Incidents and uptime statistics

This article explains the words and numbers on the Uptime pages: the status of a monitor, incidents, uptime percentage, and response time.

Status of a monitor

• PENDING: the monitor is new and has no result yet.
• UP: the last check passed.
• DOWN: the number of failed checks in a row reached the Failure threshold of the monitor.
• PAUSED: you paused the monitor. It is not checked.

One failed check does not make a monitor DOWN when the failure threshold is 2 or more. This prevents incidents from short network problems.

Incidents

An incident starts when a monitor goes DOWN. It ends at the next check that passes. One period of downtime is always one incident, however many checks fail during it.

The Incidents tab lists every incident with its monitor, status, start and end time, duration, and cause. You can filter the list by status, by monitor, and by dates.

The four Uptime tabs

• Overview: the number of monitors, how many are Up, Down, and Paused, and the active incidents.
• Report: uptime, downtime, incidents, and response times for all monitors in one table. Filter by type, status, and date range, including a custom range.
• Monitors: the list of monitors, where you pause, resume, or delete them.
• Incidents: the list of incidents.

Click a monitor to open its own page. It shows the Availability, the Response time with a chart, and the Recent checks. Choose the period at the top: the last 24 hours, 7 days, 30 days, or 90 days.

How the uptime percentage is counted

Uptime is the share of time that the monitor was not in an incident. It is counted by time, not by the number of checks. For example, one hour of downtime in a day gives 95.83%.

• Only the time since you added the monitor counts. A monitor added an hour ago is not 100% for the full 30 days.
• An incident that is still open counts as downtime up to now.
• A monitor with no checks yet shows Not enough data yet.

Response time

Each check records how long the answer took, in milliseconds. The monitor page shows the average, minimum, median, P95, and maximum for the period. P95 means that 95% of the checks were faster than this value.

• Failed checks with no answer are not in these numbers.
• If the monitor has a Slow-response threshold, a slower check shows a SLOW badge. A slow check does not open an incident.

How long the data is kept

The results of single checks are kept for 90 days. Incidents are kept for as long as the monitor exists. If you delete a monitor, its checks and incidents are deleted too.