# volume

> Manage network volumes and move files in and out of them from your terminal.

The `volume` commands create, resize and delete network volumes, and copy, sync, find and organize the files in them.

```bash
featherless-cloud volume <subcommand> [options]
```

Files in a volume are written `vol://<volume>/<path>`. Every other path is on your computer.

## Create a volume

```bash
featherless-cloud volume create datasets --gb 500 --wait
```

<table>
<thead>
  <tr>
    <th>
      Option
    </th>
    
    <th>
      Description
    </th>
  </tr>
</thead>

<tbody>
  <tr>
    <td>
      <code>
        --gb <n>
      </code>
    </td>
    
    <td>
      Capacity, 10–10,240. Default 100.
    </td>
  </tr>
  
  <tr>
    <td>
      <code>
        --wait
      </code>
    </td>
    
    <td>
      Wait up to about 30 seconds for the volume to become usable.
    </td>
  </tr>
</tbody>
</table>

## Show a volume

```bash
featherless-cloud volume get datasets
```

Prints its status, capacity, usage, price and the instances using it.

## Change a volume

```bash
featherless-cloud volume update datasets --gb 1000
featherless-cloud volume update datasets --name training-data
```

<table>
<thead>
  <tr>
    <th>
      Option
    </th>
    
    <th>
      Description
    </th>
  </tr>
</thead>

<tbody>
  <tr>
    <td>
      <code>
        --name <name>
      </code>
    </td>
    
    <td>
      New name.
    </td>
  </tr>
  
  <tr>
    <td>
      <code>
        --gb <n>
      </code>
    </td>
    
    <td>
      New capacity. It can only grow.
    </td>
  </tr>
</tbody>
</table>

## Delete a volume

```bash
featherless-cloud volume delete datasets
```

Deletes the volume and all its files. Instances using it must be stopped first.

<table>
<thead>
  <tr>
    <th>
      Option
    </th>
    
    <th>
      Description
    </th>
  </tr>
</thead>

<tbody>
  <tr>
    <td>
      <code>
        --yes
      </code>
    </td>
    
    <td>
      Skip the confirmation.
    </td>
  </tr>
</tbody>
</table>

## List files

```bash
featherless-cloud volume ls
featherless-cloud volume ls -h vol://datasets/train
featherless-cloud volume ls -r --summarize vol://datasets/train
```

With no path, `ls` lists your volumes.

<table>
<thead>
  <tr>
    <th>
      Option
    </th>
    
    <th>
      Description
    </th>
  </tr>
</thead>

<tbody>
  <tr>
    <td>
      <code>
        -h
      </code>
      
      , <code>
        --human-readable
      </code>
    </td>
    
    <td>
      Sizes like <code>
        1.5 KiB
      </code>
      
       instead of bytes.
    </td>
  </tr>
  
  <tr>
    <td>
      <code>
        -r
      </code>
      
      , <code>
        --recursive
      </code>
    </td>
    
    <td>
      Everything below the folder.
    </td>
  </tr>
  
  <tr>
    <td>
      <code>
        --summarize
      </code>
    </td>
    
    <td>
      End with the number of files and their total size.
    </td>
  </tr>
</tbody>
</table>

## Find files

```bash
featherless-cloud volume find vol://datasets --name "*.safetensors" --size +1G -l -h
featherless-cloud volume find vol://datasets/tmp --type f --older-than 30d
```

<table>
<thead>
  <tr>
    <th>
      Option
    </th>
    
    <th>
      Description
    </th>
  </tr>
</thead>

<tbody>
  <tr>
    <td>
      <code>
        --name <pattern>
      </code>
    </td>
    
