I would love Native SR-IOV graphics support, and Native implimentation of Open vSwitch and Open Virtual Network. They can be found at https://www.openvswitch.org/ and https://www.ovn.org/en/.

 

It is the one big feature that makes me re-install Proxmox every few months, I have a huge beefy server and want my whole lab contained inside it. I can fully do this in Proxmox and ESXi but Unraid right now completly lacks these features. As an example I would be able to virtualize switches and network ports and share them between VMs. This would allow me to have one virtual network running pfSense and another running a custom RHEL appliance, then using Kali Linux to test differences between them and impliment fixes for future configs, all while not exposing my actual network.

I've not read the thread to see if someone already mentioned this but:

 

When you have PCIe devices set to be passed through to a VM and then perform a BIOS update to your motherboard that may (not always, but may) change the IOMMU group ID of some or all of your passed through devices.

 

This presents an issue where the VM Managers GUI for a VM lists the devices you can pass through to the VM like normal, but the XML config still has the old entries which are not presented in the GUI because the group ID's changed vs what was in the XML previously.

 

This stops the VM from being started with an error message of missing PCI devices. The only fix is to modify the XML configuration by hand which I think many won't know what to look for or remove or even believe that's the issue due to the GUI not corresponding to what the XML has inside of it.

 

I think a fix for this would be to update the GUI to recognize missing devices that are present in the XML and ask the user and have buttons to remove missing devices.

8 hours ago, Pri said:

I've not read the thread to see if somieone already mentioned this but:

if

When you have PCIe devices set to be passed through to a VM and then perform a BIOS update to your motherboard that may (not always, but m ay) change the IOMMU group ID of some or all of your passed through devices.

 

This presents an issue where the VM Managers GUI for a VM lists the devices you can pass through to the VM like normal, but the XML config still has the old entries which are not presented in the GUI because the group ID's changed vs what was in the XML previously.

 

This stops the VM from being started with an error message of missing PCI devices. The only fix is to modify the XML configuration by hand which I think many won't know what to look for or remove or even believe that's the issue due to the GUI not corresponding to what the XML has inside of it.

 

I think a fix for this would be to update the GUI to recognize missing devices that are present in the XML and ask the user and have buttons to remove missing devices.

Hi @Pri It is something we have been thinking about, but challenge is that libvirt only stores the PCI ID and does not include any vendor product information. I was thinking that we should store the last lspci values at reboot/power off so if hardware as been changed make the VM as not being able start if it contains a PCI ID that now has a different vendor/product. Could include updating devices or maybe suggest new ID. I guess you meant PCIID rather than IOMMU group as only would be an issue if the device was put into a new IOMMU group that had other cards.

On 3/15/2025 at 3:53 PM, SimonF said:

Hi @Pri It is something we have been thinking about, but challenge is that libvirt only stores the PCI ID and does not include any vendor product information. I was thinking that we should store the last lspci values at reboot/power off so if hardware as been changed make the VM as not being able start if it contains a PCI ID that now has a different vendor/product. Could include updating devices or maybe suggest new ID. I guess you meant PCIID rather than IOMMU group as only would be an issue if the device was put into a new IOMMU group that had other cards.

Sure yeah, the PCI ID. I found when updating my BIOS both changed sometimes. I'm using EPYC which has 102 different IOMMU groups and when updating my BIOS recently half of my devices changed ID's and it has done that before on previous BIOS updates aswell.

 

One other issue I found is one of my Network card ports I had passed through to one VM had swapped ID's with an AMD CPU feature, when the VM started it took control of this: Encryption controller: Advanced Micro Devices, Inc. [AMD] Starship/Matisse PTDMA just because it had occupied the PCI ID of the network card previously. My fault for not checking before I started the VM but if you have a lot of VM's with passed through stuff this would definitely get complicated quickly.

When adding a share through the UI in Unraid VM setup, users currently have to manually mount the share from the VM OS.

The UI makes it seem like this is done automatically when adding the share but this in fact not possible since the setup has no control over VM OS files so mounting the added share from the UI must be a manual step.

 

A helpful improvement would be to provide a dynamically generated line suggestion within the template that needs to be added to the fstab file when adding a share. Possible example:

 

# Mount <sharename> share at startup using virtiofs <mountpathinvm> virtiofs defaults 0 0
<sharename> <mountpathinvm> virtiofs defaults 0 0

 

Taking into account the type of share virtiofs/9pmode and whatever other setting is needed that changes the mount script 

This would guide users through the correct mounting process and make this feature a little more polished.

 

Somewhere in here, near its respective VM would be a nice place to put the info described above

 

 

Found this thread after I already made a feature request for it: 
Dynamic Script Suggestion for Mounting Shares in Unraid VM Setup when adding a share using the UI - Feature Requests - Unraid

5 hours ago, Tareku said:

When adding a share through the UI in Unraid VM setup, users currently have to manually mount the share from the VM OS.

The UI makes it seem like this is done automatically when adding the share but this in fact not possible since the setup has no control over VM OS files so mounting the added share from the UI must be a manual step.

