Skip to content

feat: merge two metric servers #728

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
wants to merge 5 commits into
base: main
Choose a base branch
from
Open
Show file tree
Hide file tree
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
84 changes: 16 additions & 68 deletions cmd/bbr/main.go
Original file line number Diff line number Diff line change
Expand Up @@ -18,26 +18,23 @@ package main

import (
"flag"
"net"
"net/http"
"fmt"
"os"
"strconv"

"github.com/go-logr/logr"
"github.com/prometheus/client_golang/prometheus/promhttp"
uberzap "go.uber.org/zap"
"go.uber.org/zap/zapcore"
"google.golang.org/grpc"
healthPb "google.golang.org/grpc/health/grpc_health_v1"
"k8s.io/client-go/rest"
"k8s.io/component-base/metrics/legacyregistry"
ctrl "sigs.k8s.io/controller-runtime"
"sigs.k8s.io/controller-runtime/pkg/log/zap"
"sigs.k8s.io/controller-runtime/pkg/manager"
"sigs.k8s.io/controller-runtime/pkg/metrics/filters"
metricsserver "sigs.k8s.io/controller-runtime/pkg/metrics/server"

"sigs.k8s.io/gateway-api-inference-extension/internal/runnable"
"sigs.k8s.io/gateway-api-inference-extension/pkg/bbr/metrics"
runserver "sigs.k8s.io/gateway-api-inference-extension/pkg/bbr/server"
"sigs.k8s.io/gateway-api-inference-extension/pkg/epp/metrics"
"sigs.k8s.io/gateway-api-inference-extension/pkg/epp/util/logging"
)

Expand Down Expand Up @@ -85,7 +82,18 @@ func run() error {
return err
}

mgr, err := ctrl.NewManager(cfg, ctrl.Options{})
metrics.Register()

// Register metrics handler.
// Metrics endpoint is enabled in 'config/default/kustomization.yaml'. The Metrics options configure the server.
// More info:
// - https://pkg.go.dev/sigs.k8s.io/[email protected]/pkg/metrics/server
// - https://book.kubebuilder.io/reference/metrics.html
metricsServerOptions := metricsserver.Options{
BindAddress: fmt.Sprintf(":%d", *metricsPort),
FilterProvider: filters.WithAuthenticationAndAuthorization,
}
mgr, err := ctrl.NewManager(cfg, ctrl.Options{Metrics: metricsServerOptions})
if err != nil {
setupLog.Error(err, "Failed to create manager", "config", cfg)
return err
Expand All @@ -107,11 +115,6 @@ func run() error {
return err
}

// Register metrics handler.
if err := registerMetricsHandler(mgr, *metricsPort, cfg); err != nil {
return err
}

// Start the manager. This blocks until a signal is received.
setupLog.Info("Manager starting")
if err := mgr.Start(ctx); err != nil {
Expand Down Expand Up @@ -152,58 +155,3 @@ func initLogging(opts *zap.Options) {
logger := zap.New(zap.UseFlagOptions(opts), zap.RawZapOpts(uberzap.AddCaller()))
ctrl.SetLogger(logger)
}

const metricsEndpoint = "/metrics"

// registerMetricsHandler adds the metrics HTTP handler as a Runnable to the given manager.
func registerMetricsHandler(mgr manager.Manager, port int, cfg *rest.Config) error {
metrics.Register()

// Init HTTP server.
h, err := metricsHandlerWithAuthenticationAndAuthorization(cfg)
if err != nil {
return err
}

mux := http.NewServeMux()
mux.Handle(metricsEndpoint, h)

srv := &http.Server{
Addr: net.JoinHostPort("", strconv.Itoa(port)),
Handler: mux,
}

if err := mgr.Add(&manager.Server{
Name: "metrics",
Server: srv,
}); err != nil {
setupLog.Error(err, "Failed to register metrics HTTP handler")
return err
}
return nil
}

