Endpoint monitoring

Endpoint monitoring tracks the availability and response time of HTTP endpoints. The platform sends a request to your URL on a schedule, records the status code and response time, and flags the endpoint as down when the response does not match what you expect.

Any HTTP or HTTPS URL works: a public website, an API route, a health endpoint, or a status page. Checks run from the platform’s edge network, so they measure availability the way an external visitor would.

How it works

Each monitored endpoint has a URL and a check configuration. On every interval, the platform sends a request using the method, timeout, and accepted status codes you configured, and compares the response against those settings. If the status code is accepted, the check passes. If it is not, or the request times out or fails to connect, the check fails.

A single failed check does not mark the endpoint as down. The platform retries the check up to three times, with a short delay between attempts, and only opens an incident when all attempts fail. This keeps transient network blips from producing noise.

Response time is recorded on every successful check, so you get a continuous view of latency alongside uptime.

Allowing checks through your firewall

Checks originate from the platform’s edge network, so your endpoint must be reachable from the public internet. If your endpoint sits behind a firewall, a WAF, or a rate limiter, you need to allow the check traffic through.

The simplest approach is to allow the checks by user agent. Every request the platform sends carries:

User-Agent: SimpleObservability/1.0

Add a rule that permits this user agent and you will not need to maintain a list of IPs. If you prefer to allow by IP instead, the checks egress from Cloudflare’s network, and the current list of ranges is published at cloudflare.com/ips. Update your allowlist periodically, since these ranges change over time.

Troubleshooting

False downs from a WAF

A web application firewall or bot protection layer may block the platform’s requests because they carry no browser headers. Allow the SimpleObservability/1.0 user agent, or whitelist the edge network IPs, to keep the checks from being challenged or blocked.

Slow endpoints timing out

If a healthy endpoint is reported as down, the response may be exceeding the timeout. Increase the timeout value, or move expensive work behind a faster health route that returns quickly.