controller: proportionally scale paused and rolling deployments

hack/test-cmd.sh

@@ -1772,35 +1772,35 @@ __EOF__
   # Command
   # Create a deployment (revision 1)
   kubectl create -f hack/testdata/deployment-revision1.yaml "${kube_flags[@]}"
-  kube::test::get_object_assert deployment "{{range.items}}{{$id_field}}:{{end}}" 'nginx-deployment:'
+  kube::test::get_object_assert deployment "{{range.items}}{{$id_field}}:{{end}}" 'nginx:'
   kube::test::get_object_assert deployment "{{range.items}}{{$deployment_image_field}}:{{end}}" "${IMAGE_DEPLOYMENT_R1}:"
   # Rollback to revision 1 - should be no-op
-  kubectl rollout undo deployment nginx-deployment --to-revision=1 "${kube_flags[@]}"
+  kubectl rollout undo deployment nginx --to-revision=1 "${kube_flags[@]}"
   kube::test::get_object_assert deployment "{{range.items}}{{$deployment_image_field}}:{{end}}" "${IMAGE_DEPLOYMENT_R1}:"
   # Update the deployment (revision 2)
   kubectl apply -f hack/testdata/deployment-revision2.yaml "${kube_flags[@]}"
   kube::test::get_object_assert deployment.extensions "{{range.items}}{{$deployment_image_field}}:{{end}}" "${IMAGE_DEPLOYMENT_R2}:"
   # Rollback to revision 1
-  kubectl rollout undo deployment nginx-deployment --to-revision=1 "${kube_flags[@]}"
+  kubectl rollout undo deployment nginx --to-revision=1 "${kube_flags[@]}"
   sleep 1
   kube::test::get_object_assert deployment "{{range.items}}{{$deployment_image_field}}:{{end}}" "${IMAGE_DEPLOYMENT_R1}:"
   # Rollback to revision 1000000 - should be no-op
-  kubectl rollout undo deployment nginx-deployment --to-revision=1000000 "${kube_flags[@]}"
+  kubectl rollout undo deployment nginx --to-revision=1000000 "${kube_flags[@]}"
   kube::test::get_object_assert deployment "{{range.items}}{{$deployment_image_field}}:{{end}}" "${IMAGE_DEPLOYMENT_R1}:"
   # Rollback to last revision
-  kubectl rollout undo deployment nginx-deployment "${kube_flags[@]}"
+  kubectl rollout undo deployment nginx "${kube_flags[@]}"
   sleep 1
   kube::test::get_object_assert deployment "{{range.items}}{{$deployment_image_field}}:{{end}}" "${IMAGE_DEPLOYMENT_R2}:"
   # Pause the deployment
-  kubectl-with-retry rollout pause deployment nginx-deployment "${kube_flags[@]}"
+  kubectl-with-retry rollout pause deployment nginx "${kube_flags[@]}"
   # A paused deployment cannot be rolled back
-  ! kubectl rollout undo deployment nginx-deployment "${kube_flags[@]}"
+  ! kubectl rollout undo deployment nginx "${kube_flags[@]}"
   # Resume the deployment
-  kubectl-with-retry rollout resume deployment nginx-deployment "${kube_flags[@]}"
+  kubectl-with-retry rollout resume deployment nginx "${kube_flags[@]}"
   # The resumed deployment can now be rolled back
-  kubectl rollout undo deployment nginx-deployment "${kube_flags[@]}"
+  kubectl rollout undo deployment nginx "${kube_flags[@]}"
   # Clean up
-  kubectl delete deployment nginx-deployment "${kube_flags[@]}"
+  kubectl delete deployment nginx "${kube_flags[@]}"
 
   ### Set image of a deployment 
   # Pre-condition: no deployment exists

hack/testdata/deployment-revision1.yaml

 apiVersion: extensions/v1beta1
 kind: Deployment
 metadata:
-  name: nginx-deployment
+  name: nginx
   labels:
-    name: nginx-deployment
+    name: nginx-undo
 spec:
   replicas: 3
   selector:
     matchLabels:
-      name: nginx
+      name: nginx-undo
   template:
     metadata:
       labels:
-        name: nginx
+        name: nginx-undo
     spec:
       containers:
       - name: nginx

hack/testdata/deployment-revision2.yaml

 apiVersion: extensions/v1beta1
 kind: Deployment
 metadata:
-  name: nginx-deployment
+  name: nginx
   labels:
-    name: nginx-deployment
+    name: nginx-undo
 spec:
   replicas: 3
   selector:
     matchLabels:
-      name: nginx
+      name: nginx-undo
   template:
     metadata:
       labels:
-        name: nginx
+        name: nginx-undo
     spec:
       containers:
       - name: nginx

pkg/apis/extensions/validation/validation.go

@@ -147,12 +147,15 @@ var ValidateDeploymentName = apivalidation.NameIsDNSSubdomain
 
 func ValidatePositiveIntOrPercent(intOrPercent intstr.IntOrString, fldPath *field.Path) field.ErrorList {
 	allErrs := field.ErrorList{}
-	if intOrPercent.Type == intstr.String {
+	switch intOrPercent.Type {
+	case intstr.String:
 		if !validation.IsValidPercent(intOrPercent.StrVal) {
-			allErrs = append(allErrs, field.Invalid(fldPath, intOrPercent, "must be an integer or percentage (e.g '5%')"))
+			allErrs = append(allErrs, field.Invalid(fldPath, intOrPercent, "must be an integer or percentage (e.g '5%%')"))
 		}
-	} else if intOrPercent.Type == intstr.Int {
+	case intstr.Int:
 		allErrs = append(allErrs, apivalidation.ValidateNonnegativeField(int64(intOrPercent.IntValue()), fldPath)...)
+	default:
+		allErrs = append(allErrs, field.Invalid(fldPath, intOrPercent, "must be an integer or percentage (e.g '5%%')"))
 	}
 	return allErrs
 }

pkg/controller/controller_utils.go

@@ -619,7 +619,9 @@ func IsPodActive(p api.Pod) bool {
 func FilterActiveReplicaSets(replicaSets []*extensions.ReplicaSet) []*extensions.ReplicaSet {
 	active := []*extensions.ReplicaSet{}
 	for i := range replicaSets {
-		if replicaSets[i].Spec.Replicas > 0 {
+		rs := replicaSets[i]
+
+		if rs != nil && rs.Spec.Replicas > 0 {
 			active = append(active, replicaSets[i])
 		}
 	}
@@ -639,7 +641,6 @@ type ControllersByCreationTimestamp []*api.ReplicationController
 
 func (o ControllersByCreationTimestamp) Len() int      { return len(o) }
 func (o ControllersByCreationTimestamp) Swap(i, j int) { o[i], o[j] = o[j], o[i] }
-
 func (o ControllersByCreationTimestamp) Less(i, j int) bool {
 	if o[i].CreationTimestamp.Equal(o[j].CreationTimestamp) {
 		return o[i].Name < o[j].Name
@@ -647,15 +648,40 @@ func (o ControllersByCreationTimestamp) Less(i, j int) bool {
 	return o[i].CreationTimestamp.Before(o[j].CreationTimestamp)
 }
 
-// ReplicaSetsByCreationTimestamp sorts a list of ReplicationSets by creation timestamp, using their names as a tie breaker.
+// ReplicaSetsByCreationTimestamp sorts a list of ReplicaSet by creation timestamp, using their names as a tie breaker.
 type ReplicaSetsByCreationTimestamp []*extensions.ReplicaSet
 
 func (o ReplicaSetsByCreationTimestamp) Len() int      { return len(o) }
 func (o ReplicaSetsByCreationTimestamp) Swap(i, j int) { o[i], o[j] = o[j], o[i] }
-
 func (o ReplicaSetsByCreationTimestamp) Less(i, j int) bool {
 	if o[i].CreationTimestamp.Equal(o[j].CreationTimestamp) {
 		return o[i].Name < o[j].Name
 	}
 	return o[i].CreationTimestamp.Before(o[j].CreationTimestamp)
 }
+
+// ReplicaSetsBySizeOlder sorts a list of ReplicaSet by size in descending order, using their creation timestamp or name as a tie breaker.
+// By using the creation timestamp, this sorts from old to new replica sets.
+type ReplicaSetsBySizeOlder []*extensions.ReplicaSet
+
+func (o ReplicaSetsBySizeOlder) Len() int      { return len(o) }
+func (o ReplicaSetsBySizeOlder) Swap(i, j int) { o[i], o[j] = o[j], o[i] }
+func (o ReplicaSetsBySizeOlder) Less(i, j int) bool {
+	if o[i].Spec.Replicas == o[j].Spec.Replicas {
+		return ReplicaSetsByCreationTimestamp(o).Less(i, j)
+	}
+	return o[i].Spec.Replicas > o[j].Spec.Replicas
+}
+
+// ReplicaSetsBySizeNewer sorts a list of ReplicaSet by size in descending order, using their creation timestamp or name as a tie breaker.
+// By using the creation timestamp, this sorts from new to old replica sets.
+type ReplicaSetsBySizeNewer []*extensions.ReplicaSet
+
+func (o ReplicaSetsBySizeNewer) Len() int      { return len(o) }
+func (o ReplicaSetsBySizeNewer) Swap(i, j int) { o[i], o[j] = o[j], o[i] }
+func (o ReplicaSetsBySizeNewer) Less(i, j int) bool {
+	if o[i].Spec.Replicas == o[j].Spec.Replicas {
+		return ReplicaSetsByCreationTimestamp(o).Less(j, i)
+	}
+	return o[i].Spec.Replicas > o[j].Spec.Replicas
+}