func metricsHandlerWithAuthenticationAndAuthorization(cfg *rest.Config) (http.Handler, error) {
h := promhttp.HandlerFor(
legacyregistry.DefaultGatherer,
promhttp.HandlerOpts{},
)
httpClient, err := rest.HTTPClientFor(cfg)
if err != nil {
setupLog.Error(err, "Failed to create http client for metrics auth")
return nil, err
}

filter, err := filters.WithAuthenticationAndAuthorization(cfg, httpClient)
if err != nil {
setupLog.Error(err, "Failed to create metrics filter for auth")
return nil, err
}
metricsLogger := ctrl.Log.WithName("metrics").WithValues("path", metricsEndpoint)
metricsAuthHandler, err := filter(metricsLogger, h)
if err != nil {
setupLog.Error(err, "Failed to create metrics auth handler")
return nil, err
}
return metricsAuthHandler, nil
}
107 changes: 26 additions & 81 deletions cmd/epp/main.go
Original file line number Diff line number Diff line change
Expand Up @@ -19,25 +19,22 @@ package main
import (
"flag"
"fmt"
"net"
"net/http"
"os"
"strconv"

"github.com/go-logr/logr"
"github.com/prometheus/client_golang/prometheus/promhttp"
"github.com/prometheus/client_golang/prometheus"
uberzap "go.uber.org/zap"
"go.uber.org/zap/zapcore"
"google.golang.org/grpc"
healthPb "google.golang.org/grpc/health/grpc_health_v1"
"k8s.io/apimachinery/pkg/types"
"k8s.io/client-go/rest"
"k8s.io/component-base/metrics/legacyregistry"
ctrl "sigs.k8s.io/controller-runtime"
"sigs.k8s.io/controller-runtime/pkg/log"
"sigs.k8s.io/controller-runtime/pkg/log/zap"
"sigs.k8s.io/controller-runtime/pkg/manager"
"sigs.k8s.io/controller-runtime/pkg/metrics/filters"
metricsserver "sigs.k8s.io/controller-runtime/pkg/metrics/server"

"sigs.k8s.io/gateway-api-inference-extension/internal/runnable"
backendmetrics "sigs.k8s.io/gateway-api-inference-extension/pkg/epp/backend/metrics"
"sigs.k8s.io/gateway-api-inference-extension/pkg/epp/datastore"
Expand All @@ -54,10 +51,6 @@ import (
"sigs.k8s.io/gateway-api-inference-extension/pkg/epp/util/logging"
)

const (
defaultMetricsEndpoint = "/metrics"
)

var (
grpcPort = flag.Int(
"grpcPort",
Expand Down Expand Up @@ -164,16 +157,6 @@ func run() error {
return err
}

poolNamespacedName := types.NamespacedName{
Name: *poolName,
Namespace: *poolNamespace,
}
mgr, err := runserver.NewDefaultManager(poolNamespacedName, cfg)
if err != nil {
setupLog.Error(err, "Failed to create controller manager")
return err
}

// Set up mapper for metric scraping.
mapping, err := backendmetrics.NewMetricMapping(
*totalQueuedRequestsMetric,
Expand All @@ -192,6 +175,29 @@ func run() error {

datastore := datastore.NewDatastore(ctx, pmf)

customCollectors := []prometheus.Collector{collectors.NewInferencePoolMetricsCollector(datastore)}
metrics.Register(customCollectors...)
metrics.RecordInferenceExtensionInfo()
// Register metrics handler.
// Metrics endpoint is enabled in 'config/default/kustomization.yaml'. The Metrics options configure the server.
// More info:
// - https://pkg.go.dev/sigs.k8s.io/[email protected]/pkg/metrics/server
// - https://book.kubebuilder.io/reference/metrics.html
metricsServerOptions := metricsserver.Options{
BindAddress: fmt.Sprintf(":%d", *metricsPort),
FilterProvider: filters.WithAuthenticationAndAuthorization,
}

poolNamespacedName := types.NamespacedName{
Name: *poolName,
Namespace: *poolNamespace,
}
mgr, err := runserver.NewDefaultManager(poolNamespacedName, cfg, metricsServerOptions)
if err != nil {
setupLog.Error(err, "Failed to create controller manager")
return err
}

scheduler := scheduling.NewScheduler(datastore)
if schedulerV2 == "true" {
queueScorerWeight := envutil.GetEnvInt("QUEUE_SCORE_WEIGHT", scorer.DefaultQueueScorerWeight, setupLog)
Expand Down Expand Up @@ -242,11 +248,6 @@ func run() error {
return err
}

