Merge pull request #72787 from dashpole/cadvisor_prefix_whitelist

Only collect metrics for cgroups required by the summary API

cmd/kubelet/app/server.go

@@ -620,9 +620,34 @@ func run(s *options.KubeletServer, kubeDeps *kubelet.Dependencies, stopCh <-chan
 		kubeDeps.Auth = auth
 	}
 
+	var cgroupRoots []string
+
+	cgroupRoots = append(cgroupRoots, cm.NodeAllocatableRoot(s.CgroupRoot, s.CgroupDriver))
+	kubeletCgroup, err := cm.GetKubeletContainer(s.KubeletCgroups)
+	if err != nil {
+		return fmt.Errorf("failed to get the kubelet's cgroup: %v", err)
+	}
+	if kubeletCgroup != "" {
+		cgroupRoots = append(cgroupRoots, kubeletCgroup)
+	}
+
+	runtimeCgroup, err := cm.GetRuntimeContainer(s.ContainerRuntime, s.RuntimeCgroups)
+	if err != nil {
+		return fmt.Errorf("failed to get the container runtime's cgroup: %v", err)
+	}
+	if runtimeCgroup != "" {
+		// RuntimeCgroups is optional, so ignore if it isn't specified
+		cgroupRoots = append(cgroupRoots, runtimeCgroup)
+	}
+
+	if s.SystemCgroups != "" {
+		// SystemCgroups is optional, so ignore if it isn't specified
+		cgroupRoots = append(cgroupRoots, s.SystemCgroups)
+	}
+
 	if kubeDeps.CAdvisorInterface == nil {
 		imageFsInfoProvider := cadvisor.NewImageFsInfoProvider(s.ContainerRuntime, s.RemoteRuntimeEndpoint)
-		kubeDeps.CAdvisorInterface, err = cadvisor.New(imageFsInfoProvider, s.RootDirectory, cadvisor.UsingLegacyCadvisorStats(s.ContainerRuntime, s.RemoteRuntimeEndpoint))
+		kubeDeps.CAdvisorInterface, err = cadvisor.New(imageFsInfoProvider, s.RootDirectory, cgroupRoots, cadvisor.UsingLegacyCadvisorStats(s.ContainerRuntime, s.RemoteRuntimeEndpoint))
 		if err != nil {
 			return err
 		}

pkg/kubelet/cadvisor/cadvisor_linux.go

@@ -82,7 +82,7 @@ func init() {
 	}
 }
 
-func New(imageFsInfoProvider ImageFsInfoProvider, rootPath string, usingLegacyStats bool) (Interface, error) {
+func New(imageFsInfoProvider ImageFsInfoProvider, rootPath string, cgroupRoots []string, usingLegacyStats bool) (Interface, error) {
 	sysFs := sysfs.NewRealSysFs()
 
 	includedMetrics := cadvisormetrics.MetricSet{
@@ -98,10 +98,8 @@ func New(imageFsInfoProvider ImageFsInfoProvider, rootPath string, usingLegacySt
 		includedMetrics[cadvisormetrics.DiskUsageMetrics] = struct{}{}
 	}
 
-	// collect metrics for all cgroups
-	rawContainerCgroupPathPrefixWhiteList := []string{"/"}
 	// Create and start the cAdvisor container manager.
-	m, err := manager.New(memory.New(statsCacheDuration, nil), sysFs, maxHousekeepingInterval, allowDynamicHousekeeping, includedMetrics, http.DefaultClient, rawContainerCgroupPathPrefixWhiteList)
+	m, err := manager.New(memory.New(statsCacheDuration, nil), sysFs, maxHousekeepingInterval, allowDynamicHousekeeping, includedMetrics, http.DefaultClient, cgroupRoots)
 	if err != nil {
 		return nil, err
 	}

pkg/kubelet/cadvisor/cadvisor_unsupported.go

@@ -31,7 +31,7 @@ type cadvisorUnsupported struct {
 
 var _ Interface = new(cadvisorUnsupported)
 