pkg/controller/deployment/deployment_controller.go

@@ -438,13 +438,9 @@ func (dc *DeploymentController) syncDeployment(key string) error {
 	}
 
 	if d.Spec.Paused {
-		// TODO: Implement scaling for paused deployments.
-		// Don't take any action for paused deployment.
-		// But keep the status up-to-date.
-		// Ignore paused deployments
-		glog.V(4).Infof("Updating status only for paused deployment %s/%s", d.Namespace, d.Name)
-		return dc.syncPausedDeploymentStatus(d)
+		return dc.sync(d)
 	}
+
 	if d.Spec.RollbackTo != nil {
 		revision := d.Spec.RollbackTo.Revision
 		if _, err = dc.rollback(d, &revision); err != nil {
@@ -452,27 +448,135 @@ func (dc *DeploymentController) syncDeployment(key string) error {
 		}
 	}
 
+	if dc.isScalingEvent(d) {
+		return dc.sync(d)
+	}
+
 	switch d.Spec.Strategy.Type {
 	case extensions.RecreateDeploymentStrategyType:
-		return dc.syncRecreateDeployment(d)
+		return dc.rolloutRecreate(d)
 	case extensions.RollingUpdateDeploymentStrategyType:
-		return dc.syncRollingUpdateDeployment(d)
+		return dc.rolloutRolling(d)
 	}
 	return fmt.Errorf("unexpected deployment strategy type: %s", d.Spec.Strategy.Type)
 }
 
-// Updates the status of a paused deployment
-func (dc *DeploymentController) syncPausedDeploymentStatus(deployment *extensions.Deployment) error {
+// sync is responsible for reconciling deployments on scaling events or when they
+// are paused.
+func (dc *DeploymentController) sync(deployment *extensions.Deployment) error {
 	newRS, oldRSs, err := dc.getAllReplicaSetsAndSyncRevision(deployment, false)
 	if err != nil {
 		return err
 	}
-	allRSs := append(controller.FilterActiveReplicaSets(oldRSs), newRS)
+	if err := dc.scale(deployment, newRS, oldRSs); err != nil {
+		// If we get an error while trying to scale, the deployment will be requeued
+		// so we can abort this resync
+		return err
+	}
+	dc.cleanupDeployment(oldRSs, deployment)
 
-	// Sync deployment status
+	allRSs := append(oldRSs, newRS)
 	return dc.syncDeploymentStatus(allRSs, newRS, deployment)
 }
 
+// scale scales proportionally in order to mitigate risk. Otherwise, scaling up can increase the size
+// of the new replica set and scaling down can decrease the sizes of the old ones, both of which would
+// have the effect of hastening the rollout progress, which could produce a higher proportion of unavailable
+// replicas in the event of a problem with the rolled out template. Should run only on scaling events or
+// when a deployment is paused and not during the normal rollout process.
+func (dc *DeploymentController) scale(deployment *extensions.Deployment, newRS *extensions.ReplicaSet, oldRSs []*extensions.ReplicaSet) error {
+	// If there is only one active replica set then we should scale that up to the full count of the
+	// deployment. If there is no active replica set, then we should scale up the newest replica set.
+	if activeOrLatest := findActiveOrLatest(newRS, oldRSs); activeOrLatest != nil {
+		if activeOrLatest.Spec.Replicas == deployment.Spec.Replicas {
+			return nil
+		}
+		_, _, err := dc.scaleReplicaSetAndRecordEvent(activeOrLatest, deployment.Spec.Replicas, deployment)
+		return err
+	}
+
+	// If the new replica set is saturated, old replica sets should be fully scaled down.
+	// This case handles replica set adoption during a saturated new replica set.
+	if deploymentutil.IsSaturated(deployment, newRS) {
+		for _, old := range controller.FilterActiveReplicaSets(oldRSs) {
+			if _, _, err := dc.scaleReplicaSetAndRecordEvent(old, 0, deployment); err != nil {
+				return err
+			}
+		}
+		return nil
+	}
+
+	// There are old replica sets with pods and the new replica set is not saturated.
+	// We need to proportionally scale all replica sets (new and old) in case of a
+	// rolling deployment.
+	if deploymentutil.IsRollingUpdate(deployment) {
+		allRSs := controller.FilterActiveReplicaSets(append(oldRSs, newRS))
+		allRSsReplicas := deploymentutil.GetReplicaCountForReplicaSets(allRSs)
+
+		allowedSize := int32(0)
+		if deployment.Spec.Replicas > 0 {
+			allowedSize = deployment.Spec.Replicas + maxSurge(*deployment)
+		}
+
+		// Number of additional replicas that can be either added or removed from the total
+		// replicas count. These replicas should be distributed proportionally to the active
+		// replica sets.
+		deploymentReplicasToAdd := allowedSize - allRSsReplicas
+
+		// The additional replicas should be distributed proportionally amongst the active
+		// replica sets from the larger to the smaller in size replica set. Scaling direction
+		// drives what happens in case we are trying to scale replica sets of the same size.
+		// In such a case when scaling up, we should scale up newer replica sets first, and
+		// when scaling down, we should scale down older replica sets first.
+		scalingOperation := "up"
+		switch {
+		case deploymentReplicasToAdd > 0:
+			sort.Sort(controller.ReplicaSetsBySizeNewer(allRSs))
+
+		case deploymentReplicasToAdd < 0:
+			sort.Sort(controller.ReplicaSetsBySizeOlder(allRSs))
+			scalingOperation = "down"
+
+		default: /* deploymentReplicasToAdd == 0 */
+			// Nothing to add.
+			return nil
+		}
+
+		// Iterate over all active replica sets and estimate proportions for each of them.
+		// The absolute value of deploymentReplicasAdded should never exceed the absolute
+		// value of deploymentReplicasToAdd.
+		deploymentReplicasAdded := int32(0)
+		for i := range allRSs {
+			rs := allRSs[i]
+
+			proportion := getProportion(rs, *deployment, deploymentReplicasToAdd, deploymentReplicasAdded)
+
+			rs.Spec.Replicas += proportion
+			deploymentReplicasAdded += proportion
+		}
+
+		// Update all replica sets
+		for i := range allRSs {
+			rs := allRSs[i]
+
+			// Add/remove any leftovers to the largest replica set.
+			if i == 0 {
+				leftover := deploymentReplicasToAdd - deploymentReplicasAdded
+				rs.Spec.Replicas += leftover
+				if rs.Spec.Replicas < 0 {
+					rs.Spec.Replicas = 0
+				}
+			}
+
+			if _, err := dc.scaleReplicaSet(rs, rs.Spec.Replicas, deployment, scalingOperation); err != nil {
+				// Return as soon as we fail, the deployment is requeued
+				return err
+			}
+		}
+	}
+	return nil
+}
+
 // Rolling back to a revision; no-op if the toRevision is deployment's current revision
 func (dc *DeploymentController) rollback(deployment *extensions.Deployment, toRevision *int64) (*extensions.Deployment, error) {
 	newRS, allOldRSs, err := dc.getAllReplicaSetsAndSyncRevision(deployment, true)
@@ -526,13 +630,13 @@ func (dc *DeploymentController) updateDeploymentAndClearRollbackTo(deployment *e
 	return dc.updateDeployment(deployment)
 }
 
-func (dc *DeploymentController) syncRecreateDeployment(deployment *extensions.Deployment) error {
+func (dc *DeploymentController) rolloutRecreate(deployment *extensions.Deployment) error {
 	// Don't create a new RS if not already existed, so that we avoid scaling up before scaling down
 	newRS, oldRSs, err := dc.getAllReplicaSetsAndSyncRevision(deployment, false)
 	if err != nil {
 		return err
 	}
-	allRSs := append(controller.FilterActiveReplicaSets(oldRSs), newRS)
+	allRSs := append(oldRSs, newRS)
 
 	// scale down old replica sets
 	scaledDown, err := dc.scaleDownOldReplicaSetsForRecreate(controller.FilterActiveReplicaSets(oldRSs), deployment)
@@ -564,21 +668,18 @@ func (dc *DeploymentController) syncRecreateDeployment(deployment *extensions.De
 		return dc.updateDeploymentStatus(allRSs, newRS, deployment)
 	}
 
-	if deployment.Spec.RevisionHistoryLimit != nil {
-		// Cleanup old replica sets
-		dc.cleanupOldReplicaSets(oldRSs, deployment)
-	}
+	dc.cleanupDeployment(oldRSs, deployment)
 
