PYLONMON
Server & infrastructure monitoring

One system that watches
every server you run.

CPU, memory, per-mount disk, temperature and load — pushed out by a tiny open-source agent, watched from outside your network. Set a threshold and get paged when it's crossed. And if a machine stops reporting entirely, that silence is itself the alert.

Open-source agent (MIT) · Linux, Windows & Raspberry Pi · No open ports · No credit card

# install the open-source agent — one line, Linux:
curl -fsSL https://pylonmon.com/beacon.sh | sh

# ~20 seconds later the machine appears as a monitor:
🟢 web-01 — cpu 12% · mem 38% · disk 61% · 41°C

# and when something is wrong, you're paged:
🔴 db-02: disk 91% — vital threshold breached
🔴 build-runner went quiet — no vitals for 60s

How the server monitoring works

1 · PUSH

The agent reports out

One small static binary pushes vitals over outbound HTTPS every 20 seconds. No listener, no open ports, nothing to firewall — it works behind NAT, CGNAT, and networks you don't control.

2 · WATCH

Thresholds + the clock

Set vital rules like disk > 90% or temp > 80°C. PylonMon watches from outside your infrastructure — so a dead host can't take its own alarm down with it.

3 · PAGE

You hear about it

Email, Slack, Google Chat, Discord, SMS, voice calls, webhooks and more — with escalation ladders that keep climbing until someone acknowledges.

Why push-based beats scraping

The classic server performance monitoring stack — Prometheus, exporters, a dashboard to babysit — means running more servers to watch your servers, opening ports, and maintaining scrape configs. A push agent inverts it: the machine reports out, so there's nothing to reach, nothing to expose, and nothing extra to host. And because the monitoring expects a heartbeat rather than answering a poll, a machine that dies takes exactly one push interval to be noticed — the dead-man's-switch per host that a scraper structurally can't give you when the network path is the thing that failed.

What the agent watches

The agent is open source (MIT), standard-library Go, a single static binary for Linux and Windows — small enough to read before you run it. Source on GitHub →

And the same account covers what agents can't reach: website & API uptime, HTTP/S, TCP, ping, DNS, TLS-certificate and domain-expiry checks, plus heartbeats for cron jobs and backups — one system, one place that pages you.

PylonMon vs. self-hosted monitoring: a monitoring stack that lives in your rack shares fate with the rack. When the power, the switch, or the hypervisor goes, it goes too — and nobody gets paged. Off-site is the point: PylonMon still has eyes when your infrastructure doesn't.

Common questions

Do I have to install an agent?

Only for inside-the-box vitals. Uptime, HTTP/S, TCP, ping and DNS checks run from our side with nothing installed. The agent is optional, open source, and takes about a minute per machine (Linux one-liner, or a PowerShell installer on Windows).

What's the overhead?

Nearly none — it's a single small binary with no runtime and no dependencies, sleeping between 20-second pushes. It runs happily on a Raspberry Pi.

What does it cost?

The free plan includes 3 agent-monitored servers with vital-threshold alerting, plus 20 monitors overall with 1-minute intervals. Paid starts at $9/mo — 30-second intervals, agent coverage up to your full monitor cap, SMS, and more monitor types. See pricing →

My servers are behind NAT / a corporate firewall.

That's the normal case, not the exception. The agent only makes outbound HTTPS requests — if the machine can reach the internet, it can report in. Nothing inbound, ever.

Watch your first 3 servers free →

About a minute per machine. No credit card.