Linguafoeda

Hey guys - i have been experiencing a lot of instability on my WIn11 VM which i was using gvt-g to accelerate (which drastically improved Win11 VM performance, it felt ALMOST like a local pc via Windows RDP)....I documented some of the things i just went through in the below thread. I was happily using GVT-g for years until I started getting drastically bad instability issues, where WIn11 would eventually crash, then Unraid WebUI would lock up (takes 2-3 minutes just to load the dashboard, would not be able to "restart" the libvirt process) and furthermore, would not restart properly via GUI requiring me to hard pushdown the power button to restart my machine. After a lot of troubleshooting, ChatGPT thinks GVT-g is the reason for my instability. I wrote a few posts in the below thread if anyone can help me solve how to keep INtel GPU acceleration but fix my Win11 VM stability issues. I went down a rabbithole of recreating my libvirt.img but it seems libvirt is not the issue for now. I'm on 7.1.4 and my PC specs are: i7-10700T / 32GB DDR4 / Z490M-ITX. Thank you so so much for your help...

FWIW - chatGPT believes the root cause of my issues is related to Intel GVT-g acceleration that caused the original crash (just using Chrome) to begin with, that then put me in this loop of trying to recover my unstable machine post-libvirt crash. I'm not sure if ChatGPT is guiding me wrong but that's what it surmises....i'm on 7.1.4 and i don't remember specifically when i started having these issues but do rememeber Win11 VM could run for months very stable when i was on 6.10 last year. I do have plex also running with parameters: --device=/dev/dri --memory=12G if that helps.

