What to Monitor: A Practical Checklist for Web Applications
Setting up monitoring for the first time is overwhelming. There are hundreds of things you could monitor. The trick is knowing what actually matters — and in what order.
Start With What Your Customers See
Your customers do not care about your CPU usage. They care whether the app loads and whether they can do their job. Start here:
- Homepage or landing page — the first thing users (and search engines) hit. If this is down, everything else is irrelevant.
- Login or authentication endpoint — a broken login locks out every user. Monitor the actual login flow, not just the page load.
- Core application pages — the one or two pages where your users spend most of their time.
- API health endpoint — if you serve an API, create a lightweight
/healthroute and monitor it. This catches database connection issues, deployment failures, and dependency problems before your users do.
Infrastructure Layer
Once the customer-facing checks are in place, go one level deeper:
- Database connectivity — a TCP check against your database port confirms it is accepting connections. This catches firewall changes, connection pool exhaustion, and unplanned restarts.
- Mail server — if your app sends transactional email (password resets, invoices, notifications), monitor your SMTP relay with a TCP check.
- CDN or static assets — if your frontend is served from a CDN, monitor the CDN URL separately. A CDN outage looks like a blank page to your users even if your backend is healthy.
- DNS resolution — rarely fails, but when it does, nothing works. A ping check against your domain catches DNS propagation issues.
External Dependencies
Modern applications rely on third-party services. If a payment provider goes down, your checkout breaks — even though your own servers are fine.
- Payment gateway — Stripe, Mollie, PayPal. Monitor their status endpoints or your own integration endpoint.
- Authentication provider — if you use Auth0, Clerk, or similar, a TCP or HTTP check confirms the service is reachable.
- Any service your app calls synchronously — if a third-party API call blocks your request/response cycle, monitor it.
How to Prioritise
A simple rule: monitor outward in. Start with what the customer touches, then infrastructure, then dependencies. Add checks one at a time, verify they work, and move on.
Upwarden supports HTTP, TCP, and ping monitoring — which covers all three layers above. The free tier includes one monitor, enough to start with your most critical endpoint and expand from there.
Get Started
If you do not have monitoring in place yet, create a free Upwarden account and add your first check in under two minutes.