[VM GPU Passthrough, EPYC 7551, Nvidia 2080 on Supermicro H11SSL-i]

Check: 

BIOS >> Advanced >> NB Configuration >> IOMMU

--> is enabled, not auto, not disabled

[[6.10.3] Unraid windows VM SLOW after update]

I wasn't expecting the boot repair, anyway it can happen, basically is like you physically disconnected the hd from a scsi controller and connected it to a sata one.

If I were you I would make a backup copy of the vdisk and run the boot repair.

If nothing improves or if the disk will get corrupted you can revert the changes in the xml with the original disk.

[[6.10.3] Unraid windows VM SLOW after update]

From what I can see there's no irq conflict.

Can you try changing the virtual virtio controller to a sata one for the disks and check if things are better?

 

Change from this:

    <disk type='file' device='disk'>
      <driver name='qemu' type='raw' cache='writeback'/>
      <source file='/mnt/user/vm/win10/vdisk1.img' index='3'/>
      <backingStore/>
      <target dev='hdc' bus='virtio'/>
      <boot order='1'/>
      <alias name='virtio-disk2'/>
      <address type='pci' domain='0x0000' bus='0x00' slot='0x04' function='0x0'/>
    </disk>
    <disk type='block' device='disk'>
      <driver name='qemu' type='raw' cache='writeback'/>
      <source dev='/dev/disk/by-id/ata-WDC_WD20EZRX-00DC0B0_WD-WCC1T0211008' index='2'/>
      <backingStore/>
      <target dev='hdd' bus='virtio'/>
      <alias name='virtio-disk3'/>
      <address type='pci' domain='0x0000' bus='0x00' slot='0x05' function='0x0'/>
    </disk>

 

To this:

    <disk type='file' device='disk'>
      <driver name='qemu' type='raw' cache='writeback'/>
      <source file='/mnt/user/vm/win10/vdisk1.img'/>
      <backingStore/>
      <target dev='hdc' bus='sata'/>
      <boot order='1'/>
      <address type='drive' controller='0' bus='0' target='0' unit='2'/>
    </disk>
    <disk type='block' device='disk'>
      <driver name='qemu' type='raw' cache='writeback'/>
      <source dev='/dev/disk/by-id/ata-WDC_WD20EZRX-00DC0B0_WD-WCC1T0211008'/>
      <backingStore/>
      <target dev='hdd' bus='sata'/>
      <address type='drive' controller='0' bus='0' target='0' unit='3'/>
    </disk>

 

[6500XT Passthrough on X99]

I would try to enable allow unsafe interrupts in unraid settings (reboot after change).

If this doesn't work I would try kernel parameter pci=noaer (reboot after change).

 

Do not make any manual changes, you should pass only 05:00.0 and .1, do not passthrough pci bridges, set 05:00.0 and .1 in a multifunction device.

[GPU Pass through - HD7750]

17 hours ago, Sabot said:

Went to device manager and found the the video was missing

17 hours ago, Sabot said:

I tried to install AMD drivers and each time it failed for it says the hardware is missing

 

Since the gpu is quite old, and I imagine the drivers too, they may expect the sound to be on the same bus/slot but different function (which it is in the real hardware, but not in the vm), so I would give a try by modifying the vm xml from this:

    <hostdev mode='subsystem' type='pci' managed='yes'>
      <driver name='vfio'/>
      <source>
        <address domain='0x0000' bus='0x0c' slot='0x00' function='0x0'/>
      </source>
      <address type='pci' domain='0x0000' bus='0x04' slot='0x00' function='0x0'/>
    </hostdev>
    <hostdev mode='subsystem' type='pci' managed='yes'>
      <driver name='vfio'/>
      <source>
        <address domain='0x0000' bus='0x0c' slot='0x00' function='0x1'/>
      </source>
      <address type='pci' domain='0x0000' bus='0x05' slot='0x00' function='0x0'/>
    </hostdev>

 

To this:

    <hostdev mode='subsystem' type='pci' managed='yes'>
      <driver name='vfio'/>
      <source>
        <address domain='0x0000' bus='0x0c' slot='0x00' function='0x0'/>
      </source>
      <address type='pci' domain='0x0000' bus='0x04' slot='0x00' function='0x0' multifunction='on'/>
    </hostdev>
    <hostdev mode='subsystem' type='pci' managed='yes'>
      <driver name='vfio'/>
      <source>
        <address domain='0x0000' bus='0x0c' slot='0x00' function='0x1'/>
      </source>
      <address type='pci' domain='0x0000' bus='0x04' slot='0x00' function='0x1'/>
    </hostdev>

 

[Mac OS Catalina VM on unRAID 6.10.3]

2 hours ago, Zippi said:

This GPU has a UEFI custom bios because originally it was NOT UEFI

Yes, I added a note to my post above as I saw later that that gpu could be without uefi in bios and as you know mac os doesn't work without uefi gpu (at least versions of latest years).

