Documentation ¶
Index ¶
- Constants
- func GetResourceOwnershipLabelsForHostingDeployment(deployment hostingv1.HostingDeployment) map[string]string
- func GetSubresourceNamespacedName(objectName string, hostingDeployment hostingv1.HostingDeployment) types.NamespacedName
- type EndpointConfigReconciler
- type EndpointConfigReconcilerProvider
- type HostingDeploymentReconciler
- type ModelReconciler
- type ModelReconcilerProvider
Constants ¶
const ( ReconcilingEndpointStatus = "ReconcilingEndpoint" // Defines the maximum number of characters in a SageMaker Hosting Deployment Resource name MaxResourceNameLength = 63 )
Variables ¶
This section is empty.
Functions ¶
func GetResourceOwnershipLabelsForHostingDeployment ¶
func GetResourceOwnershipLabelsForHostingDeployment(deployment hostingv1.HostingDeployment) map[string]string
Helper method to get Kubernetes labels that are applied to all created models. These labels contain the namespaced-name of the HostingDeployment, allowing it to retrieve all owned models. This is necessary when the spec is updated and a model is completely deleted from the spec. In order to delete the model in Kubernetes, the HostingDeployment controller needs to be able to list all models that it created.
func GetSubresourceNamespacedName ¶
func GetSubresourceNamespacedName(objectName string, hostingDeployment hostingv1.HostingDeployment) types.NamespacedName
Get the Kubernetes name of the Model/EndpointConfig. This must be idempotent so that future reconciler invocations are able to find the object. Kubernetes resources can have names up to 253 characters long. The characters allowed in names are: digits (0-9), lower case letters (a-z), -, and .
Types ¶
type EndpointConfigReconciler ¶
type EndpointConfigReconciler interface { Reconcile(ctx context.Context, desiredDeployment *hostingv1.HostingDeployment, shouldDeleteUnusedResources bool) error GetSageMakerEndpointConfigName(ctx context.Context, desiredDeployment *hostingv1.HostingDeployment) (string, error) }
Helper type that is responsible for reconciling EndpointConfigs of an endpoint.
func NewEndpointConfigReconciler ¶
func NewEndpointConfigReconciler(client client.Client, log logr.Logger) EndpointConfigReconciler
Helper method to create a EndpointConfigReconciler.
type EndpointConfigReconcilerProvider ¶
type EndpointConfigReconcilerProvider func(client.Client, logr.Logger) EndpointConfigReconciler
Type that returns an instantiated EndpointConfigReconciler given parameters.
type HostingDeploymentReconciler ¶
type HostingDeploymentReconciler struct { client.Client Log logr.Logger PollInterval time.Duration // contains filtered or unexported fields }
HostingDeploymentReconciler reconciles a HostingDeployment object
func (*HostingDeploymentReconciler) SetupWithManager ¶
func (r *HostingDeploymentReconciler) SetupWithManager(mgr ctrl.Manager) error
type ModelReconciler ¶
type ModelReconciler interface { Reconcile(ctx context.Context, desiredDeployment *hostingv1.HostingDeployment, shouldDeleteUnusedModels bool) error GetSageMakerModelNames(ctx context.Context, desiredDeployment *hostingv1.HostingDeployment) (map[string]*string, error) }
Helper type that is responsible for reconciling models of an endpoint.
func NewModelReconciler ¶
func NewModelReconciler(client client.Client, log logr.Logger) ModelReconciler
Helper method to create a ModelReconciler.
type ModelReconcilerProvider ¶
type ModelReconcilerProvider func(client.Client, logr.Logger) ModelReconciler
Type that returns an instantiated ModelReconciler given parameters.