Collect metrics

Makefile

@@ -16,11 +16,13 @@
 
 # VERSION is based on a date stamp plus the last commit
 VERSION?=v$(shell date +%Y%m%d)-$(shell git describe --tags --match "v*")
+BRANCH?=$(shell git branch --show-current)
+SHA1?=$(shell git rev-parse HEAD)
 BUILD=$(shell date +%FT%T%z)
-LDFLAG_LOCATION=sigs.k8s.io/descheduler/cmd/descheduler/app
+LDFLAG_LOCATION=sigs.k8s.io/descheduler/pkg/version
 ARCHS = amd64 arm64
 
-LDFLAGS=-ldflags "-X ${LDFLAG_LOCATION}.version=${VERSION} -X ${LDFLAG_LOCATION}.buildDate=${BUILD}"
+LDFLAGS=-ldflags "-X ${LDFLAG_LOCATION}.version=${VERSION} -X ${LDFLAG_LOCATION}.buildDate=${BUILD} -X ${LDFLAG_LOCATION}.gitbranch=${BRANCH} -X ${LDFLAG_LOCATION}.gitsha1=${SHA1}"
 
 GOLANGCI_VERSION := v1.30.0
 HAS_GOLANGCI := $(shell ls _output/bin/golangci-lint)

README.md

@@ -45,6 +45,7 @@ Table of Contents
      * [Priority filtering](#priority-filtering)
   * [Pod Evictions](#pod-evictions)
      * [Pod Disruption Budget (PDB)](#pod-disruption-budget-pdb)
+  * [Metrics](#metrics)
   * [Compatibility Matrix](#compatibility-matrix)
   * [Getting Involved and Contributing](#getting-involved-and-contributing)
      * [Communicating With Contributors](#communicating-with-contributors)
@@ -348,9 +349,9 @@ strategies:
 
 ### RemovePodsHavingTooManyRestarts
 
-This strategy makes sure that pods having too many restarts are removed from nodes. For example a pod with EBS/PD that 
-can't get the volume/disk attached to the instance, then the pod should be re-scheduled to other nodes. Its parameters 
-include `podRestartThreshold`, which is the number of restarts at which a pod should be evicted, and `includingInitContainers`, 
+This strategy makes sure that pods having too many restarts are removed from nodes. For example a pod with EBS/PD that
+can't get the volume/disk attached to the instance, then the pod should be re-scheduled to other nodes. Its parameters
+include `podRestartThreshold`, which is the number of restarts at which a pod should be evicted, and `includingInitContainers`,
 which determines whether init container restarts should be factored into that calculation.
 
 **Parameters:**
@@ -521,6 +522,16 @@ Setting `--v=4` or greater on the Descheduler will log all reasons why any pod i
 Pods subject to a Pod Disruption Budget(PDB) are not evicted if descheduling violates its PDB. The pods
 are evicted by using the eviction subresource to handle PDB.
 
+## Metrics
+
+| name	| type	| description |
+|-------|-------|----------------|
+| build_info |	gauge |	constant 1 |
+| pods_evicted | CounterVec | total number of pods evicted |
+
+The metrics are served through https://localhost:10258/metrics by default.
+The address and port can be changed by setting `--binding-address` and `--secure-port` flags.
+
 ## Compatibility Matrix
 The below compatibility matrix shows the k8s client package(client-go, apimachinery, etc) versions that descheduler
 is compiled with. At this time descheduler does not have a hard dependency to a specific k8s release. However a

cmd/descheduler/app/options/options.go

@@ -19,20 +19,29 @@ package options
 
 import (
 	"github.com/spf13/pflag"
+
 	utilerrors "k8s.io/apimachinery/pkg/util/errors"
+	apiserveroptions "k8s.io/apiserver/pkg/server/options"
 	clientset "k8s.io/client-go/kubernetes"
-
 	"k8s.io/component-base/logs"
+
 	"sigs.k8s.io/descheduler/pkg/apis/componentconfig"
 	"sigs.k8s.io/descheduler/pkg/apis/componentconfig/v1alpha1"
 	deschedulerscheme "sigs.k8s.io/descheduler/pkg/descheduler/scheme"
 )
 
+const (
+	DefaultDeschedulerPort = 10258
+)
+
 // DeschedulerServer configuration
 type DeschedulerServer struct {
 	componentconfig.DeschedulerConfiguration
-	Client clientset.Interface
-	Logs   *logs.Options
+
+	Client         clientset.Interface
+	Logs           *logs.Options
+	SecureServing  *apiserveroptions.SecureServingOptionsWithLoopback
+	DisableMetrics bool
 }
 
 // NewDeschedulerServer creates a new DeschedulerServer with default parameters
@@ -41,9 +50,14 @@ func NewDeschedulerServer() (*DeschedulerServer, error) {
 	if err != nil {
 		return nil, err
 	}
+
+	secureServing := apiserveroptions.NewSecureServingOptions().WithLoopback()
+	secureServing.BindPort = DefaultDeschedulerPort
+
 	return &DeschedulerServer{
 		DeschedulerConfiguration: *cfg,
 		Logs:                     logs.NewOptions(),
+		SecureServing:            secureServing,
 	}, nil
 }
 
