constants

package
v0.10.2 Latest Latest
Warning

This package is not in the latest version of its module.

 Go to latest Published: Jun 25, 2023 License: Apache-2.0 Imports: 7 Imported by: 11

Documentation

Index

Constants

View Source 
const (
	RouterHeadersPropagateEnvVar = "PROPAGATE_HEADERS"
	InferenceGraphLabel          = "serving.kserve.io/inferencegraph"
)

InferenceGraph Constants

View Source 
const (
	AgentContainerName    = "agent"
	AgentConfigMapKeyName = "agent"
	AgentEnableFlag       = "--enable-puller"
	AgentConfigDirArgName = "--config-dir"
	AgentModelDirArgName  = "--model-dir"
)

Model agent Constants

View Source 
const (
	CustomSpecStorageUriEnvVarKey                     = "STORAGE_URI"
	CustomSpecProtocolEnvVarKey                       = "PROTOCOL"
	CustomSpecMultiModelServerEnvVarKey               = "MULTI_MODEL_SERVER"
	KServeContainerPrometheusMetricsPortEnvVarKey     = "KSERVE_CONTAINER_PROMETHEUS_METRICS_PORT"
	KServeContainerPrometheusMetricsPathEnvVarKey     = "KSERVE_CONTAINER_PROMETHEUS_METRICS_PATH"
	QueueProxyAggregatePrometheusMetricsPortEnvVarKey = "AGGREGATE_PROMETHEUS_METRICS_PORT"
)

InferenceService Environment Variables

View Source 
const (
	KnativeLocalGateway   = "knative-serving/knative-local-gateway"
	KnativeIngressGateway = "knative-serving/knative-ingress-gateway"
	VisibilityLabel       = "networking.knative.dev/visibility"
)

Knative constants

View Source 
const (
	InferenceServiceDefaultHttpPort     = "8080"
	InferenceServiceDefaultAgentPortStr = "9081"
	InferenceServiceDefaultAgentPort    = 9081
	CommonDefaultHttpPort               = 80
)

InferenceService Endpoint Ports

View Source 
const (
	KServiceComponentLabel = "component"
	KServiceModelLabel     = "model"
	KServiceEndpointLabel  = "endpoint"
)

Labels to put on kservice

View Source 
const (
	ParentInferenceServiceLabel = "inferenceservice"
	InferenceServiceLabel       = "serving.kserve.io/inferenceservice"
)

Labels for TrainedModel

View Source 
const (
	InferenceServiceDefault = "default"
	InferenceServiceCanary  = "canary"
)

InferenceService default/canary constants

View Source 
const (
	ArgumentModelName      = "--model_name"
	ArgumentModelDir       = "--model_dir"
	ArgumentModelClassName = "--model_class_name"
	ArgumentPredictorHost  = "--predictor_host"
	ArgumentHttpPort       = "--http_port"
	ArgumentWorkers        = "--workers"
)

InferenceService model server args

View Source 
const (
	InferenceServiceContainerName   = "kserve-container"
	StorageInitializerContainerName = "storage-initializer"
)

InferenceService container name

View Source 
const (
	ModelConfigVolumeName = "model-config"
	ModelDirVolumeName    = "model-dir"
	ModelConfigDir        = "/mnt/configs"
	ModelDir              = DefaultModelLocalMountPath
)

