iLO driver¶
Overview¶
iLO driver enables to take advantage of features of iLO management engine in
HPE ProLiant servers. The ilo
hardware type is targeted for HPE ProLiant
Gen8 and Gen9 systems which have iLO 4 management engine. From Pike
release ilo
hardware type supports ProLiant Gen10 systems which have
iLO 5 management engine. iLO5 conforms to Redfish API and hence hardware
type redfish
(see Redfish driver) is also an option for this kind of
hardware but it lacks the iLO specific features.
For more details and for up-to-date information (like tested platforms, known issues, etc), please check the iLO driver wiki page.
For enabling Gen10 systems and getting detailed information on Gen10 feature support in Ironic please check this Gen10 wiki section.
Hardware type¶
ProLiant hardware is primarily supported by the ilo
hardware type. ilo5
hardware type is only supported on ProLiant Gen10 and later systems. Both
hardware can be used with reference hardware type ipmi
(see
IPMI driver) and redfish
(see Redfish driver). For information on how
to enable the ilo
and ilo5
hardware type, see
Enabling hardware types.
Note
Only HPE ProLiant Gen10 servers supports hardware type redfish
.
Warning
It is important to note that while the HPE Edgeline series of servers may
contain iLO adapters, they are known to not be compatible with the ilo
hardware type. The redfish
hardware type should be used instead.
The hardware type ilo
supports following HPE server features:
Update Minimum Password Length security parameter as manual clean step
Update Authentication Failure Logging security parameter as manual clean step
Create Certificate Signing Request(CSR) as manual clean step
Apart from above features hardware type ilo5
also supports following
features:
Hardware interfaces¶
The ilo
hardware type supports following hardware interfaces:
- bios
Supports
ilo
andno-bios
. The default isilo
. They can be enabled by using the[DEFAULT]enabled_bios_interfaces
option inironic.conf
as given below:[DEFAULT] enabled_hardware_types = ilo enabled_bios_interfaces = ilo,no-bios
- boot
Supports
ilo-virtual-media
,ilo-pxe
andilo-ipxe
. The default isilo-virtual-media
. Theilo-virtual-media
interface provides security enhanced PXE-less deployment by using iLO virtual media to boot up the bare metal node. Theilo-pxe
andilo-ipxe
interfaces use PXE and iPXE respectively for deployment(just like PXE boot). These interfaces do not require iLO Advanced license. They can be enabled by using the[DEFAULT]enabled_boot_interfaces
option inironic.conf
as given below:[DEFAULT] enabled_hardware_types = ilo enabled_boot_interfaces = ilo-virtual-media,ilo-pxe,ilo-ipxe
- console
Supports
ilo
andno-console
. The default isilo
. They can be enabled by using the[DEFAULT]enabled_console_interfaces
option inironic.conf
as given below:[DEFAULT] enabled_hardware_types = ilo enabled_console_interfaces = ilo,no-console
- inspect
Supports
ilo
andinspector
. The default isilo
. They can be enabled by using the[DEFAULT]enabled_inspect_interfaces
option inironic.conf
as given below:[DEFAULT] enabled_hardware_types = ilo enabled_inspect_interfaces = ilo,inspector
Note
Ironic Inspector needs to be configured to use
inspector
as the inspect interface.
- management
Supports only
ilo
. It can be enabled by using the[DEFAULT]enabled_management_interfaces
option inironic.conf
as given below:[DEFAULT] enabled_hardware_types = ilo enabled_management_interfaces = ilo
- power
Supports only
ilo
. It can be enabled by using the[DEFAULT]enabled_power_interfaces
option inironic.conf
as given below:[DEFAULT] enabled_hardware_types = ilo enabled_power_interfaces = ilo
- raid
Supports
agent
andno-raid
. The default isno-raid
. They can be enabled by using the[DEFAULT]enabled_raid_interfaces
option inironic.conf
as given below:[DEFAULT] enabled_hardware_types = ilo enabled_raid_interfaces = agent,no-raid
- storage
Supports
cinder
andnoop
. The default isnoop
. They can be enabled by using the[DEFAULT]enabled_storage_interfaces
option inironic.conf
as given below:[DEFAULT] enabled_hardware_types = ilo enabled_storage_interfaces = cinder,noop
Note
The storage interface
cinder
is supported only when corresponding boot interface of theilo
hardware type based node isilo-pxe
orilo-ipxe
. Please refer to Boot From Volume for configuringcinder
as a storage interface.
- rescue
Supports
agent
andno-rescue
. The default isno-rescue
. They can be enabled by using the[DEFAULT]enabled_rescue_interfaces
option inironic.conf
as given below:[DEFAULT] enabled_hardware_types = ilo enabled_rescue_interfaces = agent,no-rescue
- vendor
Supports
ilo
,ilo-redfish
andno-vendor
. The default isilo
. They can be enabled by using the[DEFAULT]enabled_vendor_interfaces
option inironic.conf
as given below:[DEFAULT] enabled_hardware_types = ilo enabled_vendor_interfaces = ilo,ilo-redfish,no-vendor
The ilo5
hardware type supports all the ilo
interfaces described above,
except for boot
and raid
interfaces. The details of boot
and
raid
interfaces is as under:
- raid
Supports
ilo5
andno-raid
. The default isilo5
. They can be enabled by using the[DEFAULT]enabled_raid_interfaces
option inironic.conf
as given below:[DEFAULT] enabled_hardware_types = ilo5 enabled_raid_interfaces = ilo5,no-raid
- boot
Supports
ilo-uefi-https
apart from the other boot interfaces supported byilo
hardware type. This can be enabled by using the[DEFAULT]enabled_boot_interfaces
option inironic.conf
as given below:[DEFAULT] enabled_hardware_types = ilo5 enabled_boot_interfaces = ilo-uefi-https,ilo-virtual-media
The ilo
and ilo5
hardware type support all standard deploy
and
network
interface implementations, see Enabling hardware interfaces
for details.
The following command can be used to enroll a ProLiant node with
ilo
hardware type:
baremetal node create \
--driver ilo \
--deploy-interface direct \
--raid-interface agent \
--rescue-interface agent \
--driver-info ilo_address=<ilo-ip-address> \
--driver-info ilo_username=<ilo-username> \
--driver-info ilo_password=<ilo-password> \
--driver-info deploy_iso=<glance-uuid-of-deploy-iso> \
--driver-info rescue_iso=<glance-uuid-of-rescue-iso>
Note
The fields deploy_iso
and rescue_iso
used to be called
ilo_deploy_iso
and ilo_rescue_iso
before the Xena release.
The following command can be used to enroll a ProLiant node with
ilo5
hardware type:
baremetal node create \
--driver ilo5 \
--deploy-interface direct \
--raid-interface ilo5 \
--rescue-interface agent \
--driver-info ilo_address=<ilo-ip-address> \
--driver-info ilo_username=<ilo-username> \
--driver-info ilo_password=<ilo-password> \
--driver-info deploy_iso=<glance-uuid-of-deploy-iso> \
--driver-info rescue_iso=<glance-uuid-of-rescue-iso>
Please refer to Enabling drivers and hardware types for detailed explanation of hardware type.
Node configuration¶
Each node is configured for
ilo
andilo5
hardware type by setting the following ironic node object’s properties indriver_info
:ilo_address
: IP address or hostname of the iLO.ilo_username
: Username for the iLO with administrator privileges.ilo_password
: Password for the above iLO user.client_port
: (optional) Port to be used for iLO operations if you are using a custom port on the iLO. Default port used is 443.client_timeout
: (optional) Timeout for iLO operations. Default timeout is 60 seconds.ca_file
: (optional) CA certificate file to validate iLO.console_port
: (optional) Node’s UDP port for console access. Any unused port on the ironic conductor node may be used. This is required only whenilo-console
interface is used.
