Refactor Decision CRD

api/external/nova/messages.go

@@ -9,6 +9,7 @@ import (
 	"log/slog"
 	"strings"
 
+	"github.com/cobaltcore-dev/cortex/api/v1alpha1"
 	"github.com/cobaltcore-dev/cortex/internal/scheduling/lib"
 )
 
@@ -129,23 +130,21 @@ func (req ExternalSchedulerRequest) GetHypervisorType() (HypervisorType, error)
 	return "", errors.New("hypervisor type not specified in flavor extra specs")
 }
 
-type RequestIntent string
-
 const (
 	// LiveMigrationIntent indicates that the request is intended for live migration.
-	LiveMigrationIntent RequestIntent = "live_migration"
+	LiveMigrationIntent v1alpha1.SchedulingIntent = "live_migration"
 	// RebuildIntent indicates that the request is intended for rebuilding a VM.
-	RebuildIntent RequestIntent = "rebuild"
+	RebuildIntent v1alpha1.SchedulingIntent = "rebuild"
 	// ResizeIntent indicates that the request is intended for resizing a VM.
-	ResizeIntent RequestIntent = "resize"
+	ResizeIntent v1alpha1.SchedulingIntent = "resize"
 	// EvacuateIntent indicates that the request is intended for evacuating a VM.
-	EvacuateIntent RequestIntent = "evacuate"
+	EvacuateIntent v1alpha1.SchedulingIntent = "evacuate"
 	// CreateIntent indicates that the request is intended for creating a new VM.
-	CreateIntent RequestIntent = "create"
+	CreateIntent v1alpha1.SchedulingIntent = "create"
 )
 
 // GetIntent analyzes the request spec and determines the intent of the scheduling request.
-func (req ExternalSchedulerRequest) GetIntent() (RequestIntent, error) {
+func (req ExternalSchedulerRequest) GetIntent() (v1alpha1.SchedulingIntent, error) {
 	str, err := req.Spec.Data.GetSchedulerHintStr("_nova_check_type")
 	if err != nil {
 		return "", err

api/external/nova/messages_test.go

@@ -6,13 +6,15 @@ package api
 import (
 	"encoding/json"
 	"testing"
+
+	"github.com/cobaltcore-dev/cortex/api/v1alpha1"
 )
 
 func TestGetIntent(t *testing.T) {
 	tests := []struct {
 		name           string
 		schedulerHints map[string]any
-		expectedIntent RequestIntent
+		expectedIntent v1alpha1.SchedulingIntent
 		expectError    bool
 	}{
 		{

