RackManagerV2 types
Messages and enums in the rack_manager_v2 protobuf package. Proto3 has no required fields, so a field that an RPC needs says so in its description. Refer to Conventions for how nodes are described.
Messages
ConfigureScaleUpFabricManagerRequest
ConfigureScaleUpFabricManagerRequest declaratively describes the desired scale-up fabric across a set of switches. RMS reconciles the switches toward this desired state, applying only the changes that are required (for example, it does nothing when the observed configuration matches this request and controller activation is verified). Direct switch requests use each node’s host_endpoint for address and credentials.
ConfigureScaleUpFabricManagerResponse
ConfigureScaleUpFabricManagerResponse returns the handle to the reconciliation job.
NodeSystemValidationJobInfo
NodeSystemValidationJobInfo pairs a node with its async system validation job ID.
ScaleUpFabricConfig
ScaleUpFabricConfig captures the desired scale-up fabric configuration in logical terms, so RMS can apply it regardless of which switch API it uses to configure the fabric manager.
ScaleUpFabricLayout
Describes the fabrics to create from the nodes in the request. Supported sizes and placements depend on the server and physical topology. Physical positions are discovered from the supplied nodes, not specified by offsets.
ScaleUpFabricMembers
Nodes and primary selected for one requested fabric, not observed configuration.
ScaleUpFabricSpec
Target size of one independently managed fabric. RMS selects nonoverlapping node groups. The same inventory and layout produce the same selection; changing either may change the selected nodes.
StartSystemValidationRequest
StartSystemValidationRequest identifies the nodes to scan. NVSSVT reaches each Device Under Test (DUT) over its BMC (required) and, when supplied, its host OS, so every node’s bmc_endpoint must carry credentials.
StartSystemValidationResponse
StartSystemValidationResponse reports the launched scans’ job handles, one per admitted node. The NVSSVT process id and log directory live on the launcher’s side of the pod and are not addressable by callers, so they are reported in RMS’s logs rather than here; each scan’s log directory also appears in its job’s result_json once it completes.