Multi-model InferenceService

View Source 
const (
	SKLearnServer = "kserve-sklearnserver"
	MLServer      = "kserve-mlserver"
	TFServing     = "kserve-tensorflow-serving"
	XGBServer     = "kserve-xgbserver"
	TorchServe    = "kserve-torchserve"
	TritonServer  = "kserve-tritonserver"
	PMMLServer    = "kserve-pmmlserver"
	LGBServer     = "kserve-lgbserver"
	PaddleServer  = "kserve-paddleserver"
)

built-in runtime servers

View Source 
const (
	ModelClassLabel = "modelClass"
	ServiceEnvelope = "serviceEnvelope"
)
View Source 
const (
	MLServerModelClassSKLearn  = "mlserver_sklearn.SKLearnModel"
	MLServerModelClassXGBoost  = "mlserver_xgboost.XGBoostModel"
	MLServerModelClassLightGBM = "mlserver_lightgbm.LightGBMModel"
	MLServerModelClassMLFlow   = "mlserver_mlflow.MLflowRuntime"
)

allowed model class implementation in mlserver

View Source 
const (
	ServiceEnvelopeKServe   = "kserve"
	ServiceEnvelopeKServeV2 = "kservev2"
)

torchserve service envelope label allowed values

View Source 
const (
	SupportedModelSKLearn    = "sklearn"
	SupportedModelTensorflow = "tensorflow"
	SupportedModelXGBoost    = "xgboost"
	SupportedModelPyTorch    = "pytorch"
	SupportedModelONNX       = "onnx"
	SupportedModelPMML       = "pmml"
	SupportedModelLightGBM   = "lightgbm"
	SupportedModelPaddle     = "paddle"
	SupportedModelTriton     = "triton"
	SupportedModelMLFlow     = "mlflow"
)

supported model type

View Source 
const (
	RevisionLabel         = "serving.knative.dev/revision"
	RawDeploymentAppLabel = "app"
)

revision label

View Source 
const (
	StateReasonRunning          = "Running"
	StateReasonCompleted        = "Completed"
	StateReasonError            = "Error"
	StateReasonCrashLoopBackOff = "CrashLoopBackOff"
)

container state reason

View Source 
const (
	MLServerHTTPPortEnv            = "MLSERVER_HTTP_PORT"
	MLServerGRPCPortEnv            = "MLSERVER_GRPC_PORT"
	MLServerLoadModelsStartupEnv   = "MLSERVER_LOAD_MODELS_AT_STARTUP"
	MLServerModelsDirEnv           = "MODELS_DIR"
	MLServerModelImplementationEnv = "MLSERVER_MODEL_IMPLEMENTATION"
	MLServerModelNameEnv           = "MLSERVER_MODEL_NAME"
	MLServerModelURIEnv            = "MLSERVER_MODEL_URI"

	MLServerSKLearnImplementation = "mlserver_sklearn.SKLearnModel"
	MLServerXGBoostImplementation = "mlserver_xgboost.XGBoostModel"
)
View Source 
const DefaultModelLocalMountPath = "/mnt/models"

DefaultModelLocalMountPath is where models will be mounted by the storage-initializer

View Source 
const (
	NvidiaGPUResourceType = "nvidia.com/gpu"
)

GPU Constants

Variables

View Source 
var (
	KServeName                     = "kserve"
	KServeAPIGroupName             = "serving.kserve.io"
	KnativeAutoscalingAPIGroupName = "autoscaling.knative.dev"
	KnativeServingAPIGroupName     = "serving.knative.dev"
	KServeNamespace                = getEnvOrDefault("POD_NAMESPACE", "kserve")
	KServeDefaultVersion           = "v0.5.0"
)

KServe Constants

View Source 
var (
	InferenceServiceName          = "inferenceservice"
	InferenceServiceAPIName       = "inferenceservices"
	InferenceServicePodLabelKey   = KServeAPIGroupName + "/" + InferenceServiceName
	InferenceServiceConfigMapName = "inferenceservice-config"
)

