Expose job controller's workqueue rate limiting configs

ADOPTERS.md

@@ -19,4 +19,5 @@ This page contains a list of organizations who are using MPI Operator. If you'd
 | [PITS Global Data Recovery Services](https://www.pitsdatarecovery.net/) | [Benjamin Trudeau](https://github.com/benjx1990) |
 | [Polyaxon](https://polyaxon.com/) | [Mourad Mourafiq](https://github.com/mouradmourafiq) |
 | [Qutoutiao](https://www.qutoutiao.net/) | [Zhaojing Yu](https://github.com/yuzhaojing) |
+| [Run:AI](https://www.run.ai/) | [Rotem Elad](https://github.com/roteme-runai) |
 | [Tencent](http://tencent.com/en-us/) | [Lei Xue](https://github.com/carmark) |

cmd/mpi-operator/app/options/options.go

@@ -28,16 +28,18 @@ const (
 
 // ServerOption is the main context object for the controller manager.
 type ServerOption struct {
-	Kubeconfig         string
-	MasterURL          string
-	Threadiness        int
-	MonitoringPort     int
-	PrintVersion       bool
-	GangSchedulingName string
-	Namespace          string
-	LockNamespace      string
-	QPS                int
-	Burst              int
+	Kubeconfig          string
+	MasterURL           string
+	Threadiness         int
+	MonitoringPort      int
+	PrintVersion        bool
+	GangSchedulingName  string
+	Namespace           string
+	LockNamespace       string
+	QPS                 int
+	Burst               int
+	ControllerRateLimit int
+	ControllerBurst     int
 }
 
 // NewServerOption creates a new CMServer with a default config.
@@ -75,4 +77,7 @@ func (s *ServerOption) AddFlags(fs *flag.FlagSet) {
 
 	fs.IntVar(&s.QPS, "kube-api-qps", 5, "QPS indicates the maximum QPS to the master from this client.")
 	fs.IntVar(&s.Burst, "kube-api-burst", 10, "Maximum burst for throttle.")
+
+	fs.IntVar(&s.ControllerRateLimit, "controller-queue-rate-limit", 10, "Rate limit of the controller events queue .")
+	fs.IntVar(&s.ControllerBurst, "controller-queue-burst", 100, "Maximum burst of the controller events queue.")
 }

cmd/mpi-operator/app/server.go

@@ -23,6 +23,7 @@ import (
 
 	"github.com/prometheus/client_golang/prometheus"
 	"github.com/prometheus/client_golang/prometheus/promauto"
+	"golang.org/x/time/rate"
 	corev1 "k8s.io/api/core/v1"
 	"k8s.io/apimachinery/pkg/api/errors"
 	metav1 "k8s.io/apimachinery/pkg/apis/meta/v1"
@@ -38,6 +39,7 @@ import (
 	election "k8s.io/client-go/tools/leaderelection"
 	"k8s.io/client-go/tools/leaderelection/resourcelock"
 	"k8s.io/client-go/tools/record"
+	"k8s.io/client-go/util/workqueue"
 	"k8s.io/klog"
 	schedclientset "sigs.k8s.io/scheduler-plugins/pkg/generated/clientset/versioned"
 	volcanoclient "volcano.sh/apis/pkg/client/clientset/versioned"
@@ -67,6 +69,9 @@ var (
 	// allowed for timeout. Checks within the timeout period after the lease
 	// expires will still return healthy.
 	leaderHealthzAdaptorTimeout = time.Second * 20
+	//exponential workqueue rate limiting config
+	workqueueExponentialBaseDelay = 5 * time.Millisecond
+	workqueueExponentialMaxDelay  = 1000 * time.Second
 )
 
 var (
@@ -141,6 +146,11 @@ func Run(opt *options.ServerOption) error {
 		kubeInformerFactory := kubeinformers.NewSharedInformerFactoryWithOptions(kubeClient, 0, kubeInformerFactoryOpts...)
 		kubeflowInformerFactory := informers.NewSharedInformerFactoryWithOptions(mpiJobClientSet, 0, kubeflowInformerFactoryOpts...)
 
+		workqueueRateLimiter := workqueue.NewTypedMaxOfRateLimiter(
+			workqueue.NewTypedItemExponentialFailureRateLimiter[any](workqueueExponentialBaseDelay, workqueueExponentialMaxDelay),
+			&workqueue.TypedBucketRateLimiter[any]{Limiter: rate.NewLimiter(rate.Limit(opt.ControllerRateLimit), opt.ControllerBurst)},
+		)
+
 		controller, err := controllersv1.NewMPIJobController(
 			kubeClient,
 			mpiJobClientSet,
@@ -153,7 +163,8 @@ func Run(opt *options.ServerOption) error {
 			kubeInformerFactory.Core().V1().Pods(),
 			kubeInformerFactory.Scheduling().V1().PriorityClasses(),
 			kubeflowInformerFactory.Kubeflow().V2beta1().MPIJobs(),
-			namespace, opt.GangSchedulingName)
+			namespace, opt.GangSchedulingName,
+			workqueueRateLimiter)
 		if err != nil {
 			klog.Fatalf("Failed to setup the controller")
 		}

