Indicator management

https://storyboard.openstack.org/#!/story/2005342

This spec proposed creating of API for hardware indicators management.

Problem description

Depending on the hardware vendor, bare metal machines might carry on-board and expose through the baseboard management controller (BMC) various indicators, most commonly LEDs.

For example, a blade system might have LEDs on the chassis, on the blades (compute cards), on the drives, on the PSUs, on the NICs.

The use-cases when ironic-manageable LEDs might make sense include:

  • The DC staff member want to identify hardware unit in the rack by lighting up its LED in ironic

  • The Deployer wants to identify ironic node by pressing a button on the physical unit to light up the LED

  • The remote Admin user wants to be aware of the failure LEDs lighting on the bare metal nodes possibly indicating a failure

Proposed change

Overview

This RFE proposes an extension of the node ReST API endpoint that will allow reading and toggling the indicators (e.g. LEDs) on the hardware units.

Indicator management workflow

  1. An API client can choose to discover available node indicators by sending GET /v1/nodes/<node_ident>/management/indicators request. This step is optional.

  2. An API client reads the indicator on the chosen component of the bare metal node by sending GET /v1/nodes/<node_ident>/management/indicators/<component>/<ind_ident> request and presenting current indicator state to the user.

  3. The API client can change the state of the indicator(s) on the given component by sending PUT /v1/nodes/<node_ident>/management/indicators/<component> request.

Alternatives

The user can communicate with BMC independently from ironic for the same purpose. Though ironic node correlation with physical node may be challenging.

Data model impact

None.

Due to the interactive nature of indicator information, it seems that user-indicator link should be as immediate as possible, having it cached by the database can make indicators confusing.

State Machine Impact

None.

Indicators should always work the same regardless of the machine state.

REST API impact

  • GET /v1/nodes/<node_ident>/management/indicators

    Retrieves bare metal node components. Returns a JSON object listing all available node components.

    Currently known components are: system, chassis and drive. Indicator names are free-form, but hopefully descriptive.

    Error codes:

    • 404 Not Found if node is not found.

    Example response object:

    {
    "components": [
        {
            "name": "system",
            "links": [
                {
                    "href": "http://127.0.0.1:6385/v1/nodes/Compute0/
                    management/indicators/system",
                    "rel": "self"
                },
                {
                    "href": "http://127.0.0.1:6385/nodes/Compute0/
                    management/indicators/system",
                    "rel": "bookmark"
                }
            ]
        },
        {
            "name": "chassis",
            "links": [
                {
                    "href": "http://127.0.0.1:6385/v1/nodes/Compute0/
                    management/indicators/chassis",
                    "rel": "self"
                },
                {
                    "href": "http://127.0.0.1:6385/nodes/Compute0/
                    management/indicators/chassis",
                    "rel": "bookmark"
                }
            ]
        }
    ]
}

  • GET /v1/nodes/<node_ident>/management/indicators/<component>

    Retrieves indicators for a component. Returns a JSON object listing all available indicators for given hardware component along with their attributes.

    Currently known components are: system, chassis and drive. Indicator names are free-form, but hopefully descriptive.

    Error codes:

    • 404 Not Found if node or component is not found.

    Example response object:

    {
    "indicators": [
        {
            "name": "power",
            "readonly": true,
            "states": [
                "OFF",
                "ON"
            ],
            "links": [
                {
                    "href": "http://127.0.0.1:6385/v1/nodes/Compute0/
                    management/indicators/system/power",
                    "rel": "self"
                },
                {
                    "href": "http://127.0.0.1:6385/nodes/Compute0/
                    management/indicators/system/power",
                    "rel": "bookmark"
                }
            ]
        },
        {
            "name": "alert",
            "readonly": false,
            "states": [
                "OFF",
                "BLINKING",
                "UNKNOWN"
            ],
            "links": [
                {
                    "href": "http://127.0.0.1:6385/v1/nodes/Compute0/
                    management/indicators/system/alert",
                    "rel": "self"
                },
                {
                    "href": "http://127.0.0.1:6385/nodes/Compute0/
                    management/indicators/system/alert",
                    "rel": "bookmark"
                }
            ]
        },
    ]
}

  • GET /v1/nodes/<node_ident>/management/indicators/<component>/<ind_ident>

    Retrieves indicator state for the component. Returns a JSON object representing current state of the chosen indicator (ind_ident) sitting on the component.

    The field of the response object is state, the value is one of: OFF, ON, BLINKING or UNKNOWN.

