I've found more information on the host I have 2 sockets Processor Information Socket Designation: CPU1 Type: Central Processor Family: Xeon Manufacturer: Intel
Processor Information Socket Designation: CPU2 Type: Central Processor Family: Xeon According to /proc/cpuinfo I have 24 cpu's processor : 23 vendor_id : GenuineIntel cpu family : 6 model : 44 model name : Intel(R) Xeon(R) CPU X5660 @ 2.80GHz According to Cloudstack cpu count matches *Total CPU24 x 2.79 GHz**CPU Allocated for VMs97%* But *"The Number of CPU Sockets1"* On KVM counting vcpus and matching that one to the real CPU;s #of VCPUS living in - Physical CPU Number 4 - 0 1 - 1 2 - 2 2 - 3 1 - 4 6 - 5 2 - 6 1 - 10 6 - 12 5 - 14 1 - 17 1 - 22 For example 4 vcpus are living in CPU 0 so overprovisioning is working, now why is it that cloudstack sees this host as 97 % full when I'm using only *12 CPU's* Where are the rest of my CPU's ???? Any help would be really appreciate it Cheers On Tue, Jun 17, 2014 at 5:11 PM, Prashant Kumar Mishra < prashantkumar.mis...@citrix.com> wrote: > I am giving an example based on my understanding. > > Data > ------ > Host= 2 cpu ,2.3GHZ > Total capacity =2*2.3 GHZ > Total allocated: Total capacity assign to vms > > > 1-Without over provisioning you can have total cpu allocated not more > than total capacity . > 2-With overprovisioning x you can have total allocated x*total capacity , > 3-From the Cloudstack point of view If you are able to use total capacity > which is more than physical, it is over provisioned. > 4-From the kvm Point of view : I guess you are interested in Cloudstack > point of view so leaving it > > Thanks > Prashant > > -----Original Message----- > From: Ivan Rodriguez [mailto:ivan...@gmail.com] > Sent: Tuesday, June 17, 2014 11:55 AM > To: cloudstack-us...@incubator.apache.org > Subject: Re: Cloudstack 4.3 KVM CPU overprovisioning > > More information from my previous post, > > On this particular blade I only have 9 vm's running > > Total CPU24 x 2.79 GHzCPU Utilized3.9%CPU Allocated for VMs 97%Memory > Total126.02 > GBMemory Allocated53.50 GB Memory Used36.74 MBNetwork Read35.99 GBNetwork > Write > > [root@cs2-chas1-bl03 ~]# virsh list > Id Name State > ---------------------------------------------------- > 9 i-8-70-VM running > 11 i-8-75-VM running > 14 r-93-VM running > 15 i-8-72-VM running > 16 r-95-VM running > 17 i-3-84-VM running > 18 i-3-73-VM running > 19 i-3-91-VM running > 28 i-4-110-VM running > 29 i-9-112-VM running > 30 i-3-128-VM running > > My service offering is > > # of CPU Cores4CPU (in MHz)2.00 GHz > Where the number of CPU cores varies, I'm wondering if my service offering > is incorrect because on the # of cores in my understanding is the number of > vcpu's and the speed is lower than the real speed 2,000 Mhz > > Any ideas will be really appreciate it > > > > > Below is the virs vcpuinfo > > VCPU: 0 > CPU: 12 > State: running > CPU time: 11802.8s > CPU Affinity: yyyyyyyyyyyyyyyyyyyyyyyy > > VCPU: 1 > CPU: 12 > State: running > CPU time: 11317.7s > CPU Affinity: yyyyyyyyyyyyyyyyyyyyyyyy > > VCPU: 0 > CPU: 2 > State: running > CPU time: 115618.7s > CPU Affinity: yyyyyyyyyyyyyyyyyyyyyyyy > > VCPU: 1 > CPU: 2 > State: running > CPU time: 113330.9s > CPU Affinity: yyyyyyyyyyyyyyyyyyyyyyyy > > VCPU: 0 > CPU: 4 > State: running > CPU time: 3928.5s > CPU Affinity: yyyyyyyyyyyyyyyyyyyyyyyy > > VCPU: 1 > CPU: 0 > State: running > CPU time: 3459.6s > CPU Affinity: yyyyyyyyyyyyyyyyyyyyyyyy > > VCPU: 2 > CPU: 14 > State: running > CPU time: 3429.3s > CPU Affinity: yyyyyyyyyyyyyyyyyyyyyyyy > > VCPU: 3 > CPU: 12 > State: running > CPU time: 