Revendor CA against tip of k8s.io/kubernetes

cluster-autoscaler/cloudprovider/gce/gce_metrics.go

@@ -17,7 +17,8 @@ limitations under the License.
 package gce
 
 import (
-	"github.com/prometheus/client_golang/prometheus"
+	k8smetrics "k8s.io/component-base/metrics"
+	"k8s.io/component-base/metrics/legacyregistry"
 )
 
 const (
@@ -26,8 +27,8 @@ const (
 
 var (
 	/**** Metrics related to GCE API usage ****/
-	requestCounter = prometheus.NewCounterVec(
-		prometheus.CounterOpts{
+	requestCounter = k8smetrics.NewCounterVec(
+		&k8smetrics.CounterOpts{
 			Namespace: caNamespace,
 			Name:      "gce_request_count",
 			Help:      "Counter of GCE API requests for each verb and API resource.",
@@ -37,7 +38,7 @@ var (
 
 // RegisterMetrics registers all GCE metrics.
 func RegisterMetrics() {
-	prometheus.MustRegister(requestCounter)
+	legacyregistry.MustRegister(requestCounter)
 }
 
 // registerRequest registers request to GCE API.

cluster-autoscaler/cloudprovider/gce/templates_test.go

@@ -614,7 +614,7 @@ func makeResourceList2(cpu int64, memory int64, gpu int64, pods int64) (apiv1.Re
 
 func assertEqualResourceLists(t *testing.T, name string, expected, actual apiv1.ResourceList) {
 	t.Helper()
-	assert.True(t, quota.V1Equals(expected, actual),
+	assert.True(t, quota.Equals(expected, actual),
 		"%q unequal:\nExpected: %v\nActual:   %v", name, stringifyResourceList(expected), stringifyResourceList(actual))
 }
 

cluster-autoscaler/core/filter_out_schedulable.go

@@ -108,7 +108,7 @@ func filterOutSchedulableByPacking(unschedulableCandidates []*apiv1.Pod, nodes [
 	nodeNameToNodeInfo := schedulerutil.CreateNodeNameToInfoMap(nonExpendableScheduled, nodes)
 
 	sort.Slice(unschedulableCandidates, func(i, j int) bool {
-		return util.GetPodPriority(unschedulableCandidates[i]) > util.GetPodPriority(unschedulableCandidates[j])
+		return util.MoreImportantPod(unschedulableCandidates[i], unschedulableCandidates[j])
 	})
 
 	for _, pod := range unschedulableCandidates {

cluster-autoscaler/core/utils/taint_key_set.go

@@ -17,6 +17,7 @@ limitations under the License.
 package utils
 
 import (
+	v1 "k8s.io/api/core/v1"
 	schedulerapi "k8s.io/kubernetes/pkg/scheduler/api"
 )
 
@@ -26,14 +27,14 @@ type TaintKeySet map[string]bool
 var (
 	// NodeConditionTaints lists taint keys used as node conditions
 	NodeConditionTaints = TaintKeySet{
-		schedulerapi.TaintNodeNotReady:           true,
-		schedulerapi.TaintNodeUnreachable:        true,
-		schedulerapi.TaintNodeUnschedulable:      true,
-		schedulerapi.TaintNodeMemoryPressure:     true,
-		schedulerapi.TaintNodeDiskPressure:       true,
-		schedulerapi.TaintNodeNetworkUnavailable: true,
-		schedulerapi.TaintNodePIDPressure:        true,
-		schedulerapi.TaintExternalCloudProvider:  true,
-		schedulerapi.TaintNodeShutdown:           true,
+		v1.TaintNodeNotReady:                    true,
+		v1.TaintNodeUnreachable:                 true,
+		v1.TaintNodeUnschedulable:               true,
+		v1.TaintNodeMemoryPressure:              true,
+		v1.TaintNodeDiskPressure:                true,
+		v1.TaintNodeNetworkUnavailable:          true,
+		v1.TaintNodePIDPressure:                 true,
+		schedulerapi.TaintExternalCloudProvider: true,
+		schedulerapi.TaintNodeShutdown:          true,
 	}
 )

cluster-autoscaler/go.mod-extra

@@ -3,7 +3,6 @@ go 1.12
 
 require (
   github.com/rancher/go-rancher v0.1.0
-  github.com/google/go-querystring v1.0.0
   github.com/aws/aws-sdk-go v1.23.18
 )
 

cluster-autoscaler/main.go

@@ -29,7 +29,6 @@ import (
 	"syscall"
 	"time"
 
