Add interceptor

Signed-off-by: Jorge Turrado <[email protected]>

Author: JorTurFer

interceptor/config/serving.go

@@ -24,11 +24,16 @@ type Serving struct {
 	// ConfigMapCacheRsyncPeriod is the time interval
 	// for the configmap informer to rsync the local cache.
 	ConfigMapCacheRsyncPeriod time.Duration `envconfig:"KEDA_HTTP_SCALER_CONFIG_MAP_INFORMER_RSYNC_PERIOD" default:"60m"`
-	// The interceptor has an internal process that periodically fetches the state
+	// DEPRECATED: The interceptor has an internal process that periodically fetches the state
 	// of deployment that is running the servers it forwards to.
 	//
 	// This is the interval (in milliseconds) representing how often to do a fetch
 	DeploymentCachePollIntervalMS int `envconfig:"KEDA_HTTP_DEPLOYMENT_CACHE_POLLING_INTERVAL_MS" default:"250"`
+	//The interceptor has an internal process that periodically fetches the state
+	// of endpoints that is running the servers it forwards to.
+	//
+	// This is the interval (in milliseconds) representing how often to do a fetch
+	EndpointsCachePollIntervalMS int `envconfig:"KEDA_HTTP_ENDPOINTS_CACHE_POLLING_INTERVAL_MS" default:"250"`
 }
 
 // Parse parses standard configs using envconfig and returns a pointer to the

interceptor/config/timeouts.go

@@ -16,9 +16,9 @@ type Timeouts struct {
 	// ResponseHeaderTimeout is how long to wait between when the HTTP request
 	// is sent to the backing app and when response headers need to arrive
 	ResponseHeader time.Duration `envconfig:"KEDA_RESPONSE_HEADER_TIMEOUT" default:"500ms"`
-	// DeploymentReplicas is how long to wait for the backing deployment
+	// WorkloadReplicas is how long to wait for the backing workload
 	// to have 1 or more replicas before connecting and sending the HTTP request.
-	DeploymentReplicas time.Duration `envconfig:"KEDA_CONDITION_WAIT_TIMEOUT" default:"1500ms"`
+	WorkloadReplicas time.Duration `envconfig:"KEDA_CONDITION_WAIT_TIMEOUT" default:"1500ms"`
 	// ForceHTTP2 toggles whether to try to force HTTP2 for all requests
 	ForceHTTP2 bool `envconfig:"KEDA_HTTP_FORCE_HTTP2" default:"false"`
 	// MaxIdleConns is the max number of connections that can be idle in the

interceptor/config/validate.go

@@ -2,16 +2,36 @@ package config
 
 import (
 	"fmt"
+	"os"
 	"time"
+
+	"github.com/go-logr/logr"
 )
 
-func Validate(srvCfg Serving, timeoutsCfg Timeouts) error {
-	deplCachePollInterval := time.Duration(srvCfg.DeploymentCachePollIntervalMS) * time.Millisecond
-	if timeoutsCfg.DeploymentReplicas < deplCachePollInterval {
+func Validate(srvCfg *Serving, timeoutsCfg Timeouts, lggr logr.Logger) error {
+	// TODO(jorturfer): delete this on v0.8.0
+	_, deploymentEnvExist := os.LookupEnv("KEDA_HTTP_DEPLOYMENT_CACHE_POLLING_INTERVAL_MS")
+	_, endpointsEnvExist := os.LookupEnv("KEDA_HTTP_ENDPOINTS_CACHE_POLLING_INTERVAL_MS")
+	if deploymentEnvExist && endpointsEnvExist {
+		return fmt.Errorf(
+			"%s and %s are mutual exclusive",
+			"KEDA_HTTP_DEPLOYMENT_CACHE_POLLING_INTERVAL_MS",
+			"KEDA_HTTP_ENDPOINTS_CACHE_POLLING_INTERVAL_MS",
+		)
+	}
+	if deploymentEnvExist && !endpointsEnvExist {
+		srvCfg.EndpointsCachePollIntervalMS = srvCfg.DeploymentCachePollIntervalMS
+		srvCfg.DeploymentCachePollIntervalMS = 0
+		lggr.Info("WARNING: KEDA_HTTP_DEPLOYMENT_CACHE_POLLING_INTERVAL_MS has been deprecated in favor of KEDA_HTTP_ENDPOINTS_CACHE_POLLING_INTERVAL_MS and wil be removed on v0.8.0")
+	}
+	// END TODO
+
+	endpointsCachePollInterval := time.Duration(srvCfg.EndpointsCachePollIntervalMS) * time.Millisecond
+	if timeoutsCfg.WorkloadReplicas < endpointsCachePollInterval {
 		return fmt.Errorf(
-			"deployment replicas timeout (%s) should not be less than the Deployment Cache Poll Interval (%s)",
-			timeoutsCfg.DeploymentReplicas,
-			deplCachePollInterval,
+			"workload replicas timeout (%s) should not be less than the Endpoints Cache Poll Interval (%s)",
+			timeoutsCfg.WorkloadReplicas,
+			endpointsCachePollInterval,
 		)
 	}
 	return nil

