feat(allowlist): support both v1 and v1alpha2 InferencePool APIs with flag

cmd/pd-sidecar/main.go

@@ -58,6 +58,7 @@ func main() {
 	inferencePoolNamespace := flag.String("inference-pool-namespace", os.Getenv("INFERENCE_POOL_NAMESPACE"), "the Kubernetes namespace to watch for InferencePool resources (defaults to INFERENCE_POOL_NAMESPACE env var)")
 	inferencePoolName := flag.String("inference-pool-name", os.Getenv("INFERENCE_POOL_NAME"), "the specific InferencePool name to watch (defaults to INFERENCE_POOL_NAME env var)")
 	enablePrefillerSampling := flag.Bool("enable-prefiller-sampling", func() bool { b, _ := strconv.ParseBool(os.Getenv("ENABLE_PREFILLER_SAMPLING")); return b }(), "if true, the target prefill instance will be selected randomly from among the provided prefill host values")
+	poolGroup := flag.String("pool-group", proxy.DefaultPoolGroup, "group of the InferencePool this Endpoint Picker is associated with.")
 
 	opts := zap.Options{}
 	opts.BindFlags(flag.CommandLine) // optional to allow zap logging control via CLI
@@ -135,7 +136,7 @@ func main() {
 	}
 
 	// Create SSRF protection validator
-	validator, err := proxy.NewAllowlistValidator(*enableSSRFProtection, *inferencePoolNamespace, *inferencePoolName)
+	validator, err := proxy.NewAllowlistValidator(*enableSSRFProtection, *poolGroup, *inferencePoolNamespace, *inferencePoolName)
 	if err != nil {
 		logger.Error(err, "failed to create SSRF protection validator")
 		return

pkg/sidecar/proxy/allowlist.go

@@ -38,12 +38,16 @@ import (
 )
 
 const (
-	inferencePoolGroup    = "inference.networking.x-k8s.io"
-	inferencePoolVersion  = "v1alpha2"
 	inferencePoolResource = "inferencepools"
 	resyncPeriod          = 30 * time.Second
 )
 
+// InferencePool API group to version mapping
+var inferencePoolGroupToVersion = map[string]string{
+	DefaultPoolGroup: "v1",
+	LegacyPoolGroup:  "v1alpha2",
+}
+
 // AllowlistValidator manages allowed prefill targets based on InferencePool resources
 type AllowlistValidator struct {
 	logger        logr.Logger
@@ -52,6 +56,8 @@ type AllowlistValidator struct {
 	poolName      string
 	enabled       bool
 
+	gvr schema.GroupVersionResource // detected GVR
+
 	// allowedTargets maps hostport -> bool for allowed prefill targets
 	allowedTargets   set.Set[string]
 	allowedTargetsMu sync.RWMutex
@@ -65,13 +71,26 @@ type AllowlistValidator struct {
 }
 
 // NewAllowlistValidator creates a new SSRF protection validator
-func NewAllowlistValidator(enabled bool, namespace string, poolName string) (*AllowlistValidator, error) {
+func NewAllowlistValidator(enabled bool, poolGroup, namespace, poolName string) (*AllowlistValidator, error) {
 	if !enabled {
 		return &AllowlistValidator{
 			enabled: false,
 		}, nil
 	}
 
+	// Determine version based on poolGroup
+	version, exists := inferencePoolGroupToVersion[poolGroup]
+	if !exists {
+		return nil, fmt.Errorf("unsupported poolGroup: %s, "+
+			"must be one of %v", poolGroup, getSupportedPoolGroups())
+	}
+
+	gvr := schema.GroupVersionResource{
+		Group:    poolGroup,
+		Version:  version,
+		Resource: inferencePoolResource,
+	}
+
 	loadingRules := clientcmd.NewDefaultClientConfigLoadingRules()
 	overrides := &clientcmd.ConfigOverrides{}
 	config, err := clientcmd.NewNonInteractiveDeferredLoadingClientConfig(
@@ -92,39 +111,43 @@ func NewAllowlistValidator(enabled bool, namespace string, poolName string) (*Al
 		dynamicClient:  dynamicClient,
 		namespace:      namespace,
 		poolName:       poolName,
+		gvr:            gvr,
 		allowedTargets: set.New[string](),
 		podInformers:   make(map[string]cache.SharedInformer),
 		podStopChans:   make(map[string]chan struct{}),
 		stopCh:         make(chan struct{}),
 	}, nil
 }
 
