Enable setting the resource request/limits via annotations for queue-proxy side-car container

pkg/apis/serving/register.go

@@ -67,4 +67,8 @@ const (
 	// UpdaterAnnotation is the annotation key to describe the user that
 	// last updated the resource.
 	UpdaterAnnotation = GroupName + "/lastModifier"
+
+	// QueueSideCarResourcePercentageAnnotation is the percentage of user container resources to be used for queue-proxy
+	// It has to be in [0.1,100]
+	QueueSideCarResourcePercentageAnnotation = "queue.sidecar." + GroupName + "/resourcePercentage"
 )

pkg/apis/serving/v1alpha1/revision_validation.go

@@ -19,6 +19,7 @@ package v1alpha1
 import (
 	"context"
 	"fmt"
+	"strconv"
 	"strings"
 
 	"github.com/knative/pkg/apis"
@@ -81,6 +82,8 @@ func (rt *RevisionTemplateSpec) Validate(ctx context.Context) *apis.FieldError {
 		}
 	}
 
+	errs = errs.Also(validateAnnotations(rt.Annotations))
+
 	return errs
 }
 
@@ -152,6 +155,31 @@ func (rs *RevisionSpec) Validate(ctx context.Context) *apis.FieldError {
 	return errs
 }
 
+func validateAnnotations(annotations map[string]string) *apis.FieldError {
+	return validatePercentageAnnotationKey(annotations, serving.QueueSideCarResourcePercentageAnnotation)
+}
+
+func validatePercentageAnnotationKey(annotations map[string]string, resourcePercentageAnnotationKey string) *apis.FieldError {
+	if len(annotations) == 0 {
+		return nil
+	}
+
+	v, ok := annotations[resourcePercentageAnnotationKey]
+	if !ok {
+		return nil
+	}
+	value, err := strconv.ParseFloat(v, 32)
+	if err != nil {
+		return apis.ErrInvalidValue(v, apis.CurrentField).ViaKey(resourcePercentageAnnotationKey)
+	}
+
+	if value <= float64(0.1) || value > float64(100) {
+		return apis.ErrOutOfBoundsValue(value, 0.1, 100.0, resourcePercentageAnnotationKey)
+	}
+
+	return nil
+}
+
 func validateTimeoutSeconds(timeoutSeconds int64) *apis.FieldError {
 	if timeoutSeconds != 0 {
 		if timeoutSeconds > int64(networking.DefaultTimeout.Seconds()) || timeoutSeconds < 0 {

pkg/apis/serving/v1alpha1/revision_validation_test.go

@@ -27,6 +27,7 @@ import (
 	"github.com/knative/pkg/ptr"
 	"github.com/knative/serving/pkg/apis/autoscaling"
 	net "github.com/knative/serving/pkg/apis/networking"
+	"github.com/knative/serving/pkg/apis/serving"
 	corev1 "k8s.io/api/core/v1"
 	"k8s.io/apimachinery/pkg/api/resource"
 	metav1 "k8s.io/apimachinery/pkg/apis/meta/v1"
@@ -298,6 +299,42 @@ func TestRevisionTemplateSpecValidation(t *testing.T) {
 			},
 		},
 		want: nil,
+	}, {
+		name: "Queue sidecar resource percentage annotation more than 100",
+		rts: &RevisionTemplateSpec{
+			ObjectMeta: metav1.ObjectMeta{
+				Annotations: map[string]string{
+					serving.QueueSideCarResourcePercentageAnnotation: "200",
+				},
+			},
+			Spec: RevisionSpec{
+				DeprecatedContainer: &corev1.Container{
+					Image: "helloworld",
+				},
+			},
+		},
+		want: &apis.FieldError{
+			Message: "expected 0.1 <= 200 <= 100",
+			Paths:   []string{serving.QueueSideCarResourcePercentageAnnotation},
+		},
+	}, {
+		name: "Invalid queue sidecar resource percentage annotation",
+		rts: &RevisionTemplateSpec{
+			ObjectMeta: metav1.ObjectMeta{
+				Annotations: map[string]string{
+					serving.QueueSideCarResourcePercentageAnnotation: "50mx",
+				},
+			},
+			Spec: RevisionSpec{
+				DeprecatedContainer: &corev1.Container{
+					Image: "helloworld",
+				},
+			},
+		},
+		want: &apis.FieldError{
+			Message: "invalid value: 50mx",
+			Paths:   []string{fmt.Sprintf("[%s]", serving.QueueSideCarResourcePercentageAnnotation)},
+		},
 	}}
 
 	for _, test := range tests {

pkg/reconciler/revision/resources/deploy_test.go

@@ -69,7 +69,7 @@ var (
 
 	defaultQueueContainer = &corev1.Container{
 		Name:           QueueContainerName,
-		Resources:      queueResources,
+		Resources:      createQueueResources(make(map[string]string), &corev1.Container{}),
 		Ports:          append(queueNonServingPorts, queueHTTPPort),
 		ReadinessProbe: queueReadinessProbe,
 		Env: []corev1.EnvVar{{

pkg/reconciler/revision/resources/queue.go

@@ -17,8 +17,11 @@ limitations under the License.
 package resources
 
 import (
+	"math"
 	"strconv"
 