    <td>
      Names matching a pattern: <code>
        *
      </code>
      
       any characters, <code>
        ?
      </code>
      
       one, <code>
        [abc]
      </code>
      
       one of those.
    </td>
  </tr>
  
  <tr>
    <td>
      <code>
        --iname <pattern>
      </code>
    </td>
    
    <td>
      Like <code>
        --name
      </code>
      
      , ignoring case.
    </td>
  </tr>
  
  <tr>
    <td>
      <code>
        --type <f|d>
      </code>
    </td>
    
    <td>
      Only files or only folders.
    </td>
  </tr>
  
  <tr>
    <td>
      <code>
        --maxdepth <n>
      </code>
    </td>
    
    <td>
      How many levels below the folder to look.
    </td>
  </tr>
  
  <tr>
    <td>
      <code>
        --size <+n|-n>
      </code>
    </td>
    
    <td>
      Bigger (<code>
        +
      </code>
      
      ) or smaller (<code>
        -
      </code>
      
      ) than a size like <code>
        10G
      </code>
      
       or <code>
        512K
      </code>
      
      . Repeat for a range.
    </td>
  </tr>
  
  <tr>
    <td>
      <code>
        --newer-than <age>
      </code>
    </td>
    
    <td>
      Modified within an age like <code>
        30m
      </code>
      
      , <code>
        12h
      </code>
      
       or <code>
        7d
      </code>
      
      .
    </td>
  </tr>
  
  <tr>
    <td>
      <code>
        --older-than <age>
      </code>
    </td>
    
    <td>
      Modified longer ago than an age.
    </td>
  </tr>
  
  <tr>
    <td>
      <code>
        -l
      </code>
      
      , <code>
        --long
      </code>
    </td>
    
    <td>
      Print modification time and size.
    </td>
  </tr>
  
  <tr>
    <td>
      <code>
        -h
      </code>
      
      , <code>
        --human-readable
      </code>
    </td>
    
    <td>
      With <code>
        -l
      </code>
      
      , readable sizes.
    </td>
  </tr>
  
  <tr>
    <td>
      <code>
        -0
      </code>
      
      , <code>
        --print0
      </code>
    </td>
    
    <td>
      End each path with a NUL, for <code>
        xargs -0
      </code>
      
      .
    </td>
  </tr>
</tbody>
</table>

## Copy files

```bash
featherless-cloud volume cp -r ./weights vol://datasets/models/
featherless-cloud volume cp vol://datasets/models/out.bin .
featherless-cloud volume cp -r vol://datasets/models vol://archive/
```

`cp` works in either direction and between volumes. An interrupted copy resumes when you run it again. A destination ending in `/`, or an existing folder, receives each source under its own name.

<table>
<thead>
  <tr>
    <th>
      Option
    </th>
    
    <th>
      Description
    </th>
  </tr>
</thead>

<tbody>
  <tr>
    <td>
      <code>
        -r
      </code>
      
      , <code>
        --recursive
      </code>
    </td>
    
    <td>
      Copy folders and everything in them.
    </td>
  </tr>
  
  <tr>
    <td>
      <code>
        --overwrite
      </code>
    </td>
    
    <td>
      Replace files that already exist.
    </td>
  </tr>
  
  <tr>
    <td>
      <code>
        --update
      </code>
    </td>
    
    <td>
      Skip files whose size and modification time already match.
    </td>
  </tr>
  