The following properties are also required in node object’s
driver_info
ifilo-virtual-media
boot interface is used:deploy_iso
: The glance UUID of the deploy ramdisk ISO image.instance info/boot_iso
property to be either boot iso Glance UUID or a HTTP(S) URL. This is optional property and is used with Booting a Ramdisk or an ISO.Note
The
boot_iso
property used to be calledilo_boot_iso
before the Xena release.rescue_iso
: The glance UUID of the rescue ISO image. This is optional property and is used whenrescue
interface is set toagent
.
The following properties are also required in node object’s
driver_info
ifilo-pxe
orilo-ipxe
boot interface is used:deploy_kernel
: The glance UUID or a HTTP(S) URL of the deployment kernel.deploy_ramdisk
: The glance UUID or a HTTP(S) URL of the deployment ramdisk.rescue_kernel
: The glance UUID or a HTTP(S) URL of the rescue kernel. This is optional property and is used whenrescue
interface is set toagent
.rescue_ramdisk
: The glance UUID or a HTTP(S) URL of the rescue ramdisk. This is optional property and is used whenrescue
interface is set toagent
.
The following properties are also required in node object’s
driver_info
ifilo-uefi-https
boot interface is used forilo5
hardware type:deploy_kernel
: The glance UUID or a HTTPS URL of the deployment kernel.deploy_ramdisk
: The glance UUID or a HTTPS URL of the deployment ramdisk.bootloader
: The glance UUID or a HTTPS URL of the bootloader.rescue_kernel
: The glance UUID or a HTTPS URL of the rescue kernel. This is optional property and is used whenrescue
interface is set toagent
.rescue_ramdisk
: The glance UUID or a HTTP(S) URL of the rescue ramdisk. This is optional property and is used whenrescue
interface is set toagent
.
Note
ilo-uefi-https
boot interface is supported by onlyilo5
hardware type. If the images are not hosted in glance, the references must be HTTPS URLs hosted by secure webserver. This boot interface can be used only when the current boot mode isUEFI
.Note
The fields
deploy_kernel
,deploy_ramdisk
,rescue_kernel
rescue_ramdisk
andbootloader
used to have anilo_
prefix before the Xena release.The following parameters are mandatory in
driver_info
ifilo-inspect
inspect inteface is used and SNMPv3 inspection (SNMPv3 Authentication in HPE iLO4 User Guide) is desired:snmp_auth_user
: The SNMPv3 user.snmp_auth_prot_password
: The auth protocol pass phrase.snmp_auth_priv_password
: The privacy protocol pass phrase.
The following parameters are optional for SNMPv3 inspection:
snmp_auth_protocol
: The Auth Protocol. The valid values are “MD5” and “SHA”. The iLO default value is “MD5”.snmp_auth_priv_protocol
: The Privacy protocol. The valid values are “AES” and “DES”. The iLO default value is “DES”.
Note
If configuration values for ca_file
, client_port
and
client_timeout
are not provided in the driver_info
of the node,
the corresponding config variables defined under [ilo]
section in
ironic.conf will be used.
Prerequisites¶
proliantutils is a python package which contains a set of modules for managing HPE ProLiant hardware.
Install
proliantutils
module on the ironic conductor node. Minimum version required is 2.8.0:$ pip install "proliantutils>=2.8.0"
ipmitool
command must be present on the service node(s) whereironic-conductor
is running. On most distros, this is provided as part of theipmitool
package. Please refer to Hardware Inspection Support for more information on recommended version.
Different configuration for ilo hardware type¶
Glance Configuration¶
Configure Glance image service with its storage backend as Swift.
Set a temp-url key for Glance user in Swift. For example, if you have configured Glance with user
glance-swift
and tenant asservice
, then run the below command:swift --os-username=service:glance-swift post -m temp-url-key:mysecretkeyforglance
Fill the required parameters in the
[glance]
section in/etc/ironic/ironic.conf
. Normally you would be required to fill in the following details:[glance] swift_temp_url_key=mysecretkeyforglance swift_endpoint_url=https://10.10.1.10:8080 swift_api_version=v1 swift_account=AUTH_51ea2fb400c34c9eb005ca945c0dc9e1 swift_container=glance
The details can be retrieved by running the below command:
$ swift --os-username=service:glance-swift stat -v | grep -i url StorageURL: http://10.10.1.10:8080/v1/AUTH_51ea2fb400c34c9eb005ca945c0dc9e1 Meta Temp-Url-Key: mysecretkeyforglance
Swift must be accessible with the same admin credentials configured in Ironic. For example, if Ironic is configured with the below credentials in
/etc/ironic/ironic.conf
:[keystone_authtoken] admin_password = password admin_user = ironic admin_tenant_name = service
Ensure
auth_version
inkeystone_authtoken
to 2.Then, the below command should work.:
$ swift --os-username ironic --os-password password --os-tenant-name service --auth-version 2 stat Account: AUTH_22af34365a104e4689c46400297f00cb Containers: 2 Objects: 18 Bytes: 1728346241 Objects in policy "policy-0": 18 Bytes in policy "policy-0": 1728346241 Meta Temp-Url-Key: mysecretkeyforglance X-Timestamp: 1409763763.84427 X-Trans-Id: tx51de96a28f27401eb2833-005433924b Content-Type: text/plain; charset=utf-8 Accept-Ranges: bytes
Restart the Ironic conductor service:
$ service ironic-conductor restart
Web server configuration on conductor¶
The HTTP(S) web server can be configured in many ways. For apache web server on Ubuntu, refer here
Following config variables need to be set in
/etc/ironic/ironic.conf
:use_web_server_for_images
in[ilo]
section:[ilo] use_web_server_for_images = True
http_url
andhttp_root
in[deploy]
section:[deploy] # Ironic compute node's http root path. (string value) http_root=/httpboot # Ironic compute node's HTTP server URL. Example: # http://192.1.2.3:8080 (string value) http_url=http://192.168.0.2:8080
use_web_server_for_images
: If the variable is set to false
,
the ilo-virtual-media
boot interface uses swift containers to host the
intermediate floppy image and the boot ISO. If the variable is set to
true
, it uses the local web server for hosting the intermediate files.
The default value for use_web_server_for_images
is False.
http_url
: The value for this variable is prefixed with the generated
intermediate files to generate a URL which is attached in the virtual media.
http_root
: It is the directory location to which ironic conductor copies
the intermediate floppy image and the boot ISO.
Note
HTTPS is strongly recommended over HTTP web server configuration for security
enhancement. The ilo-virtual-media
boot interface will send the instance’s
configdrive over an encrypted channel if web server is HTTPS enabled. However
for ilo-uefi-https
boot interface HTTPS webserver is mandatory as this
interface only supports HTTPS URLs.
Enable driver¶
Build a deploy ISO (and kernel and ramdisk) image, see Building or downloading a deploy ramdisk image
See Glance Configuration for configuring glance image service with its storage backend as
swift
.Upload this image to Glance:
glance image-create --name deploy-ramdisk.iso --disk-format iso --container-format bare < deploy-ramdisk.iso
Enable hardware type and hardware interfaces in
/etc/ironic/ironic.conf
:[DEFAULT] enabled_hardware_types = ilo enabled_bios_interfaces = ilo enabled_boot_interfaces = ilo-virtual-media,ilo-pxe,ilo-ipxe enabled_power_interfaces = ilo enabled_console_interfaces = ilo enabled_raid_interfaces = agent enabled_management_interfaces = ilo enabled_inspect_interfaces = ilo enabled_rescue_interfaces = agent
Restart the ironic conductor service:
$ service ironic-conductor restart
Optional functionalities for the ilo
hardware type¶
Boot mode support¶
The hardware type ilo
supports automatic detection and setting
of boot mode (Legacy BIOS or UEFI).
When boot mode capability is not configured:
If config variable
default_boot_mode
in[ilo]
section of ironic configuration file is set to either ‘bios’ or ‘uefi’, then iLO driver uses that boot mode for provisioning the baremetal ProLiant servers.If the pending boot mode is set on the node then iLO driver uses that boot mode for provisioning the baremetal ProLiant servers.
