> For clean Markdown of any page, append .md to the page URL.
> For a complete documentation index, see https://docs.nvidia.com/rms/llms.txt.

# Firmware Updates


Start firmware updates on one node, many nodes, or from a firmware object.

| RPC | Execution | Summary |
| --- | --- | --- |
| [`UpdateFirmware`](#updatefirmware) | Asynchronous, one job | UpdateFirmware initiates a firmware update on a specific node asynchronously. |
| [`BatchUpdateFirmwareByNodeType`](#batchupdatefirmwarebynodetype) | Asynchronous, parent and child jobs | BatchUpdateFirmwareByNodeType initiates firmware updates on all nodes of a specific type in a rack asynchronously. |
| [`BatchUpdateFirmware`](#batchupdatefirmware) | Asynchronous, parent and child jobs | BatchUpdateFirmware initiates firmware updates on a specified list of nodes asynchronously. |
| [`ApplyStoredFirmwareObject`](#applystoredfirmwareobject) | Asynchronous, parent and child jobs | ApplyStoredFirmwareObject resolves stored firmware targets and starts firmware updates for supplied nodes. |
| [`ApplyFirmwareObject`](#applyfirmwareobject) | Asynchronous, parent and child jobs | ApplyFirmwareObject ingests a firmware object from JSON, downloads required artifacts, resolves firmware targets, and starts firmware updates for supplied nodes. |

For definitions of execution modes and job polling, refer to [Async jobs](/rms/api-reference/async-jobs).

## UpdateFirmware

`rack_manager.RackManager/UpdateFirmware`

UpdateFirmware initiates a firmware update on a specific node asynchronously. Returns a job ID immediately. Use GetFirmwareJobStatus to track progress. Works for all node types (compute, powershelf, switch).

| Property | Value |
| --- | --- |
| Execution | **Asynchronous, one job.** The response returns a job ID and the work continues after the call returns. |
| Job tracking | Poll [`GetFirmwareJobStatus`](/rms/api-reference/rpcs/utility-jobs#getfirmwarejobstatus) with the job ID in `job_id`. |
| Target | **Registered inventory.** Reads or changes nodes and racks stored in RMS inventory. |
| Behavior details | [Operations guide](/rms/documentation/operations/firmware) |
| Protobuf reference | [`UpdateFirmware`](/rms/api-reference/protobuf-reference/rack-manager/rack-manager/rack-manager-update-firmware) |

**Request:** [`UpdateFirmwareRequest`](/rms/api-reference/types/rack-manager#updatefirmwarerequest)

| Field | Type | Description |
| --- | --- | --- |
| `node_id` | `string` | Target node identifier |
| `filename` | `string` | Single firmware filename (used when firmware_targets is empty) |
| `target` | `string` | Single Redfish target URI (used when firmware_targets is empty) |
| `activate` | `bool` | Whether to immediately activate the firmware after installation |
| `rack_id` | `string` | Rack containing the target node |
| `firmware_targets` | repeated [`FirmwareTarget`](/rms/api-reference/types/rack-manager#firmwaretarget) | Multiple (target, filename) pairs for batch component updates |
| `force_update` | `bool` | Force the BMC to accept the firmware even if it's a downgrade |

**Response:** [`UpdateFirmwareResponse`](/rms/api-reference/types/rack-manager#updatefirmwareresponse)

| Field | Type | Description |
| --- | --- | --- |
| `status` | [`ReturnCode`](/rms/api-reference/types/rack-manager#returncode) | RETURN_CODE_SUCCESS if update completed, RETURN_CODE_FAILURE otherwise |
| `message` | `string` | Human-readable status or error message |
| `error_code` | [`FirmwareUpdateError`](/rms/api-reference/types/rack-manager#firmwareupdateerror) | Detailed error code for troubleshooting |
| `job_id` | `string` | Job ID for async operations (use with GetFirmwareJobStatus to track progress) |

