12.9. Configure the Compute Node

The Telemetry service monitors each node by collecting usage data from the Compute agent (openstack-ceilometer-compute) installed on that node. You can configure a node's Compute agent by replicating the /etc/ceilometer/ceilometer.conf file from another host whose Telemetry components have already been configured.
You must also configure the Compute node itself to enable notifications.

Procedure 12.7. Enabling Notifications on a Compute Node

  1. Install openstack-ceilometer-compute, python-ceilometer and python-ceilometerclient on the node: 
    # yum install openstack-ceilometer-compute python-ceilometer python-ceilometerclient
  2. Enable auditing on the node: 
    # openstack-config --set /etc/nova/nova.conf \
   DEFAULT instance_usage_audit True
  3. Configure the audit frequency: 
    # openstack-config --set /etc/nova/nova.conf \
   DEFAULT instance_usage_audit_period hour
  4. Configure what type of state changes should trigger a notification: 
    # openstack-config --set /etc/nova/nova.conf \
   DEFAULT notify_on_state_change vm_and_task_state
  5. Set the node to use the correct notification drivers. Open the /etc/nova/nova.conf file in a text editor, and add the following lines in the DEFAULT section: 
    notification_driver = messagingv2
notification_driver = ceilometer.compute.nova_notifier
    The Compute node requires two different notification drivers, which are defined using the same configuration key. You cannot use openstack-config to set these values.
  6. Start the Compute agent: 
    # systemctl start openstack-ceilometer-compute.service
  7. Configure the agent to start at boot time: 
    # systemctl enable openstack-ceilometer-compute.service
  8. Restart the openstack-nova-compute service to apply all changes to /etc/nova/nova.conf: 
    # systemctl restart openstack-nova-compute.service