@olivierlambert

Since it's a very old CPU, maybe try 1 socket with 4 cores in the topology

A Topology change to '1 socket with 4 cores per socket' finally did the trick. The VM now uses all four vCPUs.

Also, if you are concerned about your data, start with a snapshot (and ideally a backup) of this VM.

That's what I always do beforehand with planned software updates. But now it seems that I also have to prevent the OS from updating automatically at random times.

And you're absolutely right, the CPU is way outdated, but as long as it does the job in a local environment I strive against producing electronic waste.

Regarding tools, check https://docs.xcp-ng.org/vms/#windows for the best way. Windows 10 supports both.

And there I read 'It's now up to you to choose', thus my question about the standard way.

Many many thanks again for your marvelous service!

@olivierlambert Up to now I shied away from installing the VM tools after reading 'Windows 10 (1803) - crash with installed guest tools', as I can't afford to risk losing valuable business data with any Windows update.

And allow me to ask, what's the preferred and future-proof way? Installing the XCP-ng tools .zip-package, or deploying Citrix tools by simply switching on 'Windows Update tools' on XO?

Thanks for your patience.

@olivierlambert said in Paused vCPUs:

Do you have the tools installed in that VM? First time I heard about cores not detected

No. Do I have to?

Since it's a very old CPU, maybe try 1 socket with 4 cores in the topology

I'll give it a try.

@olivierlambert

Nevertheless Windows doesn't see more that two cores. When I run Prime95 I can't enter more than '2' for 'Number of cores to torture test'. With a higher number I get 'Enter an integer between 1 and 2.'.

Is something wrong with my grub parameters?

setparams 'XCP-ng'
search --label --set root root-dilqpt
multiboot2 /boot/xen.gz dom0_mem=1648M,max:1648M watchdog ucode=scan dom0_max_vcpus=1-4 crashkernel=256M,below=4G console=vga vga=mode-0x0311 sched-gran=core
module2 /boot/vmlinuz-4.19-xen root=LABSL=root-dilqpt ro nolvm hpet=disable console=hvc0 console=tty0 quiet vga=785 splash plymouth.ignore-serial-consoles
module2 /boot/initrd-4.19-xen.img

I'm at a loss. Do I really have to move to a system with a more potent processor?

Many thanks!

@olivierlambert

@olivierlambert But doesn't 'xl vcpu-list' show that 4 vCPUs are assigned to the Win10 VM, of which unfortunately only two are in use and the other two despite heavy load are paused?

@olivierlambert

Windows Taskmanager - Performance - CPU:
Maximum Speed: 2,66 GHz
Sockets: 2
Virtual processors: 2
Virtual machine: Yes
L1 cache: N/A

The Resource Monitor shows diagrams for CPU 0 and CPU 1.

Hello!

I run XCP-ng 8.2.0 on a Dell PowerEdge T110 (Xeon X3450 - 4 cores / 8 threads).

With 4 of the given 8 vCPUs assigned to dom0, half of the remaining 4 vCPUs I'd need for my Windows 10 VM are paused according to 'xl vcpu-list' and the Windows Resource Monitor, which only shows CPU 1 and CPU 2.

# xl vcpu-list
Name         ID  VCPU   CPU State   Time(s) Affinity (Hard / Soft)
Domain-0      0     0    0   -b-     114.7  0-7 / 0-7   
Domain-0      0     1    1   r--     123.4  0-7 / 0-7   
Domain-0      0     2    2   r--     123.4  0-7 / 0-7   
Domain-0      0     3    3   r--     123.4  0-7 / 0-7   
Windows 10    1     0    4   -b-      71.5  all / all
Windows 10    1     1    5   -b-      67.1  all / all
Windows 10    1     2    -   --p       0.0  all / all
Windows 10    1     3    -   --p       0.0  all / all

Any ideas how to activate those paused vCPUs? Is there a configuration checklist to get to the bottom of it?

I'm in urgent need of these two inoperative vcpus to gain performance. Running a database server on Windows with no more that one CPU core is a mess.

Thanks!