You Up?

One Incident with its evidence: when it opened and ended, who acknowledged or ended it, where the request stopped in plain words with what to do about it, and what each location saw. It opened because the failure was confirmed from more than one location. Its runs are kept for keptDays; after that kept is false and only where it stopped is left.

One Monitor: its Checks, each with its latest status, its snooze and its open Incident, and its uptime over the last 24 hours, 7 days and 30 days. failed is a problem confirmed from more than one location; warning is worth a look while the target still works, such as a certificate that expires soon; pending has no answer yet. Each uptime figure carries coveredSeconds, incidentSeconds and notMeasuredHours. Whenever notMeasuredHours is above 0, tell the user how many hours uup could not measure: they are left out of the figure.

The Organization's Plan: its name, how many Projects, Monitors, heartbeats and locations per Monitor it allows, and how many are in use. It holds no price. Only the Organization's owner can change the Plan, in the uup dashboard.

A Monitor's response times over the last 24h unless asked, 7d or 30d: each location's line of points and its median in milliseconds, and the Incidents and Regional outages in that time. A heartbeat has none.

What is broken now, and since when: the Monitors whose status is not ok, the open Incidents and the open Regional outages, across every Project the API token reaches, or one Project. Start here when the user asks whether something is down. failed is a problem confirmed from more than one location; warning is worth a look while the target still works, such as a certificate that expires soon; pending has no answer yet. An Incident's openedAt says since when. A Regional outage is one location that cannot reach a target the others can.

Uptime. With monitorId, that Monitor's uptime over the last 24 hours, 7, 30 and 90 days; add days for its uptime day by day instead, up to 90 days back, cut by the Project's timezone. Without monitorId, each of the Project's Monitors over 30 days with its days. Each uptime figure carries coveredSeconds, incidentSeconds and notMeasuredHours. Whenever notMeasuredHours is above 0, tell the user how many hours uup could not measure: they are left out of the figure.

Who is alerted, and how: a Project's Notification rules, oldest first, each with the events it sends and its Channels, and every Channel those rules send to. A Monitor's ruleIds (from get_monitor) say which rules it carries; usedBy says how many Monitors carry each. Three different acts change what is sent, and none is a rule: pausing a Monitor stops its checks; snoozing a Check holds its messages until a time; acknowledging an Incident stops its repeat alerts while its recovery is still sent.

A Monitor's Deliveries, newest first: each message to a Channel or a Webhook, what it was about and what became of it, such as sent, held while its Check was snoozed, or failed and why. Narrow them with to, and to one Incident or Regional outage with event. Use it when the user asks whether an alert went out.

A Project's Incidents, newest first. An Incident is an outage of one Check: it opens when a failure is confirmed from more than one location, and counts against uptime until it ends. Narrow them to one Monitor with monitor, and to the open ones with open: "true" or the ended ones with "false". failed is a problem confirmed from more than one location; warning is worth a look while the target still works, such as a certificate that expires soon; pending has no answer yet.

The locations a new Monitor can be checked from now, the least used first, each with its city and country. A Monitor's locations names them by id.

A Project's Monitors, in the order they were made, each with its Checks and their latest status. Find one by words in its name or target with q ("checkout" finds Checkout API and https://shop.example.com/checkout), and narrow by type and state. failed is a problem confirmed from more than one location; warning is worth a look while the target still works, such as a certificate that expires soon; pending has no answer yet.

The API token in use (its name, whether it is Read-only or Full, and the one Project it reaches, if only one) and the Projects it reaches, oldest first. Every other tool names a Project by its id, so call this first when you do not know it.

A Monitor's recent runs, newest first: each time one location checked one Check, with its outcome, status code, what went wrong and its timings. Narrow them by location, by checkType, and to failed runs only with failuresOnly: "true". One failed run is not an outage: an Incident opens only when a failure is confirmed from more than one location.