interceptor/forward_wait_func.go

@@ -5,65 +5,69 @@ import (
 	"fmt"
 
 	"github.com/go-logr/logr"
-	appsv1 "k8s.io/api/apps/v1"
+	v1 "k8s.io/api/core/v1"
 
 	"github.com/kedacore/http-add-on/pkg/k8s"
 )
 
 // forwardWaitFunc is a function that waits for a condition
 // before proceeding to serve the request.
-type forwardWaitFunc func(context.Context, string, string) (int, error)
+type forwardWaitFunc func(context.Context, string, string) (bool, error)
 
-func deploymentCanServe(depl appsv1.Deployment) bool {
-	return depl.Status.ReadyReplicas > 0
+func workloadActiveEndpoints(endpoints v1.Endpoints) int {
+	total := 0
+	for _, subset := range endpoints.Subsets {
+		total += len(subset.Addresses)
+	}
+	return total
 }
 
-func newDeployReplicasForwardWaitFunc(
+func newWorkloadReplicasForwardWaitFunc(
 	lggr logr.Logger,
-	deployCache k8s.DeploymentCache,
+	endpointCache k8s.EndpointsCache,
 ) forwardWaitFunc {
-	return func(ctx context.Context, deployNS, deployName string) (int, error) {
+	return func(ctx context.Context, endpointNS, endpointName string) (bool, error) {
 		// get a watcher & its result channel before querying the
-		// deployment cache, to ensure we don't miss events
-		watcher, err := deployCache.Watch(deployNS, deployName)
+		// endpoints cache, to ensure we don't miss events
+		watcher, err := endpointCache.Watch(endpointNS, endpointName)
 		if err != nil {
-			return 0, err
+			return false, err
 		}
 		eventCh := watcher.ResultChan()
 		defer watcher.Stop()
 
-		deployment, err := deployCache.Get(deployNS, deployName)
+		endpoints, err := endpointCache.Get(endpointNS, endpointName)
 		if err != nil {
-			// if we didn't get the initial deployment state, bail out
-			return 0, fmt.Errorf(
-				"error getting state for deployment %s/%s: %w",
-				deployNS,
-				deployName,
+			// if we didn't get the initial endpoints state, bail out
+			return false, fmt.Errorf(
+				"error getting state for endpoints %s/%s: %w",
+				endpointNS,
+				endpointName,
 				err,
 			)
 		}
-		// if there is 1 or more replica, we're done waiting
-		if deploymentCanServe(deployment) {
-			return int(deployment.Status.ReadyReplicas), nil
+		// if there is 1 or more active endpoints, we're done waiting
+		activeEndpoints := workloadActiveEndpoints(endpoints)
+		if activeEndpoints > 0 {
+			return false, nil
 		}
 
 		for {
 			select {
 			case event := <-eventCh:
-				deployment, ok := event.Object.(*appsv1.Deployment)
+				endpoints, ok := event.Object.(*v1.Endpoints)
 				if !ok {
 					lggr.Info(
-						"Didn't get a deployment back in event",
+						"Didn't get a endpoints back in event",
 					)
-				} else if deploymentCanServe(*deployment) {
-					return 0, nil
+				} else if activeEndpoints := workloadActiveEndpoints(*endpoints); activeEndpoints > 0 {
+					return true, nil
 				}
 			case <-ctx.Done():
 				// otherwise, if the context is marked done before
 				// we're done waiting, fail.
-				return 0, fmt.Errorf(
-					"context marked done while waiting for deployment %s to reach > 0 replicas: %w",
-					deployName,
+				return false, fmt.Errorf(
+					"context marked done while waiting for workload reach > 0 replicas: %w",
 					ctx.Err(),
 				)
 			}