InferenceService Constants

View Source 
var (
	InferenceServiceGKEAcceleratorAnnotationKey = KServeAPIGroupName + "/gke-accelerator"
	DeploymentMode                              = KServeAPIGroupName + "/deploymentMode"
	EnableRoutingTagAnnotationKey               = KServeAPIGroupName + "/enable-tag-routing"
	AutoscalerClass                             = KServeAPIGroupName + "/autoscalerClass"
	AutoscalerMetrics                           = KServeAPIGroupName + "/metrics"
	TargetUtilizationPercentage                 = KServeAPIGroupName + "/targetUtilizationPercentage"
	MinScaleAnnotationKey                       = KnativeAutoscalingAPIGroupName + "/min-scale"
	MaxScaleAnnotationKey                       = KnativeAutoscalingAPIGroupName + "/max-scale"
	RollOutDurationAnnotationKey                = KnativeServingAPIGroupName + "/rollout-duration"
	EnableMetricAggregation                     = KServeAPIGroupName + "/enable-metric-aggregation"
	SetPrometheusAnnotation                     = KServeAPIGroupName + "/enable-prometheus-scraping"
	KserveContainerPrometheusPortKey            = "prometheus.kserve.io/port"
	KServeContainerPrometheusPathKey            = "prometheus.kserve.io/path"
	PrometheusPortAnnotationKey                 = "prometheus.io/port"
	PrometheusPathAnnotationKey                 = "prometheus.io/path"
	DefaultPrometheusPath                       = "/metrics"
	QueueProxyAggregatePrometheusMetricsPort    = "9088"
	DefaultPodPrometheusPort                    = "9090"
)

InferenceService Annotations

View Source 
var (
	InferenceServiceInternalAnnotationsPrefix        = "internal." + KServeAPIGroupName
	StorageInitializerSourceUriInternalAnnotationKey = InferenceServiceInternalAnnotationsPrefix + "/storage-initializer-sourceuri"
	StorageSpecAnnotationKey                         = InferenceServiceInternalAnnotationsPrefix + "/storage-spec"
	StorageSpecParamAnnotationKey                    = InferenceServiceInternalAnnotationsPrefix + "/storage-spec-param"
	StorageSpecKeyAnnotationKey                      = InferenceServiceInternalAnnotationsPrefix + "/storage-spec-key"
	LoggerInternalAnnotationKey                      = InferenceServiceInternalAnnotationsPrefix + "/logger"
	LoggerSinkUrlInternalAnnotationKey               = InferenceServiceInternalAnnotationsPrefix + "/logger-sink-url"
	LoggerModeInternalAnnotationKey                  = InferenceServiceInternalAnnotationsPrefix + "/logger-mode"
	BatcherInternalAnnotationKey                     = InferenceServiceInternalAnnotationsPrefix + "/batcher"
	BatcherMaxBatchSizeInternalAnnotationKey         = InferenceServiceInternalAnnotationsPrefix + "/batcher-max-batchsize"
	BatcherMaxLatencyInternalAnnotationKey           = InferenceServiceInternalAnnotationsPrefix + "/batcher-max-latency"
	BatcherTimeoutInternalAnnotationKey              = InferenceServiceInternalAnnotationsPrefix + "/batcher-timeout"
	AgentShouldInjectAnnotationKey                   = InferenceServiceInternalAnnotationsPrefix + "/agent"
	AgentModelConfigVolumeNameAnnotationKey          = InferenceServiceInternalAnnotationsPrefix + "/configVolumeName"
	AgentModelConfigMountPathAnnotationKey           = InferenceServiceInternalAnnotationsPrefix + "/configMountPath"
	AgentModelDirAnnotationKey                       = InferenceServiceInternalAnnotationsPrefix + "/modelDir"
	PredictorHostAnnotationKey                       = InferenceServiceInternalAnnotationsPrefix + "/predictor-host"
	PredictorProtocolAnnotationKey                   = InferenceServiceInternalAnnotationsPrefix + "/predictor-protocol"
)

InferenceService Internal Annotations

View Source 
var (
	DefaultStorageSpecSecret     = "storage-config"
	DefaultStorageSpecSecretPath = "/mnt/storage-secret"
)

StorageSpec Constants

View Source 
var (
	ControllerLabelName = KServeName + "-controller-manager"
	DefaultMinReplicas  = 1
)

