Checking status…

Published numbers, with the method attached.

Probes run from Cloudflare, deliberately not from wagle's own servers: a monitor that shares a failure domain with the thing it monitors loses exactly the samples that matter, and every lost sample makes the number look better than it was. Availability is successful probes divided by probes attempted in each rolling window. Windows are trailing, never calendar, so an incident doesn't get wiped by a month rolling over.

• · A window holding less history than its span says so, and reports what it actually has. It is never presented as a full year.
• · Percentages are floored, so 100% means zero failed probes rather than a rounded 99.9995%.
• · The two checks are reported separately and never averaged. Reachability is a bare HTTP 200; config delivery crosses auth, the index and a real snapshot read.
• · Latency percentiles cover successful probes only. A timeout has no latency worth reporting.

• · One vantage point. A network problem between that location and wagle looks identical to wagle being down.
• · Probe resolution bounds it. An outage shorter than the gap between two probes can pass unseen.
• · It is self-reported. Publishing the method is the mitigation, not a claim of neutrality.
• · This is not a service-level agreement. These are measurements of what happened, not a promise about what will.