3808.8s > CPU Affinity: yyyyyyyyyyyyyyyyyyyyyyyy > > VCPU: 4 > CPU: 0 > State: running > CPU time: 3506.9s > CPU Affinity: yyyyyyyyyyyyyyyyyyyyyyyy > > VCPU: 5 > CPU: 2 > State: running > CPU time: 3695.5s > CPU Affinity: yyyyyyyyyyyyyyyyyyyyyyyy > > VCPU: 6 > CPU: 21 > State: running > CPU time: 4199.5s > CPU Affinity: yyyyyyyyyyyyyyyyyyyyyyyy > > VCPU: 7 > CPU: 0 > State: running > CPU time: 3633.2s > CPU Affinity: yyyyyyyyyyyyyyyyyyyyyyyy > > VCPU: 0 > CPU: 2 > State: running > CPU time: 650355.2s > CPU Affinity: yyyyyyyyyyyyyyyyyyyyyyyy > > VCPU: 1 > CPU: 14 > State: running > CPU time: 652099.4s > CPU Affinity: yyyyyyyyyyyyyyyyyyyyyyyy > > VCPU: 2 > CPU: 6 > State: running > CPU time: 644120.8s > CPU Affinity: yyyyyyyyyyyyyyyyyyyyyyyy > > VCPU: 3 > CPU: 1 > State: running > CPU time: 648950.4s > CPU Affinity: yyyyyyyyyyyyyyyyyyyyyyyy > > VCPU: 0 > CPU: 12 > State: running > CPU time: 10297.6s > CPU Affinity: yyyyyyyyyyyyyyyyyyyyyyyy > > VCPU: 1 > CPU: 0 > State: running > CPU time: 10297.3s > CPU Affinity: yyyyyyyyyyyyyyyyyyyyyyyy > > VCPU: 0 > CPU: 1 > State: running > CPU time: 1815.1s > CPU Affinity: yyyyyyyyyyyyyyyyyyyyyyyy > > VCPU: 1 > CPU: 1 > State: running > CPU time: 3389.8s > CPU Affinity: yyyyyyyyyyyyyyyyyyyyyyyy > > VCPU: 0 > CPU: 1 > State: running > CPU time: 409.7s > CPU Affinity: yyyyyyyyyyyyyyyyyyyyyyyy > > VCPU: 1 > CPU: 2 > State: running > CPU time: 463.9s > CPU Affinity: yyyyyyyyyyyyyyyyyyyyyyyy > > VCPU: 2 > CPU: 14 > State: running > CPU time: 602.6s > CPU Affinity: yyyyyyyyyyyyyyyyyyyyyyyy > > VCPU: 3 > CPU: 14 > State: running > CPU time: 429.9s > CPU Affinity: yyyyyyyyyyyyyyyyyyyyyyyy > > VCPU: 0 > CPU: 4 > State: running > CPU time: 10406.0s > CPU Affinity: yyyyyyyyyyyyyyyyyyyyyyyy > > VCPU: 1 > CPU: 2 > State: running > CPU time: 8263.1s > CPU Affinity: yyyyyyyyyyyyyyyyyyyyyyyy > > VCPU: 2 > CPU: 0 > State: running > CPU time: 8247.2s > CPU Affinity: yyyyyyyyyyyyyyyyyyyyyyyy > > VCPU: 3 > CPU: 0 > State: running > CPU time: 6616.0s > CPU Affinity: yyyyyyyyyyyyyyyyyyyyyyyy > > VCPU: 0 > CPU: 0 > State: running > CPU time: 33693.2s > CPU Affinity: yyyyyyyyyyyyyyyyyyyyyyyy > > VCPU: 1 > CPU: 16 > State: running > CPU time: 36008.6s > CPU Affinity: yyyyyyyyyyyyyyyyyyyyyyyy > > VCPU: 2 > CPU: 4 > State: running > CPU time: 36107.1s > CPU Affinity: yyyyyyyyyyyyyyyyyyyyyyyy > > VCPU: 3 > CPU: 16 > State: running > CPU time: 36210.2s > CPU Affinity: yyyyyyyyyyyyyyyyyyyyyyyy > > > > > On Tue, Jun 17, 2014 at 3:20 PM, Ivan Rodriguez <ivan...@gmail.com> wrote: > > > Dear Cloudstack users, > > > > Our current setup is using dell blades with 24 cpus on KVM, currently > > we have 5 servers like that(120 cpus) , and 28 Vm's currently running > > all of them linux centos with virtio modules the majority of the vm's > > have 2 vcpus per VM so that would be around 56 vcpus > > > > According to my Cloudstach Dashboard I'm already in 56 % usage of > > CPU's, it seems that CPU overprovision is not working on KVM, I've > > setup cloudstack in global preferences to have a 100 guests per hosts > > as the limit if I my understanding is correct then I should be able to > > provision around 500 vms on those 120 cpu's if I wanted to using cpu > > overprovision. > > > > How can I confirm if CPU overprovision is working as far as I can see > > cloudstack is detecting the right numper of cpu's per physical host > > > > > > Total CPU24 x 2.79 GHzCPU Utilized3.9%CPU Allocated for VMs 97%Memory > > Total126.02 GBMemory Allocated53.50 GB Memory Used36.74 MBNetwork > > Read35.99 GBNetwork Write Any help would be really appreciate it > > > > thanks > > > > > > >