If the pending boot mode is not set on the node then iLO driver uses ‘uefi’ boot mode for UEFI capable servers and “bios” when UEFI is not supported.
When boot mode capability is configured, the driver sets the pending boot mode to the configured value.
Only one boot mode (either
uefi
orbios
) can be configured for the node.If the operator wants a node to boot always in
uefi
mode orbios
mode, then they may usecapabilities
parameter withinproperties
field of an ironic node.To configure a node in
uefi
mode, then setcapabilities
as below:baremetal node set <node> --property capabilities='boot_mode:uefi'
Nodes having
boot_mode
set touefi
may be requested by adding anextra_spec
to the nova flavor:openstack flavor set ironic-test-3 --property capabilities:boot_mode="uefi" openstack server create --flavor ironic-test-3 --image test-image instance-1
If
capabilities
is used inextra_spec
as above, nova scheduler (ComputeCapabilitiesFilter
) will match only ironic nodes which have theboot_mode
set appropriately inproperties/capabilities
. It will filter out rest of the nodes.The above facility for matching in nova can be used in heterogeneous environments where there is a mix of
uefi
andbios
machines, and operator wants to provide a choice to the user regarding boot modes. If the flavor doesn’t containboot_mode
then nova scheduler will not consider boot mode as a placement criteria, hence user may get either a BIOS or UEFI machine that matches with user specified flavors.
The automatic boot ISO creation for UEFI boot mode has been enabled in Kilo.
The manual creation of boot ISO for UEFI boot mode is also supported.
For the latter, the boot ISO for the deploy image needs to be built
separately and the deploy image’s boot_iso
property in glance should
contain the glance UUID of the boot ISO. For building boot ISO, add iso
element to the diskimage-builder command to build the image. For example:
disk-image-create ubuntu baremetal iso
UEFI Secure Boot Support¶
The hardware type ilo
supports secure boot deploy, see UEFI secure boot mode
for details.
iLO specific notes:
In UEFI secure boot, digitally signed bootloader should be able to validate
digital signatures of kernel during boot process. This requires that the
bootloader contains the digital signatures of the kernel.
For the ilo-virtual-media
boot interface, it is recommended that
boot_iso
property for user image contains the glance UUID of the boot
ISO. If boot_iso
property is not updated in glance for the user image,
it would create the boot_iso
using bootloader from the deploy iso. This
boot_iso
will be able to boot the user image in UEFI secure boot
environment only if the bootloader is signed and can validate digital
signatures of user image kernel.
For HPE ProLiant Gen9 servers, one can enroll public key using iLO System
Utilities UI. Please refer to section Accessing Secure Boot options
in
HP UEFI System Utilities User Guide.
One can also refer to white paper on Secure Boot for Linux on HP ProLiant
servers for
additional details.
For more up-to-date information, refer iLO driver wiki page
Node Cleaning Support¶
The hardware type ilo
and ilo5
supports node cleaning.
For more information on node cleaning, see Node cleaning
Supported Automated Cleaning Operations¶
The automated cleaning operations supported are:
reset_bios_to_default
: Resets system ROM settings to default. By default, enabled with priority 10. This clean step is supported only on Gen9 and above servers.reset_secure_boot_keys_to_default
: Resets secure boot keys to manufacturer’s defaults. This step is supported only on Gen9 and above servers. By default, enabled with priority 20 .reset_ilo_credential
: Resets the iLO password, ifilo_change_password
is specified as part of node’s driver_info. By default, enabled with priority 30.clear_secure_boot_keys
: Clears all secure boot keys. This step is supported only on Gen9 and above servers. By default, this step is disabled.reset_ilo
: Resets the iLO. By default, this step is disabled.erase_devices
: An inband clean step that performs disk erase on all the disks including the disks visible to OS as well as the raw disks visible to Smart Storage Administrator (SSA). This step supports erasing of the raw disks visible to SSA in Proliant servers only with the ramdisk created using diskimage-builder from Ocata release. By default, this step is disabled. See Disk Erase Support for more details.
For supported in-band cleaning operations, see In-band vs out-of-band.
All the automated cleaning steps have an explicit configuration option for priority. In order to disable or change the priority of the automated clean steps, respective configuration option for priority should be updated in ironic.conf.
Updating clean step priority to 0, will disable that particular clean step and will not run during automated cleaning.
Configuration Options for the automated clean steps are listed under
[ilo]
and[deploy]
section in ironic.conf[ilo] clean_priority_reset_ilo=0 clean_priority_reset_bios_to_default=10 clean_priority_reset_secure_boot_keys_to_default=20 clean_priority_clear_secure_boot_keys=0 clean_priority_reset_ilo_credential=30 [deploy] erase_devices_priority=0
For more information on node automated cleaning, see Automated cleaning
Supported Manual Cleaning Operations¶
The manual cleaning operations supported are:
activate_license
:Activates the iLO Advanced license. This is an out-of-band manual cleaning step associated with the
management
interface. See Activating iLO Advanced license as manual clean step for user guidance on usage. Please note that this operation cannot be performed using theilo-virtual-media
boot interface as it needs this type of advanced license already active to use virtual media to boot into to start cleaning operation. Virtual media is an advanced feature. If an advanced license is already active and the user wants to overwrite the current license key, for example in case of a multi-server activation key delivered with a flexible-quantity kit or after completing an Activation Key Agreement (AKA), then the driver can still be used for executing this cleaning step.clear_ca_certificates
:Removes the CA certificates from iLO. See Removing CA certificates from iLO as manual clean step for user guidance on usage.
apply_configuration
:Applies given BIOS settings on the node. See BIOS configuration support. This step is part of the
bios
interface.factory_reset
:Resets the BIOS settings on the node to factory defaults. See BIOS configuration support. This step is part of the
bios
interface.create_configuration
:Applies RAID configuration on the node. See RAID Configuration for more information. This step is part of the
raid
interface.delete_configuration
:Deletes RAID configuration on the node. See RAID Configuration for more information. This step is part of the
raid
interface.update_firmware
:Updates the firmware of the devices. Also an out-of-band step associated with the
management
interface. See Initiating firmware update as manual clean step for user guidance on usage. The supported devices for firmware update are:ilo
,cpld
,power_pic
,bios
andchassis
. Please refer to below table for their commonly used descriptions.Device
Description
ilo
BMC for HPE ProLiant servers
cpld
System programmable logic device
power_pic
Power management controller
bios
HPE ProLiant System ROM
chassis
System chassis device
Some devices firmware cannot be updated via this method, such as: storage controllers, host bus adapters, disk drive firmware, network interfaces and Onboard Administrator (OA).
update_firmware_sum
:Updates all or list of user specified firmware components on the node using Smart Update Manager (SUM). It is an inband step associated with the
management
interface. See Smart Update Manager (SUM) based firmware update for more information on usage.security_parameters_update
:Updates the Security Parameters. See Updating security parameters as manual clean step for user guidance on usage. The supported security parameters for this clean step are:
Password_Complexity
,RequiredLoginForiLORBSU
,IPMI/DCMI_Over_LAN
,RequireHostAuthentication
andSecure_Boot
.update_minimum_password_length
:Updates the Minimum Password Length security parameter. See Update Minimum Password Length security parameter as manual clean step for user guidance on usage.
update_auth_failure_logging_threshold
:Updates the Authentication Failure Logging security parameter. See Update Authentication Failure Logging security parameter as manual clean step for user guidance on usage.
create_csr
:Creates the certificate signing request. See Create Certificate Signing Request(CSR) as manual clean step for user guidance on usage.
add_https_certificate
:Adds the signed HTTPS certificate to the iLO. See Add HTTPS Certificate as manual clean step for user guidance on usage.
iLO with firmware version 1.5 is minimally required to support all the operations.
For more information on node manual cleaning, see Manual cleaning
Node Deployment Customization¶
The hardware type ilo
and ilo5
supports customization of node
deployment via deploy templates, see Node Deployment.