 	// Sync deployment status
 	return dc.syncDeploymentStatus(allRSs, newRS, deployment)
 }
 
-func (dc *DeploymentController) syncRollingUpdateDeployment(deployment *extensions.Deployment) error {
+func (dc *DeploymentController) rolloutRolling(deployment *extensions.Deployment) error {
 	newRS, oldRSs, err := dc.getAllReplicaSetsAndSyncRevision(deployment, true)
 	if err != nil {
 		return err
 	}
-	allRSs := append(controller.FilterActiveReplicaSets(oldRSs), newRS)
+	allRSs := append(oldRSs, newRS)
 
 	// Scale up, if we can.
 	scaledUp, err := dc.reconcileNewReplicaSet(allRSs, newRS, deployment)
@@ -600,10 +701,7 @@ func (dc *DeploymentController) syncRollingUpdateDeployment(deployment *extensio
 		return dc.updateDeploymentStatus(allRSs, newRS, deployment)
 	}
 
-	if deployment.Spec.RevisionHistoryLimit != nil {
-		// Cleanup old replicas sets
-		dc.cleanupOldReplicaSets(oldRSs, deployment)
-	}
+	dc.cleanupDeployment(oldRSs, deployment)
 
 	// Sync deployment status
 	return dc.syncDeploymentStatus(allRSs, newRS, deployment)
@@ -611,11 +709,11 @@ func (dc *DeploymentController) syncRollingUpdateDeployment(deployment *extensio
 
 // syncDeploymentStatus checks if the status is up-to-date and sync it if necessary
 func (dc *DeploymentController) syncDeploymentStatus(allRSs []*extensions.ReplicaSet, newRS *extensions.ReplicaSet, d *extensions.Deployment) error {
-	totalActualReplicas, updatedReplicas, availableReplicas, _, err := dc.calculateStatus(allRSs, newRS, d)
+	newStatus, err := dc.calculateStatus(allRSs, newRS, d)
 	if err != nil {
 		return err
 	}
-	if d.Generation > d.Status.ObservedGeneration || d.Status.Replicas != totalActualReplicas || d.Status.UpdatedReplicas != updatedReplicas || d.Status.AvailableReplicas != availableReplicas {
+	if !reflect.DeepEqual(d.Status, newStatus) {
 		return dc.updateDeploymentStatus(allRSs, newRS, d)
 	}
 	return nil
@@ -626,6 +724,8 @@ func (dc *DeploymentController) syncDeploymentStatus(allRSs []*extensions.Replic
 // 2. Get new RS this deployment targets (whose pod template matches deployment's), and update new RS's revision number to (maxOldV + 1),
 //    only if its revision number is smaller than (maxOldV + 1). If this step failed, we'll update it in the next deployment sync loop.
 // 3. Copy new RS's revision number to deployment (update deployment's revision). If this step failed, we'll update it in the next deployment sync loop.
+// Note that currently the deployment controller is using caches to avoid querying the server for reads.
+// This may lead to stale reads of replica sets, thus incorrect deployment status.
 func (dc *DeploymentController) getAllReplicaSetsAndSyncRevision(deployment *extensions.Deployment, createIfNotExisted bool) (*extensions.ReplicaSet, []*extensions.ReplicaSet, error) {
 	// List the deployment's RSes & Pods and apply pod-template-hash info to deployment's adopted RSes/Pods
 	rsList, podList, err := dc.rsAndPodsWithHashKeySynced(deployment)
@@ -701,7 +801,7 @@ func (dc *DeploymentController) getNewReplicaSet(deployment *extensions.Deployme
 		return nil, err
 	} else if existingNewRS != nil {
 		// Set existing new replica set's annotation
-		if setNewReplicaSetAnnotations(deployment, existingNewRS, newRevision) {
+		if setNewReplicaSetAnnotations(deployment, existingNewRS, newRevision, true) {
 			return dc.client.Extensions().ReplicaSets(deployment.ObjectMeta.Namespace).Update(existingNewRS)
 		}
 		return existingNewRS, nil
@@ -731,8 +831,6 @@ func (dc *DeploymentController) getNewReplicaSet(deployment *extensions.Deployme
 			Template: newRSTemplate,
 		},
 	}
-	// Set new replica set's annotation
-	setNewReplicaSetAnnotations(deployment, &newRS, newRevision)
 	allRSs := append(oldRSs, &newRS)
 	newReplicasCount, err := deploymentutil.NewRSNewReplicas(deployment, allRSs, &newRS)
 	if err != nil {
@@ -740,6 +838,8 @@ func (dc *DeploymentController) getNewReplicaSet(deployment *extensions.Deployme
 	}
 
 	newRS.Spec.Replicas = newReplicasCount
+	// Set new replica set's annotation
+	setNewReplicaSetAnnotations(deployment, &newRS, newRevision, false)
 	createdRS, err := dc.client.Extensions().ReplicaSets(namespace).Create(&newRS)
 	if err != nil {
 		dc.enqueueDeployment(deployment)
@@ -881,7 +981,7 @@ func (dc *DeploymentController) addHashKeyToRSAndPods(rs extensions.ReplicaSet)
 
 // setNewReplicaSetAnnotations sets new replica set's annotations appropriately by updating its revision and
 // copying required deployment annotations to it; it returns true if replica set's annotation is changed.
-func setNewReplicaSetAnnotations(deployment *extensions.Deployment, newRS *extensions.ReplicaSet, newRevision string) bool {
+func setNewReplicaSetAnnotations(deployment *extensions.Deployment, newRS *extensions.ReplicaSet, newRevision string, exists bool) bool {
 	// First, copy deployment's annotations (except for apply and revision annotations)
 	annotationChanged := copyDeploymentAnnotationsToReplicaSet(deployment, newRS)
 	// Then, update replica set's revision annotation
@@ -894,17 +994,26 @@ func setNewReplicaSetAnnotations(deployment *extensions.Deployment, newRS *exten
 	if newRS.Annotations[deploymentutil.RevisionAnnotation] < newRevision {
 		newRS.Annotations[deploymentutil.RevisionAnnotation] = newRevision
 		annotationChanged = true
-		glog.V(4).Infof("updating replica set %q's revision to %s - %+v\n", newRS.Name, newRevision, newRS)
+		glog.V(4).Infof("Updating replica set %q revision to %s", newRS.Name, newRevision)
+	}
+	if !exists && setReplicasAnnotations(newRS, deployment.Spec.Replicas, deployment.Spec.Replicas+maxSurge(*deployment)) {
+		annotationChanged = true
 	}
 	return annotationChanged
 }
 
+var annotationsToSkip = map[string]bool{
+	annotations.LastAppliedConfigAnnotation:  true,
+	deploymentutil.RevisionAnnotation:        true,
+	deploymentutil.DesiredReplicasAnnotation: true,
+	deploymentutil.MaxReplicasAnnotation:     true,
+}
+
 // skipCopyAnnotation returns true if we should skip copying the annotation with the given annotation key
 // TODO: How to decide which annotations should / should not be copied?
 //       See https://github.com/kubernetes/kubernetes/pull/20035#issuecomment-179558615
 func skipCopyAnnotation(key string) bool {
-	// Skip apply annotations and revision annotations.
-	return key == annotations.LastAppliedConfigAnnotation || key == deploymentutil.RevisionAnnotation
+	return annotationsToSkip[key]
 }
 
 func getSkippedAnnotations(annotations map[string]string) map[string]string {
@@ -980,12 +1089,12 @@ func (dc *DeploymentController) reconcileNewReplicaSet(allRSs []*extensions.Repl
 	return scaled, err
 }
 
-func (dc *DeploymentController) getAvailablePodsForReplicaSets(deployment *extensions.Deployment, rss []*extensions.ReplicaSet, minReadySeconds int32) (int32, error) {
+func (dc *DeploymentController) getAvailablePodsForReplicaSets(deployment *extensions.Deployment, rss []*extensions.ReplicaSet) (int32, error) {
 	podList, err := dc.listPods(deployment)
 	if err != nil {
 		return 0, err
 	}
-	return deploymentutil.CountAvailablePodsForReplicaSets(podList, rss, minReadySeconds)
+	return deploymentutil.CountAvailablePodsForReplicaSets(podList, rss, deployment.Spec.MinReadySeconds)
 }
 
 func (dc *DeploymentController) reconcileOldReplicaSets(allRSs []*extensions.ReplicaSet, oldRSs []*extensions.ReplicaSet, newRS *extensions.ReplicaSet, deployment *extensions.Deployment) (bool, error) {
@@ -1002,11 +1111,7 @@ func (dc *DeploymentController) reconcileOldReplicaSets(allRSs []*extensions.Rep
 	if err != nil {
 		return false, fmt.Errorf("could not find available pods: %v", err)
 	}
-
-	_, maxUnavailable, err := deploymentutil.ResolveFenceposts(&deployment.Spec.Strategy.RollingUpdate.MaxSurge, &deployment.Spec.Strategy.RollingUpdate.MaxUnavailable, deployment.Spec.Replicas)
-	if err != nil {
-		return false, err
-	}
+	maxUnavailable := maxUnavailable(*deployment)
 
