[one-users] CPU Issue

[Ruben S. Montero]

Hi,

You are right Carsten. For Xen OpenNebula uses the credit scheduler to
distribute the CPU cycles betwen the VMs according to the CPU attribute in
the template. In the case of KVM you can achieve that with cgroups, a hook
can automatically set the share for the KVM process...

Cheers

Ruben

On Fri, Nov 26, 2010 at 12:55 AM, <Carsten.Friedrich at csiro.au> wrote:

>  As far as I can tell, for KVM, the CPU value is only used in the
> OpenNebula scheduler to prevent over-allocation. So, if you have defined the
> host as having 2 CPUs the scheduler will run e.g. 1 VM requesting CPU 2.0,
> or 2 VMs requesting CPU 1.0, or 4 VMs requesting CPU 0.5 (each) on this
> host. For KVM, it seems that once running on the hypervisor the CPU value is
> not enforced.
>
>
>
> c
>
>
>
> *From:* users-bounces at lists.opennebula.org [mailto:
> users-bounces at lists.opennebula.org] *On Behalf Of *Luca Lorenzini
> *Sent:* Friday, 26 November 2010 3:59
> *To:* users at lists.opennebula.org
> *Subject:* [one-users] CPU Issue
>
>
>
> I'm launching 3 vm with CPU=0.4 on a host dual core, but i noticed that the
> vms use more than 0.4 of the cpu. top command show me this:
>
> PID USER      PR  NI  VIRT  RES  SHR S %CPU %MEM    TIME+
> COMMAND
> 24274 root      20   0  577m 504m 2868 S   90 25.7   0:31.77
> kvm
> 24964 root      20   0  577m 511m 2864 S   60 26.1   0:29.77
> kvm
> 23724 root      20   0  577m 497m 2868 S   35 25.4   0:34.60 kvm
>
> kvm use 90% of the cpu, not just 40%
>
> _______________________________________________
> Users mailing list
> Users at lists.opennebula.org
> http://lists.opennebula.org/listinfo.cgi/users-opennebula.org
>
>


-- 
Dr. Ruben Santiago Montero
Associate Professor (Profesor Titular), Complutense University of Madrid

URL: http://dsa-research.org/doku.php?id=people:ruben
Weblog: http://blog.dsa-research.org/?author=7
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.opennebula.org/pipermail/users-opennebula.org/attachments/20101126/00f685f1/attachment-0002.htm>