**Errors:** Refer to [Conventions](/rms/api-reference/conventions#error-handling) for how RMS reports failures.

<Accordion title="Example request">

Replace each `<placeholder>` with a real value. Define `RMS_GRPCURL_FLAGS` and `RMS_ADDR` as described in [Getting started](/rms/api-reference/getting-started#call-an-rpc-with-grpcurl).

```bash
grpcurl $RMS_GRPCURL_FLAGS \
  -d '{
  "node_id": "<node_id>",
  "filename": "<filename>",
  "target": "<target>",
  "activate": false,
  "rack_id": "<rack_id>",
  "firmware_targets": [
    {
      "target": "<target>",
      "filename": "<filename>"
    }
  ],
  "force_update": false
}' \
  "$RMS_ADDR" rack_manager.RackManager/UpdateFirmware
```

The `rack_manager_client` test client offers an equivalent command:

```bash
rack_manager_client <host> <port> update_firmware <rack_id> <node_id> <target:filename>...
```

</Accordion>

## BatchUpdateFirmwareByNodeType

`rack_manager.RackManager/BatchUpdateFirmwareByNodeType`

BatchUpdateFirmwareByNodeType initiates firmware updates on all nodes of a specific type in a rack asynchronously. Returns a parent job ID for tracking overall batch status, along with individual node job IDs. Use GetFirmwareJobStatus to track progress.

| Property | Value |
| --- | --- |
| Execution | **Asynchronous, parent and child jobs.** The response returns a parent job ID and the work continues after the call returns. |
| Job tracking | Poll [`GetFirmwareJobStatus`](/rms/api-reference/rpcs/utility-jobs#getfirmwarejobstatus) with the job ID in `response.job_id`. Per-node child job IDs are in `jobs`. |
| Target | **Registered inventory.** Reads or changes nodes and racks stored in RMS inventory. |
| Behavior details | [Operations guide](/rms/documentation/operations/firmware) |
| Protobuf reference | [`BatchUpdateFirmwareByNodeType`](/rms/api-reference/protobuf-reference/rack-manager/rack-manager/rack-manager-batch-update-firmware-by-node-type) |

**Request:** [`BatchUpdateFirmwareByNodeTypeRequest`](/rms/api-reference/types/rack-manager#batchupdatefirmwarebynodetyperequest)

| Field | Type | Description |
| --- | --- | --- |
| `node_type` | [`NodeType`](/rms/api-reference/types/rack-manager#nodetype) | Type of nodes to update; non-UNSPECIFIED values override node_descriptor |
| `filename` | `string` | Single firmware filename (used when firmware_targets is empty) |
| `target` | `string` | Single Redfish target URI (used when firmware_targets is empty) |
| `rack_id` | `string` | Target rack identifier |
| `firmware_targets` | repeated [`FirmwareTarget`](/rms/api-reference/types/rack-manager#firmwaretarget) | Multiple (target, filename) pairs for batch component updates |
| `activate` | `bool` | Whether to activate firmware after all targets are flashed |
| `force_update` | `bool` | Force the BMC to accept the firmware even if it's a downgrade |
| `node_descriptor` | optional [`NodeDescriptor`](/rms/api-reference/types/rack-manager#nodedescriptor) | Flexible node identity used when node_type is NODE_TYPE_UNSPECIFIED |

**Response:** [`BatchUpdateFirmwareByNodeTypeResponse`](/rms/api-reference/types/rack-manager#batchupdatefirmwarebynodetyperesponse)

| Field | Type | Description |
| --- | --- | --- |
| `response` | [`NodeBatchResponse`](/rms/api-reference/types/rack-manager#nodebatchresponse) | - |
| `jobs` | repeated [`NodeFirmwareJobInfo`](/rms/api-reference/types/rack-manager#nodefirmwarejobinfo) | Individual job ID per node |

