Uptime monitoring8 min read
Website uptime monitoring: what to check, how often, and how to avoid false alarms
By the RenewalRemind team
Good website monitoring checks four things from outside your server: that the page answers with a success status (HTTP), how fast it answers, that the SSL certificate is valid and not close to expiry, and that DNS still resolves. Check every 5 minutes for important sites and 15–60 for the rest, retry a failure before alerting to avoid false alarms, and send alerts to the people who can actually fix the site.
Key takeaways
- Check HTTP status, response time, SSL validity/expiry and DNS — downtime has more causes than a dead server.
- 5-minute checks for shops and lead-generating sites; 15–60 minutes is fine for brochure sites.
- Retry once before alerting: one failed check is often a network blip, not an outage.
- Alert the people who can fix it, and tell them when the site recovers.
"Is the site up?" sounds like a yes-or-no question. In practice, websites fail in more ways than a server crash, and a monitor that only asks one question will miss most of them. This guide covers what to check, how often, how to avoid false alarms and who should hear about problems.
What to check
| Check | What it catches | What 'healthy' looks like |
|---|---|---|
| HTTP status | Crashed server, PHP fatal errors, 500 errors, hosting suspension, expired domain parking page | A 2xx or 3xx response from the real site |
| Response time | Overloaded server, slow database, a plugin gone wrong, a traffic spike | Consistently under about 2 seconds |
| SSL certificate | Expired or invalid certificate, wrong certificate for the name, renewal not installed | Valid, matches the domain, more than 30 days left |
| DNS | Deleted or changed records, a name server move gone wrong, an expired domain | The domain resolves to an address |
HTTP: is it really your site answering?
A plain "did the server respond?" check passes when a registrar's parking page or a host's suspension page answers instead of your site. Treat 4xx and 5xx responses as failures, and follow redirects so http:// → https:// → www chains end on a real page.
Response time: slow is the new down
Visitors leave long before a page times out. Track response time on every check and treat a site that is consistently slower than about 2 seconds as a warning, even when it technically works.
SSL: up, but unusable
A site with an expired certificate still answers — but every visitor gets a full-page security warning. Uptime-only monitors report it as up. Check validity on every run and warn ahead of expiry; see how to check SSL certificate expiry.
DNS: the outage that looks like your server
If the domain stops resolving, the site is down for everyone even though the server is fine. DNS failures usually come from record changes, a migration between DNS providers, or an expired domain.
How often to check
Interval decides how fast you find out, not how reliable the site is. Match it to what downtime costs — see what 99.9% uptime means for the maths.
- 5 minutes: online shops, booking systems, lead-generating landing pages, apps.
- 15–30 minutes: business websites, blogs, portfolios with steady traffic.
- 60 minutes: low-traffic or internal sites where a long outage is the only real risk.
How to avoid false alarms
The fastest way to make a team ignore alerts is to send wrong ones. A single failed check is often a brief network problem between the monitor and the server, not an outage. Three habits fix most of it:
- Confirm before alerting. Retry a failed check after about 30 seconds and only alert if it fails again.
- Use sensible timeouts. Around 10 seconds; anything slower is a performance problem, not a crash.
- Alert on change, not on every check. One alert when the site goes down, one when it recovers — not one per failed check.
Who should be alerted
- The person who can fix it. Usually the developer or whoever manages the hosting — not a shared inbox.
- Only the sites they look after. Teams that get alerts for everything start ignoring them.
- The client, carefully. Many freelancers tell clients after the fact with a short note on cause and fix, rather than forwarding raw alerts.
What to do when a site goes down
- Open it yourself from a different network (mobile data) to rule out a local problem.
- Check what the monitor saw: no DNS, no connection, timeout, error code or certificate problem — each points somewhere different.
- Check the host's status page and your server or hosting panel.
- Fix, wait for the recovery alert, then write down the cause so the same thing doesn't happen twice.
Frequently asked questions
How often should I check if my website is up?
Every 5 minutes for sites where downtime costs money or leads, and every 15 to 60 minutes for simple brochure sites. Shorter intervals find outages sooner but don't make the site more reliable.
What is the difference between uptime monitoring and SSL monitoring?
Uptime monitoring checks that the site responds. SSL monitoring checks that its certificate is valid and warns before it expires. A site can be up but show a certificate error to every visitor, so you need both.
Why do uptime monitors send false alarms?
A single check can fail because of a brief network problem between the monitor and the site. Good monitors retry after a short delay and only alert when the failure is confirmed.