 	// Check if we can scale down. We can scale down in the following 2 cases:
 	// * Some old replica sets have unhealthy replicas, we could safely scale down those unhealthy replicas since that won't further
@@ -1108,10 +1213,7 @@ func (dc *DeploymentController) cleanupUnhealthyReplicas(oldRSs []*extensions.Re
 // scaleDownOldReplicaSetsForRollingUpdate scales down old replica sets when deployment strategy is "RollingUpdate".
 // Need check maxUnavailable to ensure availability
 func (dc *DeploymentController) scaleDownOldReplicaSetsForRollingUpdate(allRSs []*extensions.ReplicaSet, oldRSs []*extensions.ReplicaSet, deployment *extensions.Deployment) (int32, error) {
-	_, maxUnavailable, err := deploymentutil.ResolveFenceposts(&deployment.Spec.Strategy.RollingUpdate.MaxSurge, &deployment.Spec.Strategy.RollingUpdate.MaxUnavailable, deployment.Spec.Replicas)
-	if err != nil {
-		return 0, err
-	}
+	maxUnavailable := maxUnavailable(*deployment)
 
 	// Check if we can scale down.
 	minAvailable := deployment.Spec.Replicas - maxUnavailable
@@ -1183,7 +1285,13 @@ func (dc *DeploymentController) scaleUpNewReplicaSetForRecreate(newRS *extension
 	return scaled, err
 }
 
-func (dc *DeploymentController) cleanupOldReplicaSets(oldRSs []*extensions.ReplicaSet, deployment *extensions.Deployment) error {
+// cleanupDeployment is responsible for cleaning up a deployment ie. retains all but the latest N old replica sets
+// where N=d.Spec.RevisionHistoryLimit. Old replica sets are older versions of the podtemplate of a deployment kept
+// around by default 1) for historical reasons and 2) for the ability to rollback a deployment.
+func (dc *DeploymentController) cleanupDeployment(oldRSs []*extensions.ReplicaSet, deployment *extensions.Deployment) error {
+	if deployment.Spec.RevisionHistoryLimit == nil {
+		return nil
+	}
 	diff := int32(len(oldRSs)) - *deployment.Spec.RevisionHistoryLimit
 	if diff <= 0 {
 		return nil
@@ -1209,39 +1317,31 @@ func (dc *DeploymentController) cleanupOldReplicaSets(oldRSs []*extensions.Repli
 }
 
 func (dc *DeploymentController) updateDeploymentStatus(allRSs []*extensions.ReplicaSet, newRS *extensions.ReplicaSet, deployment *extensions.Deployment) error {
-	totalActualReplicas, updatedReplicas, availableReplicas, unavailableReplicas, err := dc.calculateStatus(allRSs, newRS, deployment)
+	newStatus, err := dc.calculateStatus(allRSs, newRS, deployment)
 	if err != nil {
 		return err
 	}
-	newDeployment := *deployment
-	// TODO: Reconcile this with API definition. API definition talks about ready pods, while this just computes created pods.
-	newDeployment.Status = extensions.DeploymentStatus{
-		// TODO: Ensure that if we start retrying status updates, we won't pick up a new Generation value.
-		ObservedGeneration:  deployment.Generation,
-		Replicas:            totalActualReplicas,
-		UpdatedReplicas:     updatedReplicas,
-		AvailableReplicas:   availableReplicas,
-		UnavailableReplicas: unavailableReplicas,
-	}
-	_, err = dc.client.Extensions().Deployments(deployment.ObjectMeta.Namespace).UpdateStatus(&newDeployment)
-	if err == nil {
-		glog.V(4).Infof("Updated deployment %s status: %+v", deployment.Name, newDeployment.Status)
-	}
+	newDeployment := deployment
+	newDeployment.Status = newStatus
+	_, err = dc.client.Extensions().Deployments(deployment.Namespace).UpdateStatus(newDeployment)
 	return err
 }
 
-func (dc *DeploymentController) calculateStatus(allRSs []*extensions.ReplicaSet, newRS *extensions.ReplicaSet, deployment *extensions.Deployment) (totalActualReplicas, updatedReplicas, availableReplicas, unavailableReplicas int32, err error) {
-	totalActualReplicas = deploymentutil.GetActualReplicaCountForReplicaSets(allRSs)
-	updatedReplicas = deploymentutil.GetActualReplicaCountForReplicaSets([]*extensions.ReplicaSet{newRS})
-	minReadySeconds := deployment.Spec.MinReadySeconds
-	availableReplicas, err = dc.getAvailablePodsForReplicaSets(deployment, allRSs, minReadySeconds)
+func (dc *DeploymentController) calculateStatus(allRSs []*extensions.ReplicaSet, newRS *extensions.ReplicaSet, deployment *extensions.Deployment) (extensions.DeploymentStatus, error) {
+	availableReplicas, err := dc.getAvailablePodsForReplicaSets(deployment, allRSs)
 	if err != nil {
-		err = fmt.Errorf("failed to count available pods: %v", err)
-		return
+		return deployment.Status, fmt.Errorf("failed to count available pods: %v", err)
 	}
 	totalReplicas := deploymentutil.GetReplicaCountForReplicaSets(allRSs)
-	unavailableReplicas = totalReplicas - availableReplicas
-	return
+
+	return extensions.DeploymentStatus{
+		// TODO: Ensure that if we start retrying status updates, we won't pick up a new Generation value.
+		ObservedGeneration:  deployment.Generation,
+		Replicas:            deploymentutil.GetActualReplicaCountForReplicaSets(allRSs),
+		UpdatedReplicas:     deploymentutil.GetActualReplicaCountForReplicaSets([]*extensions.ReplicaSet{newRS}),
+		AvailableReplicas:   availableReplicas,
+		UnavailableReplicas: totalReplicas - availableReplicas,
+	}, nil
 }
 
 func (dc *DeploymentController) scaleReplicaSetAndRecordEvent(rs *extensions.ReplicaSet, newScale int32, deployment *extensions.Deployment) (bool, *extensions.ReplicaSet, error) {
@@ -1255,24 +1355,25 @@ func (dc *DeploymentController) scaleReplicaSetAndRecordEvent(rs *extensions.Rep
 	} else {
 		scalingOperation = "down"
 	}
-	newRS, err := dc.scaleReplicaSet(rs, newScale)
-	if err == nil {
-		dc.eventRecorder.Eventf(deployment, api.EventTypeNormal, "ScalingReplicaSet", "Scaled %s replica set %s to %d", scalingOperation, rs.Name, newScale)
-	} else {
-		dc.enqueueDeployment(deployment)
-	}
+	newRS, err := dc.scaleReplicaSet(rs, newScale, deployment, scalingOperation)
 	return true, newRS, err
 }
 
-func (dc *DeploymentController) scaleReplicaSet(rs *extensions.ReplicaSet, newScale int32) (*extensions.ReplicaSet, error) {
-	// TODO: Using client for now, update to use store when it is ready.
+func (dc *DeploymentController) scaleReplicaSet(rs *extensions.ReplicaSet, newScale int32, deployment *extensions.Deployment, scalingOperation string) (*extensions.ReplicaSet, error) {
 	// NOTE: This mutates the ReplicaSet passed in. Not sure if that's a good idea.
 	rs.Spec.Replicas = newScale
-	return dc.client.Extensions().ReplicaSets(rs.ObjectMeta.Namespace).Update(rs)
+	setReplicasAnnotations(rs, deployment.Spec.Replicas, deployment.Spec.Replicas+maxSurge(*deployment))
+	rs, err := dc.client.Extensions().ReplicaSets(rs.ObjectMeta.Namespace).Update(rs)
+	if err == nil {
+		dc.eventRecorder.Eventf(deployment, api.EventTypeNormal, "ScalingReplicaSet", "Scaled %s replica set %s to %d", scalingOperation, rs.Name, newScale)
+	} else {
+		glog.Warningf("Cannot update replica set %q: %v", rs.Name, err)
+		dc.enqueueDeployment(deployment)
+	}
+	return rs, err
 }
 
 func (dc *DeploymentController) updateDeployment(deployment *extensions.Deployment) (*extensions.Deployment, error) {
-	// TODO: Using client for now, update to use store when it is ready.
 	return dc.client.Extensions().Deployments(deployment.ObjectMeta.Namespace).Update(deployment)
 }
 
