Hey guys having a few issues with one of my virtual machines. I have 2 VM's configured and running, both with their own video cards. The one i'm speaking of has a GTX 980 and two USB3 hub's passed through to it. The VM's are using a shared directory for the games which is located on a 256gb SSD. So that they dont need to be installed twice. And both operating systems are installed on the 120gb SSD cache drive. The larger VM with 8 cores and 16gb of ram is having issues running a game thats not very intensive. but its causing core 0 to be very close to maxed out. Which is causing an incredible amount of lag on the PC and in game.

 

The hardware specs for the server are below

 

Intel 3970x overclocked to 4.2 Ghz & watercooled

Gigabyte GA-X79-UP4

32gb ram

256gb samsung ssd | Storage drive

120gb samsung ssd | Cache drive

2tb western digital black? | Parity drive

Nvidia Gtx 980 | Problem VM

Evga geforce 210 | second VM

 

A bit more info. If the game isnt running on the second VM then core 0 doesnt seem to be acting up. If i'm not mistaken one VM shouldnt affect the other when they're using different cores. any ideas would be greatly appreciated

I assume you have left some cores free for unRAID to run KVM (typically core 0 is used for this)?

I assume you have left some cores free for unRAID to run KVM (typically core 0 is used for this)?

 

I have not. I figured there could be something else going on in the background. I'll try freeing cores 0 and 1. and split the other 10 between the two VM's

 

Is it possible that KVM could be using all 12 cores instead of just core 0 which is why i would see the increased load across all cores? the VM in question is currently using cores 5-11 and the other using 0-4. but i still seem to be having the issue. if KVM did use core 0 then should the VM having issues be the smaller one since its using cores 0-4?

I assume you have left some cores free for unRAID to run KVM (typically core 0 is used for this)?

 

I have not. I figured there could be something else going on in the background. I'll try freeing cores 0 and 1. and split the other 10 between the two VM's

 

Is it possible that KVM could be using all 12 cores instead of just core 0 which is why i would see the increased load across all cores? the VM in question is currently using cores 5-11 and the other using 0-4. but i still seem to be having the issue. if KVM did use core 0 then should the VM having issues be the smaller one since its using cores 0-4?

I think that in theory if you have not hidden them from unRAID then all cores can be used.  However from other postings it seems that unRAID seems to preferentially use core 0.

 

In terms of which VM would have problems I would expect it to be the one that is wanting to do the most if it is continually waiting on access to core 0.  However these sorts of interaction can be mysterious and hard to predict so anything I say is simply a guess.

I assume you have left some cores free for unRAID to run KVM (typically core 0 is used for this)?

 

I have not. I figured there could be something else going on in the background. I'll try freeing cores 0 and 1. and split the other 10 between the two VM's

 

Is it possible that KVM could be using all 12 cores instead of just core 0 which is why i would see the increased load across all cores? the VM in question is currently using cores 5-11 and the other using 0-4. but i still seem to be having the issue. if KVM did use core 0 then should the VM having issues be the smaller one since its using cores 0-4?

I think that in theory if you have not hidden them from unRAID then all cores can be used.  However from other postings it seems that unRAID seems to preferentially use core 0.

 

In terms of which VM would have problems I would expect it to be the one that is wanting to do the most if it is continually waiting on access to core 0.  However these sorts of interaction can be mysterious and hard to predict so anything I say is simply a guess.

 

no problem the insight is definitely appreciated. I'm still on my free trial so i'm just trying to get used to the software and see if its viable for what i need. When i check "htop" on the server. it shows that there is one process for qemu-system-x86_64 using 286% of the cpu... I'm assuming this is my culprit so i'm going to shut down both of the vm's now. re allocate the cores. and check htop again.

Same issue unfortunately. I left 2 cores free, and the cpu usage on both of the VM's in htop is quite ridiculous. Is there a possibility I have my drives setup incorrectly? I was running 3 VM's inside of this machine using VMware before. unfortunately that didnt support pci passthrough. and thats one of the reasons i switched over to UNraid. So I know the processing power is there.

 

EDIT: Heres something insteresting which may relate to disk usage. When running crystal disk mark inside the larger VM the cpu usage also spikes on core 0 of the VM.

EDIT2: Downloading is also causing a cpu spike on both VM's

If i'm not mistaken. with 2 cores free now. that the machine should not be spiking the cpu on the VM. So what the hell could be going on? I've installed windows too many times to count and am getting a bit discouraged with the software. Since i've tried a few different configurations. with and without cache drives. and the VM with windows 10 seems to constantly be having issues compared to the smaller one which is now running windows 7.

 

I love the ability the software has and im sure that its probably my own fault. but i can't seem to find much regarding this issue.

Are you still over clocking your system? If so, I might try changing it to stock setting just to see if you are still getting weird CPU spikes. As stable as your overclocked system may be, over clocking can still introduce unpredictable and a variance in results that may be causing your issue. Just my two cents.

Are you still over clocking your system? If so, I might try changing it to stock setting just to see if you are still getting weird CPU spikes. As stable as your overclocked system may be, over clocking can still introduce unpredictable and a variance in results that may be causing your issue. Just my two cents.

 

Updated the system to the beta version. and things seem to be responding a bit better. I removed the overclock for now like suggested. But so far it seems that it hasn't cured the issue. although it has cured a few stability issues.

 

I added a second cache drive in. and now have 2 * 120 gb ssd cache drives that the windows installs are on. with a 256 gb ssd drive and a 2tb hdd parity drive. The cpu spikes seem to be happening inside the VM. And even when the program i'm using is set not to use core 0 on the VM. core 0 is still spiked.

 

it seems to be happening at random which VM is affected based on how they're started maybe?

With both VM's audio cards from nvidia disabled. The cpu usage is okay. trying to track down which of the cards is giving me issues. Is this a known problem? And maybe theres a fix for it already like i've seen for other issues in the past?