Hi Matthew, That's a very good question, thanks for bringing this up. When we implemented cap/weight for XenServer, we thought cap/weight is per VCPU not per VM. If cap is wrong, then weight is also wrong.
In all articles you referred to , they don't explicitly say if the domain is a single VCPU domain or a SMP domain. By default, the weight is 256. If weight is per domain/VM, 4VCPU VM should get the same CPU cycle as 1VCPU VM, because they have same weight 256. It doesn't make any sense to me, so we thought cap/weight is per VCPU. I'll get back to you soon with definite answer. Anthony > -----Original Message----- > From: Matthew Hartmann [mailto:[email protected]] > Sent: Thursday, June 14, 2012 7:33 AM > To: [email protected] > Subject: Re: CPU CAP XenServer > > Hello: > > I reported this issue as a bug way back when 2.2.13 was released and it > has yet to be resolved. It still exists as a bug even in 3.0.2. > > http://bugs.cloudstack.org/browse/CS-12972 > > Please feel free to submit your comments to the bug report. > > Cheers, > > Matthew > > > On 6/14/2012 10:26 AM, Diego Spinola Castro wrote: > > Hi, i found a issue and want know that it's a real issue. > > I've created a service offering with CPU CAP enabled and after > deployment > > i figured out that cap might be wrong. > > > > > > Host: Intel(R) Xeon(R) CPU E5-2630 0 @ 2.30GHz > > > > > > Service Offering: 4 x 2.0 GHz > > > > > > VM: VCPUs-params (MRW): weight: 220; cap: 86 > > VCPUs-max ( RW): 4 > > VCPUs-at-startup ( RW): 4 > > > > > > > > http://support.citrix.com/article/CTX117960 > > "The cap is expressed in percentage of one physical CPU: 100 is one > > physical CPU, 50 is half a CPU, 400 is 4 CPUs, and so on. The default, > 0 > > (zero), means there is no upper cap." > > > > > > I believe that cpu cap should be 344 instead 86. > > > > Any thoughts? > >