Okay was playing around for 30 minutes trying to "unfreeze" my GUI since i could still SSH into the terminal. I wanted to share a detailed summary of the issues I encountered and the troubleshooting steps I took after my Unraid server “froze” during a Windows 11 VM run with Intel GVT-g GPU virtualization enabled. Timeline & Key Points:The Unraid server became unresponsive (“frozen”) when attempting to start the Windows 11 VM using Intel GVT mediated devices. The libvirt daemon became stuck, and I found a defunct (<defunct>) QEMU process lingering, which was not cleared by a normal stop. I killed defunct QEMU processes manually with kill -9, then stopped all related libvirt services (libvirtd, virtlockd, virtlogd) forcibly ("killall -9 libvirtd virtlockd virtlogd"). After that, I restarted libvirt services cleanly via /etc/rc.d/rc.libvirt stop and /etc/rc.d/rc.libvirt start, which restored access to the VM tab. When trying to start the VM, I got errors in the libvirt logs about the disk image being locked, which was resolved after confirming no other process was holding it. However, upon assigning the Intel GVT mediated device to the VM, I hit a kernel error repeatedly: gvt: fail to allocate resource low GM space gvt: request 128MB avail 0MB max 128MB taken 128MB gvt: failed to create intel vgpu: -28 Checking /sys/bus/mdev/devices/ showed no active mediated devices. modprobe intel_gvt failed because the kernel module does not exist in my current Unraid kernel (6.12.24-Unraid). No GVT modules are currently loaded, and mediated devices are not created. The problem appears to be that the Intel GVT kernel module (intel_gvt / intel_vgpu_mdev) is missing from my Unraid kernel build, preventing creation and proper cleanup of vGPU devices. This causes leftover UUID references and inability to allocate vGPU memory for new VM sessions. I am currently unable to reboot the server due to physical access limitations currently. What I’ve Tried:Manually killing stale defunct QEMU processes. Restarting libvirt and associated services multiple times. Verified disk image locks with lsof and ensured no conflicts. Confirmed that no intel_gvt or related kernel modules are loaded or available. Checked kernel logs confirming the “fail to allocate resource low GM space” error and GVT memory exhaustion. Current Status & Questions:Without the proper Intel GVT kernel module, I cannot create or assign mediated devices. Is there a recommended way to manually clean up stale vGPU UUIDs or free VRAM without rebooting? Any advice on handling this issue remotely without physical reboot would be appreciated. Thanks in advance for any guidance or suggestions. I will eventually be able to physically reboot my machine in ~12-24 hours but at least Libvirt is running....even with errors as i can't restart the VM right now: Libvirt Log: 2025-07-11 05:13:58.815+0000: 271491: warning : virNetDaemonRun:880 : Make forcefull daemon shutdown 2025-07-11 05:22:33.057+0000: 730004: info : libvirt version: 11.2.0 2025-07-11 05:22:33.057+0000: 730004: info : hostname: XXXX 2025-07-11 05:22:33.057+0000: 730004: error : qemuRdpNewForHelper:103 : 'qemu-rdp' is not a suitable qemu-rdp helper name: No such file or directory 2025-07-11 05:22:35.098+0000: 730004: warning : qemuTPMVirCommandSwtpmAddTPMState:634 : This swtpm version doesn't support explicit locking 2025-07-11 05:22:37.135+0000: 730004: warning : qemuDomainObjTaintMsg:5481 : Domain id=1 name='Windows' uuid=84939d5e-3784-2728-0510-XXXX is tainted: high-privileges 2025-07-11 05:22:38.026+0000: 735140: error : qemuMonitorIORead:397 : Unable to read from monitor: Connection reset by peer 2025-07-11 05:22:38.027+0000: 735140: error : qemuProcessReportLogError:1953 : internal error: QEMU unexpectedly closed the monitor (vm='Windows'): 2025-07-11T05:22:37.961624Z qemu-system-x86_64: -blockdev {"node-name":"libvirt-1-format","read-only":false,"cache":{"direct":false,"no-flush":false},"driver":"qcow2","file":"libvirt-1-storage","backing":null}: Failed to get "write" lock Is another process using the image [/mnt/cache/domains/Windows/Windows_disk.img]? 