Enabling CPU metrics in OpenStack

Document created by dz_ny Expert on Jan 3, 2017Last modified by vivek.nandavanam on Jun 9, 2017
Version 5Show Document
  • View in full screen mode

VM CPU

Ceilometer out of the box is configured to start publishing CPU metrics [cpu_util] for the virtual machines.

 

Host CPU

Refer the following OpenStack documentation to see a list of all the available configuration options for Nova:

 

Icehouse - nova.conf - configuration options - OpenStack Configuration Reference  - icehouse 

Juno - nova.conf - configuration options - OpenStack Configuration Reference  - juno 

Kilo - nova.conf - configuration options - OpenStack Configuration Reference  - kilo 

Liberty - nova.conf - configuration options - OpenStack Configuration Reference  - liberty 

Mitaka - OpenStack Docs: nova.conf - configuration options 

Newton - OpenStack Docs: nova.conf - configuration options 

Ocata - OpenStack Docs: nova.conf - configuration options 

 

The one we will be interested in will be 'compute_monitors' to enable the compute driver to get the CPU metrics. The following entry [Based on the OpenStack version] needs to be added to the /etc/nova/nova.conf on the Compute nodes in DEFAULT section:

Icehouse and Juno

compute_monitors = ComputeDriverCPUMonitor,ComputeDriverMemoryMonitor

Kilo

compute_monitors = ComputeDriverCPUMonitor

Mitaka and above

compute_monitors = nova.compute.monitors.cpu.virt_driver

 

Restarting the nova compute service should start pushing the host CPU metrics to the controller.

     systemctl restart openstack-nova-compute.service

2 people found this helpful

Attachments

    Outcomes