Merge pull request #67555 from wgliang/opt/improve-performance

Automatic merge from submit-queue (batch tested with PRs 67555, 68196). If you want to cherry-pick this change to another branch, please follow the instructions here: https://github.com/kubernetes/community/blob/master/contributors/devel/cherry-picks.md.

Not split nodes when searching for nodes but doing it all at once

**What this PR does / why we need it**:
Not split nodes when searching for nodes but doing it all at once.

**Which issue(s) this PR fixes** *(optional, in `fixes #<issue number>(, fixes #<issue_number>, ...)` format, will close the issue(s) when PR gets merged)*:
Fixes #

**Special notes for your reviewer**:
@bsalamat 

This is a follow up PR of #66733.

https://github.com/kubernetes/kubernetes/pull/66733#discussion_r205932531

**Release note**:

```release-note
Not split nodes when searching for nodes but doing it all at once.
```

pkg/scheduler/core/generic_scheduler.go

@@ -17,6 +17,7 @@ limitations under the License.
 package core
 
 import (
+	"context"
 	"fmt"
 	"math"
 	"sort"
@@ -373,16 +374,19 @@ func (g *genericScheduler) findNodesThatFit(pod *v1.Pod, nodes []*v1.Node) ([]*v
 
 		// Create filtered list with enough space to avoid growing it
 		// and allow assigning.
-		filtered = make([]*v1.Node, 2*numNodesToFind)
+		filtered = make([]*v1.Node, numNodesToFind)
 		errs := errors.MessageCountMap{}
-		var predicateResultLock sync.Mutex
-		var filteredLen int32
+		var (
+			predicateResultLock sync.Mutex
+			filteredLen         int32
+			equivClass          *equivalence.Class
+		)
+
+		ctx, cancel := context.WithCancel(context.Background())
 
 		// We can use the same metadata producer for all nodes.
 		meta := g.predicateMetaProducer(pod, g.cachedNodeInfoMap)
 
-		var equivClass *equivalence.Class
-
 		if g.equivalenceCache != nil {
 			// getEquivalenceClassInfo will return immediately if no equivalence pod found
 			equivClass = equivalence.NewClass(pod)
@@ -412,25 +416,24 @@ func (g *genericScheduler) findNodesThatFit(pod *v1.Pod, nodes []*v1.Node) ([]*v
 				return
 			}
 			if fits {
-				filtered[atomic.AddInt32(&filteredLen, 1)-1] = g.cachedNodeInfoMap[nodeName].Node()
+				length := atomic.AddInt32(&filteredLen, 1)
+				if length > numNodesToFind {
+					cancel()
+					atomic.AddInt32(&filteredLen, -1)
+				} else {
+					filtered[length-1] = g.cachedNodeInfoMap[nodeName].Node()
+				}
 			} else {
 				predicateResultLock.Lock()
 				failedPredicateMap[nodeName] = failedPredicates
 				predicateResultLock.Unlock()
 			}
 		}
-		numNodesProcessed := int32(0)
-		for numNodesProcessed < allNodes {
-			numNodesToProcess := allNodes - numNodesProcessed
-			if numNodesToProcess > numNodesToFind {
-				numNodesToProcess = numNodesToFind
-			}
-			workqueue.Parallelize(16, int(numNodesToProcess), checkNode)
-			if filteredLen >= numNodesToFind {
-				break
-			}
-			numNodesProcessed += numNodesToProcess
-		}
+
+		// Stops searching for more nodes once the configured number of feasible nodes
+		// are found.
+		workqueue.ParallelizeUntil(ctx, 16, int(allNodes), checkNode)
+
 		filtered = filtered[:filteredLen]
 		if len(errs) > 0 {
 			return []*v1.Node{}, FailedPredicateMap{}, errors.CreateAggregateFromMessageCountMap(errs)

staging/src/k8s.io/client-go/util/workqueue/parallelizer.go

@@ -17,6 +17,7 @@ limitations under the License.
 package workqueue
 
 import (
+	"context"
 	"sync"
 
 	utilruntime "k8s.io/apimachinery/pkg/util/runtime"
@@ -24,9 +25,20 @@ import (
 
 type DoWorkPieceFunc func(piece int)
 
-// Parallelize is a very simple framework that allow for parallelizing
+// Parallelize is a very simple framework that allows for parallelizing
 // N independent pieces of work.
 func Parallelize(workers, pieces int, doWorkPiece DoWorkPieceFunc) {
+	ParallelizeUntil(nil, workers, pieces, doWorkPiece)
+}
+
+// ParallelizeUntil is a framework that allows for parallelizing N
+// independent pieces of work until done or the context is canceled.
+func ParallelizeUntil(ctx context.Context, workers, pieces int, doWorkPiece DoWorkPieceFunc) {
+	var stop <-chan struct{}
+	if ctx != nil {
+		stop = ctx.Done()
+	}
+
 	toProcess := make(chan int, pieces)
 	for i := 0; i < pieces; i++ {
 		toProcess <- i
@@ -44,7 +56,12 @@ func Parallelize(workers, pieces int, doWorkPiece DoWorkPieceFunc) {
 			defer utilruntime.HandleCrash()
 			defer wg.Done()
 			for piece := range toProcess {
-				doWorkPiece(piece)
+				select {
+				case <-stop:
+					return
+				default:
+					doWorkPiece(piece)
+				}
 			}
 		}()
 	}