# Uptime bars and percentage

Every component on a public status page shows a row of daily bars and an uptime percentage. Both come from the same rule, day by day, so the percentage always agrees with the bars. This page explains that rule.

## The days

The bars cover the page's **History window**: the last 30 or 90 days (90 by default), set in **Settings → General**. Each bar is one calendar day in UTC, and the last bar is today.

## What colours a day

For each component and each day, the first of these that applies decides the bar:

1. **An incident that affected the component.** An incident published from On-Call colours every day it was open, from its start until it was resolved (or until now, if it is still open). The colour follows its **Public impact**: **Critical** is red, **Major** orange, **Minor** yellow, and **None** blue. When several incidents overlap a day, the most severe one colours it.
2. **Scheduled maintenance that affected the component.** The day is blue. A window that was completed counts up to the time it ended, not its scheduled end.
3. **The uptime monitor's day**, for a component synced from a monitor: green when the monitor rated the day operational, yellow when degraded (Degraded Performance), red when down (Major Outage).
4. Otherwise the day is green.

Today's bar also shows a component that is not operational right now: if nothing above has coloured today, it takes the component's current status.

Hover over a bar to see what coloured it: the incident's or maintenance's title, or the status.

## How the percentage is worked out

Each day gets an uptime between 0% and 100%, and the percentage shown is the average of those days.

| Day | Its uptime |
| --- | --- |
| A day with an incident whose impact is **Minor**, **Major** or **Critical** | The share of the day the incidents did not cover. A 6-hour incident makes that day 75%. Overlapping incidents are counted once, and today counts only up to now. If the component follows a monitor, the day is never higher than the monitor measured. |
| A day with only an impact **None** notice, or with maintenance | 100%. Planned work and notices are not downtime. |
| Any other day, for a component synced from a monitor | The monitor's measured uptime for that day. |
| Any other day, for a component with no monitor | 100%. |
| A day before the monitor had any data | Skipped: it does not count either way. |

The impact matters, not the incident's severity in On-Call: a **Minor** incident costs as much time as a **Critical** one. To show an incident without counting it as downtime, publish it with impact **None**. See [Incidents and maintenance](https://docs.evohub.io/status-incidents-and-maintenance.md).

A manual component that is down right now with no incident open counts today as 0%. A component with nothing to count at all reads 100.00%.

> [!NOTE]
> A monitor reading caps a day; it never raises it. If the monitor saw the component up while an incident was open, the incident still costs its share of the day.

## Hiding the percentage

To show the bars without the percentage, untick **Show uptime percentage** in the status page's **Customization** settings and save. The bars, their colours and tooltips stay; only the "99.98% uptime" figures are left out, on component rows, groups, the history section and product cards. It is on for every page until you turn it off.

## Groups

In the **Folded groups** layout, a group's row shows one combined bar per day and one percentage:

- The combined bar is blue when the only thing affecting its components that day is maintenance or impact **None**. Otherwise its colour follows how many of the group's components were affected and how badly, and the tooltip says, for example, "2 of 5 components".
- The group's percentage is the average of its components' percentages.

## Related

- [Components and groups](https://docs.evohub.io/components-and-groups.md)
- [Incidents and maintenance](https://docs.evohub.io/status-incidents-and-maintenance.md)
- [Uptime overview](https://docs.evohub.io/uptime-overview.md)