pkg/controller/mpi_job_controller.go

@@ -269,10 +269,11 @@ func NewMPIJobController(
 	podInformer coreinformers.PodInformer,
 	priorityClassInformer schedulinginformers.PriorityClassInformer,
 	mpiJobInformer informers.MPIJobInformer,
-	namespace, gangSchedulingName string) (*MPIJobController, error) {
+	namespace, gangSchedulingName string,
+	workqueueRateLimiter workqueue.TypedRateLimiter[any]) (*MPIJobController, error) {
 	return NewMPIJobControllerWithClock(kubeClient, kubeflowClient, volcanoClient, schedClient,
 		configMapInformer, secretInformer, serviceInformer, jobInformer, podInformer,
-		priorityClassInformer, mpiJobInformer, &clock.RealClock{}, namespace, gangSchedulingName)
+		priorityClassInformer, mpiJobInformer, &clock.RealClock{}, namespace, gangSchedulingName, workqueueRateLimiter)
 }
 
 // NewMPIJobControllerWithClock returns a new MPIJob controller.
@@ -289,7 +290,8 @@ func NewMPIJobControllerWithClock(
 	priorityClassInformer schedulinginformers.PriorityClassInformer,
 	mpiJobInformer informers.MPIJobInformer,
 	clock clock.WithTicker,
-	namespace, gangSchedulingName string) (*MPIJobController, error) {
+	namespace, gangSchedulingName string,
+	workqueueRateLimiter workqueue.TypedRateLimiter[any]) (*MPIJobController, error) {
 
 	// Create event broadcaster.
 	klog.V(4).Info("Creating event broadcaster")
@@ -336,7 +338,7 @@ func NewMPIJobControllerWithClock(
 		priorityClassSynced: priorityClassSynced,
 		mpiJobLister:        mpiJobInformer.Lister(),
 		mpiJobSynced:        mpiJobInformer.Informer().HasSynced,
-		queue:               workqueue.NewTypedRateLimitingQueueWithConfig(workqueue.DefaultTypedControllerRateLimiter[any](), workqueue.TypedRateLimitingQueueConfig[any]{Name: "MPIJob"}),
+		queue:               workqueue.NewTypedRateLimitingQueueWithConfig(workqueueRateLimiter, workqueue.TypedRateLimitingQueueConfig[any]{Name: "MPIJob"}),
 		recorder:            recorder,
 		clock:               clock,
 	}

pkg/controller/mpi_job_controller_test.go

@@ -35,6 +35,7 @@ import (
 	core "k8s.io/client-go/testing"
 	"k8s.io/client-go/tools/cache"
 	"k8s.io/client-go/tools/record"
+	"k8s.io/client-go/util/workqueue"
 	"k8s.io/utils/clock"
 	clocktesting "k8s.io/utils/clock/testing"
 	"k8s.io/utils/ptr"
@@ -163,6 +164,7 @@ func (f *fixture) newController(clock clock.WithTicker) (*MPIJobController, info
 	f.kubeClient = k8sfake.NewSimpleClientset(f.kubeObjects...)
 	i := informers.NewSharedInformerFactory(f.client, noResyncPeriodFunc())
 	k8sI := kubeinformers.NewSharedInformerFactory(f.kubeClient, noResyncPeriodFunc())
+	workqueueRateLimiter := workqueue.DefaultTypedControllerRateLimiter[any]()
 
 	c, err := NewMPIJobControllerWithClock(
 		f.kubeClient,
@@ -179,6 +181,7 @@ func (f *fixture) newController(clock clock.WithTicker) (*MPIJobController, info
 		clock,
 		metav1.NamespaceAll,
 		f.gangSchedulingName,
+		workqueueRateLimiter,
 	)
 	if err != nil {
 		fmt.Println("Failed to setup the controller")

test/integration/mpi_job_controller_test.go

@@ -32,6 +32,7 @@ import (
 	kubeinformers "k8s.io/client-go/informers"
 	"k8s.io/client-go/kubernetes"
 	"k8s.io/client-go/tools/reference"
+	"k8s.io/client-go/util/workqueue"
 	"k8s.io/utils/ptr"
 	schedv1alpha1 "sigs.k8s.io/scheduler-plugins/apis/scheduling/v1alpha1"
 	schedclientset "sigs.k8s.io/scheduler-plugins/pkg/generated/clientset/versioned"
@@ -909,6 +910,7 @@ func startController(
 ) {
 	kubeInformerFactory := kubeinformers.NewSharedInformerFactory(kClient, 0)
 	mpiInformerFactory := informers.NewSharedInformerFactory(mpiClient, 0)
+	workqueueRateLimiter := workqueue.DefaultTypedControllerRateLimiter[any]()
 	var (
 		volcanoClient volcanoclient.Interface
 		schedClient   schedclientset.Interface
@@ -935,6 +937,7 @@ func startController(
 		kubeInformerFactory.Scheduling().V1().PriorityClasses(),
 		mpiInformerFactory.Kubeflow().V2beta1().MPIJobs(),
 		metav1.NamespaceAll, schedulerName,
+		workqueueRateLimiter,
 	)
 	if err != nil {
 		panic(err)