Jump to content


  • Content Count

  • Joined

  • Last visited

Everything posted by bastl

  1. I`ve asked because the german implementation never worked for me, no matter if I used the buildin webVNC client or any other like tightVNC or realVNC. None of the clients I was able to use any german key combination.
  2. If you have no access to the ISP router a way you can work around this is to setup a server somewhere in the cloud or somwhere where you have full access to it and have a VPN setup to the pfsense box. From this server you can tunnel the ports you wanna access to pfsense and port forward them to unraid.
  3. Please don't put a Unraid server into a DMZ or any server in your network, unless you really know what you're doin.
  4. What if the "domain" share isn't on the encrypted array? @SpaceInvaderOne Most users have their domain share on the cache drive which I guess for most isn't encrypted.
  5. Support for Ryzen 3000 no one really knows yet, what will change and what kernel tweaks are needed with this new chips. Maybe it works right out of the box, maybe not. Always keep in mind the TR4 platform has way more PCIe lanes available and you can directly connect more devices to the CPU. Sharing the lanes between the slots you always end up in a situation, where the devices are grouped together. For passthrough this isn't what you want. Each manufacturer is slightly different how they handle the groupings in there BIOSes and with each revision it can change. The final versions aren't out yet. On most x399 if they have multiple Nics, they are grouped together. On some boards you can split them with ACS override, on some this isn't possible. This also depends on the BIOS revs you're running. If you can, avoid the ACS patch and use a dedicated ethernet card. On TR4 no problem, you have enough slots and lanes 😁
  6. Is your Windows activated correctly? Black screen after 90 minutes, just sayin 😂
  7. Sata isn't the best option. It has more overhead and latency as using virtio or scsi. If you switch it, you need the drivers installed on Windows first or it won't boot. Add a small dummy hdd (1G) as scsi first, mount the virtio driver iso, boot up your VM and install the driver (vioscsi/win10/amd64). You can remove that disk after that. Change your xml like in the following example. <disk type='block' device='disk'> <driver name='qemu' type='raw' cache='none' discard='unmap'/> <source dev='/dev/disk/by-id/ata-CT500MX500SSD1_1807E10EA5F2'/> <target dev='hdc' bus='scsi'/> <boot order='1'/> <address type='drive' controller='0' bus='0' target='0' unit='2'/> </disk> You should see better performance with this. The hosts RAM isn't used for caching with this and trim operations performed on the guest will be passed through to the device itself which isn't possible with sata.
  8. Enable ACS override
  9. I would be really careful what you're doin. Sure, we all know you will not deal any harm to that person and this is all in his/her interest, but changing files on that persons pc in lot of countries without his permission is against the law. Just sayin.
  10. @whatdoesthisbuttondo Couple people reported using the newer Q35 as machine type fixed it for them. Newer driver revisions worked with this and I think how you boot unraid also made a difference. Switching between Legacy and UEFI boot, can't remeber which one is working. You can find your current status under Main if you click the Unraid flash device.
  11. @testdasi I guess you also tried the ACS multi-function option? Maybe?
  12. I guess it depends on the motherboard manufacturer how they configure the IOMMU groups in the BIOS. On earlier BIOS revisions on my ASRock I had the same issue like you. Onboard sound was grouped with other devices (Sata, Network, USB). ACS Override on the first BIOS doesn't helped, on the second version I could split it into it's own group and with later BIOS revisions the ACS Override Patch wasn't necesary any more. Maybe check if Gigabyte released a newer version for your board.
  13. On some boards populating the U.2 slot if available dissables one of the M.2 slots. Same for SATA. Some boards share sata ports with one of the NVME slots. Another thing to mentioned some manufactures share the PCIE slots with one of the M.2 slots. Best is to check your manual to see which ports are shared and what the limitions are if you populate all connections.
  14. Any chance you updated your nvidia driver last week? Nvidia released a fixed this weekend addressing an issue in some circumstances the cards stopped working and reporting Code43. Do you have access via a remote software like RDP or Teamviewer to your VM? If not, save your xml and add VNC as GPU and check if the VM boots to desktop and try to updfate the driver.
