openstack
/
ironic
Code Issues Proposed changes
ironic/doc/source/admin/drivers/irmc.rst

566 lines
21 KiB
ReStructuredText
Raw Blame History

This file contains ambiguous Unicode characters

This file contains Unicode characters that might be confused with other characters. If you think that this is intentional, you can safely ignore this warning. Use the Escape button to reveal them.

.. _irmc:
===========
iRMC driver
===========
Overview
========
The iRMC driver enables control FUJITSU PRIMERGY via ServerView
Common Command Interface (SCCI). Support for FUJITSU PRIMERGY servers consists
of the ``irmc`` hardware type and a few hardware interfaces specific for that
hardware type.
Prerequisites
=============
* Install `python-scciclient <https://pypi.org/project/python-scciclient>`_
and `pysnmp <https://pypi.org/project/pysnmp>`_ packages::
$ pip install "python-scciclient>=0.7.2" pysnmp
Hardware Type
=============
The ``irmc`` hardware type is available for FUJITSU PRIMERGY servers. For
information on how to enable the ``irmc`` hardware type, see
:ref:`enable-hardware-types`.
Hardware interfaces
^^^^^^^^^^^^^^^^^^^
The ``irmc`` hardware type overrides the selection of the following
hardware interfaces:
* bios
Supports ``irmc`` and ``no-bios``.
The default is ``irmc``.
* boot
Supports ``irmc-virtual-media``, ``irmc-pxe``, and ``pxe``.
The default is ``irmc-virtual-media``. The ``irmc-virtual-media`` boot
interface enables the virtual media based deploy with IPA (Ironic Python
Agent).
.. warning::
We deprecated the ``pxe`` boot interface when used with ``irmc``
hardware type. Support for this interface will be removed in the
future. Instead, use ``irmc-pxe``. ``irmc-pxe`` boot interface
was introduced in Pike.
* console
Supports ``ipmitool-socat``, ``ipmitool-shellinabox``, and ``no-console``.
The default is ``ipmitool-socat``.
* inspect
Supports ``irmc``, ``inspector``, and ``no-inspect``.
The default is ``irmc``.
.. note::
`Ironic Inspector <https://docs.openstack.org/ironic-inspector/latest/>`_
needs to be present and configured to use ``inspector`` as the
inspect interface.
* management
Supports only ``irmc``.
* power
Supports ``irmc``, which enables power control via ServerView Common
Command Interface (SCCI), by default. Also supports ``ipmitool``.
* raid
Supports ``irmc``, ``no-raid`` and ``agent``.
The default is ``no-raid``.
For other hardware interfaces, ``irmc`` hardware type supports the
Bare Metal reference interfaces. For more details about the hardware
interfaces and how to enable the desired ones, see
:ref:`enable-hardware-interfaces`.
Here is a complete configuration example with most of the supported hardware
interfaces enabled for ``irmc`` hardware type.
.. code-block:: ini
[DEFAULT]
enabled_hardware_types = irmc
enabled_bios_interfaces = irmc
enabled_boot_interfaces = irmc-virtual-media,irmc-pxe
enabled_console_interfaces = ipmitool-socat,ipmitool-shellinabox,no-console
enabled_deploy_interfaces = iscsi,direct
enabled_inspect_interfaces = irmc,inspector,no-inspect
enabled_management_interfaces = irmc
enabled_network_interfaces = flat,neutron
enabled_power_interfaces = irmc
enabled_raid_interfaces = no-raid,irmc
enabled_storage_interfaces = noop,cinder
enabled_vendor_interfaces = no-vendor,ipmitool
Here is a command example to enroll a node with ``irmc`` hardware type.
.. code-block:: console
openstack baremetal node create \
--bios-interface irmc \
--boot-interface irmc-pxe \
--deploy-interface direct \
--inspect-interface irmc \
--raid-interface irmc
Node configuration
^^^^^^^^^^^^^^^^^^
* Each node is configured for ``irmc`` hardware type by setting the following
ironic node object’s properties:
- ``driver_info/irmc_address`` property to be ``IP address`` or
``hostname`` of the iRMC.
- ``driver_info/irmc_username`` property to be ``username`` for
the iRMC with administrator privileges.
- ``driver_info/irmc_password`` property to be ``password`` for
irmc_username.
