A monitoring stack doesn't need to be a product. For a solo founder or a two-person team, five small cron jobs catch the failures that actually cost money - before customers do. Here's the exact $0 stack, copy-paste ready.
The five checks
1. Checkout heartbeat (every 15 min)
curl -s -o /dev/null -w "%{http_code}" https://yourstore.com/checkout - if it's not 200, you are losing sales in silence. Alert to your phone. This one check has caught more dead payment pages than any dashboard ever will.
2. Disk + process guard (every 30 min)
One line: if disk usage > 90% or your app process is gone, ping you. Disk-full is still the #1 silent killer of small servers.
3. Revenue canary (every hour)
Query your payment provider for the last 24h of sales. If the number you know you made doesn't appear, something between checkout and the ledger is broken - and you find out in an hour, not at tax time.
4. Backup existence + age (daily)
Check that yesterday's backup file exists AND was modified in the last 26 hours. A backup job that quietly died in March tells you nothing until you need it.
5. Restore drill (quarterly, calendar-forced)
Restore one backup to a scratch directory. If it opens, you're covered. If it doesn't, you just saved your quarter.
Why this beats a dashboard
Dashboards show you a picture; checks make decisions. The whole stack is ~80 lines of bash + one alert webhook, runs on anything, and costs nothing but attention. The failure mode it fixes is universal: the operation that runs on memory and hope.
The packaged version
I sell the finished form of this: the Agent Ops 24/7 pack (monitors + escalation ladder + 2am runbook) and the Ops Starter Kit (handoff checklists + restore drills) for teams that want it done for them.
Steal the five checks, or take the pack. Either way - run check #1 tonight. It takes four minutes and it's the one that finds the bleeding.
Top comments (0)