-	"github.com/prometheus/client_golang/prometheus"
 	"github.com/spf13/pflag"
 
 	metav1 "k8s.io/apimachinery/pkg/apis/meta/v1"
@@ -53,6 +52,7 @@ import (
 	"k8s.io/client-go/tools/leaderelection/resourcelock"
 	kube_flag "k8s.io/component-base/cli/flag"
 	componentbaseconfig "k8s.io/component-base/config"
+	"k8s.io/component-base/metrics/legacyregistry"
 	"k8s.io/klog"
 	"k8s.io/kubernetes/pkg/client/leaderelectionconfig"
 )
@@ -368,7 +368,7 @@ func main() {
 	klog.V(1).Infof("Cluster Autoscaler %s", version.ClusterAutoscalerVersion)
 
 	go func() {
-		http.Handle("/metrics", prometheus.Handler())
+		http.Handle("/metrics", legacyregistry.Handler())
 		http.Handle("/health-check", healthCheck)
 		err := http.ListenAndServe(*address, nil)
 		klog.Fatalf("Failed to start metrics: %v", err)

cluster-autoscaler/metrics/metrics.go

@@ -21,9 +21,10 @@ import (
 
 	"k8s.io/autoscaler/cluster-autoscaler/utils/errors"
 	"k8s.io/autoscaler/cluster-autoscaler/utils/gpu"
-	_ "k8s.io/kubernetes/pkg/client/metrics/prometheus" // for client-go metrics registration
+	_ "k8s.io/component-base/metrics/prometheus/restclient" // for client-go metrics registration
 
-	"github.com/prometheus/client_golang/prometheus"
+	k8smetrics "k8s.io/component-base/metrics"
+	"k8s.io/component-base/metrics/legacyregistry"
 	"k8s.io/klog"
 )
 
@@ -92,58 +93,58 @@ const (
 
 var (
 	/**** Metrics related to cluster state ****/
-	clusterSafeToAutoscale = prometheus.NewGauge(
-		prometheus.GaugeOpts{
+	clusterSafeToAutoscale = k8smetrics.NewGauge(
+		&k8smetrics.GaugeOpts{
 			Namespace: caNamespace,
 			Name:      "cluster_safe_to_autoscale",
 			Help:      "Whether or not cluster is healthy enough for autoscaling. 1 if it is, 0 otherwise.",
 		},
 	)
 
-	nodesCount = prometheus.NewGaugeVec(
-		prometheus.GaugeOpts{
+	nodesCount = k8smetrics.NewGaugeVec(
+		&k8smetrics.GaugeOpts{
 			Namespace: caNamespace,
 			Name:      "nodes_count",
 			Help:      "Number of nodes in cluster.",
 		}, []string{"state"},
 	)
 
-	nodeGroupsCount = prometheus.NewGaugeVec(
-		prometheus.GaugeOpts{
+	nodeGroupsCount = k8smetrics.NewGaugeVec(
+		&k8smetrics.GaugeOpts{
 			Namespace: caNamespace,
 			Name:      "node_groups_count",
 			Help:      "Number of node groups managed by CA.",
 		}, []string{"node_group_type"},
 	)
 
-	unschedulablePodsCount = prometheus.NewGauge(
-		prometheus.GaugeOpts{
+	unschedulablePodsCount = k8smetrics.NewGauge(
+		&k8smetrics.GaugeOpts{
 			Namespace: caNamespace,
 			Name:      "unschedulable_pods_count",
 			Help:      "Number of unschedulable pods in the cluster.",
 		},
 	)
 
 	/**** Metrics related to autoscaler execution ****/
-	lastActivity = prometheus.NewGaugeVec(
-		prometheus.GaugeOpts{
+	lastActivity = k8smetrics.NewGaugeVec(
+		&k8smetrics.GaugeOpts{
 			Namespace: caNamespace,
 			Name:      "last_activity",
 			Help:      "Last time certain part of CA logic executed.",
 		}, []string{"activity"},
 	)
 
-	functionDuration = prometheus.NewHistogramVec(
-		prometheus.HistogramOpts{
+	functionDuration = k8smetrics.NewHistogramVec(
+		&k8smetrics.HistogramOpts{
 			Namespace: caNamespace,
 			Name:      "function_duration_seconds",
 			Help:      "Time taken by various parts of CA main loop.",
 			Buckets:   []float64{0.01, 0.05, 0.1, 0.5, 1.0, 2.5, 5.0, 7.5, 10.0, 12.5, 15.0, 17.5, 20.0, 22.5, 25.0, 27.5, 30.0, 50.0, 75.0, 100.0, 1000.0},
 		}, []string{"function"},
 	)
 
