coming from just kvm/qemu(older version) iam struggling getting my VM set up the right way, before switching to proxmox i could allocate hugepages to the nodes i wanted on boot, and have qemu use them by just enabling hugepages and using cpu pinning, but on proxmox it seems to only want to allocate its own HPs (it wont use the preallocated free ones) and only use from node0, even when it runs out of HPs on node0 it wont go to node1 2 or 3 it will just fail starting.
I assume iam missing some setting or config, or maybe i need to run a different version. it uses the cpu from node i assign but i cant get the HPs working like i want
after startup, if i preallocate across the nodes like i want it just ignores those free HPs and starts allocating new ones on node0 and runs out. so i can only start the vm by removing all HPs and letting it allocate whats possible on node0 ram(less than 64gb).
I assume iam missing some setting or config, or maybe i need to run a different version. it uses the cpu from node i assign but i cant get the HPs working like i want
Code:
affinity: 0-15,128-143,16-31,144-159,32-47,160-175,48-63,176-191
args: -smp '64,sockets=1,cores=64,threads=1,maxcpus=64' -cpu 'host,topoext=on,host-cache-info=on,hv_ipi,hv_relaxed,hv_reset,hv_runtime,hv_spinlocks=0x1fff,hv_stimer,hv_synic,hv_time,hv_vapic,hv_vpindex,+kvm_pv_eoi,+kvm_pv_unhalt,+pdpe1gb'info=on,hv_ipi,hv_relaxed,hv_reset,hv_runtime,hv_spinlocks=0x1fff,hv_stimer,hv_synic,hv_time,hv>
balloon: 0
bios: ovmf
boot: order=hostpci1;hostpci2
cores: 64
hostpci0: 0000:01:00,pcie=1
hostpci1: 0000:81:00.0,pcie=1
hostpci2: 0000:82:00.0,pcie=1
hostpci3: 0000:a1:00.0,pcie=1
hostpci4: 0000:21:00.0,pcie=1
hostpci5: 0000:03:00.3,pcie=1
hostpci6: 0000:83:00.0,pcie=1
hugepages: 1024
ide2: none,media=cdrom
machine: pc-q35-7.2
memory: 49152
meta: -
name: -
net0: virtio=26:AC:18:3A:72:84,bridge=vmbr0,firewall=1
numa: 1
ostype: win10
scsihw: virtio-scsi-pci
smbios1: uuid=-
sockets: 1
vcpus: 64
vga: none
vmgenid: -
Code:
GRUB_CMDLINE_LINUX_DEFAULT="mitigations=off amd_iommu=on iommu=pt ignore_msrs=1 default_hugepagesz=1G hugepagesz=1G vfio-pci.ids=10de:2204,10de:1aef,2646:2262 module_blacklist=nvidia,nvidia_uvm,nvidia_drm,nvidia_modeset,nouveau,megaraid_sas initcall_blacklist=sysfb_init"
Code:
Token Node not in hash table.
Node 0 Node 1 Node 2 Node 3 Node 4 Node 5 Node 6 Node 7 Total
AnonHugePages 0 0 0 2 0 0 0 0 2
HugePages_Total 49152 0 0 0 0 0 0 0 49152
HugePages_Free 0 0 0 0 0 0 0 0 0
HugePages_Surp 49152 0 0 0 0 0 0 0 49152after startup, if i preallocate across the nodes like i want it just ignores those free HPs and starts allocating new ones on node0 and runs out. so i can only start the vm by removing all HPs and letting it allocate whats possible on node0 ram(less than 64gb).