interceptor/forward_wait_func_test.go

@@ -8,79 +8,63 @@ import (
 	"github.com/go-logr/logr"
 	"github.com/stretchr/testify/require"
 	"golang.org/x/sync/errgroup"
-	appsv1 "k8s.io/api/apps/v1"
-	corev1 "k8s.io/api/core/v1"
+	v1 "k8s.io/api/core/v1"
 	metav1 "k8s.io/apimachinery/pkg/apis/meta/v1"
 	"k8s.io/apimachinery/pkg/watch"
 
 	"github.com/kedacore/http-add-on/pkg/k8s"
 )
 
 // Test to make sure the wait function returns a nil error if there is immediately
-// one replica on the target deployment
+// one active endpoint on the target deployment
 func TestForwardWaitFuncOneReplica(t *testing.T) {
 	ctx := context.Background()
 
 	const waitFuncWait = 1 * time.Second
 	r := require.New(t)
 	const ns = "testNS"
-	const deployName = "TestForwardingHandlerDeploy"
-	cache := k8s.NewFakeDeploymentCache()
-	cache.AddDeployment(*newDeployment(
-		ns,
-		deployName,
-		"myimage",
-		[]int32{123},
-		nil,
-		map[string]string{},
-		corev1.PullAlways,
-	))
+	const endpointsName = "TestForwardingHandler"
+	endpoints := *newEndpoint(ns, endpointsName)
+	cache := k8s.NewFakeEndpointsCache()
+	cache.Set(endpoints)
+	cache.SetSubsets(ns, endpointsName, 1)
 
 	ctx, done := context.WithTimeout(ctx, waitFuncWait)
 	defer done()
 	group, ctx := errgroup.WithContext(ctx)
 
-	waitFunc := newDeployReplicasForwardWaitFunc(
+	waitFunc := newWorkloadReplicasForwardWaitFunc(
 		logr.Discard(),
 		cache,
 	)
 
 	group.Go(func() error {
-		_, err := waitFunc(ctx, ns, deployName)
+		_, err := waitFunc(ctx, ns, endpointsName)
 		return err
 	})
 	r.NoError(group.Wait(), "wait function failed, but it shouldn't have")
 }
 
-// Test to make sure the wait function returns an error if there are no replicas, and that doesn't change
+// Test to make sure the wait function returns an error if there are active endpoints, and that doesn't change
 // within a timeout
 func TestForwardWaitFuncNoReplicas(t *testing.T) {
 	ctx := context.Background()
 	const waitFuncWait = 1 * time.Second
 	r := require.New(t)
 	const ns = "testNS"
-	const deployName = "TestForwardingHandlerHoldsDeployment"
-	deployment := newDeployment(
-		ns,
-		deployName,
-		"myimage",
-		[]int32{123},
-		nil,
-		map[string]string{},
-		corev1.PullAlways,
-	)
-	deployment.Status.ReadyReplicas = 0
-	cache := k8s.NewFakeDeploymentCache()
-	cache.AddDeployment(*deployment)
+	const endpointsName = "TestForwardWaitFuncNoReplicas"
+	endpoints := *newEndpoint(ns, endpointsName)
+	cache := k8s.NewFakeEndpointsCache()
+	cache.Set(endpoints)
 
 	ctx, done := context.WithTimeout(ctx, waitFuncWait)
 	defer done()
-	waitFunc := newDeployReplicasForwardWaitFunc(
+	waitFunc := newWorkloadReplicasForwardWaitFunc(
 		logr.Discard(),
 		cache,
 	)
 
-	_, err := waitFunc(ctx, ns, deployName)
+	_, err := waitFunc(ctx, ns, endpointsName)
 	r.Error(err)
 }
 
