VMs are unusably slow.

[DarkMain]

Specs:

Unraid: Unraid server Pro, version 6.9.2
Motherboard: ASUSTeK COMPUTER INC. RAMPAGE IV BLACK EDITION Version Rev 1.xx -
CPU: Intel® Core™ i7-4930K CPU @ 3.40GHz

HVM: Enabled

IOMMU: Enabled

Cache:

• L1-Cache = 32 kB (max. capacity 32 kB)
• L2-Cache = 256 kB (max. capacity 256 kB)
• L3-Cache = 12 MB (max. capacity 12 MB)

Memory: 24 GB (max. installable capacity 96 GB)

• ChannelA_Dimm1 = 4 GB, 1600 MT/s
• ChannelB_Dimm1 = 4 GB, 1600 MT/s
• ChannelB_Dimm2 = 4 GB, 1600 MT/s
• ChannelC_Dimm1 = 4 GB, 1600 MT/s
• ChannelD_Dimm1 = 4 GB, 1600 MT/s
• ChannelD_Dimm2 = 4 GB, 1600 MT/s
   

This has been frustrating me for a while now and I have had no luck figuring it out (reddit posts mainly) so I thought I would try my luck here.

Every VM I create is insanely slow to the point that its unusable. Windows, Linux... doesn't matter.

Virtualization is enabled in the BIOS, the vDisk is located on an SSD through Unassigned devices.

What log files do I need to submit that might show where the problem is?

[Squid]

On 3/20/2022 at 12:32 AM, DarkMain said:

What log files do I need to submit that might show where the problem is?

Not that I'm the guy per se to help here, but anyone is going to want to see the diagnostics to be able to assist

[dlandon]

You need to provide diagnostics and more information on how you are setting up your VMs.  Memory, CPUs, machine model, etc.

[DarkMain]

30 minutes ago, dlandon said:

You need to provide diagnostics and more information on how you are setting up your VMs.  Memory, CPUs, machine model, etc.

Uploaded the diagnostics 

The VMs have been setup in multiple different way to test to see if it helps.

CPU mode = Passthrough or emulated. Both give roughly the same level of performance.
2 / 4 / 6 CPS. Performance remains about the same regardless of cores.
2 / 4 / 6 / 8 Gig of Memory. Doesn't seem to make a difference.
Machine = i440fx (multiple versions) and Q35 (Multiple versions). No difference.
Bios = OVMF or SeaBIOS. No difference.
No PCIe device are being passed through.

In all honesty, I have a feeling the issue is not related to VMs. It seems to be an I/O issue, however I can only reliably create the problem 100% when using a VM.

tower-diagnostics-20220326-0152.zip

[Squid]

Doesn't look like you've got any VMs running in those diagnostics.  Can you fire it up...

[DarkMain]

11 hours ago, Squid said:

Doesn't look like you've got any VMs running in those diagnostics.  Can you fire it up...

K, got one fired up this time.

I have also uploaded the video to my google drive - https://drive.google.com/file/d/1x4bY4DLWRgIbl_m2B5NUndsb1S5Vpssa/view?usp=sharing

tower-diagnostics-20220326-1428.zip

[dlandon]

I see a few issues right away.  The SCSI disk driver is not the best to use.  Use the VirtIO disk driver.  It looks like you are using memory balloon.  Set Initial and max memory to the same value.

 

Also post a screen shot of your Logical CPU assignments.

[DarkMain]

12 minutes ago, dlandon said:

I see a few issues right away.  The SCSI disk driver is not the best to use.  Use the VirtIO disk driver.  It looks like you are using memory balloon.  Set Initial and max memory to the same value.

 

Also post a screen shot of your Logical CPU assignments.

That's weird, because the memory is already set to the same in the GUI as is the Primary vDisk Bus.

