add priority to admit request (#2251)

* add priority to admit request

* add inference obj spec to llmrequest

* make requestObjective an internal type

* make requestObjective an internal type 2

* make requestObjective an internal type 3

* make requestObjective an internal type 4

* make requestObjective an internal type 5

Author: kaushikmitr

pkg/epp/framework/interface/scheduling/types.go

@@ -29,6 +29,11 @@ import (
 
 const nilString = "<nil>"
 
+// RequestObjectives represents the scheduling objectives parsed from the InferenceObjectiveSpec, to be used in scheduling decisions.
+type RequestObjectives struct {
+	Priority int
+}
+
 // LLMRequest is a structured representation of the fields we parse out of the LLMRequest body.
 type LLMRequest struct {
 	// RequestId is the Envoy generated Id for the request being processed
@@ -39,6 +44,8 @@ type LLMRequest struct {
 	Body *LLMRequestBody
 	// Headers is a map of the request headers.
 	Headers map[string]string
+	// Request Objective
+	Objectives RequestObjectives
 }
 
 func (r *LLMRequest) String() string {

pkg/epp/requestcontrol/director.go

@@ -149,13 +149,15 @@ func (d *Director) HandleRequest(ctx context.Context, reqCtx *handlers.RequestCo
 
 	// Parse inference objective.
 	infObjective := d.getInferenceObjective(ctx, reqCtx)
+	requestObjectives := fwksched.RequestObjectives{Priority: *infObjective.Spec.Priority}
 
 	// Prepare LLMRequest (needed for both saturation detection and Scheduler)
 	reqCtx.SchedulingRequest = &fwksched.LLMRequest{
 		RequestId:   reqCtx.Request.Headers[requtil.RequestIdHeaderKey],
 		TargetModel: reqCtx.TargetModelName,
 		Body:        requestBody,
 		Headers:     reqCtx.Request.Headers,
+		Objectives:  requestObjectives,
 	}
 
 	logger = logger.WithValues("objectiveKey", reqCtx.ObjectiveKey, "incomingModelName", reqCtx.IncomingModelName, "targetModelName", reqCtx.TargetModelName, "priority", infObjective.Spec.Priority)

pkg/epp/requestcontrol/request_control_config.go

@@ -103,6 +103,9 @@ func (c *Config) AddPlugins(pluginObjects ...plugin.Plugin) {
 		if prepareDataPlugin, ok := plugin.(fwk.PrepareDataPlugin); ok {
 			c.prepareDataPlugins = append(c.prepareDataPlugins, prepareDataPlugin)
 		}
+		if admissionPlugin, ok := plugin.(fwk.AdmissionPlugin); ok {
+			c.admissionPlugins = append(c.admissionPlugins, admissionPlugin)
+		}
 	}
 }