 

2 hours ago, Zippi said:

I installed virt manager because unlike the one I saw in the Spaceinvader One video, Macinabox didn't give me the option to choose if I wanted it or not. Maybe it's this latest version of Macinabox that no longer includes it, I don't know

Yes, the actual version of macinabox doesn't include virt-manager.

 

2 hours ago, Zippi said:

I copied the EFI partition folders of the bootloader to the EFI partition of the Mac OS installed as Spaceinvader One did in the video

Take into account that you can avoid this, you can have a separate disk with the EFI files (opencore.qcow2, opencore.img or whatever you want) and let the original apple EFI files on the disk; you will boot from the separate disk which will find and boot mac os disk; by this way you can easily replace the EFI disk without having access to the mac os disk.

Anyway both ways work.

 

2 hours ago, Zippi said:

that's why I asked if I needed to retouch some parameters like when building a Hackintosh

It's similar, but not the same; when you build a virtual machine with machinabox all the hardware is emulated, even the cpu, so the basic setup should work for all without any change. Thing change when you passthrough hardware (like gpu, ethernet, usb, etc.), when you customize things changes to the opencore config could be mandatory to make all the things work together.

 

2 hours ago, Zippi said:

I did NOT use VFIO nor ACS patch, audio and video were already on the correct bus together

In real you used vfio; when you passthrough hardware to a vm, in this case the gpu, you are attaching it to the vfio driver.

 

2 hours ago, Zippi said:

However now everything works!

And this is the important thing!!!

[Windows VM looking for USB device, won't start (no device in config)]

Open the vm in the xml mode and check at the bottom if there is something related to that usb device; if it's there manually delete the offending block of code.

[Newbie VM Confusion]

25 minutes ago, JP said:

could a VM hosted from an unraid server accomplish video editing really well?

If the vm is properly configured, without anything else running in the host (or nearly nothing running on the host) you could take into account that the vm will perform bare metal -5 - -10%; so if you assign let's say 4 cores/4 threads to the vm, it will perform like a real 4 cores/4 threads minus 5-10% in performance; this is a rough estimation.

25 minutes ago, JP said:

are there any recommendations on what I might want to consider for CPU / MB

this is only a matter of money, but make sure the cpu(s) supports vt-d and vt-x, so you will be able to passthrough hardware (descrete gpu?) through vfio to the vm.

 

About the gpu, to give you an example, I have the latest 6900xt which gives a score of about 140.000 in Geekbench (it's mac os); online results with the same gpu are higher too (about 180.000), but this may depend on cpu bottleneck as I have 2 old sandy bridge xeon cpus. I would say that also the gpu will perform near bare metal performance, vfio is good; make sure to use q35 machine type for the vm so it will be more compatible with pcie passthrough.

[GPU acceleration in Monterey VM]

35 minutes ago, HomerJ said:

I want to install Garageband and unfortunately it won't work unless I upgrade the OS to the next version

I remember garage band was preinstalled on mac os leopard in my old old old mackbook pro, cannot you use an older version or is it a no go?

[GPU acceleration in Monterey VM]

7 hours ago, HomerJ said:

Do you know if I can get the igpu to work?

Yes, it should work, but it's more complicated to make it work, first because usually the igpu is the primary gpu used by the host and isolate it to make it available for the vm can be a mess (because of iommu groups, because of the motherboard bios, because of the os system...), second you should setup opencore for your igpu and this may include framebuffer patching.

You could follow dortania's guide and hackintosh forums; unfortunately I cannot help too much for igpu, never passed through to any vm.

[Mac OS Catalina VM on unRAID 6.10.3]

On 9/8/2022 at 10:35 PM, Zippi said:

I dumped the VBIOS ROM with Techpowerup and it made me a 132Kb file. I didn't use the one made by Spaceinvaders One because it generated me a file of only 70Kb which is probably not good.

It is always recommended to dump your own bios file: the 70 kb file could be good, I would check it with an hex editor and see what it's inside.

If the dumped file is crap, you should search and solve why you are not able to dump the vbios: this is recommended, because if the gpu is in use by the host you wont be able to use it even with a proper vbios.

 

On 9/8/2022 at 10:35 PM, Zippi said:

Since I am a newbie I ask you for help: should the GPU work like this or do you have to tweak some parameters like when building a baremetal Hackintosh?

It works the same way as a bare metal hackintosh, you need lilu/whatevergreen kexts and all the workarounds (if any) needed for bare metal hackintosh.

The only thing that changes compared to bare metal is that you need also to set the topology of the vm: make sure to pass both video and audio of the gpu, set them as a multifunction device (same bus, same slot, different function) and do not use bus 0.

 

On 9/8/2022 at 10:35 PM, Zippi said:

GPU video e audio are together so no vfio, right?

It is recommended to attach them to vfio; not sure what you are asking: you need to passthrough both devices (audio and video, so 2 checkmarks), if they are alone in the same iommu group it's ok, if they are together with some other device use acs patch to split them.