2025-07-11 05:22:38.027+0000: 730004: error : qemuProcessReportLogError:1953 : internal error: process exited while connecting to monitor: 2025-07-11T05:22:37.961624Z qemu-system-x86_64: -blockdev {"node-name":"libvirt-1-format","read-only":false,"cache":{"direct":false,"no-flush":false},"driver":"qcow2","file":"libvirt-1-storage","backing":null}: Failed to get "write" lock Is another process using the image [/mnt/cache/domains/Windows/Windows_disk.img]? 2025-07-11 05:31:18.938+0000: 730000: error : virNetSocketReadWire:1782 : End of file while reading data: Input/output error 2025-07-11 05:36:31.204+0000: 760692: info : libvirt version: 11.2.0 2025-07-11 05:36:31.204+0000: 760692: info : hostname: XXXX 2025-07-11 05:36:31.204+0000: 760692: error : networkCheckRouteCollision:298 : internal error: Network is already in use by interface virbr0 2025-07-11 05:37:52.459+0000: 760676: error : qemuRdpNewForHelper:103 : 'qemu-rdp' is not a suitable qemu-rdp helper name: No such file or directory 2025-07-11 05:37:54.487+0000: 760676: error : virMediatedDeviceNew:144 : device not found: mediated device 'e3e02122-f09d-4809-adb8-XXXX' not found 2025-07-11 05:39:27.497+0000: 760678: error : qemuRdpNewForHelper:103 : 'qemu-rdp' is not a suitable qemu-rdp helper name: No such file or directory 2025-07-11 05:39:42.903+0000: 760675: error : qemuRdpNewForHelper:103 : 'qemu-rdp' is not a suitable qemu-rdp helper name: No such file or directory 2025-07-11 05:39:44.942+0000: 760675: error : virMediatedDeviceNew:144 : device not found: mediated device 'e3e02122-f09d-4809-adb8-XXXX' not found 2025-07-11 05:40:01.164+0000: 760680: error : qemuRdpNewForHelper:103 : 'qemu-rdp' is not a suitable qemu-rdp helper name: No such file or directory 2025-07-11 05:40:08.842+0000: 760679: error : qemuRdpNewForHelper:103 : 'qemu-rdp' is not a suitable qemu-rdp helper name: No such file or directory 2025-07-11 05:40:14.668+0000: 760678: error : qemuRdpNewForHelper:103 : 'qemu-rdp' is not a suitable qemu-rdp helper name: No such file or directory 2025-07-11 05:40:15.686+0000: 760723: error : udevProcessMediatedDevice:1146 : failed to wait for file '/sys/devices/pci0000:00/0000:00:02.0/fc6dbcf6-dc59-4143-95f3-XXXX/mdev_type' to appear: No such file or directory 2025-07-11 05:40:16.692+0000: 760678: error : virMediatedDeviceNew:144 : device not found: mediated device 'fc6dbcf6-dc59-4143-95f3-XXXX' not found 2025-07-11 05:43:19.114+0000: 760675: error : qemuRdpNewForHelper:103 : 'qemu-rdp' is not a suitable qemu-rdp helper name: No such file or directory 2025-07-11 05:43:26.216+0000: 760683: error : qemuRdpNewForHelper:103 : 'qemu-rdp' is not a suitable qemu-rdp helper name: No such file or directory 2025-07-11 05:45:42.828+0000: 760679: error : qemuRdpNewForHelper:103 : 'qemu-rdp' is not a suitable qemu-rdp helper name: No such file or directory 2025-07-11 05:45:43.850+0000: 760723: error : udevProcessMediatedDevice:1146 : failed to wait for file '/sys/devices/pci0000:00/0000:00:02.0/bf29780a-4561-4394-88a0-XXXX/mdev_type' to appear: No such file or directory 2025-07-11 05:45:44.859+0000: 760679: error : virMediatedDeviceNew:144 : device not found: mediated device 'bf29780a-4561-4394-88a0-XXXX' not found

