Measure uptime and response times for monitors over a period: uptime_percent (share of successful checks), estimated_downtime_minutes (failed checks times the check interval, as the dashboard counts it), failed checks, average response time, and the 95th percentile (p95_response_ms for periods of five days or less; past that only avg_daily_p95_response_ms, the average of daily p95s, which is not the period’s p95). Read-only. Give monitor_id for one monitor, or leave it out for every monitor on the page (up to 25, lowest uptime first). Paused monitors are measured too, since they can have checks for a past period. The period defaults to the last 30 days and can be up to 395 days; pass start and end as ISO 8601 times. Periods longer than five days are measured in whole days, and an end at midnight stops the day before, so 1 October to 1 November 00:00 is exactly October; start and end in the result are the period actually measured, and period_note says when that differs from the one asked for. When a monitor has no checks for the start of the period, data_starts and coverage_note say so: never present such figures as covering the whole period. Set daily for a per-day breakdown. uptime_percent is null when a monitor ran no checks in the period. This measures the monitors, not the components shown on the status page.