fix: Support scraping pods metrics that is still in scheduling status and has no assigned node

examples/daemonsetsharding/deployment-no-node-pods.yaml

@@ -0,0 +1,59 @@
+apiVersion: apps/v1
+kind: Deployment
+metadata:
+  labels:
+    app.kubernetes.io/component: exporter
+    app.kubernetes.io/name: kube-state-metrics
+    app.kubernetes.io/version: 2.10.0
+  name: kube-state-metrics-pods
+  namespace: kube-system
+spec:
+  replicas: 1
+  selector:
+    matchLabels:
+      app.kubernetes.io/name: kube-state-metrics
+  template:
+    metadata:
+      labels:
+        app.kubernetes.io/component: exporter
+        app.kubernetes.io/name: kube-state-metrics
+        app.kubernetes.io/version: 2.10.0
+    spec:
+      automountServiceAccountToken: true
+      containers:
+      - args:
+        - --resources=pods
+        - --node=""
+        - --enable-no-node-scrape
+        image: registry.k8s.io/kube-state-metrics/kube-state-metrics:v2.10.0
+        livenessProbe:
+          httpGet:
+            path: /healthz
+            port: 8080
+          initialDelaySeconds: 5
+          timeoutSeconds: 5
+        name: kube-state-metrics
+        ports:
+        - containerPort: 8080
+          name: http-metrics
+        - containerPort: 8081
+          name: telemetry
+        readinessProbe:
+          httpGet:
+            path: /
+            port: 8081
+          initialDelaySeconds: 5
+          timeoutSeconds: 5
+        securityContext:
+          allowPrivilegeEscalation: false
+          capabilities:
+            drop:
+            - ALL
+          readOnlyRootFilesystem: true
+          runAsNonRoot: true
+          runAsUser: 65534
+          seccompProfile:
+            type: RuntimeDefault
+      nodeSelector:
+        kubernetes.io/os: linux
+      serviceAccountName: kube-state-metrics

pkg/app/server.go

@@ -204,7 +204,7 @@ func RunKubeStateMetrics(ctx context.Context, opts *options.Options) error {
 
 	namespaces := opts.Namespaces.GetNamespaces()
 	nsFieldSelector := namespaces.GetExcludeNSFieldSelector(opts.NamespacesDenylist)
-	nodeFieldSelector := opts.Node.GetNodeFieldSelector()
+	nodeFieldSelector := opts.Node.GetNodeFieldSelector(opts.NoNodeScrape)
 	merged, err := storeBuilder.MergeFieldSelectors([]string{nsFieldSelector, nodeFieldSelector})
 	if err != nil {
 		return err

pkg/options/options.go

@@ -46,6 +46,7 @@ type Options struct {
 	Namespaces               NamespaceList   `yaml:"namespaces"`
 	NamespacesDenylist       NamespaceList   `yaml:"namespaces_denylist"`
 	Node                     NodeType        `yaml:"node"`
+	NoNodeScrape             bool            `yaml:"no_node_scrape"`
 	Pod                      string          `yaml:"pod"`
 	Port                     int             `yaml:"port"`
 	Resources                ResourceSet     `yaml:"resources"`
@@ -120,6 +121,7 @@ func (o *Options) AddFlags(cmd *cobra.Command) {
 
 	o.cmd.Flags().BoolVar(&o.CustomResourcesOnly, "custom-resource-state-only", false, "Only provide Custom Resource State metrics (experimental)")
 	o.cmd.Flags().BoolVar(&o.EnableGZIPEncoding, "enable-gzip-encoding", false, "Gzip responses when requested by clients via 'Accept-Encoding: gzip' header.")
+	o.cmd.Flags().BoolVar(&o.NoNodeScrape, "enable-no-node-scrape", false, "This configuration is used in conjunction with node configuration. When this configuration is true, node configuration is empty and the metric of no scheduled pods is scraped. This is experimental.")
 	o.cmd.Flags().BoolVarP(&o.Help, "help", "h", false, "Print Help text")
 	o.cmd.Flags().BoolVarP(&o.UseAPIServerCache, "use-apiserver-cache", "", false, "Sets resourceVersion=0 for ListWatch requests, using cached resources from the apiserver instead of an etcd quorum read.")
 	o.cmd.Flags().Int32Var(&o.Shard, "shard", int32(0), "The instances shard nominal (zero indexed) within the total number of shards. (default 0)")

pkg/options/types.go

@@ -108,10 +108,13 @@ func (r *ResourceSet) Type() string {
 type NodeType string
 
 // GetNodeFieldSelector returns a nodename field selector.
-func (n *NodeType) GetNodeFieldSelector() string {
+func (n *NodeType) GetNodeFieldSelector(b bool) string {
 	if string(*n) != "" {
 		return fields.OneTermEqualSelector("spec.nodeName", string(*n)).String()
 	}
+	if b {
+		return fields.OneTermEqualSelector("spec.nodeName", "").String()
+	}
 	return EmptyFieldSelector()
 }
 

pkg/options/types_test.go

@@ -175,7 +175,25 @@ func TestNodeFieldSelector(t *testing.T) {
 
 	for _, test := range tests {
 		node := test.Node
-		actual := node.GetNodeFieldSelector()
+		actual := node.GetNodeFieldSelector(false)
+		if !reflect.DeepEqual(actual, test.Wanted) {
+			t.Errorf("Test error for Desc: %s. Want: %+v. Got: %+v.", test.Desc, test.Wanted, actual)
+		}
+	}
+	tests1 := []struct {
+		Desc   string
+		Node   NodeType
+		Wanted string
+	}{
+		{
+			Desc:   "empty node name",
+			Node:   "",
+			Wanted: "spec.nodeName=",
+		},
+	}
+	for _, test := range tests1 {
+		node := test.Node
+		actual := node.GetNodeFieldSelector(true)
 		if !reflect.DeepEqual(actual, test.Wanted) {
 			t.Errorf("Test error for Desc: %s. Want: %+v. Got: %+v.", test.Desc, test.Wanted, actual)
 		}
@@ -238,7 +256,7 @@ func TestMergeFieldSelectors(t *testing.T) {
 		ns := test.Namespaces
 		deniedNS := test.DeniedNamespaces
 		selector1 := ns.GetExcludeNSFieldSelector(deniedNS)
-		selector2 := test.Node.GetNodeFieldSelector()
+		selector2 := test.Node.GetNodeFieldSelector(false)
 		actual, err := MergeFieldSelectors([]string{selector1, selector2})
 		if err != nil {
 			t.Errorf("Test error for Desc: %s. Can't merge field selector %v.", test.Desc, err)