The supported deploy steps are:
apply_configuration
:Applies given BIOS settings on the node. See BIOS configuration support. This step is part of the
bios
interface.
factory_reset
:Resets the BIOS settings on the node to factory defaults. See BIOS configuration support. This step is part of the
bios
interface.
reset_bios_to_default
:Resets system ROM settings to default. This step is supported only on Gen9 and above servers. This step is part of the
management
interface.
reset_secure_boot_keys_to_default
:Resets secure boot keys to manufacturer’s defaults. This step is supported only on Gen9 and above servers. This step is part of the
management
interface.
reset_ilo_credential
:Resets the iLO password. The password need to be specified in
ilo_password
argument of the step. This step is part of themanagement
interface.
clear_secure_boot_keys
:Clears all secure boot keys. This step is supported only on Gen9 and above servers. This step is part of the
management
interface.
reset_ilo
:Resets the iLO. This step is part of the
management
interface.
update_firmware
:Updates the firmware of the devices. This step is part of the
management
interface. See Initiating firmware update as manual clean step for user guidance on usage. The supported devices for firmware update are:ilo
,cpld
,power_pic
,bios
andchassis
. This step is part ofmanagement
interface. Please refer to below table for their commonly used descriptions.Device
Description
ilo
BMC for HPE ProLiant servers
cpld
System programmable logic device
power_pic
Power management controller
bios
HPE ProLiant System ROM
chassis
System chassis device
Some devices firmware cannot be updated via this method, such as: storage controllers, host bus adapters, disk drive firmware, network interfaces and Onboard Administrator (OA).
flash_firmware_sum
:Updates all or list of user specified firmware components on the node using Smart Update Manager (SUM). It is an inband step associated with the
management
interface. See Smart Update Manager (SUM) based firmware update for more information on usage.
apply_configuration
:Applies RAID configuration on the node. See RAID Configuration for more information. This step is part of the
raid
interface.
Example of using deploy template with the Compute service¶
Create a deploy template with a single step:
baremetal deploy template create \
CUSTOM_HYPERTHREADING_ON \
--steps '[{"interface": "bios", "step": "apply_configuration", "args": {"settings": [{"name": "ProcHyperthreading", "value": "Enabled"}]}, "priority": 150}]'
Add the trait CUSTOM_HYPERTHREADING_ON
to the node represented by $node_ident
:
baremetal node add trait $node_ident CUSTOM_HYPERTHREADING_ON
Update the flavor bm-hyperthreading-on
in the Compute service with the
following property:
openstack flavor set --property trait:CUSTOM_HYPERTHREADING_ON=required bm-hyperthreading-on
Creating a Compute instance with this flavor will ensure that the instance is
scheduled only to Bare Metal nodes with the CUSTOM_HYPERTHREADING_ON
trait.
When an instance is created using the bm-hyperthreading-on
flavor, then the
deploy steps of deploy template CUSTOM_HYPERTHREADING_ON
will be executed
during the deployment of the scheduled node, causing Hyperthreading to be
enabled in the node’s BIOS configuration.
Hardware Inspection Support¶
The hardware type ilo
supports hardware inspection.
Note
The disk size is returned by RIBCL/RIS only when RAID is preconfigured on the storage. If the storage is Direct Attached Storage, then RIBCL/RIS fails to get the disk size.
The SNMPv3 inspection gets disk size for all types of storages. If RIBCL/RIS is unable to get disk size and SNMPv3 inspection is requested, the proliantutils does SNMPv3 inspection to get the disk size. If proliantutils is unable to get the disk size, it raises an error. This feature is available in proliantutils release version >= 2.2.0.
The iLO must be updated with SNMPv3 authentication details. Pleae refer to the section SNMPv3 Authentication in HPE iLO4 User Guide for setting up authentication details on iLO. The following parameters are mandatory to be given in driver_info for SNMPv3 inspection:
snmp_auth_user
: The SNMPv3 user.snmp_auth_prot_password
: The auth protocol pass phrase.snmp_auth_priv_password
: The privacy protocol pass phrase.
The following parameters are optional for SNMPv3 inspection:
snmp_auth_protocol
: The Auth Protocol. The valid values are “MD5” and “SHA”. The iLO default value is “MD5”.snmp_auth_priv_protocol
: The Privacy protocol. The valid values are “AES” and “DES”. The iLO default value is “DES”.
The inspection process will discover the following essential properties (properties required for scheduling deployment):
memory_mb
: memory sizecpus
: number of cpuscpu_arch
: cpu architecturelocal_gb
: disk size
Inspection can also discover the following extra capabilities for iLO driver:
ilo_firmware_version
: iLO firmware versionrom_firmware_version
: ROM firmware versionsecure_boot
: secure boot is supported or not. The possible values are ‘true’ or ‘false’. The value is returned as ‘true’ if secure boot is supported by the server.server_model
: server modelpci_gpu_devices
: number of gpu devices connected to the bare metal.nic_capacity
: the max speed of the embedded NIC adapter.sriov_enabled
: true, if server has the SRIOV supporting NIC.has_rotational
: true, if server has HDD disk.has_ssd
: true, if server has SSD disk.has_nvme_ssd
: true, if server has NVME SSD disk.cpu_vt
: true, if server supports cpu virtualization.hardware_supports_raid
: true, if RAID can be configured on the server using RAID controller.nvdimm_n
: true, if server has NVDIMM_N type of persistent memory.persistent_memory
: true, if server has persistent memory.logical_nvdimm_n
: true, if server has logical NVDIMM_N configured.rotational_drive_<speed>_rpm
: The capabilitiesrotational_drive_4800_rpm
,rotational_drive_5400_rpm
,rotational_drive_7200_rpm
,rotational_drive_10000_rpm
androtational_drive_15000_rpm
are set to true if the server has HDD drives with speed of 4800, 5400, 7200, 10000 and 15000 rpm respectively.logical_raid_level_<raid_level>
: The capabilitieslogical_raid_level_0
,logical_raid_level_1
,logical_raid_level_2
,logical_raid_level_5
,logical_raid_level_6
,logical_raid_level_10
,logical_raid_level_50
andlogical_raid_level_60
are set to true if any of the raid levels among 0, 1, 2, 5, 6, 10, 50 and 60 are configured on the system.overall_security_status
:Ok
orRisk
orIgnored
as returned by iLO security dashboard. iLO computes the overall security status by evaluating the security status for each of the security parameters. Admin needs to fix the actual parameters and then re-inspect so that iLO can recompute the overall security status. If the all security params, whosesecurity_status
isRisk
, have theIgnore
field set toTrue
, then iLO sets the overall security status value asIgnored
. All the security params must have thesecurity_status
asOk
for theoverall_security_status
to have the value asOk
.last_firmware_scan_status
:Ok
orRisk
as returned by iLO security dashboard. This denotes security status of the last firmware scan done on the system. If it isRisk
, the recommendation is to run clean_stepupdate_firmware_sum
without any specific firmware components so that firmware is updated for all the components using latest SPP (Service Provider Pack) ISO and then re-inspect to get the security status again.security_override_switch
:Ok
orRisk
as returned by iLO security dashboard. This is disable/enable login to the iLO using credentials. This can be toggled only by physical visit to the bare metal.gpu_<vendor>_count
: Integer value. The capability name is dynamically formed as gpu_<vendor>_count. The vendor name is replaced in the “<vendor>”. If the vendor name is not returned by the hardware, then vendor ID in hexadecimal form is replaced in the capability name. Examples: {‘gpu_Nvidia_count’: 1}, {‘gpu_0x102b_count’: 1}.gpu_<vendor_device_name>_count
: Integer value. The capability name is formed dynamically by replacing the gpu device name as returned by ilo in “<vendor_device_name>”. Examples: {‘gpu_Nvidia_Tesla_M10_count’: 1}, {‘gpu_Embedded_Video_Controller_count’: 1}gpu_<vendor_device_name>
: Boolean. The capability name is formed dynamically by replacing the gpu device name as returned by ilo in “<vendor_device_name>”. Examples: {‘gpu_Nvidia_Tesla_M10’: True}, {‘gpu_Embedded_Video_Controller’: True}Note
The capability
nic_capacity
can only be discovered if ipmitool version >= 1.8.15 is used on the conductor. The latest version can be downloaded from here.The iLO firmware version needs to be 2.10 or above for nic_capacity to be discovered.
