Proxmox VE Shared Storage Solution - Setup and Configuration of VirtIO-FS

Preface

The author recently switched their server from Windows Server 2025 to a combination of Debian 13 + Proxmox VE. When using Windows previously, although data was categorized, Hyper-V virtual machines and other data files were still on the same disk. After switching to Proxmox VE, there was a need for multiple virtual machines to share disk access. However, there are many file sharing protocols with varying levels of configuration complexity. This article will introduce the VirtIO-FS protocol and guide you through its configuration.

Why Choose VirtIO-FS

• High Performance: Bypasses the traditional network protocol stack, uses shared memory for communication between host and virtual machines, and directly utilizes the VM's vCPU to handle file operations, achieving performance nearly identical to native disk performance.
• High Compatibility: Supports Linux kernel 5.4+ and Windows, while providing high support for POSIX protocols (such as symbolic links, hard links, permissions, locks, etc.). (Note: Some AIs seem to question the level of POSIX support)
• High Synchronization: Changes made to files by one party are immediately visible to the other
• Simple Configuration: User-friendly with straightforward setup
• Perfect Ecosystem: VirtIO-FS shares can be directly configured in the Proxmox VE Web UI, offering convenient operation and unified management

Example Environment

• OS: Debian 13
• Proxmox VE: Proxmox VE 9.1.2 x86_64

Disclaimer

Please ensure data security before proceeding. The author and this article are not responsible for any data loss incurred during operations.

---

Let's Get Started

1. Mounting Disks

• Use the following command to list disk devices

ls /dev | grep -i sd

You will see output similar to this:

sda
sda1
sdb
sdc
sdc1
sdd
sde
sde1
sdf
sdf1
sdg
sdg1
sdh
sdh1
sdh2
sdh3
sdh4
sdh5
sdh6

Here, sd<letter> represents an entire disk. For example, sda is a complete disk. sd<letter><number> represents a partition on the disk, so sda1 is partition 1 on the sda disk.

• Create a mount point in the mount directory and mount the disk

sudo mkdir /mnt/<diskName>
sudo mount /dev/<partition> /mnt/<diskName>

<diskName>: Partition name, can be filled in arbitrarily
<partition>: Partition identifier, such as sda1

Note: A mount point can only mount a single partition, not an entire disk.

• After confirming the partition, modify fstab to ensure the partition mounts automatically at boot

  sudo vim /etc/fstab

  Here's a reference to the author's fstab:

  # =========================
  # System partitions
  # =========================
  
  UUID=b21896ae-beed-4437-aff5-fb211062cbd9 /               ext4    errors=remount-ro 0 1
  UUID=AEDB-EAD3                            /boot/efi       vfat    umask=0077          0 1
  UUID=43b2adb9-e081-4b2b-897d-b2245f93348d /srv            ext4    defaults            0 2
  UUID=6a7e3626-1e06-4896-9248-55189d9702c3 /var            ext4    defaults            0 2
  UUID=c32ce32a-417f-465a-b26d-12c9e9d67a70 none            swap    sw                  0 0
  
  
  # =========================
  # Data partitions (NTFS)
  # =========================
  
  UUID=2E26E857F9477815 /mnt/Data         ntfs3 defaults,noatime,uid=1000,gid=1000,nofail 0 0
  UUID=20E085F238CE9570 /mnt/BrokenDisk   ntfs3 defaults,noatime,uid=1000,gid=1000,nofail 0 0
  UUID=51F8DF7BA9CECE74 /mnt/BackupData   ntfs3 defaults,noatime,uid=1000,gid=1000,nofail 0 0
  UUID=FB426E0A0790E389 /mnt/Media        ntfs3 defaults,noatime,uid=1000,gid=1000,nofail 0 0
  UUID=99ABE9C8F7DF98C7 /mnt/FileStorage  ntfs3 defaults,noatime,uid=1000,gid=1000,nofail 0 0

• Shut down the virtual machine you're about to operate on, save your changes, and run the following commands to apply the changes

sudo systemctl daemon-reload
sudo mount -a

• If there are no errors, the disk mounting was successful

Note: If your partition uses the NTFS filesystem and was in passthrough mode before shutting down the VM, you may encounter mounting issues as shown in Figure 1-1

If you encounter this issue, you can fix it using the following command:

sudo ntfsfix -d /dev/<partition>

2. Adding VirtIO-FS Nodes

• Log in to the Proxmox VE Web UI, navigate to datacenter, find Directory Mappings, as shown in Figure 2-1
  

  

• Click Add, fill in the parameters as needed, and add the VirtIO-FS node. You can refer to the author's configuration as shown in Figure 2-2
  

  

• After filling in the information, click Create to complete the creation of the VirtIO-FS node
• Navigate to the hardware page of the virtual machine you want to operate on, as shown in Figure 2-3
  

  

• Click Add --> Virtiofs to add the desired VirtIO-FS node

  Note: Generally, there's no need to modify parameters. Before changing parameters, make sure you understand what you're doing and the consequences.

3. Modifying VM Mount Points

• Start the virtual machine and use the following command to open fstab

sudo vim /etc/fstab

• Mount the disk in the following format

<mapName>  </path/to/mountPoint>   defaults,nofail    0 0

Notes:

• Do not include uid or gid in the mount parameters, as VirtIO-FS does not accept user and group parameters by default, which would cause mount failures.
• VirtIO permissions completely follow the host. If you want to ensure the virtual machine has sufficient permissions to access files on the disk, modify the user's uid and gid to match those of the user used to mount the disk on the host.

Here's a reference to the author's configuration:

FileStorage     /mnt/FileStorage        virtiofs        defaults,nofail 0 0
Data            /mnt/Data               virtiofs        defaults,nofail 0 0
Media           /mnt/Media              virtiofs        defaults,nofail 0 0
BackupData      /mnt/BackupData         virtiofs        defaults,nofail 0 0

• After saving, use this command to mount

sudo mount -a

If there are no error messages, the mounting was successful.

---

Conclusion

• Thank you for reading. If you found this useful, please share it with those in need
• The author is not a professional. If there are any errors or omissions, please feel free to point them out!