 

A helpful improvement would be to provide a dynamically generated line suggestion within the template that needs to be added to the fstab file when adding a share. Possible example:

 

# Mount <sharename> share at startup using virtiofs <mountpathinvm> virtiofs defaults 0 0
<sharename> <mountpathinvm> virtiofs defaults 0 0

 

Taking into account the type of share virtiofs/9pmode and whatever other setting is needed that changes the mount script 

This would guide users through the correct mounting process and make this feature a little more polished.

 

Somewhere in here, near its respective VM would be a nice place to put the info described above

 

 

Found this thread after I already made a feature request for it: 
Dynamic Script Suggestion for Mounting Shares in Unraid VM Setup when adding a share using the UI - Feature Requests - Unraid

There is some info in the help text will look to expand.

 

8 hours ago, SimonF said:

There is some info in the help text will look to expand.

 

Thank you for the reply, this is indeed helpful, but I still believe what I mentioned could be valuable. 
I am having issues with finding the help text that you posted though in the VM creation page, is it somewhere else in the docs?

EDIT: Most of the fields in the other windows have a very useful info icon to explain some of the settings. It would be nice to have that available here as well instead of hunting for the info you share!

(I could be blind but I do not see it on my page)

Edited May 16, 20251 yr by Tareku
added more info

2 hours ago, Tareku said:

Thank you for the reply, this is indeed helpful, but I still believe what I mentioned could be valuable. 
I am having issues with finding the help text that you posted though in the VM creation page, is it somewhere else in the docs?

EDIT: Most of the fields in the other windows have a very useful info icon to explain some of the settings. It would be nice to have that available here as well instead of hunting for the info you share!

(I could be blind but I do not see it on my page)

Click on the ? In a circle top right to show help text

I don't need a individual Core Selection. The Option of this VM becomes 4 Cores, this 8 and this 2. The VM/CPU, ... can handle what Core he actually used.

For People that need a individual Core Selection he can used the actually functionality.

1 hour ago, Revan335 said:

I don't need a individual Core Selection. The Option of this VM becomes 4 Cores, this 8 and this 2. The VM/CPU, ... can handle what Core he actually used.

For People that need a individual Core Selection he can used the actually functionality.

Unsure what you are requesting.

4 hours ago, SimonF said:

Unsure what you are requesting.

I must selected the CPU Cores that used by VM. For example Core 0/1+2/3 = 4 Cores

I don't type VM used 4 Cores without specific Cores.

13 minutes ago, Revan335 said:

I must selected the CPU Cores that used by VM. For example Core 0/1+2/3 = 4 Cores

I don't type VM used 4 Cores without specific Cores.

Feature was added in 7.1 for this. You can choose vCPUs without pinning.

3 hours ago, SimonF said:

Feature was added in 7.1 for this. You can choose vCPUs without pinning.

Than I'm blind. Sorry!

13 hours ago, Revan335 said:

Than I'm blind. Sorry!

changelogs ... ;)

Version 7.1.0 2025-05-05 | Unrai...

Version 7.1.0 2025-05-05 | Unraid Docs

This release adds wireless networking, the ability to import TrueNAS and other foreign pools, multiple enhancements to VMs, early steps toward making the webGUI responsive, and more.

Uploading Attachment...

better storage management GUI to facilitate the addition of shared storage.

Did you mean more than these options,

Can you provide a use case or more information?

I mean the ability to do shared storage for the OS disk. Not within the VM itself.

It would be good if we could set a manual override for the guest address for hostdevs from the UI form view. Or at least prevent the form view from removing manual address edits done in the XML view.

Even better if the form view could help you identify and pick a free address somehow.

Maybe it's an edge case, but I'm specifically thinking about iGPU address handling for Q35, as described here:

Bonus request: being able to set hostdev alias names in the form view. They also get removed if added via XML and then the VM is edited in the form view.

Many games with anti-cheat prevent you from launching if they detect you're running in a VM.

It would be great to have a feature to toggle or set the following VM settings from the Windows VM GUI.

The required VM settings, which prevent detection with most anti-cheat, are listed in the following posts:
1. https://forums.unraid.net/topic/127648-gaming-vm-cannot-run-under-virtual-machine-error-when-launching-game/#findComment-1163903
2. https://forums.unraid.net/topic/127648-gaming-vm-cannot-run-under-virtual-machine-error-when-launching-game/page/2/#findComment-1556517

Edited September 18, 2025Sep 18 by Alexb111

@SimonF

When will VirGL VirGL hardware accelerated h264/h265 be supported. I see that proxmox someone already has access to this feature.

Address: https://forum.proxmox.com/threads/virgl-hardware-accelerated-h264-h265.137023/

On 10/16/2025 at 9:54 AM, wdcew123 said:

@SimonF

When will VirGL VirGL hardware accelerated h264/h265 be supported. I see that proxmox someone already has access to this feature.

Address: https://forum.proxmox.com/threads/virgl-hardware-accelerated-h264-h265.137023/

Virgl is in 7.2

It would be really nice if Unraid had a UI for users to manage (start, shutdown, restart, snapshot) VMs. To be precise user logs in with their unraid share credentials and if assigned can control specific VMs. (maybe dockers as well).

Maybe this is a bit off topic as mostly virtualized hardware is discussed here but hey I'm literally looking for a "VM manager" for non root user!

Is there any support for Venus with this method?