The night a green dashboard lied to us

A nightly job on one of our products — the one that sends users their weekly plan — silently stopped running after a config change. No error, no crash, no alert. Our uptime monitor pinged the homepage: 200 OK, all green. Users noticed before we did.

That was the day we stopped trusting green dashboards. Ping monitors can't see the things that are supposed to happen but don't: the cron job that never fired, the checkout that breaks at step 3 while the homepage smiles.

So we built UptimeEye around two ideas:

1. Monitor real user flows (login, checkout, API calls), not just endpoints — every 30 seconds.

2. Treat cron jobs and scheduled tasks as first-class citizens with heartbeat alerts.

It's EU-hosted, has a free tier, and we run all our own products on it — which is how we keep catching our own mistakes early now.

Curious about others here: what's the worst thing that broke silently for you while every dashboard stayed green?