ugh - i was moving tabs around in chrome and my windows 11 VM crashed again, and same issue as original....main unraid GUI is now extremely unstable (will not load the VM tab, the main dashboard takes 2-3 minutes to load the docker containers and my VM section is now MISSING the windows VM from the list....). I wont be able to restart the Unraid machine without a physical hard reset of the button (which is difficult because im not physically located next to my server). So re-creating libvirt did nothing for me. what do i do now to troubleshoot this....is this is a windows 11 VM issue itself that is causing some massive failure that doesn't allow libvirt to work again after it crashes and now requires a physical hard reset? Note - i have 32GB RAM and Core i7-10700T and have otherwise before never had issues with allocating 8GB RAM / all cores/threads to my Win11 VM to function with much more intensive programs running (vs. chrome with 50 or so tabs open when it just now crashed) Libvirt Log: 2025-07-11 05:04:21.655+0000: 271495: warning : virDomainObjBeginJobInternal:434 : Cannot start job (query, none, none) in API remoteDispatchDomainGetXMLDesc for domain Windows; current job is (query, none, none) owned by (271492 remoteDispatchConnectGetAllDomainStats, 0 <null>, 0 <null> (flags=0x0)) for (165s, 0s, 0s) 2025-07-11 05:04:21.655+0000: 271495: error : virDomainObjBeginJobInternal:468 : Timed out during operation: cannot acquire state change lock (held by monitor=remoteDispatchConnectGetAllDomainStats) 2025-07-11 05:04:51.656+0000: 271494: warning : virDomainObjBeginJobInternal:434 : Cannot start job (query, none, none) in API remoteDispatchDomainGetXMLDesc for domain Windows; current job is (query, none, none) owned by (271492 remoteDispatchConnectGetAllDomainStats, 0 <null>, 0 <null> (flags=0x0)) for (195s, 0s, 0s) 2025-07-11 05:04:51.656+0000: 271494: error : virDomainObjBeginJobInternal:468 : Timed out during operation: cannot acquire state change lock (held by monitor=remoteDispatchConnectGetAllDomainStats) 2025-07-11 05:05:21.656+0000: 271493: warning : virDomainObjBeginJobInternal:434 : Cannot start job (query, none, none) in API remoteDispatchDomainGetXMLDesc for domain Windows; current job is (query, none, none) owned by (271492 remoteDispatchConnectGetAllDomainStats, 0 <null>, 0 <null> (flags=0x0)) for (225s, 0s, 0s) 2025-07-11 05:05:21.656+0000: 271493: error : virDomainObjBeginJobInternal:468 : Timed out during operation: cannot acquire state change lock (held by monitor=remoteDispatchConnectGetAllDomainStats) 2025-07-11 05:05:51.656+0000: 271496: warning : virDomainObjBeginJobInternal:434 : Cannot start job (query, none, none) in API remoteDispatchDomainGetXMLDesc for domain Windows; current job is (query, none, none) owned by (271492 remoteDispatchConnectGetAllDomainStats, 0 <null>, 0 <null> (flags=0x0)) for (255s, 0s, 0s) 2025-07-11 05:05:51.656+0000: 271496: error : virDomainObjBeginJobInternal:468 : Timed out during operation: cannot acquire state change lock (held by monitor=remoteDispatchConnectGetAllDomainStats) 2025-07-11 05:06:21.656+0000: 271495: warning : virDomainObjBeginJobInternal:434 : Cannot start job (query, none, none) in API remoteDispatchDomainGetXMLDesc for domain Windows; current job is (query, none, none) owned by (271492 remoteDispatchConnectGetAllDomainStats, 0 <null>, 0 <null> (flags=0x0)) for (285s, 0s, 0s) 2025-07-11 05:06:21.656+0000: 271495: error : virDomainObjBeginJobInternal:468 : Timed out during operation: cannot acquire state change lock (held by monitor=remoteDispatchConnectGetAllDomainStats) 2025-07-11 05:06:55.328+0000: 271494: warning : virDomainObjBeginJobInternal:434 : Cannot start job (query, none, none) in API remoteDispatchDomainGetXMLDesc for domain Windows; current job is (query, none, none) owned by (271492 remoteDispatchConnectGetAllDomainStats, 0 <null>, 0 <null> (flags=0x0)) for (319s, 0s, 0s) 2025-07-11 05:06:55.328+0000: 271494: error : virDomainObjBeginJobInternal:468 : Timed out during operation: cannot acquire state change lock (held by monitor=remoteDispatchConnectGetAllDomainStats) 2025-07-11 05:06:55.499+0000: 271496: warning : virDomainObjBeginJobInternal:434 : Cannot start job (query, none, none) in API remoteDispatchDomainGetXMLDesc for domain Windows; current job is (query, none, none) owned by (271492 remoteDispatchConnectGetAllDomainStats, 0 <null>, 0 <null> (flags=0x0)) for (319s, 0s, 0s) 2025-07-11 05:06:55.499+0000: 271496: error : virDomainObjBeginJobInternal:468 : Timed out during operation: cannot acquire state change lock (held by monitor=remoteDispatchConnectGetAllDomainStats)

