Local-first · Linux · MIT

Watch one host properly — without a monitoring fleet.

FTMON detects memory leaks, CPU hogs, disks filling, service failures, and notable journal events. Bounded history stays on the machine in SQLite.

What it is

A practical systems monitor for a desktop, workstation, VPS, or standalone server — more history and alerting than htop, without running Nagios, Zabbix, or Prometheus for a single host.

$ uv tool install ftmon && ftmon init --profile desktop

Try it without installing

The public demo uses clearly labelled synthetic data. It does not monitor the demo server or expose an operational installation.

SourceForge archive · legacy / historical

This SourceForge project is the old FTMON

What you find here — the Perl engine, the early downloads, the pink-sidebar era — is legacy and historical. Development moved on. The living project is the Python rewrite on GitHub (links above).

Twenty-five years ago I wrote FTMON as a fault-tolerance monitor in Perl and published it on SourceForge. The tagline was “Monitoring for the home user” — watch one Linux box for the slow failures that real-time tools miss: memory leaks, disks quietly filling, services flapping. It confirmed problems over consecutive cycles before alarming, escalated instead of spamming, and had a Webmin UI so you did not have to be a Perl programmer to nudge a threshold.

There were experiments that feel wonderfully of their time: Festival Lite speech alerts, MisterHouse hooks, pages for late-breaking news, even talk of Big Brother and Tivoli. Like a lot of software from 2001–2003, it then went quiet. The downloads and project page stayed; the ideas stayed with me.

The design ideas survived — delta and monotonic calcs, confirmation, baselining against the host’s own normal, one explainable incident. The Perl code did not. Today’s FTMON is a clean-room Python / MIT rewrite; this SourceForge tree remains the GPLv2 historical home.