@@ -1300,3 +1401,28 @@ func (dc *DeploymentController) rollbackToTemplate(deployment *extensions.Deploy
 	d, err = dc.updateDeploymentAndClearRollbackTo(deployment)
 	return
 }
+
+// isScalingEvent checks whether the provided deployment has been updated with a scaling event
+// by looking at the desired-replicas annotation in the active replica sets of the deployment.
+func (dc *DeploymentController) isScalingEvent(d *extensions.Deployment) bool {
+	newRS, oldRSs, err := dc.getAllReplicaSetsAndSyncRevision(d, false)
+	if err != nil {
+		return false
+	}
+	// If there is no new replica set matching this deployment and the deployment isn't paused
+	// then there is a new rollout that waits to happen
+	if newRS == nil && !d.Spec.Paused {
+		return false
+	}
+	allRSs := append(oldRSs, newRS)
+	for _, rs := range controller.FilterActiveReplicaSets(allRSs) {
+		desired, ok := getDesiredReplicasAnnotation(rs)
+		if !ok {
+			continue
+		}
+		if desired != d.Spec.Replicas {
+			return true
+		}
+	}
+	return false
+}

pkg/controller/deployment/deployment_controller_test.go

@@ -19,6 +19,7 @@ package deployment
 import (
 	"fmt"
 	"testing"
+	"time"
 
 	"k8s.io/kubernetes/pkg/api"
 	"k8s.io/kubernetes/pkg/api/testapi"
@@ -33,10 +34,16 @@ import (
 	"k8s.io/kubernetes/pkg/util/intstr"
 )
 
-func rs(name string, replicas int, selector map[string]string) *exp.ReplicaSet {
+var (
+	alwaysReady = func() bool { return true }
+	noTimestamp = unversioned.Time{}
+)
+
+func rs(name string, replicas int, selector map[string]string, timestamp unversioned.Time) *exp.ReplicaSet {
 	return &exp.ReplicaSet{
 		ObjectMeta: api.ObjectMeta{
-			Name: name,
+			Name:              name,
+			CreationTimestamp: timestamp,
 		},
 		Spec: exp.ReplicaSetSpec{
 			Replicas: int32(replicas),
@@ -47,7 +54,7 @@ func rs(name string, replicas int, selector map[string]string) *exp.ReplicaSet {
 }
 
 func newRSWithStatus(name string, specReplicas, statusReplicas int, selector map[string]string) *exp.ReplicaSet {
-	rs := rs(name, specReplicas, selector)
+	rs := rs(name, specReplicas, selector, noTimestamp)
 	rs.Status = exp.ReplicaSetStatus{
 		Replicas: int32(statusReplicas),
 	}
@@ -73,8 +80,6 @@ func deployment(name string, replicas int, maxSurge, maxUnavailable intstr.IntOr
 	}
 }
 
-var alwaysReady = func() bool { return true }
-
 func newDeployment(replicas int, revisionHistoryLimit *int) *exp.Deployment {
 	var v *int32
 	if revisionHistoryLimit != nil {
@@ -117,6 +122,13 @@ func newDeployment(replicas int, revisionHistoryLimit *int) *exp.Deployment {
 	return &d
 }
 
+// TODO: Consolidate all deployment helpers into one.
+func newDeploymentEnhanced(replicas int, maxSurge intstr.IntOrString) *exp.Deployment {
+	d := newDeployment(replicas, nil)
+	d.Spec.Strategy.RollingUpdate.MaxSurge = maxSurge
+	return d
+}
+
 func newReplicaSet(d *exp.Deployment, name string, replicas int) *exp.ReplicaSet {
 	return &exp.ReplicaSet{
 		ObjectMeta: api.ObjectMeta{
@@ -128,13 +140,262 @@ func newReplicaSet(d *exp.Deployment, name string, replicas int) *exp.ReplicaSet
 			Template: d.Spec.Template,
 		},
 	}
-
 }
 
