presence06 Posted April 25, 2019 Share Posted April 25, 2019 I noticed this yesterday... I have a Windows 10 VM, headless but I have a GTX 1660 passed through it. Works fine. I RDP into and check the CPU usage, normal. Once I disconnect though, one of my threads jumps to 100 and stays pegged... once I reconnect though it goes back down... I've restarted the VM, tried it again, still does it after I disconnect. What's going on? HTOP from unRaid WebUI console just shows the VM using CPU but that's it.. UnRaid 6.6.7. EDIT 0--- If I restart the VM And don't remote in, the CPU Usage stays low and "normal" levels. Quote Link to comment
NewDisplayName Posted April 25, 2019 Share Posted April 25, 2019 (edited) Please dont trust unraids cpu usage. Go to terminal and use htop (install nerd pack if you dont have) You should try the latest version of unraid. Edited April 25, 2019 by nuhll Quote Link to comment
presence06 Posted April 25, 2019 Author Share Posted April 25, 2019 If I go to the RC versions, can you upgrade to the stable when it's released without losing/redoing things? Quote Link to comment
NewDisplayName Posted April 25, 2019 Share Posted April 25, 2019 (edited) As far as i know you can switch back and forth between all versions (i only go forward tho) Nvm. I also one time was goin back. Works perfect. Back and forth. Edited April 25, 2019 by nuhll 1 Quote Link to comment
presence06 Posted April 25, 2019 Author Share Posted April 25, 2019 Updated to RC7 - RDP'd in, disconnected, still pegging 1 CPU/Thread. Quote Link to comment
NewDisplayName Posted April 25, 2019 Share Posted April 25, 2019 are you sure there is nothing running inside your vm? Try installing a new windows 10 vm and see if it has the same problem (dont take long time usually) Quote Link to comment
presence06 Posted April 25, 2019 Author Share Posted April 25, 2019 I'm getting this in the VM Log file - "qemu-system-x86_64: vfio: Cannot reset device 0000:0a:00.1, depends on group 28 which is not owned. IOMMU group 27: 0a:00.1 Audio device: NVIDIA Corporation Device 1aeb (rev a1) IOMMU group 28: 0a:00.2 USB controller: NVIDIA Corporation Quote Link to comment
NewDisplayName Posted April 25, 2019 Share Posted April 25, 2019 Did you passthru both of this? Quote Link to comment
presence06 Posted April 25, 2019 Author Share Posted April 25, 2019 I only have GPU and Audio passed through. USB Controller I don't have passing through. Quote Link to comment
presence06 Posted April 25, 2019 Author Share Posted April 25, 2019 (edited) So I edited the VM to not use Audio at all, restarted VM, RDP - logged out/disconnect and now it's stable without pegging any CPU's. EDIT - Nevermind.. it's still pegging 1 CPU/Thread at 100... wtf EDIT2 - ran Guru3D's DisplayDriverUninstaller - removed all Nvidia stuff..restarted, RDP - disconnected..no pegged CPU's.. Edited April 25, 2019 by presence06 Quote Link to comment
presence06 Posted April 26, 2019 Author Share Posted April 26, 2019 (edited) So after putsing with this... After removing the GPU drivers it doesn't peg any CPUs.. if I add the driver either via Windows Device Manager or Nvidia official drivers, remote in and then disconnect it pegs one CPU/thread.. I hooked up a monitor and keyboard and mouse, after I remote in and it pegs, If I log in with a monitor it doesn't peg... Something to do with remotely accessing and how it releases maybe? It doesn't happen with just the MS Generic drivers though.. I was running Windows 1903 and reverted back to 1809 and now it doesn't do it.. maybe it's something to do with 1903 build.. Edited April 26, 2019 by presence06 Quote Link to comment
NewDisplayName Posted April 26, 2019 Share Posted April 26, 2019 (edited) As far as i know you need to passthru all things which belongs together (Graphic, Audio, (does your graphic card has usb???)) - the error says you should also passthru the usb thing. (group 28) It might also be windows defender which scans the vm because its not used (you logged out)? Edited April 26, 2019 by nuhll Quote Link to comment
JonathanM Posted April 26, 2019 Share Posted April 26, 2019 Some video cards don't work properly without a valid output detected. You may need to use a dummy monitor plug. 1 Quote Link to comment
NewDisplayName Posted April 26, 2019 Share Posted April 26, 2019 1 hour ago, jonathanm said: Some video cards don't work properly without a valid output detected. You may need to use a dummy monitor plug. Good idea, but i thought thats more for old cards? I would frist try to passthru all of the devices of the graphiccard. 1 Quote Link to comment
presence06 Posted April 26, 2019 Author Share Posted April 26, 2019 I remoted in a few times now and disconnected, no pegged CPU - after downgrading from 1903 to 1809 build of Windows 10. I have the Nvidia drivers installed even and still no issue. I'm pretty sure it was the 1903 build of Windows (Windows Insider). Quote Link to comment
NewDisplayName Posted April 26, 2019 Share Posted April 26, 2019 (edited) How long you let the 1903 run at 100%? maybe it was doin some maintence or something? I know the latest windows update (i dont know if it was 1903 tho) did take a long time on my i9 9900k. Also did you disable (what ever the english name for it is) when it "saves" your windows instance and you can revert back to earlier states? When windows thinks it has time to do something it might take cpu to do what ever it thinks it needs to do...^^ Edited April 26, 2019 by nuhll Quote Link to comment