@@ -77,4 +91,7 @@ func (rs *DeschedulerServer) AddFlags(fs *pflag.FlagSet) {
 	fs.IntVar(&rs.MaxNoOfPodsToEvictPerNode, "max-pods-to-evict-per-node", rs.MaxNoOfPodsToEvictPerNode, "DEPRECATED: limits the maximum number of pods to be evicted per node by descheduler")
 	// evict-local-storage-pods allows eviction of pods that are using local storage. This is false by default.
 	fs.BoolVar(&rs.EvictLocalStoragePods, "evict-local-storage-pods", rs.EvictLocalStoragePods, "DEPRECATED: enables evicting pods using local storage by descheduler")
+	fs.BoolVar(&rs.DisableMetrics, "disable-metrics", rs.DisableMetrics, "Disables metrics. The metrics are by default served through https://localhost:10258/metrics. Secure address, resp. port can be changed through --bind-address, resp. --secure-port flags.")
+
+	rs.SecureServing.AddFlags(fs)
 }

cmd/descheduler/app/server.go

@@ -18,6 +18,7 @@ limitations under the License.
 package app
 
 import (
+	"context"
 	"flag"
 	"io"
 
@@ -26,7 +27,11 @@ import (
 
 	"github.com/spf13/cobra"
 
+	apiserver "k8s.io/apiserver/pkg/server"
+	"k8s.io/apiserver/pkg/server/mux"
+	restclient "k8s.io/client-go/rest"
 	aflag "k8s.io/component-base/cli/flag"
+	"k8s.io/component-base/metrics/legacyregistry"
 	"k8s.io/klog/v2"
 )
 
@@ -46,9 +51,30 @@ func NewDeschedulerCommand(out io.Writer) *cobra.Command {
 			s.Logs.LogFormat = s.Logging.Format
 			s.Logs.Apply()
 
+			// LoopbackClientConfig is a config for a privileged loopback connection
+			var LoopbackClientConfig *restclient.Config
+			var SecureServing *apiserver.SecureServingInfo
+			if err := s.SecureServing.ApplyTo(&SecureServing, &LoopbackClientConfig); err != nil {
+				klog.ErrorS(err, "failed to apply secure server configuration")
+				return
+			}
+
 			if err := s.Validate(); err != nil {
 				klog.ErrorS(err, "failed to validate server configuration")
+				return
+			}
+
+			if !s.DisableMetrics {
+				ctx := context.TODO()
+				pathRecorderMux := mux.NewPathRecorderMux("descheduler")
+				pathRecorderMux.Handle("/metrics", legacyregistry.HandlerWithReset())
+
+				if _, err := SecureServing.Serve(pathRecorderMux, 0, ctx.Done()); err != nil {
+					klog.Fatalf("failed to start secure server: %v", err)
+					return
+				}
 			}
+
 			err := Run(s)
 			if err != nil {
 				klog.ErrorS(err, "descheduler server")

cmd/descheduler/app/version.go

@@ -18,77 +18,19 @@ package app
 
 import (
 	"fmt"
-	"regexp"
-	"runtime"
-	"strings"
 
 	"github.com/spf13/cobra"
+	"sigs.k8s.io/descheduler/pkg/version"
 )
 
-var (
-	// version is a constant representing the version tag that
-	// generated this build. It should be set during build via -ldflags.
-	version string
-	// buildDate in ISO8601 format, output of $(date -u +'%Y-%m-%dT%H:%M:%SZ')
-	//It should be set during build via -ldflags.
-	buildDate string
-)
-
-// Info holds the information related to descheduler app version.
-type Info struct {
-	Major      string `json:"major"`
-	Minor      string `json:"minor"`
-	GitVersion string `json:"gitVersion"`
-	BuildDate  string `json:"buildDate"`
-	GoVersion  string `json:"goVersion"`
-	Compiler   string `json:"compiler"`
-	Platform   string `json:"platform"`
-}
-
-// Get returns the overall codebase version. It's for detecting
-// what code a binary was built from.
-func Get() Info {
-	majorVersion, minorVersion := splitVersion(version)
-	return Info{
-		Major:      majorVersion,
-		Minor:      minorVersion,
-		GitVersion: version,
-		BuildDate:  buildDate,
-		GoVersion:  runtime.Version(),
-		Compiler:   runtime.Compiler,
-		Platform:   fmt.Sprintf("%s/%s", runtime.GOOS, runtime.GOARCH),
-	}
-}
-
 func NewVersionCommand() *cobra.Command {
 	var versionCmd = &cobra.Command{
 		Use:   "version",
 		Short: "Version of descheduler",
 		Long:  `Prints the version of descheduler.`,
 		Run: func(cmd *cobra.Command, args []string) {
-			fmt.Printf("Descheduler version %+v\n", Get())
+			fmt.Printf("Descheduler version %+v\n", version.Get())
 		},
 	}
 	return versionCmd
 }
-
-// splitVersion splits the git version to generate major and minor versions needed.
-func splitVersion(version string) (string, string) {
-	if version == "" {
-		return "", ""
-	}
-
-	// Version from an automated container build environment for a tag. For example v20200521-v0.18.0.
-	m1, _ := regexp.MatchString(`^v\d{8}-v\d+\.\d+\.\d+$`, version)
-
-	// Version from an automated container build environment(not a tag) or a local build. For example v20201009-v0.18.0-46-g939c1c0.
-	m2, _ := regexp.MatchString(`^v\d{8}-v\d+\.\d+\.\d+-\w+-\w+$`, version)
-
-	if m1 || m2 {
-		semVer := strings.Split(version, "-")[1]
-		return strings.Trim(strings.Split(semVer, ".")[0], "v"), strings.Split(semVer, ".")[1] + "+"
-	}
-
-	// Something went wrong
-	return "", ""
-}