Uptime Checks and the Vendor Questions Behind Them: A Practical Guide

GhostHosts editorial ·

Wikimedia Foundation Servers
Illustrative image. Victorgrigas · CC BY-SA 3.0 · Cropped and resized. Image source

Uptime checks are the simplest way for a site owner to learn that something has broken before a customer tells them. GhostHosts (ghosthosts.net) is one hosting option among many, and this guide focuses on the method rather than any single provider: how to decide what to check, how often, from where, and what to ask your host about what it does and does not measure on your behalf. Treat every vendor claim as something to verify, and build your own visibility so that you are not depending on anyone's dashboard alone.

Start with what a visitor actually needs

A homepage that returns a 200 status code is not the same as a working site. Visitors log in, search, fill out forms, and complete purchases. Before choosing a check, list the three or four paths that would cost you the most if they failed. For a brochure site that may be the homepage and contact form. For a small store it is the login, product page, cart, and checkout.

Write each path down with the expected result. For example: the login page returns 200, the login POST redirects to the account page, and the account page contains a known string such as the account heading. A check that verifies content is far more useful than one that only confirms a response arrived. Content checks catch the common failure where a server answers quickly with an error page that still returns a success code.

Choose check frequency and probe location deliberately

Check frequency is a tradeoff between detection speed and noise. Checking every minute catches short outages but can produce alerts for single dropped packets. Checking every five minutes is a reasonable default for many small sites, with a confirmation rule such as requiring two or three consecutive failures before paging anyone. Decide this in advance and write it down.

Probe location matters too. A check run from one network can fail because of a local routing problem while the site is fine for everyone else. Running checks from more than one region helps separate a real outage from a network path issue. Ask whether your monitoring tool lets you choose probe locations and whether it reports which location observed each failure. If you run your own checks from a server in a single data center, remember that it shares the failure modes of that network.

Questions to ask your hosting provider

Hosts often describe their infrastructure in general terms, so turn general terms into specific questions. Ask what the provider monitors on its own side, and whether that monitoring covers your individual site or only the server it runs on. Ask how you would learn about a provider-side incident: by email, by a status page, or only by noticing your own failed checks. Ask whether you can see historical availability data for your account and how far back it goes.

Ask about the things you need to control yourself: how to change DNS records, how to renew or replace a certificate, and who on the provider side can help if your site is unreachable. Ask whether support hours match the hours your visitors use the site. If the answers are vague, that is information. Write the answers down and keep them with your runbook.

Plan details such as what each tier includes, how resources are allocated, and what recovery options exist should be confirmed in the current plan description and in writing before you buy. Do not assume that a checkout method, plan name, or support channel implies a particular service level.

Alert routing and a simple checklist

An uptime check that alerts an inbox nobody reads is not monitoring. Route alerts to a named person with a backup, and test the route by triggering a deliberate failure in a staging check. Confirm that the alert contains the failing URL, the probe location, the error, and the time, so the responder does not have to reconstruct context at 2 AM.

Use this short checklist to review your setup: list your critical user paths; confirm each has a content-based check; set frequency and a consecutive-failure threshold; run checks from at least two locations; route alerts to a named owner with a backup; test the alert path; record your host's answers to the questions above; and review the setup after any change to the site. Repeat the review each time you renew or change plans, because the features your checks depend on may change even when the site does not.

Part of the ToddMichael.tv network.