[pve-devel] cgroups implementation proposal for kvm (cpuunits support for kvm)

[Alexandre DERUMIER]

>> I am not even sure if we need to control cpu.shares - users expext 1 CPU if they assign 1CPU? 

cpu.share is working like cpuunits, it's not limiting the cpu, only give priorities between process when the host cpus are 100%


users expect 1CPU if they assign 1 CPU.

but if you assign 
2 VM with 1 cpu on an host with 1 cpu, you need to share ....




for blkio, yes, it's very difficult to limit with cgroup.(I was just talking about cpu)


----- Mail original ----- 

De: "Dietmar Maurer" <dietmar at proxmox.com> 
À: "Alexandre DERUMIER" <aderumier at odiso.com>, pve-devel at pve.proxmox.com 
Envoyé: Lundi 19 Décembre 2011 18:11:18 
Objet: RE: [pve-devel] cgroups implementation proposal for kvm (cpuunits support for kvm) 

> 3) for kvm guest,at vm start 
> create a container /dev/cgroups/fairsched/VMID/ 
> echo vmpid > /dev/cgroups/fairsched/VMID/tasks 
> echo cpuinits > /dev/cgroups/fairsched/VMID/cpu.shares 

I am not even sure if we need to control cpu.shares - users expext 1 CPU if they assign 1CPU? 

AFAIK blkio cgroup is the wrong way to limit IO for kvm, because it does not really work on (network) file systems. The plan is to do rate control inside kvm: 

http://wiki.qemu.org/Features/DiskIOLimits 

http://git.kernel.org/?p=virt/kvm/qemu-kvm.git;a=commit;h=0563e191516289c9d2f282a8c50f2eecef2fa773 

I had no time to test that so far. 

- Dietmar 





-- 

-- 




	
	Alexandre Derumier 
Ingénieur système 
e-mail : aderumier at odiso.com 
Tél : +33 (0)3 20 68 88 90 
Fax : +33 (0)3 20 68 90 81 
45 Bvd du Général Leclerc 
59100 ROUBAIX - FRANCE 













-------------- next part --------------
A non-text attachment was scrubbed...
Name: aderumier.vcf
Type: text/x-vcard
Size: 183 bytes
Desc: not available
URL: <http://lists.proxmox.com/pipermail/pve-devel/attachments/20111219/30fb6114/attachment.vcf>