infiniband

Documentation

Overview

Package infiniband monitors the infiniband status of the system. Optional, enabled if the host has NVIDIA GPUs.

Index

• Constants
• func New(ctx context.Context, cfg Config) components.Component
• type Config
  • func ParseConfig(b any, db *sql.DB) (*Config, error)
  • func (cfg *Config) Validate() error
• type ExpectedPortStates
• type Output
  • func ParseOutputJSON(data []byte) (*Output, error)
  • func ParseStateIbstat(m map[string]string) (*Output, error)
  • func ParseStatesToOutput(states ...components.State) (*Output, error)
  • func ToOutput(i *nvidia_query.Output) *Output
  • func (o *Output) Evaluate(cfg Config) (string, bool, error)
  • func (o *Output) JSON() ([]byte, error)
  • func (o *Output) States(cfg Config) ([]components.State, error)

Constants

const (
	StateNameIbstat = "ibstat"

	StateKeyIbstatData           = "data"
	StateKeyIbstatEncoding       = "encoding"
	StateValueIbstatEncodingJSON = "json"
)

Functions

func New

func New(ctx context.Context, cfg Config) components.Component

Types

type Config

type Config struct {
	Query query_config.Config `json:"query"`

	ExpectedPortStates
}

func ParseConfig

func ParseConfig(b any, db *sql.DB) (*Config, error)

func (*Config) Validate

func (cfg *Config) Validate() error

type ExpectedPortStates

type ExpectedPortStates struct {
	// The number of ports expected to be "Active" and "LinkUp".
	// If not set, it defaults to the number of GPUs.
	PortCount int `json:"port_count"`

	// The expected rate in Gb/sec.
	// If not set, it defaults to 200.
	Rate int `json:"rate"`
}

Configures the expected state of the ports.

type Output

type Output struct {
	// GPUProductName is the product name of the GPU.
	// Useful to ignore infiniband states for non-infiniband supported GPUs (e.g., GTX 4090).
	GPUProductName string `json:"gpu_product_name"`

	// Represents the number of GPUs in the system.
	// This is used to determine how many ibstat cards at certain rate are expected.
	GPUCount int `json:"gpu_count"`

	InfinibandClassExists bool                    `json:"infiniband_class_exists"`
	IbstatExists          bool                    `json:"ibstat_exists"`
	Ibstat                infiniband.IbstatOutput `json:"ibstat"`
}

func ParseOutputJSON

func ParseOutputJSON(data []byte) (*Output, error)

func ParseStateIbstat

func ParseStateIbstat(m map[string]string) (*Output, error)

func ParseStatesToOutput

func ParseStatesToOutput(states ...components.State) (*Output, error)

func ToOutput

func ToOutput(i *nvidia_query.Output) *Output

ToOutput converts nvidia_query.Output to Output. It returns an empty non-nil object, if the input or the required field is nil (e.g., i.SMI).

func (*Output) Evaluate

func (o *Output) Evaluate(cfg Config) (string, bool, error)

Returns the output evaluation reason and its healthy-ness.

func (*Output) JSON

func (o *Output) JSON() ([]byte, error)

func (*Output) States

func (o *Output) States(cfg Config) ([]components.State, error)