Monitoring without a second server

I wanted to know when something broke before the household told me, but standing up a full monitoring stack felt heavier than the things it would watch.

A lightweight agent, a handful of checks and a push notification to my phone covers it. Disk space, service health, certificate expiry and the backup job's exit code. Four things that have actually gone wrong.

When I add an alert now it is because something broke and I did not hear about it, never on the theory that I might want it one day.