-	functionDurationSummary = prometheus.NewSummaryVec(
-		prometheus.SummaryOpts{
+	functionDurationSummary = k8smetrics.NewSummaryVec(
+		&k8smetrics.SummaryOpts{
 			Namespace: caNamespace,
 			Name:      "function_duration_quantile_seconds",
 			Help:      "Quantiles of time taken by various parts of CA main loop.",
@@ -152,97 +153,97 @@ var (
 	)
 
 	/**** Metrics related to autoscaler operations ****/
-	errorsCount = prometheus.NewCounterVec(
-		prometheus.CounterOpts{
+	errorsCount = k8smetrics.NewCounterVec(
+		&k8smetrics.CounterOpts{
 			Namespace: caNamespace,
 			Name:      "errors_total",
 			Help:      "The number of CA loops failed due to an error.",
 		}, []string{"type"},
 	)
 
-	scaleUpCount = prometheus.NewCounter(
-		prometheus.CounterOpts{
+	scaleUpCount = k8smetrics.NewCounter(
+		&k8smetrics.CounterOpts{
 			Namespace: caNamespace,
 			Name:      "scaled_up_nodes_total",
 			Help:      "Number of nodes added by CA.",
 		},
 	)
 
-	gpuScaleUpCount = prometheus.NewCounterVec(
-		prometheus.CounterOpts{
+	gpuScaleUpCount = k8smetrics.NewCounterVec(
+		&k8smetrics.CounterOpts{
 			Namespace: caNamespace,
 			Name:      "scaled_up_gpu_nodes_total",
 			Help:      "Number of GPU nodes added by CA, by GPU name.",
 		}, []string{"gpu_name"},
 	)
 
-	failedScaleUpCount = prometheus.NewCounterVec(
-		prometheus.CounterOpts{
+	failedScaleUpCount = k8smetrics.NewCounterVec(
+		&k8smetrics.CounterOpts{
 			Namespace: caNamespace,
 			Name:      "failed_scale_ups_total",
 			Help:      "Number of times scale-up operation has failed.",
 		}, []string{"reason"},
 	)
 
-	scaleDownCount = prometheus.NewCounterVec(
-		prometheus.CounterOpts{
+	scaleDownCount = k8smetrics.NewCounterVec(
+		&k8smetrics.CounterOpts{
 			Namespace: caNamespace,
 			Name:      "scaled_down_nodes_total",
 			Help:      "Number of nodes removed by CA.",
 		}, []string{"reason"},
 	)
 
-	gpuScaleDownCount = prometheus.NewCounterVec(
-		prometheus.CounterOpts{
+	gpuScaleDownCount = k8smetrics.NewCounterVec(
+		&k8smetrics.CounterOpts{
 			Namespace: caNamespace,
 			Name:      "scaled_down_gpu_nodes_total",
 			Help:      "Number of GPU nodes removed by CA, by reason and GPU name.",
 		}, []string{"reason", "gpu_name"},
 	)
 
-	evictionsCount = prometheus.NewCounter(
-		prometheus.CounterOpts{
+	evictionsCount = k8smetrics.NewCounter(
+		&k8smetrics.CounterOpts{
 			Namespace: caNamespace,
 			Name:      "evicted_pods_total",
 			Help:      "Number of pods evicted by CA",
 		},
 	)
 
-	unneededNodesCount = prometheus.NewGauge(
-		prometheus.GaugeOpts{
+	unneededNodesCount = k8smetrics.NewGauge(
+		&k8smetrics.GaugeOpts{
 			Namespace: caNamespace,
 			Name:      "unneeded_nodes_count",
 			Help:      "Number of nodes currently considered unneeded by CA.",
 		},
 	)
 
-	scaleDownInCooldown = prometheus.NewGauge(
-		prometheus.GaugeOpts{
+	scaleDownInCooldown = k8smetrics.NewGauge(
+		&k8smetrics.GaugeOpts{
 			Namespace: caNamespace,
 			Name:      "scale_down_in_cooldown",
 			Help:      "Whether or not the scale down is in cooldown. 1 if its, 0 otherwise.",
 		},
 	)
 