  15. I am facing a couple issues right now. I have a NVME i have passed trough to a Windows VM. So far so good, no issues with passing through the whole controller. But if I passthrought the whole controller, all partitions of that disk aren't available for other VMs anymore. My idea was to have 100GB of that NVME available for a Linux distro (dual boot) in my case PopOS. In the documentation of PopOS for dual boot they stated "select the OS you wanna boot from the boot menu". The problem is during the boot process the OVMF Bios isn't accesible with the passed through USB Keyboard and mouse. I have no chance to select which efi partition I wanna use for booting. I guess with VNC this isn't a issue, but with a GPU only passthrough at this stage there are no device drivers for the keyboard loaded. Does this change if I passthrough a whole USB controller? Will the keyboard work in this case, someone maybe tried this or has a setup and can check before I messup my complete setup? Another question: What happens if I pass through a xxxx-devive-part2 partition to a VM and install a OS on it? During the process the partition is split into a efi partition, swap, root and home partition. Did I end up in a scenario, where I have to adjust the XML later to add the partitions 3, 4, and 5 back in? And how is the performance compared to controller passed through if I choose a partition and set it to SCSI to have trim still working? My current Win10 VM is setup, if I reboot the Server and boot it straight from the NVME, all drivers are installed for a bare metal use and everthing is working. Just in case I need all the cores or I have to check the Unraid stick for errors. I only have that one box. If I have the PopOS installed on the same NVME during the boot I see the second efi partition and can select it in the motherboards bios. Same keyboard, same USB port. Maybe there is a way to have a earlier access to the passed through keyboard in a Q35 OVMF VM somehow that I'am missing. Any help or ideas are highly appreciated!
  16. @ToXIc Should be fixed with the MSI_FIX https://wiki.unraid.net/index.php/UnRAID_6/VM_Guest_Support#Enable_MSI_for_Interrupts_to_Fix_HDMI_Audio_Support
  17. First of all backup all your important data inside the VM. If there are sections of the disk with errors which are used by the VM and files which aren't used that often sitting on them you might not see a error. You can try to use chkdsk inside the VM. I never used it inside a VM but in theory Windows should find bad sections and should try to repair it, move the files to good areas of the disk. After that you could try to use a imaging software like Acronis or Clonezilla inside the VM to save the vdisk and later write back the image to a new vdisk or you directly clone it to a new attached vdisk.
  18. Not 100% sure how Unraid handles it. Maybe it depends on the template you have choosen during the VM setup or the cores you selected. In general I manual adjust it for the cores I have selected.
  19. @dnLL the "topology" section basically does this in the xml. 1 core with HT: <topology sockets='1' cores='1' threads='2'/> 2 cores without HT: <topology sockets='1' cores='2' threads='1'/>
  20. Not actual sure what you mean by that. You can define a core topology in the xml to "emulate" different kind of CPU models and core topolgies. For example you can emulate a 2 socket CPU with let's say 2 cores each or specific CPU features. Good start if you wanna dive deeper into the topic is the RedHat documentation. https://access.redhat.com/documentation/en-us/red_hat_enterprise_linux/7/html/virtualization_deployment_and_administration_guide/sect-manipulating_the_domain_xml-cpu_model_and_topology
  21. @dnLL Keep in mind, if a VM fully utilises let's say a HT core, the performance of that physical core is also affected. If you don't have such high workloads you might be fine and won't see any issues but in general it's the best idea to use the physical and the ht core at the same time in a VM. Also core 0 is always used by unraid itself. If you fully utilise this core you kinda affecting all other VMs. All the IO load, networking and storage access for example is handled by unraid and can kinda easily reduce your overall system performance in this way. 10 VMs running on 8 threads sooner or later you will have some issues.
  22. @toastman W10 is doin a lot of stuff in the backround out of the box. For example there is a update option which is set as default to provide microsoft updates for other PCs in your network and even to deliver it via p2p to other windows 10 machines via the internet. https://support.microsoft.com/en-us/help/4040215/window-10-delivery-optimization Turn this piece of crap of or try to disable your nic in the VM for a while and check if this reduces the load.
  23. As you already noticed the 590 same as the vega cards suffers of the reset bug. Nothing you really can do at this point. If you have to restart the VM, you have to reset the whole server. Another thing to mention, newer drivers for AMD GPU's showed issues where the passthrough ends in a black screen, were the windows driver or older versions from end of last year are working. For some people the GPU works in a Q35 machine type VM.