A scheduled probe runs every 5 minutes and writes one record per component: whether the
check succeeded, and how long it took. The percentages above are counted from those records over the
last 30 days, and from nothing else. There is no way to seed this history, and no
figure here has ever been typed in by hand.
No checks have been recorded yet. This page will report "no data" rather than a percentage until
the probe has run, which is the honest answer to a question nobody has measured.
The probe runs inside the same infrastructure it measures, which is the standard weakness of
self-hosted monitoring: if the application is down, the probe does not run, and the outage shows up
as an absence of checks rather than as failed ones. That is why coverage is
published beside uptime. Read the two together — 100% uptime at 60% coverage means four hours of the
day were not observed, not that they were fine.
The upstream check asks the GST record provider for no GSTIN at all. It measures that the endpoint is
alive and how quickly it answers; it performs no verification, so it costs nothing and touches no
customer data. Latency figures are the round trip our probe sees, which is not the same as the
latency of your own requests from your own network — treat them as a floor, and
measure
from your side before you depend on a number.
