Hi Community,
I have a Cisco UCS c220m4 server (2 socket, 10 cores each, 128GB memory) with NUMA enabled, no affinity and no hotplug configured on the VM (16vCPU, 96GB).
esxtop with "m" selected shows there is 2 NUMA node detected with 64GB memory each.
Have tried all combination of configuration but esxtop with "m" and "f" and "g" to get NUMA stats selected always shows NHN to be a "-" instead of "0,1" for the configured VM.
Hope some one from the community or VMware can shed some light on how the NUMA scheduler works here since the VM crashes (hangs) whenever the application tries to consume more than 64GB of memory even though it is configured to have 96GB (wide VM) and RHEL7 installed does detect the 96GB.
Thanks.
Regards,
ChewCC