+func getSupportedPoolGroups() []string {
+	groups := make([]string, 0, len(inferencePoolGroupToVersion))
+	for group := range inferencePoolGroupToVersion {
+		groups = append(groups, group)
+	}
+	return groups
+}
+
 // Start begins watching InferencePool resources and managing the allowlist
 func (av *AllowlistValidator) Start(ctx context.Context) error {
 	if !av.enabled {
 		return nil
 	}
 
 	av.logger = log.FromContext(ctx).WithName("allowlist-validator")
-	av.logger.Info("starting SSRF protection allowlist validator", "namespace", av.namespace, "poolName", av.poolName)
-
-	gvr := schema.GroupVersionResource{
-		Group:    inferencePoolGroup,
-		Version:  inferencePoolVersion,
-		Resource: inferencePoolResource,
-	}
+	av.logger.Info("starting SSRF protection allowlist validator",
+		"namespace", av.namespace, "poolName", av.poolName, "gvr", av.gvr.String())
 
 	// Create informer for the specific InferencePool resource
 	lw := &cache.ListWatch{
-		ListFunc: func(options metav1.ListOptions) (runtime.Object, error) {
+		ListWithContextFunc: func(ctx context.Context, options metav1.ListOptions) (runtime.Object, error) {
 			// List with field selector to get only the specific InferencePool
 			options.FieldSelector = "metadata.name=" + av.poolName
-			return av.dynamicClient.Resource(gvr).Namespace(av.namespace).List(ctx, options)
+			return av.dynamicClient.Resource(av.gvr).Namespace(av.namespace).List(ctx, options)
 		},
-		WatchFunc: func(options metav1.ListOptions) (watch.Interface, error) {
+		WatchFuncWithContext: func(ctx context.Context, options metav1.ListOptions) (watch.Interface, error) {
 			// Watch the specific InferencePool by name using field selector
 			options.FieldSelector = "metadata.name=" + av.poolName
-			return av.dynamicClient.Resource(gvr).Namespace(av.namespace).Watch(ctx, options)
+			return av.dynamicClient.Resource(av.gvr).Namespace(av.namespace).Watch(ctx, options)
 		},
 	}
 
@@ -142,7 +165,7 @@ func (av *AllowlistValidator) Start(ctx context.Context) error {
 
 	// Wait for cache sync
 	if !cache.WaitForCacheSync(av.stopCh, av.poolInformer.HasSynced) {
-		return fmt.Errorf("failed to sync InferencePool cache within timeout (check RBAC permissions for inferencepools.%s and that pool '%s' exists)", inferencePoolGroup, av.poolName)
+		return fmt.Errorf("failed to sync InferencePool cache within timeout (check RBAC permissions for inferencepools.%s and that pool '%s' exists)", av.gvr.String(), av.poolName)
 	}
 
 	av.logger.Info("allowlist validator started successfully")

pkg/sidecar/proxy/allowlist_test.go

@@ -28,7 +28,7 @@ var _ = Describe("AllowlistValidator", func() {
 
 		BeforeEach(func() {
 			var err error
-			validator, err = NewAllowlistValidator(false, "test-namespace", "test-pool")
+			validator, err = NewAllowlistValidator(false, DefaultPoolGroup, "test-namespace", "test-pool")
 			Expect(err).ToNot(HaveOccurred())
 		})
 

pkg/sidecar/proxy/data_parallel_test.go

@@ -59,7 +59,7 @@ var _ = Describe("Data Parallel support", func() {
 				DataParallelSize:          testDataParallelSize,
 			}
 			theProxy := NewProxy(strconv.Itoa(fakeProxyPort), decodeURL, cfg)
-			theProxy.allowlistValidator, err = NewAllowlistValidator(false, "", "")
+			theProxy.allowlistValidator, err = NewAllowlistValidator(false, DefaultPoolGroup, "", "")
 			Expect(err).ToNot(HaveOccurred())
 
 			err = theProxy.startDataParallel(ctx, nil, grp)

pkg/sidecar/proxy/proxy.go

@@ -60,6 +60,11 @@ const (
 
 	// ConnectorSGLang enables SGLang P/D disaggregation protocol
 	ConnectorSGLang = "sglang"
+
+	// DefaultPoolGroup is the default pool group name
+	DefaultPoolGroup = "inference.networking.k8s.io"
+	// LegacyPoolGroup is the legacy pool group name
+	LegacyPoolGroup = "inference.networking.x-k8s.io"
 )
 
 // Config represents the proxy server configuration