presence06 Posted April 26, 2019 Author Share Posted April 26, 2019 I guess I don't know how long it was running with the CPU at 100, but I just noticed it was doing that.. It would only do it when it was on the lock screen though.. I'll stick with this build for now and maybe setup another VM with Insider Builds and see if I can replicate it. Quote Link to comment
presence06 Posted April 26, 2019 Author Share Posted April 26, 2019 Installed a Insiders Preview Edition of Win 10, setup passthrough and installed GPU drivers, RDP, disconnect, CPU's are pegged.. Quote Link to comment
NewDisplayName Posted April 26, 2019 Share Posted April 26, 2019 I would let it work for some time. Maybe its doin something. 1 Quote Link to comment
laterv Posted August 5, 2019 Share Posted August 5, 2019 Having the same issue. Upgraded some Windows 10 1809 VMs on a Hyper-V host (Server 2016) to Windows 10 1903 yesterday. When disconnecting an RDP session (no logout, just disconnect), the dwm.exe of the RDP host surges cpu usage until the RDP session gets connected again. Upgrading Graphics drivers (Intel HD Graphics P630) didn't help. There are quite a few other reports on the same issue: english: https://answers.microsoft.com/en-us/windows/forum/all/dwmexe-high-cpu-one-core-on-target-system-after/dbce0938-60c5-4051-81ef-468e51d743ab https://answers.microsoft.com/en-us/windows/forum/windows_10-performance/after-exiting-a-remote-desktop-session-cpu-load/e21e7da0-9a64-4a14-a671-b7cb1b61f66e?auth=1 german: https://www.mikrocontroller.net/topic/476003 https://www.heise.de/forum/heise-online/News-Kommentare/Juni-Sicherheitsupdates-sabotieren-Ereignisanzeige-in-allen-Windows-Versionen/OT-Sei-1903-20-CPU-Load-von-DWM-exe-nachdem-RDP-Session-beendet-wurde/posting-34690755/show/ italian: https://social.technet.microsoft.com/Forums/it-IT/fc9ffe13-47d3-47ac-a1fc-c247d8f4ca8f/utilizzo-cpu-elevata-quando-si-esce-da-rdp?forum=win10itprogeneralIT#61fb43a6-1726-4e35-9ea8-74bc268dbd79 I've tried to report the issue to ms rds dev team ([email protected]). Please reach out to Microsoft as well to report the issue, e.g. via feedback hub. Quote Link to comment
JonathanM Posted August 5, 2019 Share Posted August 5, 2019 1 hour ago, laterv said: Upgraded some Windows 10 1809 VMs on a Hyper-V host (Server 2016) What does that have to do with a KVM VM hosted on Unraid? This is not a generic tech support forum, it is for users of unraid. Quote Link to comment
laterv Posted August 6, 2019 Share Posted August 6, 2019 13 hours ago, jonathanm said: What does that have to do with a KVM VM hosted on Unraid? This is not a generic tech support forum, it is for users of unraid. The issue the OP has observed is neither related to Unraid nor Hyper-V. It is just a bug in Windows 10 1903. I've been in contact to the rds dev team at Microsoft. The issue is under investigation. As of now, either stay at Windows 10 1809 or wait for a fix/update to resolve the issue in Windows 10 1903. Please have a look at the given background info instead of ranting at a newbie. Quote Link to comment
JonathanM Posted August 6, 2019 Share Posted August 6, 2019 27 minutes ago, laterv said: The issue the OP has observed is neither related to Unraid nor Hyper-V. It is just a bug in Windows 10 1903. I've been in contact to the rds dev team at Microsoft. The issue is under investigation. As of now, either stay at Windows 10 1809 or wait for a fix/update to resolve the issue in Windows 10 1903. Please have a look at the given background info instead of ranting at a newbie. I wasn't ranting, I was asking why you were posting on an unraid forum. Do you use unraid? Quote Link to comment
JonathanM Posted August 6, 2019 Share Posted August 6, 2019 28 minutes ago, laterv said: It is just a bug in Windows 10 1903. I've been in contact to the rds dev team at Microsoft Then there was no need to post here. Quote Link to comment
JorgeB Posted August 6, 2019 Share Posted August 6, 2019 (edited) 1 hour ago, jonathanm said: Then there was no need to post here. Seems a little harsh no? 1 hour ago, laterv said: The issue the OP has observed is neither related to Unraid nor Hyper-V. It is just a bug in Windows 10 1903. I've been in contact to the rds dev team at Microsoft. The issue is under investigation. As of now, either stay at Windows 10 1809 or wait for a fix/update to resolve the issue in Windows 10 1903. Please have a look at the given background info instead of ranting at a newbie. Since updating my Win10 VM to 1903 I was having some issues but I wasn't associating them with RDP, I usually leave the session open but I would lose connection after some time and then the VM would be practically unresponsive, trying to reconnect RDP took very long and failed, and even when using Unraid VNC it stayed unresponsive, needed to force a shutdown. After reading this I tried just closing the RDP session and that leaves the VM in the unresponsive state, so I think it's related, I'm trying the workaround linked above and appears to fix it for now, so thanks for posting. Edited August 6, 2019 by johnnie.black Quote Link to comment
Recommended Posts
Join the conversation
You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.