stevehopkinson

Happy new year! I've recently migrated my Unraid setup to new hardware, but I've had issues getting my existing Unraid configuration to boot cleanly and consistently. I've had no success fixing the existing config, but the new hardware has no issues booting into a clean install of Unraid, so my plan is to create a clean configuration and then only copy over the minimum required changes from my existing configuration in order to migrate the shares (pool and array) and key Docker containers. What process would I need to follow to achieve this? All I want to maintain is the shares, a few Docker containers, and the data for those containers. Are there any potential issues/gotchas with this approach that I should be aware of? Thanks in advance!

I've continued debugging for a couple of days but I still can't get Unraid to boot reliably. Half the time it hangs; half the time it boots successfully. I can't see any obvious variable that leads it to go one way or the other. @trurl - is there anything in the diagnostics that I'm missing? Until I can get it booting successfully 100% of the time on the new hardware I won't be migrating – tearing my hair out trying to understand what's going on!

I booted into safe mode and set Unraid to bind my GPU's IOMMU group to Unraid at boot. When it prompted me to restart so that changes to system devices could take effect, I did, and then it restarted into full-fat Unraid - successfully! I have no idea whether the VFIO binding actually changed anything, but this was my first successful boot without safe mode. I tried to do a clean boot into full-fat Unraid with the new config, but it hung at the same place again. After another restart, it managed to boot successfully. There seems to be a variable I can't pin down, which is causing it to fail some of the time but boot successfully the rest of the time. I've attached the diagnostics from a successful boot into full-fat Unraid, in case they shed any light on things. steeju-diagnostics-20221219-1928.zip

@trurl - not booting in GUI mode (I only ever use the web interface).

@trurl - I've booted the system in safe mode and downloaded the diagnostics. I've attached them to this post. @Hoopster - both the old and new hardware are in UEFI mode, so I don't believe that should be an issue. But thanks - I've had that problem before so I know what a pain it can be! I've also noticed that even when I boot in safe mode, the system will hang if I have a display connected to the discrete GPU. I usually keep a dummy plug in the GPU for running it headlessly, but if I tried to boot with the dummy plug connected the system hanged on 'Loading /bzroot… OK' again. On the old hardware I had to do some pretty funky tweaks to get the iGPU passed through to a VM, so it's possible there's still some hardware-specific config left over that's causing an issue with the new hardware… Edit: The GPU display may be a red herring. It seems to intermittently hang/not hang even in safe mode, regardless of whether or not the dummy plug is entered. steeju-diagnostics-20221219-1750.zip

Hi there, I've been running Unraid flawlessly on an HP EliteDesk Mini for a couple of years. I'm now trying to migrate to new hardware, but for some reason the new server only boots in safe mode. When I try to boot into full-fat Unraid, it hangs at 'Loading /bzroot... OK' and never progresses further. I've tried booting it from a newly-created Unraid USB and the system boots fine. I suspect there's some config that causes the new hardware to hang when it's loaded, but which isn't executed in safe mode. Is there any way to persist the boot logs to the flash drive so I can review them for any possible issues? Would there be any value in sharing diagnostic files from the system after booting in safe mode? Thanks in advance! Note: I'm booting without any of the pool/array discs connected. Could that cause any issues? I would have assumed Unraid only looks for those discs after booting successfully. Edit 1: Both the new and old systems are booting in UEFI mode. Edit 2: It also intermittently hangs at the same stage, even in safe mode. I thought this might have been related to booting with a display attached to the discrete GPU, but that may have been a red herring. Edit 3: I've exported the diagnostics after booting in safe mode - they're attached to this post now. steeju-diagnostics-20221219-1750.zip

Embarrassingly I've fixed it - it's was the classic MSI priority issue for passed-through devices. I ran the MSI util and the audio hardware on the GPU was unchecked; I checked it, applied and rebooted and now everything runs without a hitch (literally). Feel free to lock this thread!

Hi all – I've been building Unraid 'gaming servers' with GPUs passed through to VMs for a few years, but I'm having an issue with my latest build that's got me stumped. I recently finished a build with a 10700 and an RTX3060, and on the whole performance is as expected - but when playing games or other GPU-intensive applications, the frame rate will stutter every six seconds like clockwork. Everything runs fine, then no new frames get rendered for 2-3 tenths of a second, then everything runs fine again until the next stutter. You can see this clearly on the 3DMark results posted below: in the two GPU tests, the framerate drops every six seconds like clockwork. Any thoughts on what the issue might be? I've tried all my usual tricks but nothing's working. I've attached my diagnostics in case that helps. Thanks! smolboi-diagnostics-20220824-1535.zip

Yes, this was all using the latest Nvidia drivers (496.49). I'm not sure which diagnostics file to attach, as I've run in many configuration variations and am having the same issue with a wide range of setups. The key limitation seems to be that there's no combination of values that will get the passthrough working correctly when boot mode is UEFI.

Context: my partner and I are planning to go travelling for a year, and we want to be able to take a single tiny computer with us that can be both a portable server and a gaming rig when we need it. The NUC 11 Enthusiast seemed perfect for that purpose - it's the size of a paperback, has a punchy 4-core/8-thread processor, an Intel XE iGPU that can handle hardware transcoding for Plex duties, and a dedicated RTX2060 to pass through to a Windows VM that we can spin up for gaming. I've had a couple of PCs set up this way in the past, and I've always been able to resolve any issues that came up. Getting everything set up on the NUC was surprisingly easy, but when it came to passing through the RTX2060 to a Windows 10 VM we ran into the dreaded Code 43 issue. We've tried absolutely every solution out there (ripping VBIOS, removing headers, every BIOS/machine type combination, allowing unsafe interrupts, setting PCIe ACS override to 'both', ensuring all the card's functions are in the same IOMMU group, stubbing the USB/serial controllers, setting multifunction='on', installing Nvidia drivers while the card is disabled, etc). It's reached the stage where the only thing I haven't tried is setting the boot mode to Legacy, which seems to resolve the issue for most users. Unfortunately the 11th-gen NUCs are the first Intel systems to have removed support for legacy booting entirely. It is only possible to boot in UEFI mode. And as far as I can tell, there's no bulletproof solution to fixing Code 43 errors when Unraid is booting in UEFI mode. Every guide I read ends with '…and then I set boot mode to Legacy and everything worked'. But that's not an option for us, and it probably won't be an option for a lot of users as manufactures continue to end support for legacy booting. Are there any tried-and-tested solutions for Nvidia GPU passthrough that don't require users to switch to legacy boot mode? Any thoughts would be greatly appreciated. Thanks!