Uptime Kuma · monitor 4 · 1–15 September 2026 · times in

Datatractor Registry latency regimes

Two response-time states, 151 ms and 282 ms, switched in a single check. Fourteen elevated episodes in fifteen days, each 4h 19m long; eleven of them end with the host briefly unreachable.

Fifteen days at ten-minute resolution

Median response time per ten-minute bucket. Axis clipped at 400 ms; carets mark buckets peaking above it.

Median response time (10-min buckets) Elevated episode Unreachable

The switch itself

All fourteen transitions overlaid at the raw 60-second sample rate, aligned on the first elevated check. No ramp — the step happens between two consecutive checks.

Individual episode (14 overlaid) Median across episodes

When the episodes land, and how often

Each episode against the time of day it occupies. Start times slide across the clock; from 10 September two run per day instead of one every day and a half.

Elevated episode (start → end) Host unreachable at the end

Every normal → elevated transition

All episodes, newest first, in local . Step is the jump from the preceding check; gap is the time since the previous episode began; ends down is the run of failed checks immediately after it.

#	Entered elevated	Returned to normal	Duration	Step	Median	p95	Peak	Gap	Ends down