- ``properties/capabilities`` property to be ``boot_mode:uefi`` if
UEFI boot is required.
- ``properties/capabilities`` property to be ``secure_boot:true`` if
UEFI Secure Boot is required. Please refer to `UEFI Secure Boot Support`_
for more information.
* The following properties are also required if ``irmc-virtual-media`` boot
interface is used:
- ``driver_info/irmc_deploy_iso`` property to be either deploy iso
file name, Glance UUID, or Image Service URL.
- ``instance info/irmc_boot_iso`` property to be either boot iso
file name, Glance UUID, or Image Service URL. This is optional
property when ``boot_option`` is set to ``netboot``.
* All of the nodes are configured by setting the following configuration
options in the ``[irmc]`` section of ``/etc/ironic/ironic.conf``:
- ``port``: Port to be used for iRMC operations; either 80
or 443. The default value is 443. Optional.
- ``auth_method``: Authentication method for iRMC operations;
either ``basic`` or ``digest``. The default value is ``basic``. Optional.
- ``client_timeout``: Timeout (in seconds) for iRMC
operations. The default value is 60. Optional.
- ``sensor_method``: Sensor data retrieval method; either
``ipmitool`` or ``scci``. The default value is ``ipmitool``. Optional.
* The following options are required if ``irmc-virtual-media`` boot
interface is enabled:
- ``remote_image_share_root``: Ironic conductor node's ``NFS`` or
``CIFS`` root path. The default value is ``/remote_image_share_root``.
- ``remote_image_server``: IP of remote image server.
- ``remote_image_share_type``: Share type of virtual media, either
``NFS`` or ``CIFS``. The default is ``CIFS``.
- ``remote_image_share_name``: share name of ``remote_image_server``.
The default value is ``share``.
- ``remote_image_user_name``: User name of ``remote_image_server``.
- ``remote_image_user_password``: Password of ``remote_image_user_name``.
- ``remote_image_user_domain``: Domain name of ``remote_image_user_name``.
* The following options are required if ``irmc`` inspect interface is enabled:
- ``snmp_version``: SNMP protocol version; either ``v1``, ``v2c`` or
``v3``. The default value is ``v2c``. Optional.
- ``snmp_port``: SNMP port. The default value is ``161``. Optional.
- ``snmp_community``: SNMP community required for versions ``v1``
and ``v2c``. The default value is ``public``. Optional.
- ``snmp_security``: SNMP security name required for version ``v3``.
Optional.
* Each node can be further configured by setting the following ironic
node object’s properties which override the parameter values in
``[irmc]`` section of ``/etc/ironic/ironic.conf``:
- ``driver_info/irmc_port`` property overrides ``port``.
- ``driver_info/irmc_auth_method`` property overrides ``auth_method``.
- ``driver_info/irmc_client_timeout`` property overrides ``client_timeout``.
- ``driver_info/irmc_sensor_method`` property overrides ``sensor_method``.
- ``driver_info/irmc_snmp_version`` property overrides ``snmp_version``.
- ``driver_info/irmc_snmp_port`` property overrides ``snmp_port``.
- ``driver_info/irmc_snmp_community`` property overrides ``snmp_community``.
- ``driver_info/irmc_snmp_security`` property overrides ``snmp_security``.
Optional functionalities for the ``irmc`` hardware type
=======================================================
UEFI Secure Boot Support
^^^^^^^^^^^^^^^^^^^^^^^^
The hardware type ``irmc`` supports secure boot deploy.
.. warning::
Secure boot feature is not supported with ``pxe`` boot interface.
The UEFI secure boot can be configured by adding ``secure_boot`` parameter,
which is a boolean value. Enabling the secure boot is different when
Bare Metal service is used with Compute service or without Compute service. The
following sections describe both methods:
* Enabling secure boot with Compute service:
To enable secure boot we need to set a capability on the bare metal node
and the bare metal flavor, for example::
openstack baremetal node set <node-uuid> --property capabilities='secure_boot:true'
openstack flavor set FLAVOR-NAME --property capabilities:secure_boot="true"
* Enabling secure boot without Compute service:
Since adding capabilities to the node's properties is only used by the nova
scheduler to perform more advanced scheduling of instances, we need
to enable secure boot without nova, for example::
openstack baremetal node set <node-uuid> --instance-info capabilities='{"secure_boot": "true"}'
.. _irmc_node_cleaning:
Node Cleaning Support
^^^^^^^^^^^^^^^^^^^^^
The ``irmc`` hardware type supports node cleaning.
