Proxmox vCPU flags

Good luck with that. Qemu gives you what it gives you and there really isn't a work-around short of rebuilding qemu and figuring out how to support each and every one. Maybe try converting the old container to the new lxc format would be simpler?

It seems your software is highly optimized for one particular CPU and so even changing to a different real CPU will be problematic. For instance, I have here a Xeon D-1541. On the bare metal it has identical flags to your VM case with cpu=host. It is a Broadwell-generation processor and some flags on newer CPU's simply didn't exist then.

I'm thinking the qemu version you use is old enough that the flags you want didn't exist at the time. They can't simply be passed through because some of them could represent privileged operations that a hypervisor would need to intercept in order safely visualize them.
 
Good luck with that. Qemu gives you what it gives you and there really isn't a work-around short of rebuilding qemu and figuring out how to support each and every one. Maybe try converting the old container to the new lxc format would be simpler?

It seems your software is highly optimized for one particular CPU and so even changing to a different real CPU will be problematic. For instance, I have here a Xeon D-1541. On the bare metal it has identical flags to your VM case with cpu=host. It is a Broadwell-generation processor and some flags on newer CPU's simply didn't exist then.

I'm thinking the qemu version you use is old enough that the flags you want didn't exist at the time. They can't simply be passed through because some of them could represent privileged operations that a hypervisor would need to intercept in order safely visualize them.
Thank you for the reply.

So if I understand correct here, it si QEMU that is an issue due to age. If i had a newer version then maybe I would not face this?

The reason we want to move away from containers is so we can use the ha feature in our upcoming proxmox refresh.

I do need to also add that the container and the vm is on the same host. Is it due to the versions that it is so different and possibly the virutualisation type?
 
I don't know that it is the age of Qemu, but it fits the symptoms. Possibly some or all of the features represented by those flags would be safe to use in a VM, but there is no way for five-years-ago Qemu to know that. You could try setting up PVE 6 on a different PC and see if it supports more of the flags you need.

From https://www.qemu.org/docs/master/system/target-i386.html

Code:
Host passthrough

This passes the host CPU model features, model, stepping, exactly to the guest. Note
that KVM may filter out some host CPU model features if they cannot be supported with
virtualization.

Also, I think PVE HA supports containers now (I don't use HA though).
 
We have containers that have these flags enabled

You cannot enable or disable cpu flags in a container, so this cannot be compared.

It seems your software is highly optimized for one particular CPU and so even changing to a different real CPU will be problematic.

That's what I thought and also wrote. Have you checked what I suggested? Running the container inside of a VM just to see how fast or slow the code runs?
 

About

The Proxmox community has been around for many years and offers help and support for Proxmox VE, Proxmox Backup Server, and Proxmox Mail Gateway.
We think our community is one of the best thanks to people like you!

Get your subscription!

The Proxmox team works very hard to make sure you are running the best software and getting stable updates and security enhancements, as well as quick enterprise support. Tens of thousands of happy customers have a Proxmox subscription. Get yours easily in our online shop.

Buy now!