**Errors:** Refer to [Conventions](/rms/api-reference/conventions#error-handling) for how RMS reports failures.

<Accordion title="Example request">

Replace each `<placeholder>` with a real value. Define `RMS_GRPCURL_FLAGS` and `RMS_ADDR` as described in [Getting started](/rms/api-reference/getting-started#call-an-rpc-with-grpcurl).

```bash
grpcurl $RMS_GRPCURL_FLAGS \
  -d '{
  "node_type": "COMPUTE_GB200_NVIDIA",
  "filename": "<filename>",
  "target": "<target>",
  "rack_id": "<rack_id>",
  "firmware_targets": [
    {
      "target": "<target>",
      "filename": "<filename>"
    }
  ],
  "activate": false,
  "force_update": false,
  "node_descriptor": {
    "attributes": {
      "<key>": "<attributes>"
    }
  }
}' \
  "$RMS_ADDR" rack_manager.RackManager/BatchUpdateFirmwareByNodeType
```

The `rack_manager_client` test client offers an equivalent command:

```bash
rack_manager_client <host> <port> batch_update_firmware_by_node_type <rack_id> <node_type> <target:filename>...
```

</Accordion>

## BatchUpdateFirmware

`rack_manager.RackManager/BatchUpdateFirmware`

BatchUpdateFirmware initiates firmware updates on a specified list of nodes asynchronously. Each node is identified by its NodeInfo. Separate firmware targets are provided for compute trays, switch trays, and power shelves, and the appropriate target is applied to each node based on its node type. Returns a parent job ID for tracking overall batch status, along with individual node job IDs. Use GetFirmwareJobStatus to track progress.

| Property | Value |
| --- | --- |
| Execution | **Asynchronous, parent and child jobs.** The response returns a parent job ID and the work continues after the call returns. |
| Job tracking | Poll [`GetFirmwareJobStatus`](/rms/api-reference/rpcs/utility-jobs#getfirmwarejobstatus) with the job ID in `response.job_id`. Per-node child job IDs are in `jobs`. |
| Target | **Caller-supplied nodes.** The request carries each node's endpoints and credentials, and RMS does not store them. |
| Behavior details | [Operations guide](/rms/documentation/operations/firmware) |
| Protobuf reference | [`BatchUpdateFirmware`](/rms/api-reference/protobuf-reference/rack-manager/rack-manager/rack-manager-batch-update-firmware) |

**Request:** [`BatchUpdateFirmwareRequest`](/rms/api-reference/types/rack-manager#batchupdatefirmwarerequest)

| Field | Type | Description |
| --- | --- | --- |
| `nodes` | [`NodeSet`](/rms/api-reference/types/rack-manager#nodeset) | Nodes to update |
| `firmware_targets` | map of `int32` to [`FirmwareTargetList`](/rms/api-reference/types/rack-manager#firmwaretargetlist) | Firmware targets keyed by NodeType enum value |
| `activate` | `bool` | Whether to activate firmware after installation |
| `force_update` | `bool` | Force the BMC to accept the firmware even if it's a downgrade |
| `node_descriptor_firmware_targets` | repeated [`NodeDescriptorFirmwareTargetList`](/rms/api-reference/types/rack-manager#nodedescriptorfirmwaretargetlist) | Firmware targets keyed by flexible node identity |

**Response:** [`BatchUpdateFirmwareResponse`](/rms/api-reference/types/rack-manager#batchupdatefirmwareresponse)

| Field | Type | Description |
| --- | --- | --- |
| `response` | [`NodeBatchResponse`](/rms/api-reference/types/rack-manager#nodebatchresponse) | - |
| `jobs` | repeated [`NodeFirmwareJobInfo`](/rms/api-reference/types/rack-manager#nodefirmwarejobinfo) | Individual job ID per node |