    Error codes:

    • 404 Not Found if node, component or indicator is not found.

    Example response object:

    {
  "state": "ON"
}

  • PUT /v1/nodes/<node_ident>/management/indicators/<component>/<ind_ident>

    Set the state of the desired indicators of the component. The endpoint accepts a JSON object. The following field is mandatory:

    • state requested indicator state

    • 400 Bad Request if state is not an accepted value

    • 404 Not Found if node, component or indicator is not found.

    Example request object:

    {
  "state": "ON"
}

Client (CLI) impact

“ironic” CLI

None.

“openstack baremetal” CLI

The following commands will be created:

openstack baremetal node indicator list <node> [component]
openstack baremetal node indicator show <node> <component> indicator
openstack baremetal node indicator set <node> <component> <indicator> --state {ON,OFF,BLINKING}

The first command lists all indicators for the specified component, or for all components if specific component is not given.

RPC API impact

The new RPC calls are introduced:

  • Listing the indicators

    def get_supported_indicators(self, context, node_id, component=None):
   """Get node hardware components and their indicators.

   :param context: request context.
   :param node_id: node id or uuid.
   :param component: The hardware component, one of
       :mod:`ironic.common.components` or `None` to return all
       available components.
   :returns: a `dict` holding indicator IDs as keys, indicator properties
       as values. Indicator properties is a `dict` that includes:
       `readonly` bool, `states` list containing zero or more values from
       mod:`ironic.common.indicator_states`.
   """

  • Reading the indicator

    def get_indicator_state(self, context, node_id, component, indicator):
    """Get node hardware component indicator state.

    :param context: request context.
    :param node_id: node id or uuid.
    :param component: The hardware component, one of
        :mod:`ironic.common.components`.
    :param indicator: Indicator IDs, as
        reported by `get_supported_indicators`
    :returns: current indicator state. One of the values from
        mod:`ironic.common.indicator_states`.
    """"

  • Setting the indicator

    def set_indicator_state(self, context, node_id, component,
                       indicator, state):
    """Set node hardware components indicator to the desired state.

    :param context: request context.
    :param node_id: node id or uuid.
    :param component: The hardware component, one of
        :mod:`ironic.common.components`.
    :param indicator: Indicator IDs, as
        reported by `get_supported_indicators`)
    :param state: Indicator state, one of
        mod:`ironic.common.indicator_states`.
    """

Driver API impact

Optional indicator API methods is added to ManagementInterface:

  • Listing the indicators

    def get_supported_indicators(self, task, component=None):
    """Get a map of the supported indicators (e.g. LEDs).

    :param task: A task from TaskManager.
    :returns: A dictionary of hardware components
        (:mod:`ironic.common.components`) as keys with indicator
        properties as values. Indicator properties is a `dict`
        that includes: `readonly` bool, `states` list containing
        zero or more values from mod:`ironic.common.indicator_states`.
    """

  • Reading the indicator

    def get_indicator_state(self, task, component, indicator):
    """Get current state of the indicator of the hardware component.

    :param task: A task from TaskManager.
    :param component: The hardware component, one of
        :mod:`ironic.common.components`.
    :param indicator: Indicator ID (as reported by
        `get_supported_indicators`).
    :returns: current indicator state. One of the values from
        mod:`ironic.common.indicator_states`.
    """

  • Setting the indicator

    def set_indicator_state(self, task, component, indicator, state):
    """Set indicator on the hardware component to the desired state.

    :param task: A task from TaskManager.
    :param component: The hardware component, one of
        :mod:`ironic.common.components`.
    :param indicator: Indicator ID (as reported by
        `get_supported_indicators`).
    :state: Desired state of the indicator, one of
        :mod:`ironic.common.indicator_states`.
    """

The above methods are implemented for Redfish and IPMI hardware types.

Nova driver impact

None.

Ramdisk impact

None.

Security impact

None.

Other end user impact

The indicators can be made accessible through Horizon or other UI tools.

Scalability impact

None.

Performance Impact

None.

Other deployer impact

None.

Developer impact

None.

Implementation

Assignee(s)

Primary assignee:

<etingof>

Work Items

  • Add indicator management methods to ironic management interface

  • Add indicator management to ironic ipmi and redfish hardware types

  • Add RPC for indicator management

  • Add REST API endpoint for indicator management

Dependencies

None.

Testing

  • Unit tests and Tempest API will be provided

Upgrades and Backwards Compatibility

This change is fully backward compatible.

Documentation Impact

API reference will be provided.

References