[pve-devel] Qemu numa setting

[Stefan Priebe - Profihost AG]

Am 19.02.2014 07:57, schrieb Alexandre DERUMIER:
>>> Is there any reason we don't add it by default? 
> 
> I didn't read qemu numa implementation yet.
> 
> I don't known :
> 
> - does it work if host is a non numa architecture ? 
> - do we need to map numa guest/host ?
> - how does it play with auto numa balacing from host kernel ?

Yeah that's the question i would like to know too. But i think as long
as you don't pin each VM CPU to a specific core on the host it makes no
sense at all as you don't know which CPU will do the stuff next time...

> (Note that I'm using only numa architecture, quad socket opteron servers, so i'm interested to improve memory access performance)

Same for me but i haven't found any guides...

Stefan

> 
> 
> 
> ----- Mail original ----- 
> 
> De: "Stefan Priebe - Profihost AG" <s.priebe at profihost.ag> 
> À: "pve-devel" <pve-devel at pve.proxmox.com> 
> Envoyé: Mercredi 19 Février 2014 07:23:21 
> Objet: [pve-devel] Qemu numa setting 
> 
> 
> Hi, 
> 
> 
> at least windows needs numa setting to have memory hot plug working. 
> 
> 
> Is there any reason we don't add it by default? 
> 
> Greets, 
> Stefan 
> 
> 
> Excuse my typo s ent from my mobile phone. 
> _______________________________________________ 
> pve-devel mailing list 
> pve-devel at pve.proxmox.com 
> http://pve.proxmox.com/cgi-bin/mailman/listinfo/pve-devel 
>