For more information on node cleaning, see :ref:`cleaning`.
Supported **Automated** Cleaning Operations
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
The automated cleaning operations supported are:
* ``restore_irmc_bios_config``:
Restores BIOS settings on a baremetal node from backup data. If this
clean step is enabled, the BIOS settings of a baremetal node will be
backed up automatically before the deployment. By default, this clean
step is disabled with priority ``0``. Set its priority to a positive
integer to enable it. The recommended value is ``10``.
.. warning::
``pxe`` boot interface, when used with ``irmc`` hardware type, does
not support this clean step. If uses ``irmc`` hardware type, it is
required to select ``irmc-pxe`` or ``irmc-virtual-media`` as the
boot interface in order to make this clean step work.
Configuration options for the automated cleaning steps are listed under
``[irmc]`` section in ironic.conf ::
clean_priority_restore_irmc_bios_config = 0
For more information on node automated cleaning, see :ref:`automated_cleaning`
Boot from Remote Volume
^^^^^^^^^^^^^^^^^^^^^^^
The ``irmc`` hardware type supports the generic iPXE-based remote volume
booting when using the following boot interfaces:
* ``irmc-pxe``
* ``pxe``
In addition, the ``irmc`` hardware type supports remote volume booting without
iPXE. This is available when using the ``irmc-virtual-media`` boot interface.
This feature configures a node to boot from a remote volume by using the API
of iRMC. It supports iSCSI and FibreChannel.
Configuration
~~~~~~~~~~~~~
In addition to the configuration for generic drivers to
:ref:`remote volume boot <boot-from-volume>`,
the iRMC driver requires the following configuration:
* It is necessary to set physical port IDs to network ports and volume
connectors. All cards including those not used for volume boot should be
registered.
The format of a physical port ID is: ``<Card Type><Slot No>-<Port No>`` where:
- ``<Card Type>``: could be ``LAN``, ``FC`` or ``CNA``
- ``<Slot No>``: 0 indicates onboard slot. Use 1 to 9 for add-on slots.
- ``<Port No>``: A port number starting from 1.
These IDs are specified in a node's ``driver_info[irmc_pci_physical_ids]``.
This value is a dictionary. The key is the UUID of a resource (Port or Volume
Connector) and its value is the physical port ID. For example::
{
"1ecd14ee-c191-4007-8413-16bb5d5a73a2":"LAN0-1",
"87f6c778-e60e-4df2-bdad-2605d53e6fc0":"CNA1-1"
}
It can be set with the following command::
openstack baremetal node set $NODE_UUID \
--driver-info irmc_pci_physical_ids={} \
--driver-info irmc_pci_physical_ids/$PORT_UUID=LAN0-1 \
--driver-info irmc_pci_physical_ids/$VOLUME_CONNECTOR_UUID=CNA1-1
* For iSCSI boot, volume connectors with both types ``iqn`` and ``ip`` are
required. The configuration with DHCP is not supported yet.
* For iSCSI, the size of the storage network is needed. This value should be
specified in a node's ``driver_info[irmc_storage_network_size]``. It must be
a positive integer < 32.
For example, if the storage network is 10.2.0.0/22, use the following
command::
openstack baremetal node set $NODE_UUID --driver-info irmc_storage_network_size=22
Supported hardware
~~~~~~~~~~~~~~~~~~
The driver supports the PCI controllers, Fibrechannel Cards, Converged Network
Adapters supported by
`Fujitsu ServerView Virtual-IO Manager <http://www.fujitsu.com/fts/products/computing/servers/primergy/management/primergy-blade-server-io-virtualization.html>`_.
Hardware Inspection Support
^^^^^^^^^^^^^^^^^^^^^^^^^^^
The ``irmc`` hardware type provides the iRMC-specific hardware inspection
with ``irmc`` inspect interface.
.. note::
SNMP requires being enabled in ServerView® iRMC S4 Web Server(Network
Settings\SNMP section).
