Unable to allocate RAM to VM as despite of having enough memory

[Sujith Arangan]

I got 1024 GB RAM in my host and I am just running 1 VM. But I am unable to allocate more RAMs. I am getting an error of

TASK ERROR: Total memory is bigger than 524288MB

When I check the free memory in the proxmox terminal I see enough memory left.

free -m
total used free shared buff/cache available
Mem: 1031760 10953 997234 77 29464 1020806
Swap: 8191 0 8191

The VM is with NUMA enabled CPU. If the NUMA is disabled. The VM won't start.

Kindly help me on this.

 

[bbgeek17]

It seems like Max Memory may be CPU model dependent. What type of CPU are you using for this VM? You may want to try with "host".

Good luck

---

Blockbridge : Ultra low latency all-NVME shared storage for Proxmox - https://www.blockbridge.com/proxmox

 

[Sujith Arangan]

It seems like Max Memory may be CPU model dependent. What type of CPU are you using for this VM? You may want to try with "host".

Good luck

---

Blockbridge : Ultra low latency all-NVME shared storage for Proxmox - https://www.blockbridge.com/proxmox

It's an AMD processor.

Vendor ID: AuthenticAMD
BIOS Vendor ID: Advanced Micro Devices, Inc.
Model name: AMD EPYC 9274F 24-Core Processor
BIOS Model name: AMD EPYC 9274F 24-Core Processor Unknown CPU @ 4.0GHz
BIOS CPU family: 107

And the VM configuration is as below

bios: ovmf
boot: order=scsi0;ide2;net0
cores: 96
cpu: EPYC
efidisk0: local-lvm:vm-100-disk-0,efitype=4m,pre-enrolled-keys=1,size=4M
ide2: none,media=cdrom
memory: 953674
meta: creation-qemu=8.1.5,ctime=1715882860
name: strata-2
net0: virtio=BC:24:11:61:A1:42,bridge=vmbr0
numa: 0
ostype: l26
parent: forcloning
runningcpu: EPYC,enforce,+kvm_pv_eoi,+kvm_pv_unhalt,vendor=AuthenticAMD
runningmachine: pc-i440fx-8.1+pve0
scsi0: local-lvm:vm-100-disk-1,iothread=1,size=500G
scsi1: local-lvm:vm-100-disk-2,iothread=1,size=50G
scsihw: virtio-scsi-single
smbios1: uuid=3e9ae1e0-2879-480c-82c1-c5baeffae580
snaptime: 1716355811
sockets: 1
vmgenid: 56947432-0c0b-461f-a6c2-b7f4d324bc76
vmstate: local-lvm:vm-100-state-newclone

I made the CPU type to host and still I get the error TASK ERROR: memory size (972800) must be aligned to 16384 for hotplugging

 

[bbgeek17]

I made the CPU type to host and still I get the error TASK ERROR: memory size (972800) must be aligned to 16384 for hotplugging

972800 does not divide by 16384 by a whole number. Try 983040

---

Blockbridge : Ultra low latency all-NVME shared storage for Proxmox - https://www.blockbridge.com/proxmox

 

[Sujith Arangan]

I am getting the error below. I am unable to assign more than 16 GB to the VM
TASK ERROR: memory size (983040) must be aligned to 16384 for hotplugging

 

[bbgeek17]

Have you tried to shutdown the VM and make your changes offline?

---

Blockbridge : Ultra low latency all-NVME shared storage for Proxmox - https://www.blockbridge.com/proxmox

 

[Sujith Arangan]

Have you tried to shutdown the VM and make your changes offline?

---

Blockbridge : Ultra low latency all-NVME shared storage for Proxmox - https://www.blockbridge.com/proxmox

Yes that's exactly what I am doing. The VMs don;t even turn on after making the changes.

 

[emptness]

I join the topic. The same problem. I can't assign RAM more than 512GB.
memory: hotplug problem - you cannot add more memory than max mem 524288 MB

If you set the host processor type, it will work. But this is not how live migration to nodes with a different type of processor works. This all extremely reduces the functionality of the cluster.

Is there a solution to this problem with large amounts of RAM for a VM?

 

[bbgeek17]

You may want to read through this thread/bug/etc:
https://forum.proxmox.com/threads/problem-with-hotplug-and-64gb-ram.73621/

---

Blockbridge : Ultra low latency all-NVME shared storage for Proxmox - https://www.blockbridge.com/proxmox

 

[emptness]

Unfortunately, there is no solution to the problem on the links. It only helps to either disable the hotplug RAM or set the Host processor type. Both options disable the necessary functionality.

I would like to get a response from the developer's representatives. Is there a solution at the moment and is it planned to fix it and when?