# Update, resize, rebuild and delete a node

> Signed node software updates, changing the machine size, rebuilding on a new machine and deleting a node safely.

Everything you can do to a node is in **Actions** on the **Node** page. You never need SSH or Hetzner Console for it.

| Action | Who | Confirmation |
|---|---|---|
| **Upgrade to …** | Owners, admins | — |
| **Change machine size** | Owners, admins | Choose the size and confirm |
| **Replace machine & restore** | Owners, admins | Authenticator code and the node's name |
| **Delete node** | Owners | Authenticator code and the node's name, then a 15-minute wait you can undo |

With **Strict** token custody, resizing, replacing and deleting ask you for a temporary Hetzner token first — see [Connect Hetzner](https://belvador.dev/en/docs/node/connection#3-connect-it-in-belvador).

## Node software updates

Belvador publishes node software as **signed releases**. The node's agent checks for them every 15 minutes and installs only releases whose signatures and checksums verify against Belvador's signing keys; a release that fails to start is rolled back.

When a newer version is available for your node, the page shows **Upgrade to {version}**. The status turns **Upgrading** while it installs. A backup is taken before an upgrade.

The **Security** section shows when the node last received a signed release manifest; if that is older than 24 hours, the node turns **Degraded**.

## Change machine size

Pick a bigger machine when the disk or memory runs high, or a smaller one when you need less. Only sizes with the same CPU architecture are offered.

The machine powers off for about **1–2 minutes**. Errors sent meanwhile are held at the edge and delivered afterwards. Hetzner bills the new size from then on.

## Replace machine and restore

Use **Replace machine & restore** when the machine is lost or broken — deleted by accident, a failed disk, a node that stays Failed.

1. The current server is deleted in Hetzner, including anything on its disk.
2. Belvador creates a new server and installs the node software.
3. Your data is restored from the **latest successful backup**. Errors sent in the meantime stay queued at the edge; errors received after that backup and not held at the edge are lost.

It takes about **5–10 minutes**. Issues cannot be viewed until it finishes. If you hold your own recovery kit, the timeline asks you to upload it — see [Backups](https://belvador.dev/en/docs/node/backups#restoring).

## Delete a node

**Delete node** removes the node and everything on it:

- all error data on the node is wiped;
- the server and its firewall are deleted from your Hetzner project;
- every DSN that sends to this node stops accepting errors.

Deletion starts after a **15-minute** delay. Every owner gets an email, and any owner or admin can choose **Undo** until then. A final backup is started when deletion is prepared.

After deletion, check your Hetzner project: if the node's server or firewall is still there, delete it in Hetzner Console so it is no longer billed. The complete deletion run against Hetzner has not yet been verified end to end on a live account.

Deleting a node does not delete the workspace or its projects. To leave Belvador entirely, an owner can also delete the workspace from **Settings** — see [Account, two-factor and roles](https://belvador.dev/en/docs/account-security#delete-a-workspace).