> ## Documentation Index
> Fetch the complete documentation index at: https://instabox.sh/docs/llms.txt
> Use this file to discover all available pages before exploring further.

# Status and incidents

> What the status page shows, how incidents are posted, how to follow them, and how uptime is calculated.

[status.instabox.sh](https://status.instabox.sh) shows whether each part of InstaBox is working
now, and its uptime over the last 90 days. Check it when requests start to fail, to see whether the
problem is on our side.

Most incidents on the page are posted automatically by our control plane and an outside check, with
no person in the loop:

* A problem must persist before it is posted: the control plane waits 2 to 3 minutes, and the
  outside check fails twice before it posts.
* The rows judged from our users' requests move only when failures come from several customers, so
  one customer's failing requests do not open an incident. Our team still handles those.
* Automation posts at most a partial outage. Only a person on our team posts a full outage, after
  confirming it.
* Our team also posts incidents by hand for problems the checks do not cover.

## What the page shows

Each row is one part of InstaBox, with its status now and its uptime over the last 90 days.

| Component | What an incident means for you |
| - | - |
| Dashboard and API | The CLI, the SDKs or the dashboard cannot reach InstaBox. Machines that are already running keep running. |
| Sandboxes: Create, Delete, Pause & Resume | Creating, deleting, pausing, resuming, starting or stopping a machine fails more often than usual, or takes longer. |
| Sandboxes: SSH | You cannot connect to a new machine over SSH or run a command in it. |
| Sandboxes: Public URLs | A web server in a new machine cannot be reached through its public URL. |
| CI runners: Instabox runners | Your GitHub Actions jobs wait longer than usual for an InstaBox runner, or do not start. |

**Sandboxes** and **CI runners** are groups. A group shows the combined uptime of its rows.

| Level | What it means for you | Counts as downtime |
| - | - | - |
| Degraded performance | It works, but slower or with more failures than usual. Retrying usually works. | No |
| Partial outage | Many requests fail, or one kind of request fails while others still work. | Yes |
| Full outage | Nothing works. Only a person on our team sets this level. | Yes |
| Scheduled maintenance | Planned work on a component. | No |

## Follow incidents

To read past incidents, open **View history** on the status page.

To be told about incidents, use **Subscribe to updates** on the status page, or add a feed to your
reader. Email and Slack subscriptions only get the incidents a person on our team posts. The feeds
carry every incident, including the automatic ones:

* RSS: `https://status.instabox.sh/feed.rss`
* Atom: `https://status.instabox.sh/feed.atom`

## What to expect during an incident

* **Posted after a few minutes.** A problem shows on the page after it lasts 3 minutes at degraded
  performance, or 2 minutes at an outage. Slow CI jobs take longer to show, because jobs must first
  wait several minutes. The outside check runs about every 10 minutes, so the rows it covers can take
  longer too.
* **Monitored before it is resolved.** When such a component recovers, the incident says
  "Recovered. Monitoring to confirm." for 30 minutes. If the problem comes back in that time, it
  continues in the same incident. The final update says how long the incident lasted and what you
  need to do, or "No action needed."
* **Resolved by the next good check.** Incidents on Dashboard and API, SSH and Public URLs are resolved
  by the next run of the outside check that passes.
* **A person takes over.** When our team posts an incident on a component, the automation leaves that
  component alone until the incident is resolved.

## How uptime is calculated

The status page is hosted on incident.io, which calculates uptime as described in
[How we calculate uptime](https://docs.incident.io/status-pages/uptime-calculation):

```text theme={null}
downtime = partial outage time + full outage time
uptime   = 1 - downtime / the last 90 days
```

* A partial outage and a full outage count in full as downtime. Degraded performance and scheduled
  maintenance count as up.
* Downtime counts only while the component shows the outage. The minutes before an incident is posted
  and the 30 minutes of monitoring after it recovers do not count.
* A group counts as up only when every row in it is up. Time when two rows are down at once counts
  once.

Our checks measure what you see, so a failure of a provider we run on shows like any other failure.

## Report a problem

Check the status page first, so you know whether we already have an open incident. Then email
[support@instabox.sh](mailto:support@instabox.sh) or tell us on [Discord](https://discord.gg/instabox).

Enterprise SLA terms are set in the customer's agreement, not on the status page.

> [card] [Errors](https://instabox.sh/docs/reference/errors.md)
>
> Every error code, with whether to retry and what to do.

> [card] [Limits](https://instabox.sh/docs/account/limits.md)
>
> What each plan allows, and the limits inside a machine.