To discover IPMI based attributes you need to enable iLO feature ‘IPMI/DCMI over LAN Access’ on iLO4 and iLO5 management engine.
The proliantutils returns only active NICs for Gen10 ProLiant HPE servers. The user would need to delete the ironic ports corresponding to inactive NICs for Gen8 and Gen9 servers as proliantutils returns all the discovered (active and otherwise) NICs for Gen8 and Gen9 servers and ironic ports are created for all of them. Inspection logs a warning if the node under inspection is Gen8 or Gen9.
The security dashboard capabilities are applicable only for Gen10 ProLiant HPE servers and above. To fix the security dashboard parameters value from
Risk
toOk
, user need to fix the parameters separately and re-inspect to see the security status of the parameters.
The operator can specify these capabilities in nova flavor for node to be selected for scheduling:
openstack flavor set my-baremetal-flavor --property capabilities:server_model="<in> Gen8"
openstack flavor set my-baremetal-flavor --property capabilities:nic_capacity="10Gb"
openstack flavor set my-baremetal-flavor --property capabilities:ilo_firmware_version="<in> 2.10"
openstack flavor set my-baremetal-flavor --property capabilities:has_ssd="true"
See Capabilities discovery for more details and examples.
Swiftless deploy for intermediate images¶
The hardware type ilo
with ilo-virtual-media
as boot interface
can deploy and boot the server with and without swift
being used for
hosting the intermediate temporary floppy image (holding metadata for
deploy kernel and ramdisk) and the boot ISO. A local HTTP(S) web server on
each conductor node needs to be configured.
Please refer to Web server configuration on conductor for more information.
The HTTPS web server needs to be enabled (instead of HTTP web server) in order
to send management information and images in encrypted channel over HTTPS.
Note
This feature assumes that the user inputs are on Glance which uses swift as backend. If swift dependency has to be eliminated, please refer to HTTP(S) Based Deploy Support also.
Deploy Process¶
Please refer to Swiftless deploy for intermediate images.
HTTP(S) Based Deploy Support¶
The user input for the images given in driver_info
like deploy_iso
,
deploy_kernel
and deploy_ramdisk
and in instance_info
like
image_source
, kernel
, ramdisk
and boot_iso
may also be given as
HTTP(S) URLs.
The HTTP(S) web server can be configured in many ways. For the Apache web server on Ubuntu, refer here. The web server may reside on a different system than the conductor nodes, but its URL must be reachable by the conductor and the bare metal nodes.
Deploy Process¶
Please refer to HTTP(S) based deploy.
Support for iLO driver with Standalone Ironic¶
It is possible to use ironic as standalone services without other
OpenStack services. The ilo
hardware type can be used in standalone ironic.
This feature is referred to as iLO driver with standalone ironic
in this document.
Configuration¶
The HTTP(S) web server needs to be configured as described in HTTP(S) Based Deploy Support and Web server configuration on conductor needs to be configured for hosting intermediate images on conductor as described in Swiftless deploy for intermediate images.
Deploy Process¶
Glance and swift for partition images¶
Glance and swift with whole-disk images¶
Swiftless deploy¶
HTTP(S) based deploy¶
Standalone ironic¶
Activating iLO Advanced license as manual clean step¶
iLO driver can activate the iLO Advanced license key as a manual cleaning
step. Any manual cleaning step can only be initiated when a node is in the
manageable
state. Once the manual cleaning is finished, the node will be
put in the manageable
state again. User can follow steps from
Manual cleaning to initiate manual cleaning operation on a node.
An example of a manual clean step with activate_license
as the only clean
step could be:
"clean_steps": [{
"interface": "management",
"step": "activate_license",
"args": {
"ilo_license_key": "ABC12-XXXXX-XXXXX-XXXXX-YZ345"
}
}]
The different attributes of activate_license
clean step are as follows:
Attribute |
Description |
---|---|
|
Interface of clean step, here |
|
Name of clean step, here |
|
Keyword-argument entry (<name>: <value>) being passed to clean step |
|
iLO Advanced license key to activate enterprise features. This is mandatory. |
Removing CA certificates from iLO as manual clean step¶
iLO driver can remove the invalidated CA certificates as a manual step.
Any manual cleaning step can only be initiated when a node is in the
manageable
state. Once the manual cleaning is finished, the node will be
put in the manageable
state again. User can follow steps from
Manual cleaning to initiate manual cleaning operation on a node.
An example of a manual clean step with clear_ca_certificates
as the only clean
step could be:
"clean_steps": [{
"interface": "management",
"step": "clear_ca_certificates",
"args": {
"certificate_files" : ["/path/to/certsA", "/path/to/certsB"]
}
}]
The different attributes of clear_ca_certificates
clean step are as follows:
Attribute |
Description |
---|---|
|
Interface of clean step, here |
|
Name of clean step, here |
|
Keyword-argument entry (<name>: <value>) being passed to clean step |
|
List of CA certificates which are to be removed. |
This is mandatory. |
Initiating firmware update as manual clean step¶
iLO driver can invoke secure firmware update as a manual cleaning step. Any
manual cleaning step can only be initiated when a node is in the manageable
state. Once the manual cleaning is finished, the node will be put in the
manageable
state again. A user can follow steps from Manual cleaning
to initiate manual cleaning operation on a node.
An example of a manual clean step with update_firmware
as the only clean
step could be:
"clean_steps": [{
"interface": "management",
"step": "update_firmware",
"args": {
"firmware_update_mode": "ilo",
"firmware_images":[
{
"url": "file:///firmware_images/ilo/1.5/CP024444.scexe",
"checksum": "a94e683ea16d9ae44768f0a65942234d",
"component": "ilo"
},
{
"url": "swift://firmware_container/cpld2.3.rpm",
"checksum": "<md5-checksum-of-this-file>",
"component": "cpld"
},
{
"url": "http://my_address:port/firmwares/bios_vLatest.scexe",
"checksum": "<md5-checksum-of-this-file>",
"component": "bios"
},
{
"url": "https://my_secure_address_url/firmwares/chassis_vLatest.scexe",
"checksum": "<md5-checksum-of-this-file>",
"component": "chassis"
},
{
"url": "file:///home/ubuntu/firmware_images/power_pic/pmc_v3.0.bin",
"checksum": "<md5-checksum-of-this-file>",
"component": "power_pic"
}
]
}
}]
The different attributes of update_firmware
clean step are as follows:
Attribute |
Description |
---|---|
|
Interface of clean step, here |
|
Name of clean step, here |
|
Keyword-argument entry (<name>: <value>) being passed to clean step |
|
Mode (or mechanism) of out-of-band firmware update. Supported value is |
|
Ordered list of dictionaries of images to be flashed. This is mandatory. |
Each firmware image block is represented by a dictionary (JSON), in the form:
{
"url": "<url of firmware image file>",
"checksum": "<md5 checksum of firmware image file to verify the image>",
"component": "<device on which firmware image will be flashed>"
}
All the fields in the firmware image block are mandatory.
