|
Hello! I am testing CloudStack 4.20.1.0 in our development environment, and I have encountered a problem. I configured the cpu.overprovisioning.factor to 4.0 in the cluster settings and 4 in the global settings. After making these changes, I restarted the cloudstack-management service. However, the host's available cores have not updated to reflect the new overprovisioning values. Did I miss any steps? 
|
Replies: 1 comment 3 replies
|
@leo79901 The most specific configuration takes precedence when set at different levels. Once you set the value at the Cluster level, that takes precedence over the global configuration. Updating the value at the cluster level does not require a restart. For this to take effect, you need to stop and start existing instances since they were created with the previous cp overcommit value. |
|
Thank you. I remember in previous versions, after modifying the CPU overcommit ratio, the number of CPUs and their frequency displayed on the host interface would change. Has this behavior changed in the new version? |
|
I noticed that if I allocate 16 GHz to a virtual machine, only 4 GHz is shown as allocated in the host interface. This might indicate that the new overcommit parameter causes the host interface to display less allocated usage, rather than showing the host's frequency as being higher. |
|
@leo79901 your observations are correct, but I am not sure whether the implementation has changed recently. Irrespectively, the result is the same. On the cluster level, I can see that the total capacity is doubled when I changed the CPU overprovisioning factor to 4.0 from 2.0. |
@leo79901 The most specific configuration takes precedence when set at different levels. Once you set the value at the Cluster level, that takes precedence over the global configuration. Updating the value at the cluster level does not require a restart. For this to take effect, you need to stop and start existing instances since they were created with the previous cp overcommit value.