# Rebuild an instance > Erase an instance's disk and reinstall it from an operating system image, keeping its addresses and plan. Source: https://docs.coritan.com/cloud-compute/rebuild/ A rebuild erases the instance's disk and installs an operating system image on it from scratch. Use it to start again with a clean system, or to move to another distribution or release. The instance keeps its plan, hostname and IP addresses, and its billing does not change. ## What a rebuild keeps Kept : The plan with its vCPU, memory and disk size, the hostname, the floating IPs attached to the instance, and its [backups](/cloud-compute/backups/). Erased : Every file on the disk, the instance's [snapshots](/cloud-compute/snapshots/), and the SSH keys saved on the **Access** tab. The instance also gets a new random password, which we do not show. > [!WARNING] > The confirmation dialog says your snapshots stay. They do not: they go with the old disk. To keep a copy of the disk, [create a backup](/cloud-compute/backups/) before you rebuild. ## Before you begin - Back up anything on the disk that you want to keep. A rebuild cannot be undone. - The instance is not suspended. - Leave [rescue mode](/cloud-compute/rescue-mode/) first if the instance is in it. - Choose an image that fits the instance. Each image shows the disk it needs, such as "Needs 20 GB", and the instance's disk size is on the **Disk** card of the **Console** tab. ## Rebuild the instance 1. In the [dashboard](https://www.coritan.com/dashboard/compute), go to **Cloud Compute**, open the instance and select the **Access** tab. The **Rebuild operating system** card is at the bottom. 2. Choose an image. The images are grouped by distribution, and each one names the user you sign in as and the disk it needs. **Current** marks the image the instance runs now; choose it to reinstall the same release. 3. Select **Rebuild instance…**. 4. In the dialog, type the instance's hostname to confirm, then select **Erase and rebuild**. The **Danger zone** card on the **Settings** tab has the same rebuild: its **Open Access** button brings you to this card. ## Result We stop the instance, replace its disk with a fresh copy of the image, and start it again with the same hostname and primary address. A message confirms that the rebuild has started. When it has finished, reload the page: the header names the new image. To sign in to the rebuilt instance, do one of these: - Save your SSH keys again on the **Access** tab, then select **Restart** in the header, as [Manage SSH keys](/cloud-compute/access/#manage-ssh-keys) describes. - [Reset the password](/cloud-compute/access/#reset-the-password). This needs the QEMU guest agent, which not every image includes. The rebuilt instance has new SSH host keys, so your SSH client warns that the host identification has changed. [Connect to an instance over SSH](/cloud-compute/connect-to-an-instance/#troubleshooting) shows how to clear the warning. ## Troubleshooting `Cannot rebuild a suspended instance` : The service behind the instance is suspended. Settle it from the **Billing** tab first. `Template not available (ready) on this node` : The image is not ready on the host that runs your instance, although it is ready elsewhere. Choose another image, or try again later. `No templates are ready` : None of the images is ready right now. Try again later. `Too many requests for this action. Please wait and try again.` : You started more than 8 rebuilds of this instance in 10 minutes. Wait for the time the response gives in `Retry-After`. A second floating IP does not answer after the rebuild : The rebuild sets up only the instance's primary address. For each other floating IP on the instance, [detach it and attach it again](/floating-ips/attach-and-detach/), then restart the instance from the header. ## Related - [Operating system images](/cloud-compute/images/) describes what each image sets up on its first boot. - [Back up and restore an instance](/cloud-compute/backups/) to keep a copy of the disk first. - [Repair an instance in rescue mode](/cloud-compute/rescue-mode/) to fix a system without erasing it. ## With the API `POST /api/v1/client/vps/{uuid}/rebuild` starts a rebuild. Send the image's `template_id`, which `GET /api/v1/client/vps/templates` lists, as [Operating system images](/cloud-compute/images/#with-the-api) describes. ```bash curl -X POST https://api.coritan.com/api/v1/client/vps/$INSTANCE_UUID/rebuild \ -H "Authorization: Bearer $CORITAN_TOKEN" \ -H "Content-Type: application/json" \ -d '{"template_id": 7}' ``` The response is `{"status": "rebuild_queued"}`, and the rebuild runs in the background. Follow it with `GET /api/v1/client/vps/{uuid}/tasks`: the host destroys the old machine in a `qmdestroy` task and starts the new one in a `qmstart` task, and the rebuild is done when that `qmstart` reads `OK`. A suspended instance answers `403` with `Cannot rebuild a suspended instance`, and one we are moving answers `409` with `Instance is migrating`. An image that is not ready on the instance's host answers `400` with `Template not available (ready) on this node`. More than 8 rebuilds in 10 minutes answer `429`, as [Rate limits](/api/rate-limits/) describes. ## API - `POST /api/v1/client/vps/{uuid}/rebuild`: Rebuild the VPS with a new OS template (https://docs.coritan.com/api/reference/client/cloud-compute/#op-post-api-v1-client-vps-uuid-rebuild)