Appendix B. Power Management Drivers

Although IPMI is the main method the director uses for power management control, the director also supports other power management types. This appendix provides a list of the supported power management features. Use these power management settings for Section 5.1, “Registering Nodes for the Overcloud”.

B.1. Dell Remote Access Controller (DRAC)

DRAC is an interface that provides out-of-band remote management features including power management and server monitoring.

pm_type
Set this option to pxe_drac.
pm_user; pm_password
The DRAC username and password.
pm_addr
The IP address of the DRAC host.

B.2. Integrated Lights-Out (iLO)

iLO from Hewlett-Packard is an interface that provides out-of-band remote management features including power management and server monitoring.

pm_type
Set this option to pxe_ilo.
pm_user; pm_password
The iLO username and password.
pm_addr

The IP address of the iLO interface.

  • Edit the /etc/ironic/ironic.conf file and add pxe_ilo to the enabled_drivers option to enable this driver.
  • The director also requires an additional set of utilities for iLo. Install the python-proliantutils package and restart the openstack-ironic-conductor service:

    $ sudo yum install python-proliantutils
    $ sudo systemctl restart openstack-ironic-conductor.service
  • HP nodes must a 2015 firmware version for successful introspection. The director has been successfully tested with nodes using firmware version 1.85 (May 13 2015).
  • Using a shared iLO port is not supported.

B.3. Cisco Unified Computing System (UCS)

UCS from Cisco is a data center platform that unites compute, network, storage access, and virtualization resources. This driver focuses on the power management for bare metal systems connected to the UCS.

pm_type
Set this option to pxe_ucs.
pm_user; pm_password
The UCS username and password.
pm_addr
The IP address of the UCS interface.
pm_service_profile

The UCS service profile to use. Usually takes the format of org-root/ls-[service_profile_name]. For example:

"pm_service_profile": "org-root/ls-Nova-1"
  • Edit the /etc/ironic/ironic.conf file and add pxe_ucs to the enabled_drivers option to enable this driver.
  • The director also requires an additional set of utilities for UCS. Install the python-UcsSdk package and restart the openstack-ironic-conductor service:

    $ sudo yum install python-UcsSdk
    $ sudo systemctl restart openstack-ironic-conductor.service

B.4. Fujitsu Integrated Remote Management Controller (iRMC)

Fujitsu’s iRMC is a Baseboard Management Controller (BMC) with integrated LAN connection and extended functionality. This driver focuses on the power management for bare metal systems connected to the iRMC.

Important

iRMC S4 or higher is required.

pm_type
Set this option to pxe_irmc.
pm_user; pm_password
The username and password for the iRMC interface.
pm_addr
The IP address of the iRMC interface.
pm_port (Optional)
The port to use for iRMC operations. The default is 443.
pm_auth_method (Optional)
The authentication method for iRMC operations. Use either basic or digest. The default is basic
pm_client_timeout (Optional)
Timeout (in seconds) for iRMC operations. The default is 60 seconds.
pm_sensor_method (Optional)

Sensor data retrieval method. Use either ipmitool or scci. The default is ipmitool.

  • Edit the /etc/ironic/ironic.conf file and add pxe_irmc to the enabled_drivers option to enable this driver.
  • The director also requires an additional set of utilities if you enabled SCCI as the sensor method. Install the python-scciclient package and restart the openstack-ironic-conductor service:

    $ yum install python-scciclient
    $ sudo systemctl restart openstack-ironic-conductor.service

B.5. SSH and Virsh

The director can access a host running libvirt through SSH and use virtual machines as nodes. The director uses virsh to control the power management of these nodes.

Important

This option is available for testing and evaluation purposes only. It is not recommended for Red Hat OpenStack Platform enterprise environments.

pm_type
Set this option to pxe_ssh.
pm_user; pm_password

The SSH username and contents of the SSH private key. If using the CLI tools to register your nodes, the private key must be on one line with new lines replaced with escape characters (\n). For example:

-----BEGIN RSA PRIVATE KEY-----\nMIIEogIBAAKCAQEA .... kk+WXt9Y=\n-----END RSA PRIVATE KEY-----

Add the SSH public key to the libvirt server’s authorized_keys collection.

pm_addr

The IP address of the virsh host.

  • The server hosting libvirt requires an SSH key pair with the public key set as the pm_password attribute.
  • Ensure the chosen pm_user has full access to the libvirt environment.

B.6. Fake PXE Driver

This driver provides a method to use bare metal devices without power management. This means the director does not control the registered bare metal devices and as such require manual control of power at certain points in the introspect and deployment processes.

Important

This option is available for testing and evaluation purposes only. It is not recommended for Red Hat OpenStack Platform enterprise environments.

pm_type

Set this option to fake_pxe.

  • This driver does not use any authentication details because it does not control power management.
  • Edit the /etc/ironic/ironic.conf file and add fake_pxe to the enabled_drivers option to enable this driver. Restart the baremetal services after editing the file:

    $ sudo systemctl restart openstack-ironic-api openstack-ironic-conductor
  • When performing introspection on nodes, manually power the nodes after running the openstack baremetal introspection bulk start command.
  • When performing overcloud deployment, check the node status with the ironic node-list command. Wait until the node status changes from deploying to deploy wait-callback and then manually power the nodes.
  • After the overcloud provisioning process completes, reboot the nodes. To check the completion of provisioning, check the node status with the ironic node-list command, wait until the node status changes to active, then manually reboot all overcloud nodes.