Uptime monitoring answers one question continuously: can my users reach my service right now? This guide walks through setting it up well.
1. Decide what to monitor
Start with the pages and endpoints that make you money or generate support tickets when broken: the homepage, login, checkout, your API health endpoint and any customer-facing subdomains.
2. Pick intervals
Interval × failure threshold ≈ worst-case detection time. One-minute checks with a threshold of 2 detect an outage in about two minutes; five-minute checks in about ten. Match the interval to how much an outage costs.
3. Define "healthy"
Most sites should accept 200–299. Decide whether redirects count as healthy - Upmonora follows up to five redirects and evaluates the final response, validating every hop for security.
4. Route alerts to people who can act
Email the on-call person or a shared inbox, and send signed webhooks into your incident tooling. Always send recovery alerts too, so nobody keeps investigating a resolved problem.
5. Publish a status page
During an outage a status page answers "is it just me?" for customers and reduces duplicate tickets. Keep it honest: show real history.
6. Report uptime honestly
Show both check-based and duration-based uptime, and exclude planned maintenance windows explicitly rather than silently.
7. Review insights weekly
Outages are usually preceded by warning signs: slowdowns, repeated short incidents, recurring error types. Upmonora's AI insights surface these automatically - five minutes a week reviewing them prevents many incidents.