@@ -90,100 +74,58 @@ func TestWaitFuncWaitsUntilReplicas(t *testing.T) {
 	totalWaitDur := 500 * time.Millisecond
 
 	const ns = "testNS"
-	const deployName = "TestForwardingHandlerHoldsDeployment"
-	deployment := newDeployment(
-		ns,
-		deployName,
-		"myimage",
-		[]int32{123},
-		nil,
-		map[string]string{},
-		corev1.PullAlways,
-	)
-	deployment.Spec.Replicas = k8s.Int32P(0)
-	cache := k8s.NewFakeDeploymentCache()
-	cache.AddDeployment(*deployment)
+	const endpointsName = "TestForwardingHandlerHolds"
+
+	endpoints := *newEndpoint(ns, endpointsName)
+	cache := k8s.NewFakeEndpointsCache()
+	cache.Set(endpoints)
 	// create a watcher first so that the goroutine
 	// can later fetch it and send a message on it
-	_, err := cache.Watch(ns, deployName)
+	_, err := cache.Watch(ns, endpointsName)
 	r.NoError(err)
 
 	ctx, done := context.WithTimeout(ctx, totalWaitDur)
-	waitFunc := newDeployReplicasForwardWaitFunc(
+	waitFunc := newWorkloadReplicasForwardWaitFunc(
 		logr.Discard(),
 		cache,
 	)
 
-	// this channel will be closed immediately after the replicas were increased
-	replicasIncreasedCh := make(chan struct{})
+	// this channel will be closed immediately after the active endpoints were increased
+	activeEndpointsIncreasedCh := make(chan struct{})
 	go func() {
 		time.Sleep(totalWaitDur / 2)
-		watcher := cache.GetWatcher(ns, deployName)
+		watcher := cache.GetWatcher(ns, endpointsName)
 		r.NotNil(watcher, "watcher was not found")
-		modifiedDeployment := deployment.DeepCopy()
-		modifiedDeployment.Spec.Replicas = k8s.Int32P(1)
-		watcher.Action(watch.Modified, modifiedDeployment)
-		close(replicasIncreasedCh)
+		modifiedEndpoints := endpoints.DeepCopy()
+		modifiedEndpoints.Subsets = []v1.EndpointSubset{
+			{
+				Addresses: []v1.EndpointAddress{
+					{IP: "1.2.3.4"},
+				},
+			},
+		}
+		watcher.Action(watch.Modified, modifiedEndpoints)
+		close(activeEndpointsIncreasedCh)
 	}()
-	_, err = waitFunc(ctx, ns, deployName)
+	_, err = waitFunc(ctx, ns, endpointsName)
 	r.NoError(err)
 	done()
 }
 
-// newDeployment creates a new deployment object
+// newEndpoint creates a new endpoints object
 // with the given name and the given image. This does not actually create
-// the deployment in the cluster, it just creates the deployment object
+// the endpoints in the cluster, it just creates the endpoints object
 // in memory
-func newDeployment(
+func newEndpoint(
 	namespace,
-	name,
-	image string,
-	ports []int32,
-	env []corev1.EnvVar,
-	labels map[string]string,
-	pullPolicy corev1.PullPolicy,
-) *appsv1.Deployment {
-	containerPorts := make([]corev1.ContainerPort, len(ports))
-	for i, port := range ports {
-		containerPorts[i] = corev1.ContainerPort{
-			ContainerPort: port,
-		}
-	}
-	deployment := &appsv1.Deployment{
-		TypeMeta: metav1.TypeMeta{
-			Kind: "Deployment",
-		},
+	name string,
+) *v1.Endpoints {
+	endpoints := &v1.Endpoints{
 		ObjectMeta: metav1.ObjectMeta{
 			Name:      name,
 			Namespace: namespace,
-			Labels:    labels,
-		},
-		Spec: appsv1.DeploymentSpec{
-			Selector: &metav1.LabelSelector{
-				MatchLabels: labels,
-			},
-			Replicas: k8s.Int32P(1),
-			Template: corev1.PodTemplateSpec{
-				ObjectMeta: metav1.ObjectMeta{
-					Labels: labels,
-				},
-				Spec: corev1.PodSpec{
-					Containers: []corev1.Container{
-						{
-							Image:           image,
-							Name:            name,
-							ImagePullPolicy: pullPolicy,
-							Ports:           containerPorts,
-							Env:             env,
-						},
-					},
-				},
-			},
-		},
-		Status: appsv1.DeploymentStatus{
-			ReadyReplicas: 1,
 		},
 	}
 
-	return deployment
+	return endpoints
 }