The different types of firmware url schemes supported are:
file
,http
,https
andswift
.Note
This feature assumes that while using
file
url scheme the file path is on the conductor controlling the node.Note
The
swift
url scheme assumes the swift account of theservice
project. Theservice
project (tenant) is a special project created in the Keystone system designed for the use of the core OpenStack services. When Ironic makes use of Swift for storage purpose, the account is generallyservice
and the container is generallyironic
andilo
driver uses a container namedironic_ilo_container
for their own purpose.Note
While using firmware files with a
.rpm
extension, make sure the commandsrpm2cpio
andcpio
are present on the conductor, as they are utilized to extract the firmware image from the package.The firmware components that can be updated are:
ilo
,cpld
,power_pic
,bios
andchassis
.The firmware images will be updated in the order given by the operator. If there is any error during processing of any of the given firmware images provided in the list, none of the firmware updates will occur. The processing error could happen during image download, image checksum verification or image extraction. The logic is to process each of the firmware files and update them on the devices only if all the files are processed successfully. If, during the update (uploading and flashing) process, an update fails, then the remaining updates, if any, in the list will be aborted. But it is recommended to triage and fix the failure and re-attempt the manual clean step
update_firmware
for the abortedfirmware_images
.The devices for which the firmwares have been updated successfully would start functioning using their newly updated firmware.
As a troubleshooting guidance on the complete process, check Ironic conductor logs carefully to see if there are any firmware processing or update related errors which may help in root causing or gain an understanding of where things were left off or where things failed. You can then fix or work around and then try again. A common cause of update failure is HPE Secure Digital Signature check failure for the firmware image file.
To compute
md5
checksum for your image file, you can use the following command:$ md5sum image.rpm 66cdb090c80b71daa21a67f06ecd3f33 image.rpm
Smart Update Manager (SUM) based firmware update¶
The firmware update based on SUM is an inband clean/deploy step supported by iLO driver. The firmware update is performed on all or list of user specified firmware components on the node. Refer to SUM User Guide to get more information on SUM based firmware update.
Note
update_firmware_sum
clean step requires the agent ramdisk with
Proliant Hardware Manager
from the proliantutils version 2.5.0 or
higher. See DIB support for Proliant Hardware Manager to create the
agent ramdisk with Proliant Hardware Manager
.
Note
flash_firmware_sum
deploy step requires the agent ramdisk with
Proliant Hardware Manager
from the proliantutils version 2.9.5 or
higher. See DIB support for Proliant Hardware Manager to create the
agent ramdisk with Proliant Hardware Manager
.
The attributes of update_firmware_sum
/flash_firmware_sum
step are as
follows:
Attribute |
Description |
---|---|
|
Interface of the clean step, here |
|
Name of the clean step, here |
|
Keyword-argument entry (<name>: <value>) being passed to the clean step |
The keyword arguments used for the step are as follows:
url
: URL of SPP (Service Pack for Proliant) ISO. It is mandatory. The URL schemes supported arehttp
,https
andswift
.checksum
: MD5 checksum of SPP ISO to verify the image. It is mandatory.components
: List of filenames of the firmware components to be flashed. It is optional. If not provided, the firmware update is performed on all the firmware components.
The step performs an update on all or a list of firmware components and
returns the SUM log files. The log files include hpsum_log.txt
and
hpsum_detail_log.txt
which holds the information about firmware components,
firmware version for each component and their update status. The log object
will be named with the following pattern:
<node>[_<instance-uuid>]_update_firmware_sum_<timestamp yyyy-mm-dd-hh-mm-ss>.tar.gz
or
<node>[_<instance-uuid>]_flash_firmware_sum_<timestamp yyyy-mm-dd-hh-mm-ss>.tar.gz
Refer to Retrieving logs from the deploy ramdisk for more information on enabling and viewing the logs returned from the ramdisk.
An example of update_firmware_sum
clean step:
{
"interface": "management",
"step": "update_firmware_sum",
"args":
{
"url": "http://my_address:port/SPP.iso",
"checksum": "abcdefxyz",
"components": ["CP024356.scexe", "CP008097.exe"]
}
}
The step fails if there is any error in the processing of step arguments. The processing error could happen during validation of components’ file extension, image download, image checksum verification or image extraction. In case of a failure, check Ironic conductor logs carefully to see if there are any validation or firmware processing related errors which may help in root cause analysis or gaining an understanding of where things were left off or where things failed. You can then fix or work around and then try again.
Warning
This feature is officially supported only with RHEL and SUSE based IPA ramdisk. Refer to SUM for supported OS versions for specific SUM version.
Note
Refer Guidelines for SPP ISO for steps to get SPP (Service Pack for ProLiant) ISO.
Updating security parameters as manual clean step¶
iLO driver can invoke security parameters update as a manual clean step. Any
manual cleaning step can only be initiated when a node is in the manageable
state. Once the manual cleaning is finished, the node will be put in the
manageable
state again. A user can follow steps from Manual cleaning
to initiate manual cleaning operation on a node. This feature is only supported
for iLO5 based hardware.
An example of a manual clean step with security_parameters_update
as the
only clean step could be:
"clean_steps": [{
"interface": "management",
"step": "security_parameters_update",
"args": {
"security_parameters":[
{
"param": "password_complexity",
"enable": "True",
"ignore": "False"
},
{
"param": "require_login_for_ilo_rbsu",
"enable": "True",
"ignore": "False"
},
{
"param": "ipmi_over_lan",
"enable": "True",
"ignore": "False"
},
{
"param": "secure_boot",
"enable": "True",
"ignore": "False"
},
{
"param": "require_host_authentication",
"enable": "True",
"ignore": "False"
}
]
}
}]
The different attributes of security_parameters_update
clean step are as follows:
Attribute |
Description |
---|---|
|
Interface of clean step, here |
|
Name of clean step, here |
|
Keyword-argument entry (<name>: <value>) being passed to clean step |
|
Ordered list of dictionaries of security parameters to be updated. This is mandatory. |
Each security parameter block is represented by a dictionary (JSON), in the form:
{
"param": "<security parameter name>",
"enable": "security parameter to be enabled/disabled",
"ignore": "security parameter status to be ignored or not"
}
In all of these fields, param
field is mandatory. Remaining fields are boolean and are optional. If user doesn’t
pass any value then for enable
field the default will be True and for ignore
field default will be False.
The Security Parameters which are supported for this clean step are:
Password_Complexity
,RequiredLoginForiLORBSU
,RequireHostAuthentication
,IPMI/DCMI_Over_LAN
andSecure_Boot
.
Update Minimum Password Length security parameter as manual clean step¶
iLO driver can invoke Minimum Password Length
security parameter update as a manual clean
step. This feature is only supported for iLO5 based hardware.
An example of a manual clean step with update_minimum_password_length
as the
only clean step could be:
"clean_steps": [{
"interface": "management",
"step": "update_minimum_password_length",
"args": {
"password_length": "8",
"ignore": "False"
}
}]
Both the arguments password_length
and ignore
are optional. The accepted values for password_length are
0 to 39. If user doesn’t pass any value, the default value for password_length will be 8 and for ignore the default
value be False.
Update Authentication Failure Logging security parameter as manual clean step¶
iLO driver can invoke Authentication Failure Logging
security parameter update as a manual clean
step. This feature is only supported for iLO5 based hardware.
An example of a manual clean step with Authentication Failure Logging
as the
only clean step could be:
"clean_steps": [{
"interface": "management",
"step": "update_auth_failure_logging_threshold",
"args": {
"logging_threshold": "1",
"ignore": "False"
}
}]
Both the arguments logging_threshold
and ignore
are optional. The accepted values for logging_threshold are
0 to 5. If user doesn’t pass any value, the default value for logging_threshold will be 1 and for ignore the default
value be False. If user passes the value of logging_threshold as 0, the Authentication Failure Logging security
parameter will be disabled.
Create Certificate Signing Request(CSR) as manual clean step¶
iLO driver can invoke create_csr
request as a manual clean step. This step is only supported for iLO5 based hardware.
