cloudstack
cloudstack copied to clipboard
CPU overprovisioning factor doesn't seem to work
ISSUE TYPE
- Other
COMPONENT NAME
CPU overprovisioning
CLOUDSTACK VERSION
4.19.0.1
CONFIGURATION
OS / ENVIRONMENT
SUMMARY
Setting the overprovisioning factor on a cluster does not seem to work as expected.
STEPS TO REPRODUCE
Set the overprovisioning factor on the cluster to 1
Try to create vms totaling a higher number of cpus than the actual number the host has
EXPECTED RESULTS
Cloudstack should not let me create new vms with offerings that exceed the number of actual cpus.
ACTUAL RESULTS
I could create a new vm without any issues.
If we follow the documentation, setting a value of 1 to the cpu overprovisioning factor of the cluster should only let me create vms totaling the actual ammount of cpus of the host, not more. So if the host has 64 cpus (hyperthreading enabled), I shouldn't be able to create, let's say, 2 vms with 40 cpus each, yet I can, and the dashboard reflects this, showing more than 100% cpus allocated. I don't know if this is expected behavior or I'm supposed to do something more than just setting that value, what I am trying to do is limit the availability of the host cpus to the actual, real, number.
Thanks for opening your first issue here! Be sure to follow the issue template!
@tdtmusic2
If you have a single host with 64 vCPUs of 2000 MHz clock speed and you set CPU over the provisioning value of 1 at the cluster, cloudstack will calculate the total CPU capacity as 64 * 2000 * 1 = 128000. Cloudstack will allow you to deploy instances as long as their vCPU count* CPU MHZz value in the offering is less than the total (remaining capacity )value above.
@tdtmusic2 I tried to reproduce this issue on the latest 4.19 branch and it's working as expected. As @rajujith suggested, can you also check the speed of your hosts and the speed of the VMs you created for testing? If you use the same speed in the service offering as the host, then you won't be able to launch VMs using more cores than available on the host.
@tdtmusic2 I am closing the issue. If you still face this issue, please feel free to re-open.