# Manage instances

> Pay only for the instances you're using, and change or remove them as your work changes.

This page explains what each action does to an instance, its disk and its billing.

<table>
<thead>
  <tr>
    <th>
      Action
    </th>
    
    <th>
      Console
    </th>
    
    <th>
      CLI
    </th>
  </tr>
</thead>

<tbody>
  <tr>
    <td>
      Start
    </td>
    
    <td>
      <strong>
        Start
      </strong>
    </td>
    
    <td>
      <code>
        featherless-cloud instance start <name>
      </code>
    </td>
  </tr>
  
  <tr>
    <td>
      Stop
    </td>
    
    <td>
      <strong>
        Stop
      </strong>
    </td>
    
    <td>
      <code>
        featherless-cloud instance stop <name>
      </code>
    </td>
  </tr>
  
  <tr>
    <td>
      Restart
    </td>
    
    <td>
      <strong>
        Restart instance
      </strong>
    </td>
    
    <td>
      <code>
        featherless-cloud instance restart <name>
      </code>
    </td>
  </tr>
  
  <tr>
    <td>
      Edit
    </td>
    
    <td>
      <strong>
        Edit instance
      </strong>
    </td>
    
    <td>
      <code>
        featherless-cloud instance update <name>
      </code>
    </td>
  </tr>
  
  <tr>
    <td>
      Delete
    </td>
    
    <td>
      <strong>
        Delete instance
      </strong>
    </td>
    
    <td>
      <code>
        featherless-cloud instance delete <name>
      </code>
    </td>
  </tr>
</tbody>
</table>

To act on several instances at once, select them in the **Instances** list and use **Bulk actions**.

## Stop an instance

Stopping ends all processes and compute billing. The instance keeps its configuration, and its network volumes stay attached for the next start.

<warning>

Stopping erases the container disk. Anything not on a network volume is lost.

</warning>

## Start an instance

Starting runs the instance again with its last configuration. It needs a free GPU of the same model and credit for one hour of its rate, so it can fail when capacity is short.

## Restart an instance

Restarting runs the container again on the same GPU. Like a stop, it erases the container disk.

## Edit an instance

Saving changes to a running instance restarts it, which erases the container disk. Public ports and the tmux setting apply without a restart.

Secret variables and network volumes can't be changed after you deploy. To change them, deploy a new instance.

With the CLI, `--env` replaces the whole list of environment variables.

## Delete an instance

Deleting removes the instance and its configuration. Network volumes are kept, with their data, and keep billing until you delete them.

## When an instance keeps crashing

If the container crashes 10 times in a row, each time within 10 minutes of starting, the instance is stopped automatically and every member of the organization is emailed. Check its **Logs** and start command before starting it again.

## Next steps

<card-group>
<card icon="i-lucide-chart-line" title="Logs and metrics" to="/instances/logs-and-metrics">

See what an instance is doing.

</card>

<card icon="i-lucide-plug" title="Connect to an instance" to="/instances/connect">

Open a terminal or connect with SSH.

</card>
</card-group>