An example of a manual clean step with create_csr
as the only clean step could be:
"clean_steps": [{
"interface": "management",
"step": "create_csr",
"args": {
"csr_params": {
"City": "Bengaluru",
"CommonName": "1.1.1.1",
"Country": "India",
"OrgName": "HPE",
"State": "Karnataka"
}
}
}]
The [ilo]cert_path
option in ironic.conf
is used as the directory path for
creating the CSR, which defaults to /var/lib/ironic/ilo
. The CSR is created in the directory location
given in [ilo]cert_path
in node_uuid
directory as <node_uuid>.csr.
Add HTTPS Certificate as manual clean step¶
iLO driver can invoke add_https_certificate
request as a manual clean step. This step is only supported for
iLO5 based hardware.
An example of a manual clean step with add_https_certificate
as the only clean step could be:
"clean_steps": [{
"interface": "management",
"step": "add_https_certificate",
"args": {
"cert_file": "/test1/iLO.crt"
}
}]
Argument cert_file
is mandatory. The cert_file
takes the path or url of the certificate file.
The url schemes supported are: file
, http
and https
.
The CSR generated in step create_csr
needs to be signed by a valid CA and the resultant HTTPS certificate should
be provided in cert_file
. It copies the cert_file
to [ilo]cert_path
under node.uuid
as <node_uuid>.crt
before adding it to iLO.
RAID Support¶
The inband RAID functionality is supported by iLO driver. See RAID Configuration for more information. Bare Metal service update node with following information after successful configuration of RAID:
Node
properties/local_gb
is set to the size of root volume.Node
properties/root_device
is filled withwwn
details of root volume. It is used by iLO driver as root device hint during provisioning.The value of raid level of root volume is added as
raid_level
capability to the node’scapabilities
parameter withinproperties
field. The operator can specify theraid_level
capability in nova flavor for node to be selected for scheduling:openstack flavor set ironic-test --property capabilities:raid_level="1+0" openstack server create --flavor ironic-test --image test-image instance-1
DIB support for Proliant Hardware Manager¶
Install ironic-python-agent-builder
following the guide 1
To create an agent ramdisk with Proliant Hardware Manager
,
use the proliant-tools
element in DIB:
ironic-python-agent-builder -o proliant-agent-ramdisk -e proliant-tools fedora
Disk Erase Support¶
erase_devices
is an inband clean step supported by iLO driver. It
performs erase on all the disks including the disks visible to OS as
well as the raw disks visible to the Smart Storage Administrator (SSA).
This inband clean step requires ssacli
utility starting from version
2.60-19.0
to perform the erase on physical disks. See the
ssacli documentation for more information on ssacli utility and different
erase methods supported by SSA.
The disk erasure via shred
is used to erase disks visible to the OS
and its implementation is available in Ironic Python Agent. The raw disks
connected to the Smart Storage Controller are erased using Sanitize erase
which is a ssacli supported erase method. If Sanitize erase is not supported
on the Smart Storage Controller the disks are erased using One-pass
erase (overwrite with zeros).
This clean step is supported when the agent ramdisk contains the
Proliant Hardware Manager
from the proliantutils version 2.3.0 or higher.
This clean step is performed as part of automated cleaning and it is disabled
by default. See In-band vs out-of-band for more information on
enabling/disabling a clean step.
Install ironic-python-agent-builder
following the guide 1
To create an agent ramdisk with Proliant Hardware Manager
, use the
proliant-tools
element in DIB:
ironic-python-agent-builder -o proliant-agent-ramdisk -e proliant-tools fedora
See the proliant-tools for more information on creating agent ramdisk with
proliant-tools
element in DIB.
Firmware based UEFI iSCSI boot from volume support¶
With Gen9 (UEFI firmware version 1.40 or higher) and Gen10 HPE Proliant servers, the driver supports firmware based UEFI boot of an iSCSI cinder volume.
This feature requires the node to be configured to boot in UEFI
boot mode,
as well as user image should be UEFI
bootable image, and PortFast
needs to be enabled in switch configuration for immediate spanning tree
forwarding state so it wouldn’t take much time setting the iSCSI target as
persistent device.
The driver does not support this functionality when in bios
boot mode. In
case the node is configured with ilo-pxe
or ilo-ipxe
as boot interface
and the boot mode configured on the bare metal is bios
, the iscsi boot
from volume is performed using iPXE. See Boot From Volume
for more details.
To use this feature, configure the boot mode of the bare metal to uefi
and
configure the corresponding ironic node using the steps given in Boot From Volume.
In a cloud environment with nodes configured to boot from bios
and uefi
boot
modes, the virtual media driver only supports uefi boot mode, and that attempting to
use iscsi boot at the same time with a bios volume will result in an error.
BIOS configuration support¶
The ilo
and ilo5
hardware types support ilo
BIOS interface.
The support includes providing manual clean steps apply_configuration and
factory_reset to manage supported BIOS settings on the node.
See BIOS Configuration for more details and examples.
Note
Prior to the Stein release the user is required to reboot the node manually in order for the settings to take into effect. Starting with the Stein release, iLO drivers reboot the node after running clean steps related to the BIOS configuration. The BIOS settings are cached and the clean step is marked as success only if all the requested settings are applied without any failure. If application of any of the settings fails, the clean step is marked as failed and the settings are not cached.
Configuration¶
Following are the supported BIOS settings and the corresponding brief description for each of the settings. For a detailed description please refer to HPE Integrated Lights-Out REST API Documentation.
AdvancedMemProtection
: Configure additional memory protection with ECC (Error Checking and Correcting). Allowed values areAdvancedEcc
,OnlineSpareAdvancedEcc
,MirroredAdvancedEcc
.AutoPowerOn
: Configure the server to automatically power on when AC power is applied to the system. Allowed values areAlwaysPowerOn
,AlwaysPowerOff
,RestoreLastState
.BootMode
: Select the boot mode of the system. Allowed values areUefi
,LegacyBios
BootOrderPolicy
: Configure how the system attempts to boot devices per the Boot Order when no bootable device is found. Allowed values areRetryIndefinitely
,AttemptOnce
,ResetAfterFailed
.CollabPowerControl
: Enables the Operating System to request processor frequency changes even if the Power Regulator option on the server configured for Dynamic Power Savings Mode. Allowed values areEnabled
,Disabled
.DynamicPowerCapping
: Configure when the System ROM executes power calibration during the boot process. Allowed values areEnabled
,Disabled
,Auto
.DynamicPowerResponse
: Enable the System BIOS to control processor performance and power states depending on the processor workload. Allowed values areFast
,Slow
.IntelligentProvisioning
: Enable or disable the Intelligent Provisioning functionality. Allowed values areEnabled
,Disabled
.IntelPerfMonitoring
: Exposes certain chipset devices that can be used with the Intel Performance Monitoring Toolkit. Allowed values areEnabled
,Disabled
.IntelProcVtd
: Hypervisor or operating system supporting this option can use hardware capabilities provided by Intel’s Virtualization Technology for Directed I/O. Allowed values areEnabled
,Disabled
.IntelQpiFreq
: Set the QPI Link frequency to a lower speed. Allowed values areAuto
,MinQpiSpeed
.IntelTxt
: Option to modify Intel TXT support. Allowed values areEnabled
,Disabled
.PowerProfile
: Set the power profile to be used. Allowed values areBalancedPowerPerf
,MinPower
,MaxPerf
,Custom
.PowerRegulator
: Determines how to regulate the power consumption. Allowed values areDynamicPowerSavings
,StaticLowPower
,StaticHighPerf
,OsControl
.ProcAes
: Enable or disable the Advanced Encryption Standard Instruction Set (AES-NI) in the processor. Allowed values areEnabled
,Disabled
.ProcCoreDisable
: Disable processor cores using Intel’s Core Multi-Processing (CMP) Technology. Allowed values are Integers ranging from0
to24
.ProcHyperthreading
: Enable or disable Intel Hyperthreading. Allowed values areEnabled
,Disabled
.ProcNoExecute
: Protect your system against malicious code and viruses. Allowed values areEnabled
,Disabled
.ProcTurbo
: Enables the processor to transition to a higher frequency than the processor’s rated speed using Turbo Boost Technology if the processor has available power and is within temperature specifications. Allowed values areEnabled
,Disabled
.ProcVirtualization
: Enables or Disables a hypervisor or operating system supporting this option to use hardware capabilities provided by Intel’s Virtualization Technology. Allowed values areEnabled
,Disabled
.SecureBootStatus
: The current state of Secure Boot configuration. Allowed values areEnabled
,Disabled
.Note
This setting is read-only and can’t be modified with
apply_configuration
clean step.Sriov
: If enabled, SR-IOV support enables a hypervisor to create virtual instances of a PCI-express device, potentially increasing performance. If enabled, the BIOS allocates additional resources to PCI-express devices. Allowed values areEnabled
,Disabled
.ThermalConfig
: select the fan cooling solution for the system. Allowed values areOptimalCooling
,IncreasedCooling
,MaxCooling
ThermalShutdown
: Control the reaction of the system to caution level thermal events. Allowed values areEnabled
,Disabled
.TpmState
: Current TPM device state. Allowed values areNotPresent
,PresentDisabled
,PresentEnabled
.Note
This setting is read-only and can’t be modified with
apply_configuration
clean step.TpmType
: Current TPM device type. Allowed values areNoTpm
,Tpm12
,Tpm20
,Tm10
.Note
This setting is read-only and can’t be modified with
apply_configuration
clean step.UefiOptimizedBoot
: Enables or Disables the System BIOS boot using native UEFI graphics drivers. Allowed values areEnabled
,Disabled
.WorkloadProfile
: Change the Workload Profile to accomodate your desired workload. Allowed values areGeneralPowerEfficientCompute
,GeneralPeakFrequencyCompute
,GeneralThroughputCompute
,Virtualization-PowerEfficient
,Virtualization-MaxPerformance
,LowLatency
,MissionCritical
,TransactionalApplicationProcessing
,HighPerformanceCompute
,DecisionSupport
,GraphicProcessing
,I/OThroughput
,Custom
Note
This setting is only applicable to ProLiant Gen10 servers with iLO 5 management systems.