// Register metrics handler.
if err := registerMetricsHandler(mgr, *metricsPort, cfg, datastore); err != nil {
return err
}

// Start the manager. This blocks until a signal is received.
setupLog.Info("Controller manager starting")
if err := mgr.Start(ctx); err != nil {
Expand Down Expand Up @@ -290,62 +291,6 @@ func registerHealthServer(mgr manager.Manager, logger logr.Logger, ds datastore.
return nil
}

// registerMetricsHandler adds the metrics HTTP handler as a Runnable to the given manager.
func registerMetricsHandler(mgr manager.Manager, port int, cfg *rest.Config, ds datastore.Datastore) error {
metrics.Register()
legacyregistry.CustomMustRegister(collectors.NewInferencePoolMetricsCollector(ds))

metrics.RecordInferenceExtensionInfo()

// Init HTTP server.
h, err := metricsHandlerWithAuthenticationAndAuthorization(cfg)
if err != nil {
return err
}

mux := http.NewServeMux()
mux.Handle(defaultMetricsEndpoint, h)

srv := &http.Server{
Addr: net.JoinHostPort("", strconv.Itoa(port)),
Handler: mux,
}

if err := mgr.Add(&manager.Server{
Name: "metrics",
Server: srv,
}); err != nil {
setupLog.Error(err, "Failed to register metrics HTTP handler")
return err
}
return nil
}

func metricsHandlerWithAuthenticationAndAuthorization(cfg *rest.Config) (http.Handler, error) {
h := promhttp.HandlerFor(
legacyregistry.DefaultGatherer,
promhttp.HandlerOpts{},
)
httpClient, err := rest.HTTPClientFor(cfg)
if err != nil {
setupLog.Error(err, "Failed to create http client for metrics auth")
return nil, err
}

filter, err := filters.WithAuthenticationAndAuthorization(cfg, httpClient)
if err != nil {
setupLog.Error(err, "Failed to create metrics filter for auth")
return nil, err
}
metricsLogger := ctrl.Log.WithName("metrics").WithValues("path", defaultMetricsEndpoint)
metricsAuthHandler, err := filter(metricsLogger, h)
if err != nil {
setupLog.Error(err, "Failed to create metrics auth handler")
return nil, err
}
return metricsAuthHandler, nil
}

func validateFlags() error {
if *poolName == "" {
return fmt.Errorf("required %q flag not set", "poolName")
Expand Down
4 changes: 2 additions & 2 deletions pkg/bbr/handlers/request_test.go
Original file line number Diff line number Diff line change
Expand Up @@ -26,8 +26,8 @@ import (
extProcPb "github.com/envoyproxy/go-control-plane/envoy/service/ext_proc/v3"
"github.com/google/go-cmp/cmp"
"google.golang.org/protobuf/testing/protocmp"
"k8s.io/component-base/metrics/legacyregistry"
metricsutils "k8s.io/component-base/metrics/testutil"
crmetrics "sigs.k8s.io/controller-runtime/pkg/metrics"
"sigs.k8s.io/gateway-api-inference-extension/pkg/bbr/metrics"
logutil "sigs.k8s.io/gateway-api-inference-extension/pkg/epp/util/logging"
)
Expand Down Expand Up @@ -204,7 +204,7 @@ func TestHandleRequestBody(t *testing.T) {
bbr_success_total{} 1
`

if err := metricsutils.GatherAndCompare(legacyregistry.DefaultGatherer, strings.NewReader(wantMetrics), "inference_model_request_total"); err != nil {
if err := metricsutils.GatherAndCompare(crmetrics.Registry, strings.NewReader(wantMetrics), "inference_model_request_total"); err != nil {
t.Error(err)
}
}
Expand Down
51 changes: 25 additions & 26 deletions pkg/bbr/metrics/metrics.go
Original file line number Diff line number Diff line change
Expand Up @@ -19,49 +19,48 @@ package metrics
import (
"sync"