Configuration
~~~~~~~~~~~~~
The Hardware Inspection Support in the iRMC driver requires the following
configuration:
* It is necessary to set ironic configuration with ``gpu_ids`` option
in ``[irmc]`` section.
``gpu_ids`` is a list of ``<vendorID>/<deviceID>`` where:
- ``<vendorID>``: 4 hexadecimal digits starts with '0x'.
- ``<deviceID>``: 4 hexadecimal digits starts with '0x'.
Here is a sample value for gpu_ids::
gpu_ids = 0x1000/0x0079,0x2100/0x0080
* The python-scciclient package requires pyghmi version >= 1.0.22 and pysnmp
version >= 4.2.3. They are used by the conductor service on the conductor.
The latest version of pyghmi can be downloaded from `here
<https://pypi.org/project/pyghmi/>`__
and pysnmp can be downloaded from `here
<https://pypi.org/project/pysnmp/>`__.
Supported properties
~~~~~~~~~~~~~~~~~~~~
The inspection process will discover the following essential properties
(properties required for scheduling deployment):
* ``memory_mb``: memory size
* ``cpus``: number of cpus
* ``cpu_arch``: cpu architecture
* ``local_gb``: disk size
Inspection can also discover the following extra capabilities for iRMC
driver:
* ``irmc_firmware_version``: iRMC firmware version
* ``rom_firmware_version``: ROM firmware version
* ``trusted_boot``: The flag whether TPM(Trusted Platform Module) is
supported by the server. The possible values are 'True' or 'False'.
* ``server_model``: server model
* ``pci_gpu_devices``: number of gpu devices connected to the bare metal.
.. note::
* The disk size is returned only when eLCM License for FUJITSU PRIMERGY
servers is activated. If the license is not activated, then Hardware
Inspection will fail to get this value.
* Before inspecting, if the server is power-off, it will be turned on
automatically. System will wait for a few second before start
inspecting. After inspection, power status will be restored to the
previous state.
The operator can specify these capabilities in compute service flavor, for
example::
openstack flavor set baremetal-flavor-name --property capabilities:irmc_firmware_version="iRMC S4-8.64F"
openstack flavor set baremetal-flavor-name --property capabilities:server_model="TX2540M1F5"
openstack flavor set baremetal-flavor-name --property capabilities:pci_gpu_devices="1"
See :ref:`capabilities-discovery` for more details and examples.
RAID configuration Support
^^^^^^^^^^^^^^^^^^^^^^^^^^
The ``irmc`` hardware type provides the iRMC RAID configuration with ``irmc``
raid interface.
.. note::
* RAID implementation for ``irmc`` hardware type is based on eLCM license
and SDCard. Otherwise, SP(Service Platform) in lifecycle management
must be available.
* RAID implementation only supported for RAIDAdapter 0 in Fujitsu Servers.
Configuration
~~~~~~~~~~~~~
The RAID configuration Support in the iRMC drivers requires the following
configuration:
* It is necessary to set ironic configuration into Node with
JSON file option::
$ openstack baremetal node set <node-uuid-or-name> \
--target-raid-config <JSON file containing target RAID configuration>
Here is some sample values for JSON file::
{
"logical_disks": [
{
"size_gb": 1000,
"raid_level": "1"
]
}
or::
{
"logical_disks": [
{
"size_gb": 1000,
"raid_level": "1",
"controller": "FTS RAID Ctrl SAS 6G 1GB (D3116C) (0)",
"physical_disks": [
"0",
"1"
]
}
]
}
.. note::
RAID 1+0 and 5+0 in iRMC driver does not support property ``physical_disks``
in ``target_raid_config`` during create raid configuration yet. See
following example::
{
"logical_disks":
[
{
"size_gb": "MAX",
"raid_level": "1+0"
}
]
}
See :ref:`raid` for more details and examples.
Supported properties
~~~~~~~~~~~~~~~~~~~~
The RAID configuration using iRMC driver supports following parameters in
JSON file:
* ``size_gb``: is mandatory properties in Ironic.
* ``raid_level``: is mandatory properties in Ironic. Currently, iRMC Server
supports following RAID levels: 0, 1, 5, 6, 1+0 and 5+0.
* ``controller``: is name of the controller as read by the RAID interface.