-func New(imageFsInfoProvider ImageFsInfoProvider, rootPath string, usingLegacyStats bool) (Interface, error) {
+func New(imageFsInfoProvider ImageFsInfoProvider, rootPath string, cgroupsRoots []string, usingLegacyStats bool) (Interface, error) {
 	return &cadvisorUnsupported{}, nil
 }
 

pkg/kubelet/cadvisor/cadvisor_windows.go

@@ -33,7 +33,7 @@ type cadvisorClient struct {
 var _ Interface = new(cadvisorClient)
 
 // New creates a cAdvisor and exports its API on the specified port if port > 0.
-func New(imageFsInfoProvider ImageFsInfoProvider, rootPath string, usingLegacyStats bool) (Interface, error) {
+func New(imageFsInfoProvider ImageFsInfoProvider, rootPath string, cgroupRoots []string, usingLegacyStats bool) (Interface, error) {
 	client, err := winstats.NewPerfCounterClient()
 	return &cadvisorClient{
 		rootPath:       rootPath,

pkg/kubelet/cm/helpers_linux.go

@@ -236,3 +236,37 @@ func getCgroupProcs(dir string) ([]int, error) {
 func GetPodCgroupNameSuffix(podUID types.UID) string {
 	return podCgroupNamePrefix + string(podUID)
 }
+
+// NodeAllocatableRoot returns the literal cgroup path for the node allocatable cgroup
+func NodeAllocatableRoot(cgroupRoot, cgroupDriver string) string {
+	root := ParseCgroupfsToCgroupName(cgroupRoot)
+	nodeAllocatableRoot := NewCgroupName(root, defaultNodeAllocatableCgroupName)
+	if libcontainerCgroupManagerType(cgroupDriver) == libcontainerSystemd {
+		return nodeAllocatableRoot.ToSystemd()
+	}
+	return nodeAllocatableRoot.ToCgroupfs()
+}
+
+// GetKubeletContainer returns the cgroup the kubelet will use
+func GetKubeletContainer(kubeletCgroups string) (string, error) {
+	if kubeletCgroups == "" {
+		cont, err := getContainer(os.Getpid())
+		if err != nil {
+			return "", err
+		}
+		return cont, nil
+	}
+	return kubeletCgroups, nil
+}
+
+// GetRuntimeContainer returns the cgroup used by the container runtime
+func GetRuntimeContainer(containerRuntime, runtimeCgroups string) (string, error) {
+	if containerRuntime == "docker" {
+		cont, err := getContainerNameForProcess(dockerProcessName, dockerPidFile)
+		if err != nil {
+			return "", fmt.Errorf("failed to get container name for docker process: %v", err)
+		}
+		return cont, nil
+	}
+	return runtimeCgroups, nil
+}

pkg/kubelet/cm/helpers_unsupported.go

@@ -59,3 +59,18 @@ func getCgroupProcs(dir string) ([]int, error) {
 func GetPodCgroupNameSuffix(podUID types.UID) string {
 	return ""
 }
+
+// NodeAllocatableRoot returns the literal cgroup path for the node allocatable cgroup
+func NodeAllocatableRoot(cgroupRoot, cgroupDriver string) string {
+	return ""
+}
+
+// GetKubeletContainer returns the cgroup the kubelet will use
+func GetKubeletContainer(kubeletCgroups string) (string, error) {
+	return "", nil
+}
+
+// GetRuntimeContainer returns the cgroup used by the container runtime
+func GetRuntimeContainer(containerRuntime, runtimeCgroups string) (string, error) {
+	return "", nil
+}

test/e2e_node/environment/conformance.go

@@ -99,7 +99,7 @@ func containerRuntime() error {
 	}
 
 	// Setup cadvisor to check the container environment
-	c, err := cadvisor.New(cadvisor.NewImageFsInfoProvider("docker", ""), "/var/lib/kubelet", false)
+	c, err := cadvisor.New(cadvisor.NewImageFsInfoProvider("docker", ""), "/var/lib/kubelet", []string{"/"}, false)
 	if err != nil {
 		return printError("Container Runtime Check: %s Could not start cadvisor %v", failed, err)
 	}