Its the same in the XML

  <memory unit='KiB'>6291456</memory>
  <currentMemory unit='KiB'>6291456</currentMemory>

    <disk type='file' device='disk'>
      <driver name='qemu' type='raw' cache='writeback'/>
      <source file='/mnt/disks/VMs/domains/Windows 10/vdisk1.img'/>
      <target dev='hdc' bus='virtio'/>
      <boot order='1'/>
      <address type='pci' domain='0x0000' bus='0x00' slot='0x05' function='0x0'/>
    </disk>

 

Edited March 26, 2022 by DarkMain

[dlandon]

Some observations:

• Your xml shows 6GB of memory but your video shows 4GB?  That's not what your xml shows.
• I can't understand how you can peg your CPU usage at 100% with 3 cores and 6 CPUs doing a file copy.  Did you have the xml setup you posted when you did the video?
• You have the vdisk on a UD device.  Is it a SSD or a spinner disk?  How is it formatted?
• Why not put that disk in a Pool and do your VMs there?

As you said before, you need to concentrate on hardware issues.

[DarkMain]

4 hours ago, dlandon said:

Some observations:

• Your xml shows 6GB of memory but your video shows 4GB?  That's not what your xml shows.
• I can't understand how you can peg your CPU usage at 100% with 3 cores and 6 CPUs doing a file copy.  Did you have the xml setup you posted when you did the video?
• You have the vdisk on a UD device.  Is it a SSD or a spinner disk?  How is it formatted?
• Why not put that disk in a Pool and do your VMs there?

As you said before, you need to concentrate on hardware issues.

I have multiple VMs.
The video is just showing the issue that is present in all of them. (It was a brand new install of Windows 11 for testing hence the low ram). This one had the same number of CPUs (same config, no other VMs running). 4GB memory (Initial and max both the same size), Machine was Q35-6.2, Bios was OVMF TPM, the vDisk was qcow with the VirtIO bus.

Like I said, its 100% repeatable within a VM regardless of how the VM is set up. I can make another video showing whatever you like on the screen with any VM if that will help.
If I get lucky I MIGHT be able to do a file transfer or two before the issue starts.
(Its not just restricted to file transfers either, its any HDD activity, hence the unusable VMs)

The vDisk is on a UD because it was installed there before cache pools were a thing.
I have been using Unraid since very late version 3, so somethings may not be 'standard'. (MY docker app data is called 'applications' because 'appdata' never existed in earlier versions, Same with the 'System', 'Domain' and 'ISO' dir.)

Regardless, With the Windows 11 VM I actually made a new pool to test it out before making this post.

Yes the UD is an SSD and its formatted in XFS.


You're probably right about it being Hardware but I have no idea where to start with diagnosing it, and like I said, VMs are the only thing I have found that I can use to reliably create the issue.
(It used to be a problem in Krusader, but that was random and also involved an old spinning drive so I just lived with it, however, I haven't seen the problem in Krusader in at least 3 months.).
 

Edited March 27, 2022 by DarkMain

[Natcoso9955]

Did you get anywhere with this, im not sure if im having the same issue or something similar, just trying to install windows server 2016 took like 24+ hrs with it finalizing config and the entire time i couldnt load the vm tab in unraid.

[DarkMain]

Nope, sorry.

Still no solution.

I've just installed a new RAID card and pulled all the drives off the mother boards to see if one of the controllers on there was causing the issue.

Its honestly driving me nuts.

[showstopper]

Sorry to bump such an old thread but... Does anyone else have this issue? Al VM's are incredibly slow, exactly like @DarkMain

[PSYCHOPATHiO]

On 3/20/2022 at 7:32 AM, DarkMain said:

Specs:

Unraid: Unraid server Pro, version 6.9.2
Motherboard: ASUSTeK COMPUTER INC. RAMPAGE IV BLACK EDITION Version Rev 1.xx -
CPU: Intel® Core™ i7-4930K CPU @ 3.40GHz

 

i know this is old, but this is an amazing CPU unfortunatly the motherboard i have is dying and cant find a decent replacement for a decent price.... so had to go for a complete new system..... do not throw away this server... one of the most stable CPUs I have ever came accross.

[nitrossub]

i just fix my slow vms it was a connection problem for me if you are running pfsense and unraid connect, turn on pure nat reflection .  i was having big lag problems and unresponsive it was just i long vnc connection going out to the internet and back again. may  not be related to your problem but may help someone.