  <tr>
    <td>
      <code>
        --concurrency <n>
      </code>
    </td>
    
    <td>
      Files at a time, 1–32. Default 16.
    </td>
  </tr>
  
  <tr>
    <td>
      <code>
        --dry-run
      </code>
    </td>
    
    <td>
      Show what would be copied.
    </td>
  </tr>
  
  <tr>
    <td>
      <code>
        -v
      </code>
      
      , <code>
        --verbose
      </code>
    </td>
    
    <td>
      Print each file.
    </td>
  </tr>
  
  <tr>
    <td>
      <code>
        --progress
      </code>
    </td>
    
    <td>
      Show a running total.
    </td>
  </tr>
</tbody>
</table>

## Move files

```bash
featherless-cloud volume mv vol://datasets/tmp/run1 vol://datasets/runs/
```

<table>
<thead>
  <tr>
    <th>
      Option
    </th>
    
    <th>
      Description
    </th>
  </tr>
</thead>

<tbody>
  <tr>
    <td>
      <code>
        --overwrite
      </code>
    </td>
    
    <td>
      Replace files at the destination.
    </td>
  </tr>
  
  <tr>
    <td>
      <code>
        --concurrency <n>
      </code>
    </td>
    
    <td>
      Files at a time, 1–32. Default 16.
    </td>
  </tr>
  
  <tr>
    <td>
      <code>
        -v
      </code>
      
      , <code>
        --verbose
      </code>
    </td>
    
    <td>
      Print each file.
    </td>
  </tr>
  
  <tr>
    <td>
      <code>
        --progress
      </code>
    </td>
    
    <td>
      Show a running total.
    </td>
  </tr>
</tbody>
</table>

## Sync a folder

```bash
featherless-cloud volume sync --dry-run ./weights vol://datasets/models
featherless-cloud volume sync ./weights vol://datasets/models
featherless-cloud volume sync --mode download --mkdirs ./results vol://datasets/results
```

`sync` sends what changed on each side since the last sync and never deletes. The folder must exist on both sides unless you pass `--mkdirs`. A file changed on both sides is a conflict: `sync` reports it and exits with code 1. Choose a side with `--mode upload --force` or `--mode download --force`.

<table>
<thead>
  <tr>
    <th>
      Option
    </th>
    
    <th>
      Description
    </th>
  </tr>
</thead>

<tbody>
  <tr>
    <td>
      <code>
        --mode <upload|download|both>
      </code>
    </td>
    
    <td>
      Direction. Default <code>
        both
      </code>
      
      .
    </td>
  </tr>
  
  <tr>
    <td>
      <code>
        --dry-run
      </code>
    </td>
    
    <td>
      Show what would be sent and received.
    </td>
  </tr>
  
  <tr>
    <td>
      <code>
        --mkdirs
      </code>
    </td>
    
    <td>
      Create missing folders on either side.
    </td>
  </tr>
  
  <tr>
    <td>
      <code>
        --force
      </code>
    </td>
    
    <td>
      Send even when the chosen side is older. Needs <code>
        --mode upload
      </code>
      
       or <code>
        --mode download
      </code>
      
      .
    </td>
  </tr>
  
  <tr>
    <td>
      <code>
        --include <glob>
      </code>
    </td>
    
    <td>
      Only paths matching a glob. Repeatable.
    </td>
  </tr>
  
  <tr>
    <td>
      <code>
        --exclude <glob>
      </code>
    </td>
    
    <td>
      Skip paths matching a glob. Repeatable.
    </td>
  </tr>
  
  <tr>
    <td>
      <code>
        --concurrency <n>
      </code>
    </td>
    
    <td>
      Files at a time, 1–32. Default 16.
    </td>
  </tr>
  
  <tr>
    <td>
      <code>
        -v
      </code>
      
      , <code>
        --verbose
      </code>
    </td>
    
    <td>
      Print each file.
    </td>
  </tr>
  
  <tr>
    <td>
      <code>
        --progress
      </code>
    </td>
    
    <td>
      Show a running total.
    </td>
  </tr>
</tbody>
</table>

## Create folders

```bash
featherless-cloud volume mkdir -p vol://datasets/runs/2026-11
```

<table>
<thead>
  <tr>
    <th>
      Option
    </th>
    
    <th>
      Description
    </th>
  </tr>
</thead>

<tbody>
  <tr>
    <td>
      <code>
        -p
      </code>
      
      , <code>
        --parents
      </code>
    </td>
    
    <td>
      Create missing parent folders.
    </td>
  </tr>
</tbody>
</table>

## Delete files

```bash
featherless-cloud volume rm vol://datasets/old.bin
featherless-cloud volume rm -r vol://datasets/tmp
```

<table>
<thead>
  <tr>
    <th>
      Option
    </th>
    
    <th>
      Description
    </th>
  </tr>
</thead>

<tbody>
  <tr>
    <td>
      <code>
        -r
      </code>
      
      , <code>
        --recursive
      </code>
    </td>
    
    <td>
      Delete folders and everything in them.
    </td>
  </tr>
  
  <tr>
    <td>
      <code>
        --yes
      </code>
    </td>
    
    <td>
      Skip the confirmation.
    </td>
  </tr>
</tbody>
</table>