 func newListOptions() api.ListOptions {
 	return api.ListOptions{}
 }
 
+// TestScale tests proportional scaling of deployments. Note that fenceposts for
+// rolling out (maxUnavailable, maxSurge) have no meaning for simple scaling other
+// than recording maxSurge as part of the max-replicas annotation that is taken
+// into account in the next scale event (max-replicas is used for calculating the
+// proportion of a replica set).
+func TestScale(t *testing.T) {
+	newTimestamp := unversioned.Date(2016, 5, 20, 2, 0, 0, 0, time.UTC)
+	oldTimestamp := unversioned.Date(2016, 5, 20, 1, 0, 0, 0, time.UTC)
+	olderTimestamp := unversioned.Date(2016, 5, 20, 0, 0, 0, 0, time.UTC)
+
+	tests := []struct {
+		name          string
+		deployment    *exp.Deployment
+		oldDeployment *exp.Deployment
+
+		newRS  *exp.ReplicaSet
+		oldRSs []*exp.ReplicaSet
+
+		expectedNew *exp.ReplicaSet
+		expectedOld []*exp.ReplicaSet
+
+		desiredReplicasAnnotations map[string]int32
+	}{
+		{
+			name:          "normal scaling event: 10 -> 12",
+			deployment:    newDeployment(12, nil),
+			oldDeployment: newDeployment(10, nil),
+
+			newRS:  rs("foo-v1", 10, nil, newTimestamp),
+			oldRSs: []*exp.ReplicaSet{},
+
+			expectedNew: rs("foo-v1", 12, nil, newTimestamp),
+			expectedOld: []*exp.ReplicaSet{},
+		},
+		{
+			name:          "normal scaling event: 10 -> 5",
+			deployment:    newDeployment(5, nil),
+			oldDeployment: newDeployment(10, nil),
+
+			newRS:  rs("foo-v1", 10, nil, newTimestamp),
+			oldRSs: []*exp.ReplicaSet{},
+
+			expectedNew: rs("foo-v1", 5, nil, newTimestamp),
+			expectedOld: []*exp.ReplicaSet{},
+		},
+		{
+			name:          "proportional scaling: 5 -> 10",
+			deployment:    newDeployment(10, nil),
+			oldDeployment: newDeployment(5, nil),
+
+			newRS:  rs("foo-v2", 2, nil, newTimestamp),
+			oldRSs: []*exp.ReplicaSet{rs("foo-v1", 3, nil, oldTimestamp)},
+
+			expectedNew: rs("foo-v2", 4, nil, newTimestamp),
+			expectedOld: []*exp.ReplicaSet{rs("foo-v1", 6, nil, oldTimestamp)},
+		},
+		{
+			name:          "proportional scaling: 5 -> 3",
+			deployment:    newDeployment(3, nil),
+			oldDeployment: newDeployment(5, nil),
+
+			newRS:  rs("foo-v2", 2, nil, newTimestamp),
+			oldRSs: []*exp.ReplicaSet{rs("foo-v1", 3, nil, oldTimestamp)},
+
+			expectedNew: rs("foo-v2", 1, nil, newTimestamp),
+			expectedOld: []*exp.ReplicaSet{rs("foo-v1", 2, nil, oldTimestamp)},
+		},
+		{
+			name:          "proportional scaling: 9 -> 4",
+			deployment:    newDeployment(4, nil),
+			oldDeployment: newDeployment(9, nil),
+
+			newRS:  rs("foo-v2", 8, nil, newTimestamp),
+			oldRSs: []*exp.ReplicaSet{rs("foo-v1", 1, nil, oldTimestamp)},
+
+			expectedNew: rs("foo-v2", 4, nil, newTimestamp),
+			expectedOld: []*exp.ReplicaSet{rs("foo-v1", 0, nil, oldTimestamp)},
+		},
+		{
+			name:          "proportional scaling: 7 -> 10",
+			deployment:    newDeployment(10, nil),
+			oldDeployment: newDeployment(7, nil),
+
+			newRS:  rs("foo-v3", 2, nil, newTimestamp),
+			oldRSs: []*exp.ReplicaSet{rs("foo-v2", 3, nil, oldTimestamp), rs("foo-v1", 2, nil, olderTimestamp)},
+
+			expectedNew: rs("foo-v3", 3, nil, newTimestamp),
+			expectedOld: []*exp.ReplicaSet{rs("foo-v2", 4, nil, oldTimestamp), rs("foo-v1", 3, nil, olderTimestamp)},
+		},
+		{
+			name:          "proportional scaling: 13 -> 8",
+			deployment:    newDeployment(8, nil),
+			oldDeployment: newDeployment(13, nil),
+
+			newRS:  rs("foo-v3", 2, nil, newTimestamp),
+			oldRSs: []*exp.ReplicaSet{rs("foo-v2", 8, nil, oldTimestamp), rs("foo-v1", 3, nil, olderTimestamp)},
+
+			expectedNew: rs("foo-v3", 1, nil, newTimestamp),
+			expectedOld: []*exp.ReplicaSet{rs("foo-v2", 5, nil, oldTimestamp), rs("foo-v1", 2, nil, olderTimestamp)},
+		},
+		// Scales up the new replica set.
+		{
+			name:          "leftover distribution: 3 -> 4",
+			deployment:    newDeployment(4, nil),
+			oldDeployment: newDeployment(3, nil),
+
+			newRS:  rs("foo-v3", 1, nil, newTimestamp),
+			oldRSs: []*exp.ReplicaSet{rs("foo-v2", 1, nil, oldTimestamp), rs("foo-v1", 1, nil, olderTimestamp)},
+
+			expectedNew: rs("foo-v3", 2, nil, newTimestamp),
+			expectedOld: []*exp.ReplicaSet{rs("foo-v2", 1, nil, oldTimestamp), rs("foo-v1", 1, nil, olderTimestamp)},
+		},
+		// Scales down the older replica set.
+		{
+			name:          "leftover distribution: 3 -> 2",
+			deployment:    newDeployment(2, nil),
+			oldDeployment: newDeployment(3, nil),
+
+			newRS:  rs("foo-v3", 1, nil, newTimestamp),
+			oldRSs: []*exp.ReplicaSet{rs("foo-v2", 1, nil, oldTimestamp), rs("foo-v1", 1, nil, olderTimestamp)},
+
+			expectedNew: rs("foo-v3", 1, nil, newTimestamp),
+			expectedOld: []*exp.ReplicaSet{rs("foo-v2", 1, nil, oldTimestamp), rs("foo-v1", 0, nil, olderTimestamp)},
+		},
+		// Scales up the latest replica set first.
+		{
+			name:          "proportional scaling (no new rs): 4 -> 5",
+			deployment:    newDeployment(5, nil),
+			oldDeployment: newDeployment(4, nil),
+
+			newRS:  nil,
+			oldRSs: []*exp.ReplicaSet{rs("foo-v2", 2, nil, oldTimestamp), rs("foo-v1", 2, nil, olderTimestamp)},
+
+			expectedNew: nil,
+			expectedOld: []*exp.ReplicaSet{rs("foo-v2", 3, nil, oldTimestamp), rs("foo-v1", 2, nil, olderTimestamp)},
+		},
+		// Scales down to zero
+		{
+			name:          "proportional scaling: 6 -> 0",
+			deployment:    newDeployment(0, nil),
+			oldDeployment: newDeployment(6, nil),
+
+			newRS:  rs("foo-v3", 3, nil, newTimestamp),
+			oldRSs: []*exp.ReplicaSet{rs("foo-v2", 2, nil, oldTimestamp), rs("foo-v1", 1, nil, olderTimestamp)},
+
+			expectedNew: rs("foo-v3", 0, nil, newTimestamp),
+			expectedOld: []*exp.ReplicaSet{rs("foo-v2", 0, nil, oldTimestamp), rs("foo-v1", 0, nil, olderTimestamp)},
+		},
+		// Scales up from zero
+		{
+			name:          "proportional scaling: 0 -> 6",
+			deployment:    newDeployment(6, nil),
+			oldDeployment: newDeployment(0, nil),
+
+			newRS:  rs("foo-v3", 0, nil, newTimestamp),
+			oldRSs: []*exp.ReplicaSet{rs("foo-v2", 0, nil, oldTimestamp), rs("foo-v1", 0, nil, olderTimestamp)},
+
+			expectedNew: rs("foo-v3", 6, nil, newTimestamp),
+			expectedOld: []*exp.ReplicaSet{rs("foo-v2", 0, nil, oldTimestamp), rs("foo-v1", 0, nil, olderTimestamp)},
+		},
+		// Scenario: deployment.spec.replicas == 3 ( foo-v1.spec.replicas == foo-v2.spec.replicas == foo-v3.spec.replicas == 1 )
+		// Deployment is scaled to 5. foo-v3.spec.replicas and foo-v2.spec.replicas should increment by 1 but foo-v2 fails to
+		// update.
+		{
+			name:          "failed rs update",
+			deployment:    newDeployment(5, nil),
+			oldDeployment: newDeployment(5, nil),
+
+			newRS:  rs("foo-v3", 2, nil, newTimestamp),
+			oldRSs: []*exp.ReplicaSet{rs("foo-v2", 1, nil, oldTimestamp), rs("foo-v1", 1, nil, olderTimestamp)},
+
+			expectedNew: rs("foo-v3", 2, nil, newTimestamp),
+			expectedOld: []*exp.ReplicaSet{rs("foo-v2", 2, nil, oldTimestamp), rs("foo-v1", 1, nil, olderTimestamp)},
+
+			desiredReplicasAnnotations: map[string]int32{"foo-v2": int32(3)},
+		},
+		{
+			name:          "deployment with surge pods",
+			deployment:    newDeploymentEnhanced(20, intstr.FromInt(2)),
+			oldDeployment: newDeploymentEnhanced(10, intstr.FromInt(2)),
+
+			newRS:  rs("foo-v2", 6, nil, newTimestamp),
+			oldRSs: []*exp.ReplicaSet{rs("foo-v1", 6, nil, oldTimestamp)},
+
+			expectedNew: rs("foo-v2", 11, nil, newTimestamp),
+			expectedOld: []*exp.ReplicaSet{rs("foo-v1", 11, nil, oldTimestamp)},
+		},
+		{
+			name:          "change both surge and size",
+			deployment:    newDeploymentEnhanced(50, intstr.FromInt(6)),
+			oldDeployment: newDeploymentEnhanced(10, intstr.FromInt(3)),
+
+			newRS:  rs("foo-v2", 5, nil, newTimestamp),
+			oldRSs: []*exp.ReplicaSet{rs("foo-v1", 8, nil, oldTimestamp)},
+
+			expectedNew: rs("foo-v2", 22, nil, newTimestamp),
+			expectedOld: []*exp.ReplicaSet{rs("foo-v1", 34, nil, oldTimestamp)},
+		},
+	}
+
+	for _, test := range tests {
+		_ = olderTimestamp
+		t.Log(test.name)
+		fake := fake.Clientset{}
+		dc := &DeploymentController{
+			client:        &fake,
+			eventRecorder: &record.FakeRecorder{},
+		}
+
+		if test.newRS != nil {
+			desiredReplicas := test.oldDeployment.Spec.Replicas
+			if desired, ok := test.desiredReplicasAnnotations[test.newRS.Name]; ok {
+				desiredReplicas = desired
+			}
+			setReplicasAnnotations(test.newRS, desiredReplicas, desiredReplicas+maxSurge(*test.oldDeployment))
+		}
+		for i := range test.oldRSs {
+			rs := test.oldRSs[i]
+			if rs == nil {
+				continue
+			}
+			desiredReplicas := test.oldDeployment.Spec.Replicas
+			if desired, ok := test.desiredReplicasAnnotations[rs.Name]; ok {
+				desiredReplicas = desired
+			}
+			setReplicasAnnotations(rs, desiredReplicas, desiredReplicas+maxSurge(*test.oldDeployment))
+		}
+
+		if err := dc.scale(test.deployment, test.newRS, test.oldRSs); err != nil {
+			t.Errorf("%s: unexpected error: %v", test.name, err)
+			continue
+		}
+		if test.expectedNew != nil && test.newRS != nil && test.expectedNew.Spec.Replicas != test.newRS.Spec.Replicas {
+			t.Errorf("%s: expected new replicas: %d, got: %d", test.name, test.expectedNew.Spec.Replicas, test.newRS.Spec.Replicas)
+			continue
+		}
+		if len(test.expectedOld) != len(test.oldRSs) {
+			t.Errorf("%s: expected %d old replica sets, got %d", test.name, len(test.expectedOld), len(test.oldRSs))
+			continue
+		}
+		for n := range test.oldRSs {
+			rs := test.oldRSs[n]
+			exp := test.expectedOld[n]
+			if exp.Spec.Replicas != rs.Spec.Replicas {
+				t.Errorf("%s: expected old (%s) replicas: %d, got: %d", test.name, rs.Name, exp.Spec.Replicas, rs.Spec.Replicas)
+			}
+		}
+	}
+}
+
 func TestDeploymentController_reconcileNewReplicaSet(t *testing.T) {
 	tests := []struct {
 		deploymentReplicas  int
@@ -188,8 +449,8 @@ func TestDeploymentController_reconcileNewReplicaSet(t *testing.T) {
 
 	for i, test := range tests {
 		t.Logf("executing scenario %d", i)
-		newRS := rs("foo-v2", test.newReplicas, nil)
-		oldRS := rs("foo-v2", test.oldReplicas, nil)
+		newRS := rs("foo-v2", test.newReplicas, nil, noTimestamp)
+		oldRS := rs("foo-v2", test.oldReplicas, nil, noTimestamp)
 		allRSs := []*exp.ReplicaSet{newRS, oldRS}
 		deployment := deployment("foo", test.deploymentReplicas, test.maxSurge, intstr.FromInt(0), nil)
 		fake := fake.Clientset{}
@@ -289,8 +550,8 @@ func TestDeploymentController_reconcileOldReplicaSets(t *testing.T) {
 
 		newSelector := map[string]string{"foo": "new"}
 		oldSelector := map[string]string{"foo": "old"}
-		newRS := rs("foo-new", test.newReplicas, newSelector)
-		oldRS := rs("foo-old", test.oldReplicas, oldSelector)
+		newRS := rs("foo-new", test.newReplicas, newSelector, noTimestamp)
+		oldRS := rs("foo-old", test.oldReplicas, oldSelector, noTimestamp)
 		oldRSs := []*exp.ReplicaSet{oldRS}
 		allRSs := []*exp.ReplicaSet{oldRS, newRS}
 