+	"k8s.io/apimachinery/pkg/api/resource"
+
 	"k8s.io/apimachinery/pkg/util/intstr"
 
 	"github.com/knative/pkg/logging"
@@ -38,12 +41,6 @@ import (
 const requestQueueHTTPPortName = "queue-port"
 
 var (
-	queueResources = corev1.ResourceRequirements{
-		Requests: corev1.ResourceList{
-			corev1.ResourceName("cpu"): queueContainerCPU,
-		},
-	}
-
 	queueHTTPPort = corev1.ContainerPort{
 		Name:          requestQueueHTTPPortName,
 		ContainerPort: int32(networking.BackendHTTPPort),
@@ -83,6 +80,82 @@ var (
 	}
 )
 
+func createQueueResources(annotations map[string]string, userContainer *corev1.Container) corev1.ResourceRequirements {
+	resources := corev1.ResourceRequirements{}
+	resourceRequests := corev1.ResourceList{corev1.ResourceCPU: queueContainerCPU}
+	resourceLimits := corev1.ResourceList{}
+	ok := false
+	var requestCPU, limitCPU, requestMemory, limitMemory resource.Quantity
+	var resourcePercentage float32
+
+	if ok, resourcePercentage = createResourcePercentageFromAnnotations(annotations, serving.QueueSideCarResourcePercentageAnnotation); ok {
+
+		if ok, requestCPU = computeResourceRequirements(userContainer.Resources.Requests.Cpu(), resourcePercentage, queueContainerRequestCPU); ok {
+			resourceRequests[corev1.ResourceCPU] = requestCPU
+		}
+
+		if ok, limitCPU = computeResourceRequirements(userContainer.Resources.Limits.Cpu(), resourcePercentage, queueContainerLimitCPU); ok {
+			resourceLimits[corev1.ResourceCPU] = limitCPU
+		}
+
+		if ok, requestMemory = computeResourceRequirements(userContainer.Resources.Requests.Memory(), resourcePercentage, queueContainerRequestMemory); ok {
+			resourceRequests[corev1.ResourceMemory] = requestMemory
+		}
+
+		if ok, limitMemory = computeResourceRequirements(userContainer.Resources.Limits.Memory(), resourcePercentage, queueContainerLimitMemory); ok {
+			resourceLimits[corev1.ResourceMemory] = limitMemory
+		}
+
+	}
+
+	resources.Requests = resourceRequests
+
+	if len(resourceLimits) != 0 {
+		resources.Limits = resourceLimits
+	}
+
+	return resources
+}
+
+func computeResourceRequirements(resourceQuantity *resource.Quantity, percentage float32, boundary resourceBoundary) (bool, resource.Quantity) {
+	if resourceQuantity.IsZero() {
+		return false, resource.Quantity{}
+	}
+
+	// Incase the resourceQuantity MilliValue overflow in we use MaxInt64
+	// https://github.com/kubernetes/apimachinery/blob/master/pkg/api/resource/quantity.go
+	scaledValue := resourceQuantity.Value()
+	scaledMilliValue := int64(math.MaxInt64 - 1)
+	if scaledValue < (math.MaxInt64 / 1000) {
+		scaledMilliValue = resourceQuantity.MilliValue()
+	}
+
+	// float64(math.MaxInt64) > math.MaxInt64, to avoid overflow
+	percentageValue := float64(scaledMilliValue) * float64(percentage)
+	var newValue int64
+	if percentageValue >= math.MaxInt64 {
+		newValue = math.MaxInt64
+	} else {
+		newValue = int64(percentageValue)
+	}
+
+	newquantity := *resource.NewMilliQuantity(newValue, resource.BinarySI)
+	newquantity = boundary.applyBoundary(newquantity)
+	return true, newquantity
+}
+
+func createResourcePercentageFromAnnotations(m map[string]string, k string) (bool, float32) {
+	v, ok := m[k]
+	if !ok {
+		return false, 0
+	}
+	value, err := strconv.ParseFloat(v, 32)
+	if err != nil {
+		return false, 0
+	}
+	return true, float32(value / 100)
+}
+
 // makeQueueContainer creates the container spec for the queue sidecar.
 func makeQueueContainer(rev *v1alpha1.Revision, loggingConfig *logging.Config, observabilityConfig *metrics.ObservabilityConfig,
 	autoscalerConfig *autoscaler.Config, deploymentConfig *deployment.Config) *corev1.Container {
@@ -116,7 +189,7 @@ func makeQueueContainer(rev *v1alpha1.Revision, loggingConfig *logging.Config, o
 	return &corev1.Container{
 		Name:           QueueContainerName,
 		Image:          deploymentConfig.QueueSidecarImage,
-		Resources:      queueResources,
+		Resources:      createQueueResources(rev.GetAnnotations(), rev.Spec.GetContainer()),
 		Ports:          ports,
 		ReadinessProbe: queueReadinessProbe,
 		Env: []corev1.EnvVar{{