# Rebooting your servers

Why some updates need a reboot, and how Hatchbox reboots your servers for you

Most updates take effect as soon as they install. A few don&#39;t: a new Linux kernel, and updates to core libraries like `libc6`, `systemd` and `dbus`, only take effect when the server reboots. Until then, the server keeps running the old versions, security holes included.

Ubuntu ships a new kernel every few weeks, so expect a server to need a reboot every two to four weeks.

## How you&#39;ll know

- The server&#39;s page shows **Reboot required since** the date the first update asked for one. Hover over it to see which packages are waiting.
- If a server has waited two weeks, your account&#39;s admins get one email listing the servers that need a reboot.
- Over SSH, Ubuntu&#39;s login message says `*** System restart required ***`.

## Let Hatchbox reboot in your maintenance window

The easiest way is to let Hatchbox do it. Set a [maintenance window](https://hatchbox.io/docs/articles/70-maintenance-windows) on your cluster, and leave **Reboot servers when an update needs it** ticked. It&#39;s under the window in your cluster&#39;s **Settings**, and it&#39;s on by default.

Each week in the window, Hatchbox updates your servers one at a time. If a server needs a reboot after its update, Hatchbox reboots it and waits for it to come back, with its services started, before moving on to the next server. Your account&#39;s admins get a notification naming the servers it rebooted. **Run maintenance** on the cluster page reboots the same way.

Hatchbox leaves the reboot to you, and the server keeps showing **Reboot required**, when:

- The window closed before the server&#39;s update finished
- Apps on another cluster use a database on the server, since rebooting it would take their database away outside their own window
- A database backup, deploy or other job started on the server during its update

If a rebooted server doesn&#39;t come back within 10 minutes, the run stops and your account&#39;s admins are notified.

Windows set up before November 21, 2026 have reboots turned off, since they were set up when the window didn&#39;t reboot. Tick the box to turn them on.

## Reboot a server yourself

Open the server in Hatchbox and click **Reboot**, ideally when traffic is low. Your apps, databases and Caddy start back up on their own. The server page stops showing **Reboot required** right away.

You can also reboot over SSH as `root`:

```
reboot
```

Hatchbox notices the reboot the next time it updates, deploys to or maintains the server.

## What happens during a reboot

A reboot usually takes about a minute.

- **One server running everything:** your apps are down for that minute.
- **Several web servers behind a load balancer:** Hatchbox reboots them one at a time, and the load balancer sends traffic to the others, so your apps stay up.
- **A separate database server:** your apps can&#39;t reach the database while it reboots. Most frameworks reconnect on their own once it&#39;s back.

Pick a maintenance window when your traffic is lowest, and most of your users won&#39;t notice.