@@ -429,7 +690,7 @@ func TestDeploymentController_cleanupUnhealthyReplicas(t *testing.T) {
 
 	for i, test := range tests {
 		t.Logf("executing scenario %d", i)
-		oldRS := rs("foo-v2", test.oldReplicas, nil)
+		oldRS := rs("foo-v2", test.oldReplicas, nil, noTimestamp)
 		oldRSs := []*exp.ReplicaSet{oldRS}
 		deployment := deployment("foo", 10, intstr.FromInt(2), intstr.FromInt(2), nil)
 		fakeClientset := fake.Clientset{}
@@ -538,7 +799,7 @@ func TestDeploymentController_scaleDownOldReplicaSetsForRollingUpdate(t *testing
 
 	for i, test := range tests {
 		t.Logf("executing scenario %d", i)
-		oldRS := rs("foo-v2", test.oldReplicas, nil)
+		oldRS := rs("foo-v2", test.oldReplicas, nil, noTimestamp)
 		allRSs := []*exp.ReplicaSet{oldRS}
 		oldRSs := []*exp.ReplicaSet{oldRS}
 		deployment := deployment("foo", test.deploymentReplicas, intstr.FromInt(0), test.maxUnavailable, map[string]string{"foo": "bar"})
@@ -610,7 +871,7 @@ func TestDeploymentController_scaleDownOldReplicaSetsForRollingUpdate(t *testing
 	}
 }
 
-func TestDeploymentController_cleanupOldReplicaSets(t *testing.T) {
+func TestDeploymentController_cleanupDeployment(t *testing.T) {
 	selector := map[string]string{"foo": "bar"}
 
 	tests := []struct {
@@ -669,7 +930,7 @@ func TestDeploymentController_cleanupOldReplicaSets(t *testing.T) {
 		}
 
 		d := newDeployment(1, &tests[i].revisionHistoryLimit)
-		controller.cleanupOldReplicaSets(test.oldRSs, d)
+		controller.cleanupDeployment(test.oldRSs, d)
 
 		gotDeletions := 0
 		for _, action := range fake.Actions() {

pkg/controller/deployment/util.go

+/*
+Copyright 2016 The Kubernetes Authors All rights reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+*/
+
+package deployment
+
+import (
+	"fmt"
+	"sort"
+	"strconv"
+
+	"github.com/golang/glog"
+
+	"k8s.io/kubernetes/pkg/apis/extensions"
+	"k8s.io/kubernetes/pkg/controller"
+	deploymentutil "k8s.io/kubernetes/pkg/util/deployment"
+	"k8s.io/kubernetes/pkg/util/integer"
+)
+
+// findActiveOrLatest returns the only active or the latest replica set in case there is at most one active
+// replica set. If there are more active replica sets, then we should proportionally scale them.
+func findActiveOrLatest(newRS *extensions.ReplicaSet, oldRSs []*extensions.ReplicaSet) *extensions.ReplicaSet {
+	if newRS == nil && len(oldRSs) == 0 {
+		return nil
+	}
+
+	sort.Sort(sort.Reverse(controller.ReplicaSetsByCreationTimestamp(oldRSs)))
+	allRSs := controller.FilterActiveReplicaSets(append(oldRSs, newRS))
+
+	switch len(allRSs) {
+	case 0:
+		// If there is no active replica set then we should return the newest.
+		if newRS != nil {
+			return newRS
+		}
+		return oldRSs[0]
+	case 1:
+		return allRSs[0]
+	default:
+		return nil
+	}
+}
+
+func getDesiredReplicasAnnotation(rs *extensions.ReplicaSet) (int32, bool) {
+	return getIntFromAnnotation(rs, deploymentutil.DesiredReplicasAnnotation)
+}
+
+func getMaxReplicasAnnotation(rs *extensions.ReplicaSet) (int32, bool) {
+	return getIntFromAnnotation(rs, deploymentutil.MaxReplicasAnnotation)
+}
+
+func getIntFromAnnotation(rs *extensions.ReplicaSet, annotationKey string) (int32, bool) {
+	annotationValue, ok := rs.Annotations[annotationKey]
+	if !ok {
+		return int32(0), false
+	}
+	intValue, err := strconv.Atoi(annotationValue)
+	if err != nil {
+		glog.Warningf("Cannot convert the value %q with annotation key %q for the replica set %q",
+			annotationValue, annotationKey, rs.Name)
+		return int32(0), false
+	}
+	return int32(intValue), true
+}
+
+func setReplicasAnnotations(rs *extensions.ReplicaSet, desiredReplicas, maxReplicas int32) bool {
+	updated := false
+	if rs.Annotations == nil {
+		rs.Annotations = make(map[string]string)
+	}
+	desiredString := fmt.Sprintf("%d", desiredReplicas)
+	if hasString := rs.Annotations[deploymentutil.DesiredReplicasAnnotation]; hasString != desiredString {
+		rs.Annotations[deploymentutil.DesiredReplicasAnnotation] = desiredString
+		updated = true
+	}
+	maxString := fmt.Sprintf("%d", maxReplicas)
+	if hasString := rs.Annotations[deploymentutil.MaxReplicasAnnotation]; hasString != maxString {
+		rs.Annotations[deploymentutil.MaxReplicasAnnotation] = maxString
+		updated = true
+	}
+	return updated
+}
+
+// maxUnavailable returns the maximum unavailable pods a rolling deployment can take.
+func maxUnavailable(deployment extensions.Deployment) int32 {
+	if !deploymentutil.IsRollingUpdate(&deployment) {
+		return int32(0)
+	}
+	// Error caught by validation
+	_, maxUnavailable, _ := deploymentutil.ResolveFenceposts(&deployment.Spec.Strategy.RollingUpdate.MaxSurge, &deployment.Spec.Strategy.RollingUpdate.MaxUnavailable, deployment.Spec.Replicas)
+	return maxUnavailable
+}
+
+// maxSurge returns the maximum surge pods a rolling deployment can take.
+func maxSurge(deployment extensions.Deployment) int32 {
+	if !deploymentutil.IsRollingUpdate(&deployment) {
+		return int32(0)
+	}
+	// Error caught by validation
+	maxSurge, _, _ := deploymentutil.ResolveFenceposts(&deployment.Spec.Strategy.RollingUpdate.MaxSurge, &deployment.Spec.Strategy.RollingUpdate.MaxUnavailable, deployment.Spec.Replicas)
+	return maxSurge
+}
+
+// getProportion will estimate the proportion for the provided replica set using 1. the current size
+// of the parent deployment, 2. the replica count that needs be added on the replica sets of the
+// deployment, and 3. the total replicas added in the replica sets of the deployment so far.
+func getProportion(rs *extensions.ReplicaSet, d extensions.Deployment, deploymentReplicasToAdd, deploymentReplicasAdded int32) int32 {
+	if rs == nil || rs.Spec.Replicas == 0 || deploymentReplicasToAdd == 0 || deploymentReplicasToAdd == deploymentReplicasAdded {
+		return int32(0)
+	}
+
+	rsFraction := getReplicaSetFraction(*rs, d)
+	allowed := deploymentReplicasToAdd - deploymentReplicasAdded
+
+	if deploymentReplicasToAdd > 0 {
+		// Use the minimum between the replica set fraction and the maximum allowed replicas
+		// when scaling up. This way we ensure we will not scale up more than the allowed
+		// replicas we can add.
+		return integer.Int32Min(rsFraction, allowed)
+	}
+	// Use the maximum between the replica set fraction and the maximum allowed replicas
+	// when scaling down. This way we ensure we will not scale down more than the allowed
+	// replicas we can remove.
+	return integer.Int32Max(rsFraction, allowed)
+}
+
+// getReplicaSetFraction estimates the fraction of replicas a replica set can have in
+// 1. a scaling event during a rollout or 2. when scaling a paused deployment.
+func getReplicaSetFraction(rs extensions.ReplicaSet, d extensions.Deployment) int32 {
+	// If we are scaling down to zero then the fraction of this replica set is its whole size (negative)
+	if d.Spec.Replicas == int32(0) {
+		return -rs.Spec.Replicas
+	}
+
+	deploymentReplicas := d.Spec.Replicas + maxSurge(d)
+	annotatedReplicas, ok := getMaxReplicasAnnotation(&rs)
+	if !ok {
+		// If we cannot find the annotation then fallback to the current deployment size. Note that this
+		// will not be an accurate proportion estimation in case other replica sets have different values
+		// which means that the deployment was scaled at some point but we at least will stay in limits
+		// due to the min-max comparisons in getProportion.
+		annotatedReplicas = d.Status.Replicas
+	}
+
+	// We should never proportionally scale up from zero which means rs.spec.replicas and annotatedReplicas
+	// will never be zero here.
+	newRSsize := (float64(rs.Spec.Replicas * deploymentReplicas)) / float64(annotatedReplicas)
+	return integer.RoundToInt32(newRSsize) - rs.Spec.Replicas
+}

pkg/util/deployment/deployment.go

@@ -41,6 +41,14 @@ import (
 const (
 	// The revision annotation of a deployment's replica sets which records its rollout sequence
 	RevisionAnnotation = "deployment.kubernetes.io/revision"
+	// DesiredReplicasAnnotation is the desired replicas for a deployment recorded as an annotation
+	// in its replica sets. Helps in separating scaling events from the rollout process and for
+	// determining if the new replica set for a deployment is really saturated.
+	DesiredReplicasAnnotation = "deployment.kubernetes.io/desired-replicas"
+	// MaxReplicasAnnotation is the maximum replicas a deployment can have at a given point, which
+	// is deployment.spec.replicas + maxSurge. Used by the underlying replica sets to estimate their
+	// proportions in case the deployment has surge replicas.
+	MaxReplicasAnnotation = "deployment.kubernetes.io/max-replicas"
 