**Errors:** Refer to [Conventions](/rms/api-reference/conventions#error-handling) for how RMS reports failures.

<Accordion title="Example request">

Replace each `<placeholder>` with a real value. Define `RMS_GRPCURL_FLAGS` and `RMS_ADDR` as described in [Getting started](/rms/api-reference/getting-started#call-an-rpc-with-grpcurl).

```bash
grpcurl $RMS_GRPCURL_FLAGS \
  -d '{
  "nodes": {
    "nodes": [
      {
        "node_id": "<node_id>",
        "rack_id": "<rack_id>",
        "type": "COMPUTE_GB200_NVIDIA",
        "bmc_endpoint": {
          "interface": {
            "ip_address": "<ip_address>",
            "mac_address": "<mac_address>",
            "host_name": "<host_name>"
          },
          "port": 0,
          "credentials": {
            "user_pass": {
              "username": "<username>",
              "password": "<password>"
            }
          }
        },
        "host_endpoint": {
          "interface": {
            "ip_address": "<ip_address>",
            "mac_address": "<mac_address>",
            "host_name": "<host_name>"
          },
          "port": 0,
          "credentials": {
            "user_pass": {
              "username": "<username>",
              "password": "<password>"
            }
          }
        },
        "node_descriptor": {
          "attributes": {
            "<key>": "<attributes>"
          }
        },
        "additional_host_endpoints": [
          {
            "interface": {
              "ip_address": "<ip_address>",
              "mac_address": "<mac_address>",
              "host_name": "<host_name>"
            },
            "port": 0,
            "credentials": {
              "user_pass": {
                "username": "<username>",
                "password": "<password>"
              }
            }
          }
        ]
      }
    ]
  },
  "firmware_targets": {
    "<key>": {
      "targets": [
        {
          "target": "<target>",
          "filename": "<filename>"
        }
      ]
    }
  },
  "activate": false,
  "force_update": false,
  "node_descriptor_firmware_targets": [
    {
      "node_descriptor": {
        "attributes": {
          "<key>": "<attributes>"
        }
      },
      "firmware_targets": {
        "targets": [
          {
            "target": "<target>",
            "filename": "<filename>"
          }
        ]
      }
    }
  ]
}' \
  "$RMS_ADDR" rack_manager.RackManager/BatchUpdateFirmware
```

The `rack_manager_client` test client offers an equivalent command:

```bash
rack_manager_client <host> <port> batch_update_firmware <nodes.csv> <targets.csv>
```

</Accordion>

## ApplyStoredFirmwareObject

`rack_manager.RackManager/ApplyStoredFirmwareObject`

ApplyStoredFirmwareObject resolves stored firmware targets and starts firmware updates for supplied nodes.

| Property | Value |
| --- | --- |
| Execution | **Asynchronous, parent and child jobs.** The response returns a parent job ID and the work continues after the call returns. |
| Job tracking | Poll [`GetFirmwareJobStatus`](/rms/api-reference/rpcs/utility-jobs#getfirmwarejobstatus) with the job ID in `response.job_id`. Per-node child job IDs are in `jobs`. |
| Target | **Caller-supplied nodes.** The request carries each node's endpoints and credentials, and RMS does not store them. |
| Behavior details | [Operations guide](/rms/documentation/operations/firmware) |
| Protobuf reference | [`ApplyStoredFirmwareObject`](/rms/api-reference/protobuf-reference/rack-manager/rack-manager/rack-manager-apply-stored-firmware-object) |

**Request:** [`ApplyStoredFirmwareObjectRequest`](/rms/api-reference/types/rack-manager#applystoredfirmwareobjectrequest)

| Field | Type | Description |
| --- | --- | --- |
| `rack_id` | `string` | - |
| `object_id` | `string` | Empty means use default for hardware_type |
| `firmware_type` | `string` | "dev" or "prod" |
| `hardware_type` | `string` | - |
| `components` | repeated `string` | Deprecated: global component filter for all node types |
| `nodes` | [`NodeSet`](/rms/api-reference/types/rack-manager#nodeset) | Caller-supplied nodes |
| `force_update` | `bool` | - |
| `component_filters` | map of `int32` to [`FirmwareObjectComponentFilter`](/rms/api-reference/types/rack-manager#firmwareobjectcomponentfilter) | Component filters keyed by NodeType. If non-empty, only listed node types are updated. A listed node type with an empty components list means all components for that node type. |
| `node_descriptor_component_filters` | repeated [`NodeDescriptorFirmwareObjectComponentFilter`](/rms/api-reference/types/rack-manager#nodedescriptorfirmwareobjectcomponentfilter) | Component filters keyed by flexible node identity |