 

On top of this make sure the vbios of the gpu is uefi, otherwise nothing will work.

[[6.10.3] Unraid windows VM SLOW after update]

I would check if there is some irq conflict with the msi utility, in particular check for shared irq with the virtual sata controller:

https://forums.unraid.net/topic/76035-help-struggling-with-nvidia-audio-passthrough/#comment-1076592

 

On 9/10/2022 at 2:33 PM, BoKKeR said:

updated to i440fx-6.2

This may change irq arrangement, alternatively try to restore the older version

[GPU acceleration in Monterey VM]

Monterey has no drivers for 1070, so there's no way to get acceleration or any video output from that gpu; stick to high sierra (with installed nvidia drivers), change gpu with a compatible one, or do not use gpu passthrough.

[Grafigkarte NVIDIA 1080 TI]

Monitor is attached to the gtx, but the gpu is bound to vfio, so it's perfectly normal that at some time during unraid boot and after some video output the screen will look like it's frozen. The gpu is isolated and the os cannot use it for its video output anymore, because you set it to be reserved for something else (vm for example).

Connect to unraid from another device and you will find that it's booting and it's not crashing.

[Way to manually remove or delete VMs?]

On 8/10/2022 at 4:50 PM, dlfz.Network said:

Requested operation is not valid: cannot undefine domain with nvram ! Not valid!

Type this in unraid terminal:

virsh undefine --nvram "Macinabox BigSur"

 

[hard disk usb installazione]

Ciao, credo che unassigned devices possa esserti di aiuto (sempre che tu non voglia assegnarli all'array, ma non credo proprio..). Lo scarichi da community applications.

Qui la discussione specifica:

 

[Hibernating (and then UN-hibernating) Ubuntu?]

And this?

https://askubuntu.com/questions/6769/hibernate-and-resume-from-a-swap-file

 

First reply may contain some useful info.

[My qcow2 VM image for Home Assistant seems corrupted, thoughts?]

data should be in the larger partition, nbd0p8, try to mount it somewhere.

[My qcow2 VM image for Home Assistant seems corrupted, thoughts?]

1 hour ago, RyanServer711 said:

fdisk -1

it's -l not -1

[My qcow2 VM image for Home Assistant seems corrupted, thoughts?]

is something using the qcow2 file?is the vm completely shutdown?

If so, try to remove the "dirty bit".

[[Support] SpaceinvaderOne - Macinabox]

1 hour ago, JamesJB said:

then I let it download which took me like 46min.

the recovery needs to download more or less 12 gb, unless you have a fast connection it sounds like you have a corrupted installer and that's probably why it reboots to the bootscreen. I would start from scratch again.

[Windows PC drive images]

You are the only one that knows what your hard drives contain, so make a copy of what you need.

Or if the quesrion is if you can clone several drives into one image, as far as I know...no. As far as the vm, to run it only needs the os partion image(s) (and the efi partition (if efi)). Usually you clone the whole drive containing all the os partitions.

If I were you I would attach the os hard drive in the server and pass it to the vm directly, or if you want an image file for the disk, I would attach the os hard drive in the server and use linux dd command to clone it as a raw image; or another alternative is to boot your old pc from a linux live usb pedrive/cd/dvd and use dd command to clone the hd as a raw image and then transfer it somewhere in the server.

Take into account that you need space to save image(s), on a disk different than that you are cloning, at least a free space equal to the whole disk capacity you will clone.

I would avoid any virtio device and use legacy ones when you create the vm.

[My qcow2 VM image for Home Assistant seems corrupted, thoughts?]

You can use qemu-nbd command line.

First backup your qcow2 file.

Then in a terminal:

Load kernel module:

modprobe nbd max_part=8

Create nbd0 block device:

qemu-nbd --connect=/dev/nbd0 /replace/with/path/to/file.qcow2

Use fdisk to list partitions and see if you get any output:

fdisk -l /dev/nbd0

Yiu can also use fsck to try to repair partitions:

fsck /dev/nbd0p1

nbd0p1 being partition 1.

 

You can also mount the partition (in this example nbd0p1) somewhere:

mount /dev/nbd0p1 /mnt/somepoint/

 

Once done, disconnect:

qemu-nbd --disconnect /dev/nbd0

 

[Virtual machine networking help]

11 hours ago, Mongoose0318 said:

If I open up a browser on the virtual machine itself and type in the ip address I can access the web gui just fine. 

sounds like you have a firewall installed into the vm which prevents other hosts on the same lan to access that port; check iptables or any other firewall rules.

Or you may have a firewall in the router.

[[Support] SpaceinvaderOne - Macinabox]

7 hours ago, JamesJB said:

Not sure what I am doing wrong Please help.

follow the tutorial, your installation is not completed.

mac os base system is the recovery (first boot), you need to select mac os installer to complete the installation, or better, let the bootpicker timeout without selecting anything and it will automatically select and boot the proper partition to complete the installation, all is managed by nvram entries.