Linux Process Sentinel

What it does

The sentinel checks a target service every 10 seconds with systemctl is-active. If the service is down it logs an alert, restarts it, waits three seconds and confirms the recovery. If the service is still down it logs a failure. Every event goes to ~/sentinel-logs/sentinel.log and to the system journal, with a running restart count.

How it’s built

• The sentinel runs as its own systemd service with Restart=always.
• The service it watches has Restart=no, so the sentinel is the only thing that brings it back.
• The target in the repo is a simulated service, dummy-app, that writes a heartbeat every five seconds. It stands in for a real service such as nginx or a Python app, which is the intended next step.

How I tested it

I followed the sentinel’s journal with journalctl -u sentinel -f, stopped the dummy app with sudo systemctl stop dummy-app and watched the alert, restart and recovery appear within about ten seconds.

What I’d change

The test was manual. A real version needs a configurable list of services, thresholds for repeated failures so it doesn’t restart in a loop, and a notification when it gives up.