / ERROR in last 15 min
Cron Freshness
Job ran on schedule
HTTPS Certificate
Expiry > 14 days, auto-renew trigger
Log Patterns
Custom regex for your stack
Disk Free
Alert at <10% free
Memory
OOM-risk detection
CPU Runaway
Per-process anomaly
External API
Stripe / Telegram / Anthropic up
Backup Freshness
Last B2 upload < 25h
TLS Key Files
Secrets readable, permissions OK
Response JSON Shape
API returns expected schema
Queue Depth
Celery / Redis / custom
Custom Per-Service
Whatever matters for you

Real Incident Timeline—From Our Own Logs

Guardian is running on AskOracle right now. Here's what happened during a recent test crash:

07:50:13
✅ Guardian started. 17 checks, all OK.
07:50:30
💥 systemctl stop askoracle — simulated crash.
07:51:13
🔴 CRIT detected: http_root, http_store → connection refused. Telegram alert sent.
07:51:13
🛠 Auto-remediation: systemctl restart askoracle
07:52:14
✅ Recovery: http_root, http_store, admin_queue — all back. Telegram: "recovered".

Total human intervention required: zero. You get one alert and one recovery notification.

What Makes Guardian Different

🏠 Lives on Your VPS
Not SaaS. No data leaves your server. Works during network splits.
🛠 Actually Fixes Issues
Most monitors just alert. Guardian restarts services, renews certs, and clears caches—without waking you up.
📱 Telegram Alerts
Only on state changes. No spam. Escalates after 3 failures.
🎯 Tuned Per-Service
We write checks specific to your stack, not generic ping probes.
📝 Source Code is Yours
Clean, auditable Python. You can fork, modify, or do whatever you want. No vendor lock-in.
💾 256MB RAM
Runs on any VPS, including $4/mo droplets. Zero impact on your apps.

Pricing

One-time installation. Optional monthly tuning. No per-check, per-alert, or "usage tier" nonsense.

MOST POPULAR
Setup
$70 one-time
For 1 site or a small stack (1-3 services). Installed, tuned, and documented.
  • 1-3 services monitored
  • Up to 10 custom checks
  • Auto-restart on crash (whitelist)
  • Telegram alerts (state transitions)
  • systemd service + logs
  • Installed in 24h, tested with you
  • Monthly tuning
Request Install →
Full
$249 one-time
For complex stacks, multiple sites, and custom business-logic checks.
  • Up to 10 services / 25 checks
  • Custom log-pattern detection
  • External API health checks
  • Backup freshness verification
  • Queue / disk / CPU anomaly checks
  • 3 months post-launch tuning
  • Priority support
Request Quote →
Enterprise
$699+
For multiple servers, custom integrations, SLAs, and a dedicated channel.
  • Unlimited services / checks
  • Multi-server deployments
  • Integration with PagerDuty / Opsgenie
  • Grafana / Prometheus export
  • Signed NDA + Slack channel
  • 6 months tuning + 99.9% SLA
  • On-call escalation setup
Talk to Us →

Guardian Monitor · $29/mo

We watch the watcher. We review your alert stream weekly, add new checks as your stack evolves, and tune thresholds to kill false positives.

Subscribe →

FAQ

Can't I just use UptimeRobot or Better Uptime?

You can—for basic "is the site up?" pings. But those services ping from the outside: they see "DOWN" 5-10 minutes after the fact, can't see why, and definitely can't restart anything. Guardian sits on your VPS, sees the real state (systemd, logs, DB), and fixes it.

What if Guardian itself crashes?

It's a systemd service with Restart=always, plus a watchdog cron job that checks every 10 minutes that the Guardian process is alive and restarts it if not. It's fully redundant.

What stacks do you support?

Any Linux VPS. We've run Guardian on Python/Flask, Node.js/Express, PHP/Laravel, Ruby/Rails, and static sites on Nginx, plus ad-hoc stacks with cron jobs and background workers.

What services can it auto-restart?

By default: the monitored service itself (via systemctl restart), PostgreSQL/MySQL (if a DB check fails), and certbot renew (if a cert has < 7 days). Everything else is alert-only. You approve the whitelist during setup.

Does this replace my APM or error tracker?

No, it complements them. APMs (like Sentry or Honeybadger) track code errors. Guardian tracks infrastructure health. Run both if you can.

Can I self-install from the source code?

Yes—the code is yours after payment (MIT-like terms). Some clients prefer a DIY approach with our consulting. If that's you, the $70 Setup becomes a 1-hour review and documentation hand-off.

What is the turnaround time?

$70 Setup: same day to 24 hours. $249 Full: 3-5 days. Enterprise: 2 weeks for scoping + 1-2 weeks for installation.

See Our Own Guardian in Action

We eat our own dog food. Guardian watches AskOracle right now—you can see it in the logs:

systemctl status oracle-guardian
Active: active (running)
Memory: 38.2M / 256M limit
# /root/oracle_guardian/guardian.py monitors 17 checks:
✓ askoracle_http · last OK 34s ago · 0.02s
✓ askoracle_pg · last OK 34s ago · 1 activations
✓ bot_nexus · active + HTTP 200
✓ bot_paper · active + HTTP 200
✓ bot_apex · active + HTTP 200
✓ bot_magma · active + HTTP 200
✓ ...12 more