 	// Here are the possible rollback event reasons
 	RollbackRevisionNotFound  = "DeploymentRollbackRevisionNotFound"
@@ -434,6 +442,21 @@ func NewRSNewReplicas(deployment *extensions.Deployment, allRSs []*extensions.Re
 	}
 }
 
+// IsSaturated checks if the new replica set is saturated by comparing its size with its deployment size.
+// Both the deployment and the replica set have to believe this replica set can own all of the desired
+// replicas in the deployment and the annotation helps in achieving that.
+func IsSaturated(deployment *extensions.Deployment, rs *extensions.ReplicaSet) bool {
+	if rs == nil {
+		return false
+	}
+	desiredString := rs.Annotations[DesiredReplicasAnnotation]
+	desired, err := strconv.Atoi(desiredString)
+	if err != nil {
+		return false
+	}
+	return rs.Spec.Replicas == deployment.Spec.Replicas && int32(desired) == deployment.Spec.Replicas
+}
+
 // Polls for deployment to be updated so that deployment.Status.ObservedGeneration >= desiredGeneration.
 // Returns error if polling timesout.
 func WaitForObservedDeployment(getDeploymentFunc func() (*extensions.Deployment, error), desiredGeneration int64, interval, timeout time.Duration) error {

pkg/util/intstr/intstr.go

@@ -144,5 +144,5 @@ func getIntOrPercentValue(intOrStr *IntOrString) (int, bool, error) {
 		}
 		return int(v), true, nil
 	}
-	return 0, false, fmt.Errorf("invalid value: neither int nor percentage")
+	return 0, false, fmt.Errorf("invalid type: neither int nor percentage")
 }

test/e2e/deployment.go

@@ -80,6 +80,9 @@ var _ = framework.KubeDescribe("Deployment", func() {
 	It("deployment should label adopted RSs and pods", func() {
 		testDeploymentLabelAdopted(f)
 	})
+	It("paused deployment should be able to scale", func() {
+		testScalePausedDeployment(f)
+	})
 })
 
 func newRS(rsName string, replicas int32, rsPodLabels map[string]string, imageName string, image string) *extensions.ReplicaSet {
@@ -569,6 +572,8 @@ func testPausedDeployment(f *framework.Framework) {
 	podLabels := map[string]string{"name": nginxImageName}
 	d := newDeployment(deploymentName, 1, podLabels, nginxImageName, nginxImage, extensions.RollingUpdateDeploymentStrategyType, nil)
 	d.Spec.Paused = true
+	tgps := int64(20)
+	d.Spec.Template.Spec.TerminationGracePeriodSeconds = &tgps
 	framework.Logf("Creating paused deployment %s", deploymentName)
 	_, err := c.Extensions().Deployments(ns).Create(d)
 	Expect(err).NotTo(HaveOccurred())
@@ -622,21 +627,34 @@ func testPausedDeployment(f *framework.Framework) {
 	err = framework.WaitForObservedDeployment(c, ns, deploymentName, deployment.Generation)
 	Expect(err).NotTo(HaveOccurred())
 
-	newRS, err := deploymentutil.GetNewReplicaSet(deployment, c)
+	// Update the deployment template - the new replicaset should stay the same
+	framework.Logf("Updating paused deployment %q", deploymentName)
+	newTGPS := int64(40)
+	deployment, err = framework.UpdateDeploymentWithRetries(c, ns, d.Name, func(update *extensions.Deployment) {
+		update.Spec.Template.Spec.TerminationGracePeriodSeconds = &newTGPS
+	})
 	Expect(err).NotTo(HaveOccurred())
-	Expect(framework.DeleteReplicaSet(unversionedClient, ns, newRS.Name)).NotTo(HaveOccurred())
 
-	deployment, err = c.Extensions().Deployments(ns).Get(deploymentName)
+	err = framework.WaitForObservedDeployment(c, ns, deploymentName, deployment.Generation)
 	Expect(err).NotTo(HaveOccurred())
 
-	if !deployment.Spec.Paused {
-		err = fmt.Errorf("deployment %q should be paused", deployment.Name)
+	framework.Logf("Looking for new replicaset for paused deployment %q (there should be none)", deploymentName)
+	newRS, err := deploymentutil.GetNewReplicaSet(deployment, c)
+	Expect(err).NotTo(HaveOccurred())
+	if newRS != nil {
+		err = fmt.Errorf("No replica set should match the deployment template but there is %q", newRS.Name)
 		Expect(err).NotTo(HaveOccurred())
 	}
-	shouldBeNil, err := deploymentutil.GetNewReplicaSet(deployment, c)
+
+	_, allOldRs, err := deploymentutil.GetOldReplicaSets(deployment, c)
 	Expect(err).NotTo(HaveOccurred())
-	if shouldBeNil != nil {
-		err = fmt.Errorf("deployment %q shouldn't have a replica set but there is %q", deployment.Name, shouldBeNil.Name)
+	if len(allOldRs) != 1 {
+		err = fmt.Errorf("expected an old replica set")
+		Expect(err).NotTo(HaveOccurred())
+	}
+	framework.Logf("Comparing deployment diff with old replica set %q", allOldRs[0].Name)
+	if *allOldRs[0].Spec.Template.Spec.TerminationGracePeriodSeconds == newTGPS {
+		err = fmt.Errorf("TerminationGracePeriodSeconds on the replica set should be %d but is %d", tgps, newTGPS)
 		Expect(err).NotTo(HaveOccurred())
 	}
 }
@@ -944,3 +962,54 @@ func testDeploymentLabelAdopted(f *framework.Framework) {
 	Expect(err).NotTo(HaveOccurred())
 	Expect(int32(len(pods.Items))).Should(Equal(replicas))
 }
+
+func testScalePausedDeployment(f *framework.Framework) {
+	ns := f.Namespace.Name
+	c := adapter.FromUnversionedClient(f.Client)
+
+	podLabels := map[string]string{"name": nginxImageName}
+	replicas := int32(3)
+
+	// Create a nginx deployment.
+	deploymentName := "nginx-deployment"
+	d := newDeployment(deploymentName, replicas, podLabels, nginxImageName, nginxImage, extensions.RollingUpdateDeploymentStrategyType, nil)
+	framework.Logf("Creating deployment %q", deploymentName)
+	_, err := c.Extensions().Deployments(ns).Create(d)
+	Expect(err).NotTo(HaveOccurred())
+	defer stopDeployment(c, f.Client, ns, deploymentName)
+
+	// Check that deployment is created fine.
+	deployment, err := c.Extensions().Deployments(ns).Get(deploymentName)
+	Expect(err).NotTo(HaveOccurred())
+
+	err = framework.WaitForObservedDeployment(c, ns, deploymentName, deployment.Generation)
+	Expect(err).NotTo(HaveOccurred())
+
+	rs, err := deploymentutil.GetNewReplicaSet(deployment, c)
+	Expect(err).NotTo(HaveOccurred())
+
+	// Pause the deployment and try to scale it.
+	deployment, err = framework.UpdateDeploymentWithRetries(c, ns, d.Name, func(update *extensions.Deployment) {
+		update.Spec.Paused = true
+	})
+	Expect(err).NotTo(HaveOccurred())
+
+	// Scale the paused deployment.
+	framework.Logf("Scaling up the paused deployment %q", deploymentName)
+	newReplicas := int32(5)
+	deployment, err = framework.UpdateDeploymentWithRetries(c, ns, deployment.Name, func(update *extensions.Deployment) {
+		update.Spec.Replicas = newReplicas
+	})
+	Expect(err).NotTo(HaveOccurred())
+
+	err = framework.WaitForObservedDeployment(c, ns, deploymentName, deployment.Generation)
+	Expect(err).NotTo(HaveOccurred())
+
+	rs, err = deploymentutil.GetNewReplicaSet(deployment, c)
+	Expect(err).NotTo(HaveOccurred())
+
+	if rs.Spec.Replicas != newReplicas {
+		err = fmt.Errorf("Expected %d replicas for the new replica set, got %d", newReplicas, rs.Spec.Replicas)
+		Expect(err).NotTo(HaveOccurred())
+	}
+}