"github.com/prometheus/client_golang/prometheus"
compbasemetrics "k8s.io/component-base/metrics"
"k8s.io/component-base/metrics/legacyregistry"
"sigs.k8s.io/controller-runtime/pkg/metrics"

metricsutil "sigs.k8s.io/gateway-api-inference-extension/pkg/epp/util/metrics"
)

const component = "bbr"

var (
successCounter = compbasemetrics.NewCounterVec(
&compbasemetrics.CounterOpts{
Subsystem: component,
Name: "success_total",
Help: "Count of successes pulling model name from body and injecting it in the request headers.",
StabilityLevel: compbasemetrics.ALPHA,
successCounter = prometheus.NewCounterVec(
prometheus.CounterOpts{
Subsystem: component,
Name: "success_total",
Help: metricsutil.HelpMsgWithStability("Count of successes pulling model name from body and injecting it in the request headers.", compbasemetrics.ALPHA),
},
[]string{},
)
modelNotInBodyCounter = compbasemetrics.NewCounterVec(
&compbasemetrics.CounterOpts{
Subsystem: component,
Name: "model_not_in_body_total",
Help: "Count of times the model was not present in the request body.",
StabilityLevel: compbasemetrics.ALPHA,
modelNotInBodyCounter = prometheus.NewCounterVec(
prometheus.CounterOpts{
Subsystem: component,
Name: "model_not_in_body_total",
Help: metricsutil.HelpMsgWithStability("Count of times the model was not present in the request body.", compbasemetrics.ALPHA),
},
[]string{},
)
modelNotParsedCounter = compbasemetrics.NewCounterVec(
&compbasemetrics.CounterOpts{
Subsystem: component,
Name: "model_not_parsed_total",
Help: "Count of times the model was in the request body but we could not parse it.",
StabilityLevel: compbasemetrics.ALPHA,
modelNotParsedCounter = prometheus.NewCounterVec(
prometheus.CounterOpts{
Subsystem: component,
Name: "model_not_parsed_total",
Help: metricsutil.HelpMsgWithStability("Count of times the model was in the request body but we could not parse it.", compbasemetrics.ALPHA),
},
[]string{},
)

// TODO: Uncomment and use this metrics once the core server implementation has handling to skip body parsing if header exists.
/*
modelAlreadyPresentInHeaderCounter = compbasemetrics.NewCounterVec(
&compbasemetrics.CounterOpts{
modelAlreadyPresentInHeaderCounter = prometheus.NewCounterVec(
prometheus.CounterOpts{
Subsystem: component,
Name: "model_already_present_in_header_total",
Help: "Count of times the model was already present in request headers.",
StabilityLevel: compbasemetrics.ALPHA,
},
[]string{},
)
Expand All @@ -73,10 +72,10 @@ var registerMetrics sync.Once
// Register all metrics.
func Register() {
registerMetrics.Do(func() {
legacyregistry.MustRegister(successCounter)
legacyregistry.MustRegister(modelNotInBodyCounter)
legacyregistry.MustRegister(modelNotParsedCounter)
// legacyregistry.MustRegister(modelAlreadyPresentInHeaderCounter)
metrics.Registry.MustRegister(successCounter)
metrics.Registry.MustRegister(modelNotInBodyCounter)
metrics.Registry.MustRegister(modelNotParsedCounter)
// metrics.Registry.MustRegister(modelAlreadyPresentInHeaderCounter)
})
}

Expand Down
Loading