 	/**** Metrics related to NodeAutoprovisioning ****/
-	napEnabled = prometheus.NewGauge(
-		prometheus.GaugeOpts{
+	napEnabled = k8smetrics.NewGauge(
+		&k8smetrics.GaugeOpts{
 			Namespace: caNamespace,
 			Name:      "nap_enabled",
 			Help:      "Whether or not Node Autoprovisioning is enabled. 1 if it is, 0 otherwise.",
 		},
 	)
 
-	nodeGroupCreationCount = prometheus.NewCounter(
-		prometheus.CounterOpts{
+	nodeGroupCreationCount = k8smetrics.NewCounter(
+		&k8smetrics.CounterOpts{
 			Namespace: caNamespace,
 			Name:      "created_node_groups_total",
 			Help:      "Number of node groups created by Node Autoprovisioning.",
 		},
 	)
 
-	nodeGroupDeletionCount = prometheus.NewCounter(
-		prometheus.CounterOpts{
+	nodeGroupDeletionCount = k8smetrics.NewCounter(
+		&k8smetrics.CounterOpts{
 			Namespace: caNamespace,
 			Name:      "deleted_node_groups_total",
 			Help:      "Number of node groups deleted by Node Autoprovisioning.",
@@ -252,25 +253,25 @@ var (
 
 // RegisterAll registers all metrics.
 func RegisterAll() {
-	prometheus.MustRegister(clusterSafeToAutoscale)
-	prometheus.MustRegister(nodesCount)
-	prometheus.MustRegister(nodeGroupsCount)
-	prometheus.MustRegister(unschedulablePodsCount)
-	prometheus.MustRegister(lastActivity)
-	prometheus.MustRegister(functionDuration)
-	prometheus.MustRegister(functionDurationSummary)
-	prometheus.MustRegister(errorsCount)
-	prometheus.MustRegister(scaleUpCount)
-	prometheus.MustRegister(gpuScaleUpCount)
-	prometheus.MustRegister(failedScaleUpCount)
-	prometheus.MustRegister(scaleDownCount)
-	prometheus.MustRegister(gpuScaleDownCount)
-	prometheus.MustRegister(evictionsCount)
-	prometheus.MustRegister(unneededNodesCount)
-	prometheus.MustRegister(scaleDownInCooldown)
-	prometheus.MustRegister(napEnabled)
-	prometheus.MustRegister(nodeGroupCreationCount)
-	prometheus.MustRegister(nodeGroupDeletionCount)
+	legacyregistry.MustRegister(clusterSafeToAutoscale)
+	legacyregistry.MustRegister(nodesCount)
+	legacyregistry.MustRegister(nodeGroupsCount)
+	legacyregistry.MustRegister(unschedulablePodsCount)
+	legacyregistry.MustRegister(lastActivity)
+	legacyregistry.MustRegister(functionDuration)
+	legacyregistry.MustRegister(functionDurationSummary)
+	legacyregistry.MustRegister(errorsCount)
+	legacyregistry.MustRegister(scaleUpCount)
+	legacyregistry.MustRegister(gpuScaleUpCount)
+	legacyregistry.MustRegister(failedScaleUpCount)
+	legacyregistry.MustRegister(scaleDownCount)
+	legacyregistry.MustRegister(gpuScaleDownCount)
+	legacyregistry.MustRegister(evictionsCount)
+	legacyregistry.MustRegister(unneededNodesCount)
+	legacyregistry.MustRegister(scaleDownInCooldown)
+	legacyregistry.MustRegister(napEnabled)
+	legacyregistry.MustRegister(nodeGroupCreationCount)
+	legacyregistry.MustRegister(nodeGroupDeletionCount)
 }
 
 // UpdateDurationFromStart records the duration of the step identified by the

cluster-autoscaler/simulator/cluster.go

@@ -233,7 +233,7 @@ func findPlaceFor(removedNode string, pods []*apiv1.Pod, nodes []*apiv1.Node, no
 
 	loggingQuota := glogx.PodsLoggingQuota()
 
-	tryNodeForPod := func(nodename string, pod *apiv1.Pod, predicateMeta predicates.PredicateMetadata) bool {
+	tryNodeForPod := func(nodename string, pod *apiv1.Pod, predicateMeta predicates.Metadata) bool {
 		nodeInfo, found := newNodeInfos[nodename]
 		if found {
 			if nodeInfo.Node() == nil {