Hey guys - thanks so much for the help. So i attempted to follow both of your steps of just trying to create a new TPM / "fresh" re-install using my old XML/vdisk location, instead of backing up and restoring the backed up .fd file from /etc/libvirt/qemu/nvram; As soon as i restored the XML, clicked save, then re-went to edit VM -> form view -> under BIOS, I switched to "OVMF", clicked save, then re-went back to form a 3rd time, and changed it back to OVMF TPM and clicked save again, I fixed the intel GVT-g issue (step 9 in this post) and then attempted to launch but got stuck in the BIOS with the error similar to the below: Do you know what i did wrong that caused this? I would like to know how to fix this TPM issue to fresh install in future, and i'm sure maybe someone similar to me runs into a similar issue trying to re-create TPM that might not have backed up their .fd files. For now - i just restore the .fd file and it's working, not sure if that will cause any instability but seems okay so far for last 1 hour running the VM again.

okay i got this working and was able to recreate the libvirt image. below were the steps: backup my libvirt.img, Windows vdisk.img and /etc/libvirt/qemu/nvram (had two .fd files, one for win11 VM with UUID matching the WIn11 XML and one for my macOS VM) Copy and paste the XML into notepad from my existing VMs Turn off VM in settings -> VM Manager Use Krusader or MC to delete libvirt.img in mnt/user/system/libvirt Turn on VM in settings -> VM Manager Copy the two .fd files from step 1 back to /etc/libvirt/qemu/nvram Go to VM tab -> add VM -> WIndows 11 (so i get proper logo) -> form view switch on top right -> paste the XML for my previous Win11 VM saved in step 2 At this stage, if i try to launch the VM, it gives me an error "device not found: mediated device 'XXXX-XXXX-XXXX-XXXX-XXXX' not found" which relates to the Intel GPU acceleration Go to settings -> intel gvt-g plugin settings, delete the existing assignment in the bottom drop-down then in the middle dropdown ("To create another vGPU assignment for 0000:00:02.0 select mode and VM:"), I selected "i915-GVTg_V5_4" and assigned it to the Windows 11 VM i added in step 7 Start VM like normal (i use "Remote Desktop Connection" and "Anydesk" to connect) note my libvirt log still have a few errors below but i'll monitor my VM over the next few days/weeks to see if it causes a failure / crash like original post 2025-07-11 03:53:47.801+0000: 271492: warning : qemuTPMVirCommandSwtpmAddTPMState:634 : This swtpm version doesn't support explicit locking 2025-07-11 03:53:49.832+0000: 271492: warning : qemuDomainObjTaintMsg:5481 : Domain id=5 name='Windows' uuid=XXXX-XXXX-XXXX-XXXX-XXXX is tainted: high-privileges 2025-07-11 04:00:59.367+0000: 271494: error : qemuRdpNewForHelper:103 : 'qemu-rdp' is not a suitable qemu-rdp helper name: No such file or directory

hmm i see. so if i don't backup the .fd files living in the /etc folder, i can just recreate the TPM and all i would have to do is reset my pin? can i store the tpm-related files in the ISO folder where the vdisk lives to keep it simple in the future for backing up?

Docker storage driver: native I beleive i'm using just the native one in the docker settings tab. this has been an issue since i was on v6 too, but it takes like 30 mins for me to wait until the noVNC can even access krusader.

I don't believe so, how would I know if I'm using that? Is it something I unintentionally turned on?

any idea what could be causing the lag on my end where container doesn't launch until 30mins+ in?

Don't believe i use bitlocker, though when i attempted to start the libvirt i got the above error message? My VM actually just crashed while using it to import music into itunes, causing unraid to almost completely freeze. below is the gist: Gistgist:c38c104eb9da46f62a8d5cab73f90219GitHub Gist: instantly share code, notes, and snippets.

Hey @ich777 I'm having trouble with my Win11 VM using TPM. see my thread below. is there a better way to setup my WIn11 VM (i'm trying to reset my libvirt.img due to crashing) so that i don't run into unncessarily complex backups of .fd files in etc/ folder when my VM goes caput next time? i'm running 7.1.1, wondering if the new unraid after 6.10 built in the TPM type security needed to run windows updates.