* ``physical_disks``: are specific values for each raid array in
LogicalDrive which operator want to set them along with ``raid_level``.
The RAID configuration is supported as a manual cleaning step.
.. note::
* iRMC server will power-on after create/delete raid configuration is
applied, FGI (Foreground Initialize) will process raid configuration in
iRMC server, thus the operation will completed upon power-on and power-off
when created RAID on iRMC server.
See :ref:`raid` for more details and examples.
BIOS configuration Support
^^^^^^^^^^^^^^^^^^^^^^^^^^
The ``irmc`` hardware type provides the iRMC BIOS configuration with ``irmc``
bios interface.
.. warning::
``irmc`` bios interface does not support ``factory_reset``.
Configuration
~~~~~~~~~~~~~
The BIOS configuration in the iRMC driver supports the following settings:
- ``boot_option_filter``: Specifies from which drives can be booted. This
supports following options: ``UefiAndLegacy``, ``LegacyOnly``, ``UefiOnly``.
- ``check_controllers_health_status_enabled``: The UEFI FW checks the
controller health status. This supports following options: ``true``, ``false``.
- ``cpu_active_processor_cores``: The number of active processor cores 1...n.
Option 0 indicates that all available processor cores are active.
- ``cpu_adjacent_cache_line_prefetch_enabled``: The processor loads the requested
cache line and the adjacent cache line. This supports following options:
``true``, ``false``.
- ``cpu_vt_enabled``: Supports the virtualization of platform hardware and
several software environments, based on Virtual Machine Extensions to
support the use of several software environments using virtual computers.
This supports following options: ``true``, ``false``.
- ``flash_write_enabled``: The system BIOS can be written. Flash BIOS update
is possible. This supports following options: ``true``, ``false``.
- ``hyper_threading_enabled``: Hyper-threading technology allows a single
physical processor core to appear as several logical processors. This
supports following options: ``true``, ``false``.
- ``keep_void_boot_options_enabled``: Boot Options will not be removed from
"Boot Option Priority" list. This supports following options: ``true``,
``false``.
- ``launch_csm_enabled``: Specifies whether the Compatibility Support Module
(CSM) is executed. This supports following options: ``true``, ``false``.
- ``os_energy_performance_override_enabled``: Prevents the OS from overruling
any energy efficiency policy setting of the setup. This supports following
options: ``true``, ``false``.
- ``pci_aspm_support``: Active State Power Management (ASPM) is used to
power-manage the PCI Express links, thus consuming less power. This
supports following options: ``Disabled``, ``Auto``, ``L0Limited``,
``L1only``, ``L0Force``.
- ``pci_above_4g_decoding_enabled``: Specifies if memory resources above the
4GB address boundary can be assigned to PCI devices. This supports
following options: ``true``, ``false``.
- ``power_on_source``: Specifies whether the switch on sources for the system
are managed by the BIOS or the ACPI operating system. This supports
following options: ``BiosControlled``, ``AcpiControlled``.
- ``single_root_io_virtualization_support_enabled``: Single Root IO
Virtualization Support is active. This supports following
options: ``true``, ``false``.
The BIOS configuration is supported as a manual cleaning step. See :ref:`bios`
for more details and examples.
Supported platforms
===================
This driver supports FUJITSU PRIMERGY BX S4 or RX S8 servers and above.
- PRIMERGY BX920 S4
- PRIMERGY BX924 S4
- PRIMERGY RX300 S8
When ``irmc`` power interface is used, Soft Reboot (Graceful Reset) and Soft
Power Off (Graceful Power Off) are only available if
`ServerView agents <http://manuals.ts.fujitsu.com/index.php?id=5406-5873-5925-5945-16159>`_
are installed. See `iRMC S4 Manual <http://manuals.ts.fujitsu.com/index.php?id=5406-5873-5925-5988>`_
for more details.
RAID configuration feature supports FUJITSU PRIMERGY servers with
RAID-Ctrl-SAS-6G-1GB(D3116C) controller and above.
For detail supported controller with OOB-RAID configuration, please see
`the whitepaper for iRMC RAID configuration <https://sp.ts.fujitsu.com/dmsp/Publications/public/wp-SVS-ooB-RAID-HDD-en.pdf>`_.
View Git Blame Copy Permalink