Controller Constants

View Source 
var (
	ServiceAnnotationDisallowedList = []string{
		autoscaling.MinScaleAnnotationKey,
		autoscaling.MaxScaleAnnotationKey,
		StorageInitializerSourceUriInternalAnnotationKey,
		"kubectl.kubernetes.io/last-applied-configuration",
	}

	RevisionTemplateLabelDisallowedList = []string{
		VisibilityLabel,
	}
)
View Source 
var (
	MLServerISGRPCPort = int32(9000)
	MLServerISRestPort = int32(8080)
)
View Source 
var AutoScalerKPAMetricsAllowedList = []AutoScalerKPAMetricsType{
	AutoScalerKPAMetricsConcurrency,
	AutoScalerKPAMetricsRPS,
}

Autoscaler KPA Metrics Allowed List

View Source 
var AutoscalerAllowedClassList = []AutoscalerClassType{
	AutoscalerClassHPA,
}

Autoscaler Class Allowed List

View Source 
var AutoscalerAllowedMetricsList = []AutoscalerMetricsType{
	AutoScalerMetricsCPU,
	AutoScalerMetricsMemory,
}

Autoscaler Metrics Allowed List

View Source 
var (
	DefaultAutoscalerClass = AutoscalerClassHPA
)

Autoscaler Default Class

View Source 
var (
	DefaultCPUUtilization int32 = 80
)

Autoscaler Default Metrics Value

View Source 
var (
	LocalGatewayHost = "knative-local-gateway.istio-system.svc." + network.GetClusterDomainName()
)
View Source 
var (
	ModelConfigFileName = "models.json"
)

InferenceService MultiModel Constants

View Source 
var (
	PodMutatorWebhookName = KServeName + "-pod-mutator-webhook"
)

Webhook Constants

View Source 
var (
	TrainedModelAllocated = KServeAPIGroupName + "/" + "trainedmodel-allocated"
)

TrainedModel Constants

Functions

func CanaryExplainerServiceName added in v0.1.3 

func CanaryExplainerServiceName(name string) string

func CanaryPredictorServiceName added in v0.1.3 

func CanaryPredictorServiceName(name string) string

func CanaryServiceName added in v0.1.3 

func CanaryServiceName(name string, component InferenceServiceComponent) string

func CanaryTransformerServiceName added in v0.1.3 

func CanaryTransformerServiceName(name string) string

func DefaultExplainerServiceName added in v0.1.3 

func DefaultExplainerServiceName(name string) string

func DefaultPredictorServiceName added in v0.1.3 

func DefaultPredictorServiceName(name string) string

func DefaultServiceName added in v0.1.3 

func DefaultServiceName(name string, component InferenceServiceComponent) string

func DefaultTransformerServiceName added in v0.1.3 

func DefaultTransformerServiceName(name string) string

func ExplainPath added in v0.7.0 

func ExplainPath(name string) string

func ExplainPrefix added in v0.3.0 

func ExplainPrefix() string

func GetRawServiceLabel added in v0.7.0 

func GetRawServiceLabel(service string) string

GetRawServiceLabel generate native service label

func HostRegExp added in v0.3.0 

func HostRegExp(host string) string

HostRegExp returns an ECMAScript regular expression to match either host or host:<any port> for clusterLocalHost, we will also match the prefixes.