Certificate based validation in iLO¶
The driver supports validation of certificates on the HPE Proliant servers.
The path to certificate file needs to be appropriately set in ca_file
in
the node’s driver_info
. To update SSL certificates into iLO,
refer to HPE Integrated Lights-Out Security Technology Brief.
Use iLO hostname or IP address as a ‘Common Name (CN)’ while
generating Certificate Signing Request (CSR). Use the same value as
ilo_address while enrolling node to Bare Metal service to avoid SSL
certificate validation errors related to hostname mismatch.
Rescue mode support¶
The hardware type ilo
supports rescue functionality. Rescue operation can
be used to boot nodes into a rescue ramdisk so that the rescue
user can
access the node.
Please refer to Rescue Mode for detailed explanation of rescue feature.
Inject NMI support¶
The management interface ilo
supports injection of non-maskable
interrupt (NMI) to a bare metal. Following command can be used to inject
NMI on a server:
baremetal node inject nmi <node>
Following command can be used to inject NMI via Compute service:
openstack server dump create <server>
Note
This feature is supported on HPE ProLiant Gen9 servers and beyond.
Soft power operation support¶
The power interface ilo
supports soft power off and soft reboot
operations on a bare metal. Following commands can be used to perform
soft power operations on a server:
baremetal node reboot --soft \
[--power-timeout <power-timeout>] <node>
baremetal node power off --soft \
[--power-timeout <power-timeout>] <node>
Note
The configuration [conductor]soft_power_off_timeout
is used as a
default timeout value when no timeout is provided while invoking
hard or soft power operations.
Note
Server POST state is used to track the power status of HPE ProLiant Gen9 servers and beyond.
Out of Band RAID Support¶
With Gen10 HPE Proliant servers and later the ilo5
hardware type supports
firmware based RAID configuration as a clean step. This feature requires the
node to be configured to ilo5
hardware type and its raid interface to be
ilo5
. See RAID Configuration for more information.
After a successful RAID configuration, the Bare Metal service will update the node with the following information:
Node
properties/local_gb
is set to the size of root volume.Node
properties/root_device
is filled withwwn
details of root volume. It is used by iLO driver as root device hint during provisioning.
Later the value of raid level of root volume can be added in
baremetal-with-RAID10
(RAID10 for raid level 10) resource class.
And consequently flavor needs to be updated to request the resource class
to create the server using selected node:
baremetal node set test_node --resource-class \
baremetal-with-RAID10
openstack flavor set --property \
resources:CUSTOM_BAREMETAL_WITH_RAID10=1 test-flavor
openstack server create --flavor test-flavor --image test-image instance-1
Note
Supported raid levels for ilo5
hardware type are: 0, 1, 5, 6, 10, 50, 60
IPv6 support¶
With the IPv6 support in proliantutils>=2.8.0
, nodes can be enrolled
into the baremetal service using the iLO IPv6 addresses.
baremetal node create --driver ilo --deploy-interface direct \
--driver-info ilo_address=2001:0db8:85a3:0000:0000:8a2e:0370:7334 \
--driver-info ilo_username=test-user \
--driver-info ilo_password=test-password \
--driver-info deploy_iso=test-iso \
--driver-info rescue_iso=test-iso
Note
No configuration changes (in e.g. ironic.conf) are required in order to support IPv6.
Out of Band Sanitize Disk Erase Support¶
With Gen10 HPE Proliant servers and later the ilo5
hardware type supports
firmware based sanitize disk erase as a clean step. This feature requires the
node to be configured to ilo5
hardware type and its management interface
to be ilo5
.
The possible erase pattern its supports are:
For HDD - ‘overwrite’, ‘zero’, ‘crypto’
For SSD - ‘block’, ‘zero’, ‘crypto’
The default erase pattern are, for HDD, ‘overwrite’ and for SSD, ‘block’.
Note
In average 300GB HDD with default pattern “overwrite” would take approx. 9 hours and 300GB SSD with default pattern “block” would take approx. 30 seconds to complete the erase.
UEFI-HTTPS Boot support¶
The UEFI firmware on Gen10 HPE Proliant servers supports booting from secured URLs.
With this capability ilo5
hardware with ilo-uefi-https
boot interface supports
deploy/rescue features in more secured environments.
If swift is used as glance backend and ironic is configured to use swift to store temporary images, it is required that swift is configured on HTTPS so that the tempurl generated is HTTPS URL.
If the webserver is used for hosting the temporary images, then the webserver is required to serve requests on HTTPS.
If the images are hosted on a HTTPS webserver or swift configured with HTTPS with custom certificates, the user is required to export SSL certificates into iLO. Refer to HPE Integrated Lights-Out Security Technology Brief for more information.
The following command can be used to enroll a ProLiant node with ilo5
hardware type
and ilo-uefi-https
boot interface:
baremetal node create \
--driver ilo5 \
--boot-interface ilo-uefi-https \
--deploy-interface direct \
--raid-interface ilo5 \
--rescue-interface agent \
--driver-info ilo_address=<ilo-ip-address> \
--driver-info ilo_username=<ilo-username> \
--driver-info ilo_password=<ilo-password> \
--driver-info deploy_kernel=<glance-uuid-of-deploy-kernel> \
--driver-info deploy_ramdisk=<glance-uuid-of-rescue-ramdisk> \
--driver-info bootloader=<glance-uuid-of-bootloader>
Layer 3 or DHCP-less ramdisk booting¶
DHCP-less deploy is supported by ilo
and ilo5
hardware types.
However it would work only with ilo-virtual-media boot interface. See
Layer 3 or DHCP-less ramdisk booting for more information.
Events subscription¶
Events subscription is supported by ilo
and ilo5
hardware types with
ilo
vendor interface for Gen10 and Gen10 Plus servers. See
Node Vendor Passthru Methods for more information.
- 1(1,2)
ironic-python-agent-builder: https://docs.openstack.org/ironic-python-agent-builder/latest/install/index.html