iRMC driver¶

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 Automated cleaning

Configuration¶

In addition to the configuration for generic drivers to remote volume boot, 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.

Configuration¶

The Hardware Inspection Support in the iRMC driver requires the following configuration:

• It is necessary to set ironic configuration with gpu_ids and fpga_ids options in [irmc] section.

  gpu_ids and fpga_ids are lists of <vendorID>/<deviceID> where:

  • <vendorID>: 4 hexadecimal digits starts with ‘0x’.
  • <deviceID>: 4 hexadecimal digits starts with ‘0x’.

  Here are sample values for gpu_ids and fpga_ids:

  gpu_ids = 0x1000/0x0079,0x2100/0x0080
  fpga_ids = 0x1000/0x005b,0x1100/0x0180
  

• 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 and pysnmp can be downloaded from here.

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.

Inspection can also set/unset node’s traits with the following cpu type for iRMC driver:

• CUSTOM_CPU_FPGA: The bare metal contains fpga cpu type.

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 Capabilities discovery for more details and examples.

The operator can add a trait in compute service flavor, for example:

openstack baremetal node add trait $NODE_UUID CUSTOM_CPU_FPGA

A valid trait must be no longer than 255 characters. Standard traits are defined in the os_traits library. A custom trait must start with the prefix CUSTOM_ and use the following characters: A-Z, 0-9 and _.

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"
              ]
          }
      ]
  }
  

{
  "logical_disks":
    [
      {
        "size_gb": "MAX",
        "raid_level": "1+0"
      }
    ]
}

See RAID Configuration 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.

See RAID Configuration for more details and examples.

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 BIOS Configuration for more details and examples.