func InferenceServiceHostName added in v0.3.0 

func InferenceServiceHostName(name string, namespace string, domain string) string

func InferenceServicePrefix added in v0.3.0 

func InferenceServicePrefix(name string) string

func InferenceServiceURL added in v0.3.0 

func InferenceServiceURL(scheme, name, namespace, domain string) string

func ModelConfigName added in v0.7.0 

func ModelConfigName(inferenceserviceName string, shardId int) string

func PredictPath added in v0.7.0 

func PredictPath(name string, protocol InferenceServiceProtocol) string

func PredictPrefix added in v0.3.0 

func PredictPrefix() string

func PredictorURL added in v0.3.0 

func PredictorURL(metadata metav1.ObjectMeta, isCanary bool) string

func TransformerURL added in v0.3.0 

func TransformerURL(metadata metav1.ObjectMeta, isCanary bool) string

func VirtualServiceHostname added in v0.7.0 

func VirtualServiceHostname(name string, predictorHostName string) string

Types

type AutoScalerKPAMetricsType added in v0.9.0 

type AutoScalerKPAMetricsType string
var (
	AutoScalerKPAMetricsRPS         AutoScalerKPAMetricsType = "rps"
	AutoScalerKPAMetricsConcurrency AutoScalerKPAMetricsType = "concurrency"
)

type AutoscalerClassType added in v0.7.0 

type AutoscalerClassType string
var (
	AutoscalerClassHPA AutoscalerClassType = "hpa"
)

Autoscaler Class

type AutoscalerMetricsType added in v0.7.0 

type AutoscalerMetricsType string
var (
	AutoScalerMetricsCPU AutoscalerMetricsType = "cpu"
)

Autoscaler Metrics 

var (
	AutoScalerMetricsMemory AutoscalerMetricsType = "memory"
)

Autoscaler Memory metrics

type CheckResultType added in v0.7.0 

type CheckResultType int

CheckResultType raw k8s deployment, resource exist check result 

const (
	CheckResultCreate  CheckResultType = 0
	CheckResultUpdate  CheckResultType = 1
	CheckResultExisted CheckResultType = 2
	CheckResultUnknown CheckResultType = 3
)

type DeploymentModeType added in v0.7.0 

type DeploymentModeType string
const (
	Serverless          DeploymentModeType = "Serverless"
	RawDeployment       DeploymentModeType = "RawDeployment"
	ModelMeshDeployment DeploymentModeType = "ModelMesh"
)

type InferenceServiceComponent added in v0.3.0 

type InferenceServiceComponent string
const (
	Predictor   InferenceServiceComponent = "predictor"
	Explainer   InferenceServiceComponent = "explainer"
	Transformer InferenceServiceComponent = "transformer"
)

InferenceService Component enums

func (InferenceServiceComponent) String added in v0.3.0 

func (e InferenceServiceComponent) String() string

type InferenceServiceProtocol added in v0.7.0 

type InferenceServiceProtocol string
const (
	ProtocolV1      InferenceServiceProtocol = "v1"
	ProtocolV2      InferenceServiceProtocol = "v2"
	ProtocolGRPCV1  InferenceServiceProtocol = "grpc-v1"
	ProtocolGRPCV2  InferenceServiceProtocol = "grpc-v2"
	ProtocolUnknown InferenceServiceProtocol = ""
)

InferenceService protocol enums

func GetProtocolVersionString added in v0.9.0 

func GetProtocolVersionString(protocol ProtocolVersion) InferenceServiceProtocol

type InferenceServiceVerb added in v0.3.0 

type InferenceServiceVerb string
const (
	Predict InferenceServiceVerb = "predict"
	Explain InferenceServiceVerb = "explain"
)

InferenceService verb enums

func (InferenceServiceVerb) String added in v0.3.0 

func (v InferenceServiceVerb) String() string

type ProtocolVersion added in v0.9.0 

type ProtocolVersion int
const (
	V1 ProtocolVersion
	V2
	GRPCV1
	GRPCV2
	Unknown
)

func GetProtocolVersionInt added in v0.9.0 

func GetProtocolVersionInt(protocol InferenceServiceProtocol) ProtocolVersion

Source Files

View all Source files

Jump to

Keyboard shortcuts

? : This menu
/ : Search site
f or F : Jump to
y or Y : Canonical URL
go.dev uses cookies from Google to deliver and enhance the quality of its services and to analyze traffic. Learn more.