I thought I had a pretty good understanding of virtual machine queuing but I am getting unexpected results that I do not quite understand.
I have a system that has two processors in it with each processor having 22 logical cores (hyperthreading is DISABLED). This would give me a total of 44 cores. Now I configured the first 10Gbps NIC to use cores 1 through 22 with a maximum of 16 processors (31 receive queues total since I'm using SET) and I configured the second 10Gbps port to use cores 23 through 44.
However, if you look at the assignments below, it is putting some of them on GROUP 1 core 0 which doesn't make any sense. The process groups are only supposed to be getting used if you have more than 64 cores and I do not. Has anyone seen this before or know what is going on?
Hyper-V 2019 running on DL360 Gen9 with 2021-09 CU running switch embedded teaming.
Now when I go and look at "Get-NetAdapterRss" you can see the second NIC has a MaxProcessor of 1:0 which explains why everything one NIC 2 is getting put on 1:0. But why? With VMQ it shows each NIC has 31 queues and the Get-NetadapterRss shows that there is 128 number of receive queues but I can't increase the MaxProcessors above 16 (16 is the max for some reason even though this is a 44 core server):
InterfaceDescription : HPE Ethernet 10Gb 2-port 560FLR-SFP+ Adapter
Enabled : True
NumberOfReceiveQueues : 128
Profile : NUMAStatic
BaseProcessor: [Group:Number] : 0:23
MaxProcessor: [Group:Number] : 1:0
MaxProcessors : 1
RssProcessorArray: [Group:Number/NUMA Distance] : 1:0/32767
IndirectionTable: [Group:Number] : 1:0```