api/v1alpha1/history_types.go

@@ -0,0 +1,123 @@
+// Copyright SAP SE
+// SPDX-License-Identifier: Apache-2.0
+
+package v1alpha1
+
+import (
+	corev1 "k8s.io/api/core/v1"
+	metav1 "k8s.io/apimachinery/pkg/apis/meta/v1"
+)
+
+// SchedulingIntent defines the intent of a scheduling decision.
+type SchedulingIntent string
+
+// Other intents can be defined by the operators.
+const (
+	// Used as default intent if the operator does not specify one.
+	SchedulingIntentUnknown SchedulingIntent = "Unknown"
+)
+
+type SchedulingHistoryEntry struct {
+	// The timestamp of when the decision was made.
+	Timestamp metav1.Time `json:"timestamp"`
+	// The pipeline that was used for the decision.
+	PipelineRef corev1.ObjectReference `json:"pipelineRef"`
+	// The intent of the decision (e.g., initial scheduling, rescheduling, etc.).
+	Intent SchedulingIntent `json:"intent"`
+	// The top hosts ordered by score for the decision (limited to 3).
+	// This is not a complete list of all candidates — only the highest-ranked
+	// hosts are retained to keep the history compact.
+	// +kubebuilder:validation:Optional
+	// +kubebuilder:validation:MaxItems=3
+	OrderedHosts []string `json:"orderedHosts,omitempty"`
+	// Whether the scheduling decision was successful.
+	// +kubebuilder:validation:Optional
+	Successful bool `json:"successful"`
+}
+
+type HistorySpec struct {
+	// The scheduling domain this object with the history belongs to.
+	SchedulingDomain SchedulingDomain `json:"schedulingDomain"`
+	// The resource ID this history belongs to (e.g., the UUID of a nova instance).
+	ResourceID string `json:"resourceID"`
+	// The availability zone of the resource, if known. Only set for scheduling
+	// domains that provide AZ information (e.g., Nova).
+	// +kubebuilder:validation:Optional
+	AvailabilityZone *string `json:"availabilityZone,omitempty"`
+}
+
+// CurrentDecision holds the full context of the most recent scheduling
+// decision. When a new decision arrives the previous CurrentDecision is
+// compacted into a SchedulingHistoryEntry and appended to History.
+type CurrentDecision struct {
+	// The timestamp of when the decision was made.
+	Timestamp metav1.Time `json:"timestamp"`
+	// The pipeline that was used for the decision.
+	PipelineRef corev1.ObjectReference `json:"pipelineRef"`
+	// The intent of the decision (e.g., initial scheduling, rescheduling, etc.).
+	Intent SchedulingIntent `json:"intent"`
+	// Whether the scheduling decision was successful.
+	Successful bool `json:"successful"`
+	// The target host selected for the resource. nil when no host was found.
+	// +kubebuilder:validation:Optional
+	TargetHost *string `json:"targetHost,omitempty"`
+	// A human-readable explanation of the scheduling decision.
+	// +kubebuilder:validation:Optional
+	Explanation string `json:"explanation,omitempty"`
+	// The top hosts ordered by score (limited to 3).
+	// +kubebuilder:validation:Optional
+	// +kubebuilder:validation:MaxItems=3
+	OrderedHosts []string `json:"orderedHosts,omitempty"`
+}
+
+type HistoryStatus struct {
+	// Current represents the latest scheduling decision with full context.
+	// +kubebuilder:validation:Optional
+	Current CurrentDecision `json:"current,omitempty"`
+	// History of past scheduling decisions (limited to last 10).
+	// +kubebuilder:validation:Optional
+	History []SchedulingHistoryEntry `json:"history,omitempty"`
+
+	// Conditions represent the latest available observations of the history's state.
+	// +kubebuilder:validation:Optional
+	Conditions []metav1.Condition `json:"conditions,omitempty" patchStrategy:"merge" patchMergeKey:"type"`
+}
+
+// +kubebuilder:object:root=true
+// +kubebuilder:subresource:status
+// +kubebuilder:resource:scope=Cluster
+// +kubebuilder:printcolumn:name="Domain",type="string",JSONPath=".spec.schedulingDomain"
+// +kubebuilder:printcolumn:name="Resource ID",type="string",JSONPath=".spec.resourceID"
+// +kubebuilder:printcolumn:name="AZ",type="string",JSONPath=".spec.availabilityZone"
+// +kubebuilder:printcolumn:name="Target Host",type="string",JSONPath=".status.current.targetHost"
+// +kubebuilder:printcolumn:name="Status",type="string",JSONPath=".status.conditions[?(@.type=='Ready')].reason"
+// +kubebuilder:printcolumn:name="Created",type="date",JSONPath=".metadata.creationTimestamp"
+
+// History is the Schema for the history API
+type History struct {
+	metav1.TypeMeta `json:",inline"`
+
+	// Standard object metadata.
+	// +optional
+	metav1.ObjectMeta `json:"metadata,omitempty"`
+
+	// Spec defines the desired state of History.
+	// +required
+	Spec HistorySpec `json:"spec"`
+	// Status defines the observed state of History.
+	// +optional
+	Status HistoryStatus `json:"status,omitempty"`
+}
+
+// +kubebuilder:object:root=true
+
+// HistoryList contains a list of History
+type HistoryList struct {
+	metav1.TypeMeta `json:",inline"`
+	metav1.ListMeta `json:"metadata,omitempty"`
+	Items           []History `json:"items"`
+}
+
+func init() {
+	SchemeBuilder.Register(&History{}, &HistoryList{})
+}

api/v1alpha1/pipeline_types.go

@@ -80,6 +80,9 @@ type PipelineSpec struct {
 
 	// If this pipeline should create decision objects.
 	// When this is false, the pipeline will still process requests.
+	// NOTE: This flag is intentionally kept as "createDecisions" to avoid
+	// breaking changes. It will be renamed when the deprecated Decision CRD
+	// is fully replaced in a future refactoring.
 	// +kubebuilder:default=false
 	CreateDecisions bool `json:"createDecisions,omitempty"`
 

cmd/main.go

@@ -41,7 +41,6 @@ import (
 	"github.com/cobaltcore-dev/cortex/internal/knowledge/extractor"
 	"github.com/cobaltcore-dev/cortex/internal/knowledge/kpis"
 	"github.com/cobaltcore-dev/cortex/internal/scheduling/cinder"
-	"github.com/cobaltcore-dev/cortex/internal/scheduling/explanation"
 	schedulinglib "github.com/cobaltcore-dev/cortex/internal/scheduling/lib"
 	"github.com/cobaltcore-dev/cortex/internal/scheduling/machines"
 	"github.com/cobaltcore-dev/cortex/internal/scheduling/manila"
@@ -443,19 +442,6 @@ func main() {
 			os.Exit(1)
 		}
 	}
-	if slices.Contains(mainConfig.EnabledControllers, "explanation-controller") {
-		// Setup a controller which will reconcile the history and explanation for
-		// decision resources.
-		explanationControllerConfig := conf.GetConfigOrDie[explanation.ControllerConfig]()
-		explanationController := &explanation.Controller{
-			Client: multiclusterClient,
-			Config: explanationControllerConfig,
-		}
-		if err := explanationController.SetupWithManager(mgr, multiclusterClient); err != nil {
-			setupLog.Error(err, "unable to create controller", "controller", "ExplanationController")
-			os.Exit(1)
-		}
-	}
 	if slices.Contains(mainConfig.EnabledControllers, "reservations-controller") {
 		monitor := reservationscontroller.NewControllerMonitor(multiclusterClient)
 		metrics.Registry.MustRegister(&monitor)