What does "Not measured" mean?
It means the target refused our probe rather than failing. Bot protection, a WAF rule or a per-IP rate limit returned a challenge or a refusal, so we have no measurement of whether your service is healthy — and reporting that as downtime would be a lie in the more damaging direction.
A monitor in this state is not counted as down and does not open an incident for availability. It does raise a probe-blocked incident, because a monitor that cannot measure anything needs your attention just as much as one that is failing.
How do I allowlist the PageLantern probe?
Match on the User-Agent substring PageLantern-Probe. Our probes send a User-Agent of the form Mozilla/5.0 (compatible; PageLantern-Probe/1.0; +https://pagelantern.com/docs/website-monitoring).
Match the substring, never the whole string. Token order and the contact URL can change, and an exact-match rule would silently stop matching on the day we change it — which is the worst possible failure mode for an allowlist, because nothing breaks visibly until you need it.
Cloudflare WAF → Custom rules → Skip
(http.user_agent contains "PageLantern-Probe")
nginx # before your bot-blocking rules
if ($http_user_agent ~* "PageLantern-Probe") { set $skip_bot_check 1; }
SiteGround Site Tools → Security → allowlist the probe egress addresses,
or disable the anti-bot challenge for the monitored pathWhat is second-region confirmation?
An alert policy that re-runs a failing check from a different region before an incident opens. If the second region also fails, the incident opens as normal. If it succeeds, the failure is treated as a local network artefact and nobody is paged.
It is the right answer for targets behind rate limiting or geographic routing, where a single vantage point produces failures that are real from where we stand and irrelevant to your users. Available on HTTP, HTTPS, API and content (DOM) checks — the types a satellite in another region can reproduce as a plain HTTP exchange. SSL, domain, DNS, TCP, heartbeat and browser checks run hub-local and cannot be re-run remotely, so the option is not offered for them.
PageLantern does not run every check continuously from many locations. Confirmation is a targeted second opinion at the moment it matters, not multi-region probing.
Why did my monitor go down right after a PageLantern deploy?
Our probe egress address can change when we deploy. A target that allowlists IP addresses rather than the User-Agent will start refusing the new address, and the monitor will report failures that have nothing to do with your service.
This is the strongest argument for matching on the User-Agent substring. If you must allowlist by address, expect to revisit it, and pair it with the probe-blocked state so a refusal is not scored as downtime.