**Response:** [`ApplyStoredFirmwareObjectResponse`](/rms/api-reference/types/rack-manager#applystoredfirmwareobjectresponse)

| Field | Type | Description |
| --- | --- | --- |
| `response` | [`NodeBatchResponse`](/rms/api-reference/types/rack-manager#nodebatchresponse) | response.job_id is parent job id |
| `object_id` | `string` | - |
| `jobs` | repeated [`NodeFirmwareJobInfo`](/rms/api-reference/types/rack-manager#nodefirmwarejobinfo) | - |

**Errors:** Refer to [Conventions](/rms/api-reference/conventions#error-handling) for how RMS reports failures.

<Accordion title="Example request">

Replace each `<placeholder>` with a real value. Define `RMS_GRPCURL_FLAGS` and `RMS_ADDR` as described in [Getting started](/rms/api-reference/getting-started#call-an-rpc-with-grpcurl).

```bash
grpcurl $RMS_GRPCURL_FLAGS \
  -d '{
  "rack_id": "<rack_id>",
  "object_id": "<object_id>",
  "firmware_type": "<firmware_type>",
  "hardware_type": "<hardware_type>",
  "nodes": {
    "nodes": [
      {
        "node_id": "<node_id>",
        "rack_id": "<rack_id>",
        "type": "COMPUTE_GB200_NVIDIA",
        "bmc_endpoint": {
          "interface": {
            "ip_address": "<ip_address>",
            "mac_address": "<mac_address>",
            "host_name": "<host_name>"
          },
          "port": 0,
          "credentials": {
            "user_pass": {
              "username": "<username>",
              "password": "<password>"
            }
          }
        },
        "host_endpoint": {
          "interface": {
            "ip_address": "<ip_address>",
            "mac_address": "<mac_address>",
            "host_name": "<host_name>"
          },
          "port": 0,
          "credentials": {
            "user_pass": {
              "username": "<username>",
              "password": "<password>"
            }
          }
        },
        "node_descriptor": {
          "attributes": {
            "<key>": "<attributes>"
          }
        },
        "additional_host_endpoints": [
          {
            "interface": {
              "ip_address": "<ip_address>",
              "mac_address": "<mac_address>",
              "host_name": "<host_name>"
            },
            "port": 0,
            "credentials": {
              "user_pass": {
                "username": "<username>",
                "password": "<password>"
              }
            }
          }
        ]
      }
    ]
  },
  "force_update": false,
  "component_filters": {
    "<key>": {
      "components": [
        "<components>"
      ]
    }
  },
  "node_descriptor_component_filters": [
    {
      "node_descriptor": {
        "attributes": {
          "<key>": "<attributes>"
        }
      },
      "component_filter": {
        "components": [
          "<components>"
        ]
      }
    }
  ]
}' \
  "$RMS_ADDR" rack_manager.RackManager/ApplyStoredFirmwareObject
```

</Accordion>

## ApplyFirmwareObject

`rack_manager.RackManager/ApplyFirmwareObject`

ApplyFirmwareObject ingests a firmware object from JSON, downloads required artifacts, resolves firmware targets, and starts firmware updates for supplied nodes.

RMS treats the ingested firmware object as ephemeral for this request. It uses access_token only for artifact download and does not persist the token or object.

| Property | Value |
| --- | --- |
| Execution | **Asynchronous, parent and child jobs.** The response returns a parent job ID and the work continues after the call returns. |
| Job tracking | Poll [`GetFirmwareJobStatus`](/rms/api-reference/rpcs/utility-jobs#getfirmwarejobstatus) with the job ID in `response.job_id`. Per-node child job IDs are in `jobs`. |
| Target | **Caller-supplied nodes.** The request carries each node's endpoints and credentials, and RMS does not store them. |
| Behavior details | [Operations guide](/rms/documentation/operations/firmware) |
| Protobuf reference | [`ApplyFirmwareObject`](/rms/api-reference/protobuf-reference/rack-manager/rack-manager/rack-manager-apply-firmware-object) |