Hi - just wanted to update on this. i finally got around to attempting to recreate my libvirt and came across a whole host of issues. I initially had created a OVMF TPM bios for my windows so i could update to 2H22...when i recreated my libvirt.img and pasted the backed up XML....I think this is related to the .fd files in /etc/libvirt/qemu/nvram/, because i got error message: operation failed: unable to find any master var store for loader: /usr/share/qemu/ovmf-x64/OVMF_CODE-pure-efi-tpm.fd Then i had some sort of device issue which i believe is related to intel gvt-g plugin i'm using to accelerate the VM? device not found: mediated device '5b1865eb-1ab5-4e5a-xxxx-be46c922xxxx' not found At this point, i reverted back because i think i'm whack a moling a bunch of issues. any idea what else i should be aware of since it seems not so simple to just delete libvirt and re-paste XML. One question i have - if i don't have the backup .fd files for OVMF TPM, is it possible to easily recreate TPM files to use my old Win11 vdisk?

anyone have an update for this?

still having same issue 2025-07-05 19:57:47,518 DEBG 'start-script' stdout output: [warn] Failed to bring 'up' WireGuard kernel implementation

https://bbs.archlinux.org/viewtopic.php?pid=2249137#p2249137 should we implement this fix suggested in the above thread or wait for binhex-qbittorrent to get fixed in a new release? Can confirm the same issue, the funny thing is that in my case they are complaining just because of the header as in both cases resolv.conf points to 127.0.0.1 but sometimes is generated by Network manager and others by resolvconf depending on which tool run last. # Generated by NetworkManager nameserver 127.0.0.1# Generated by resolvconf nameserver 127.0.0.1#Solution Add this file in conf.d of networkmanager cat /etc/NetworkManager/conf.d/rc-manager.conf [main] rc-manager=resolvconfRegenerate the config with resolvconf -u Reboot the system

also having this same wireguard kernel issue, Should i just roll back (binhex/arch-qbittorrentvpn:5.1.1-1-01) or is there a fix to implement? 2025-07-01 00:22:34,901 DEBG 'start-script' stdout output: [warn] Failed to bring 'up' WireGuard kernel implementation

crisis averted - it's likely a CGNAT issue....ISP switched up on me for some reason.

I just restarted the server and wireguard still isn't connecting, so i guess i can start with that. What could possibly be causing wireguard to malfunction? any tips to start troubleshooting from the start?

Recently I keep losing remote access to my server, emby is going down and Wireguard is failing when out of the home. I can only connect when on local lan (wifi), it feels like either my unRAID install itself is failing or wireguard is failing. This follows recent issues where my VM has started to crash (libvert failed to start) causing entire system to hang when I try reboot. And a few weeks ago I lost access to first remote access via wireguard, then completely lost access to the unRAID home LAN address, which then required a physical hard reset. If I restart my server right now, wireguard will start working again but then I lose it after a few days of uptime. I did change my wifi router recently to eero from Google wifi, but I triple checked all ports / DHCP reservations were exactly copied over (which explains why things work initially). I'm hardwired to one of the eero routers which connects back via a 24 port switch to the main eero router. Any idea how to troubleshoot this? Is there a way to to check if the flash USB is failing? Was curious if that might be the issue because of the coincidence of VM failure (libvert failed to start), etc.

wanted to bump this re- the questions above?

bump

still having this major issue with VM crashing eventually and causing entire system to hang, thus not able to restart and have to manually power cycle. i haven't done anything to my libvirt image or replaced my VM .xmls. I have a few questions: 1. Is there any risk to deleting the libvirt image file? 2. is literally copy and pasting the xml file into word then re-pasting it back in after the libvirt delete keep things exactly as they were, no extra steps needed? 3. is there any way to troubleshoot the Unraid USB to know if the flash drive is causing issues? @ELP1 were you able to ever fix your similar issue to mine?

What's the proper step to back up my VM xml, just copy and paste?

I just went into settings, stopped VM then tried to turn it on again and now am back to square 1 of the error message "Libvirt Service failed to start." this person above is also seemingly having same issue and claiming he has to do a fresh install of unraid to fix it (has tried to recreate his libvirt image 3 times and failed). I'd really like to avoid a fresh unraid install if possible. I haven't done anything to touch my libvirt file - i'm assuming that will be the next step? and figure out why my server is not properly restarting requiring a physical power off and power on for me to restart the server.