feat: record events for workload

Author: 0x5457

cmd/main.go

@@ -258,6 +258,7 @@ func main() {
 		Client:    mgr.GetClient(),
 		Scheme:    mgr.GetScheme(),
 		Scheduler: scheduler,
+		Recorder:  mgr.GetEventRecorderFor("tensorfusionworkload"),
 	}).SetupWithManager(mgr); err != nil {
 		setupLog.Error(err, "unable to create controller", "controller", "TensorFusionWorkload")
 		os.Exit(1)

internal/controller/tensorfusionworkload_controller.go

@@ -25,6 +25,7 @@ import (
 	"k8s.io/apimachinery/pkg/api/errors"
 	"k8s.io/apimachinery/pkg/runtime"
 	"k8s.io/apimachinery/pkg/types"
+	"k8s.io/client-go/tools/record"
 	ctrl "sigs.k8s.io/controller-runtime"
 	"sigs.k8s.io/controller-runtime/pkg/client"
 	"sigs.k8s.io/controller-runtime/pkg/log"
@@ -46,6 +47,7 @@ type TensorFusionWorkloadReconciler struct {
 	client.Client
 	Scheme    *runtime.Scheme
 	Scheduler scheduler.Scheduler
+	Recorder  record.EventRecorder
 }
 
 // +kubebuilder:rbac:groups=tensor-fusion.ai,resources=tensorfusionworkloads,verbs=get;list;watch;create;update;patch;delete
@@ -277,6 +279,7 @@ func (r *TensorFusionWorkloadReconciler) scaleUpWorkers(ctx context.Context, wor
 		// Schedule GPU for the worker
 		gpu, err := r.Scheduler.Schedule(ctx, workload.Spec.PoolName, workload.Spec.Resources.Requests)
 		if err != nil {
+			r.Recorder.Eventf(workload, corev1.EventTypeWarning, "ScheduleGPUFailed", "Failed to schedule GPU: %v", err)
 			return fmt.Errorf("schedule GPU: %w", err)
 		}
 

internal/controller/tensorfusionworkload_controller_test.go

@@ -27,6 +27,7 @@ import (
 	"k8s.io/apimachinery/pkg/api/resource"
 	metav1 "k8s.io/apimachinery/pkg/apis/meta/v1"
 	"k8s.io/apimachinery/pkg/types"
+	"k8s.io/client-go/tools/record"
 	"sigs.k8s.io/controller-runtime/pkg/client"
 	"sigs.k8s.io/controller-runtime/pkg/reconcile"
 
@@ -90,6 +91,7 @@ var _ = Describe("TensorFusionWorkload Controller", func() {
 			Client:    k8sClient,
 			Scheme:    k8sClient.Scheme(),
 			Scheduler: scheduler.NewScheduler(k8sClient),
+			Recorder:  record.NewFakeRecorder(3),
 		}
 
 		// Clean up any pods from previous tests