**Request:** [`ApplyFirmwareObjectRequest`](/rms/api-reference/types/rack-manager#applyfirmwareobjectrequest)

| Field | Type | Description |
| --- | --- | --- |
| `rack_id` | `string` | - |
| `config_json` | `string` | SOT JSON containing the firmware object ID and artifacts |
| `access_token` | optional `string` | Used only for artifact download; RMS does not persist it |
| `firmware_type` | `string` | "dev" or "prod" |
| `hardware_type` | `string` | - |
| `nodes` | [`NodeSet`](/rms/api-reference/types/rack-manager#nodeset) | Caller-supplied nodes |
| `force_update` | `bool` | - |
| `component_filters` | map of `int32` to [`FirmwareObjectComponentFilter`](/rms/api-reference/types/rack-manager#firmwareobjectcomponentfilter) | Component filters keyed by NodeType. If non-empty, only listed node types are updated. A listed node type with an empty components list means all components for that node type. |
| `node_descriptor_component_filters` | repeated [`NodeDescriptorFirmwareObjectComponentFilter`](/rms/api-reference/types/rack-manager#nodedescriptorfirmwareobjectcomponentfilter) | Component filters keyed by flexible node identity |

**Response:** [`ApplyFirmwareObjectResponse`](/rms/api-reference/types/rack-manager#applyfirmwareobjectresponse)

| Field | Type | Description |
| --- | --- | --- |
| `response` | [`NodeBatchResponse`](/rms/api-reference/types/rack-manager#nodebatchresponse) | response.job_id is parent job id |
| `object_id` | `string` | - |
| `jobs` | repeated [`NodeFirmwareJobInfo`](/rms/api-reference/types/rack-manager#nodefirmwarejobinfo) | - |

**Errors:** Refer to [Conventions](/rms/api-reference/conventions#error-handling) for how RMS reports failures.

<Accordion title="Example request">

Replace each `<placeholder>` with a real value. Define `RMS_GRPCURL_FLAGS` and `RMS_ADDR` as described in [Getting started](/rms/api-reference/getting-started#call-an-rpc-with-grpcurl).

```bash
grpcurl $RMS_GRPCURL_FLAGS \
  -d '{
  "rack_id": "<rack_id>",
  "config_json": "<config_json>",
  "access_token": "<access_token>",
  "firmware_type": "<firmware_type>",
  "hardware_type": "<hardware_type>",
  "nodes": {
    "nodes": [
      {
        "node_id": "<node_id>",
        "rack_id": "<rack_id>",
        "type": "COMPUTE_GB200_NVIDIA",
        "bmc_endpoint": {
          "interface": {
            "ip_address": "<ip_address>",
            "mac_address": "<mac_address>",
            "host_name": "<host_name>"
          },
          "port": 0,
          "credentials": {
            "user_pass": {
              "username": "<username>",
              "password": "<password>"
            }
          }
        },
        "host_endpoint": {
          "interface": {
            "ip_address": "<ip_address>",
            "mac_address": "<mac_address>",
            "host_name": "<host_name>"
          },
          "port": 0,
          "credentials": {
            "user_pass": {
              "username": "<username>",
              "password": "<password>"
            }
          }
        },
        "node_descriptor": {
          "attributes": {
            "<key>": "<attributes>"
          }
        },
        "additional_host_endpoints": [
          {
            "interface": {
              "ip_address": "<ip_address>",
              "mac_address": "<mac_address>",
              "host_name": "<host_name>"
            },
            "port": 0,
            "credentials": {
              "user_pass": {
                "username": "<username>",
                "password": "<password>"
              }
            }
          }
        ]
      }
    ]
  },
  "force_update": false,
  "component_filters": {
    "<key>": {
      "components": [
        "<components>"
      ]
    }
  },
  "node_descriptor_component_filters": [
    {
      "node_descriptor": {
        "attributes": {
          "<key>": "<attributes>"
        }
      },
      "component_filter": {
        "components": [
          "<components>"
        ]
      }
    }
  ]
}' \
  "$RMS_ADDR" rack_manager.RackManager/ApplyFirmwareObject
```

</Accordion>