claron-cloud-diagnostics-20260519-1647.zip
Since updating to 7.3.0, when my VM is running either core 0 or 16 is pinned at 100%

I have traced the issue to QEMU but cannot resolve. I have tried various settings in the VM config. This issue is not linked to the VM OS as the core is pinned at 100% before the OS VM boots.

• VM: Windows Server 2016 running Blue Iris/CCTV workload

• Initial symptom:

  • One host CPU thread permanently pinned at ~100%

  • qemu-system-x86 consuming up to ~220% host CPU

  • Guest Windows CPU usage only ~9%

  • Host overall remained mostly idle (~88% idle)

VM configuration changes attempted

• Reduced VM from:

  • 16 vCPUs (8c/16t)

  • to 8 vCPUs (4c/8t)

• Applied explicit CPU pinning:

  • VM vCPUs pinned to cores 1-4,17-20

  • Emulator/iothread pinned to 0,16

• Added:

  • <kvm><hint-dedicated state='on'/></kvm>

• Disabled Hyper-V enlightenments:

  • removed entire <hyperv> block

• Disabled Windows dynamic tick:

  • bcdedit /set disabledynamictick yes

• Set Windows power plan to High Performance

Storage changes attempted

• Original VM used SATA emulation:

  • <target dev='hdc' bus='sata'/>

• Installed VirtIO drivers inside Windows

• Migrated all disks to VirtIO-SCSI

• Added:

  • <controller type='scsi' model='virtio-scsi'>

  • <driver queues='8'/>

• Retained:

  • cache='none'

  • io='native'

• Tested with:

  • qcow2 system disk

  • raw physical disk passthrough devices

Network changes attempted

• Original NIC config:

  • type='direct'

  • macvtap

  • eth2 mode='bridge'

• Changed to:

  • <interface type='bridge'>

  • <source bridge='br0'/>

  • virtio-net

Graphics/device emulation changes attempted

Removed:

• VNC graphics

• QXL video

• USB tablet

• watchdog

• VirtIO ISO/CD-ROM

• unused SATA/IDE devices

Do you see the same if you create a new VM using a new vdisk (same OS)

10 minutes ago, JorgeB said:

Do you see the same if you create a new VM using a new vdisk (same OS)

I've just created a new VM and installed the same OS. It does not appear to be causing the CPU core to pin at 100%

That suggests a specific issue with that VM or vdisk, it may not be easy to troubleshoot. If you can, copy the data to the new VM and start using that one.

Give a look at this: https://product.unraid.net/p/7-3-0-qemu-iothreads-pinned-cores-show-100-iowait-at-idle?b=unraid-os-bugs