fix: operator defaults (#2398)

Signed-off-by: mohammedabdulwahhab <furkhan324@berkeley.edu>

deploy/cloud/operator/internal/dynamo/component_frontend.go

+/*
+ * SPDX-FileCopyrightText: Copyright (c) 2025 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
+ * SPDX-License-Identifier: Apache-2.0
+ */
+
+package dynamo
+
+import (
+	"fmt"
+
+	commonconsts "github.com/ai-dynamo/dynamo/deploy/cloud/operator/internal/consts"
+	corev1 "k8s.io/api/core/v1"
+	"k8s.io/apimachinery/pkg/api/resource"
+	"k8s.io/apimachinery/pkg/util/intstr"
+)
+
+// FrontendDefaults implements ComponentDefaults for Frontend components
+type FrontendDefaults struct {
+	*BaseComponentDefaults
+}
+
+func NewFrontendDefaults() *FrontendDefaults {
+	return &FrontendDefaults{&BaseComponentDefaults{}}
+}
+
+func (f *FrontendDefaults) GetBaseContainer(numberOfNodes int32) (corev1.Container, error) {
+	// Frontend doesn't need backend-specific config
+	container := f.getCommonContainer()
+
+	// Add HTTP port
+	container.Ports = []corev1.ContainerPort{
+		{
+			Protocol:      corev1.ProtocolTCP,
+			Name:          commonconsts.DynamoContainerPortName,
+			ContainerPort: int32(commonconsts.DynamoServicePort),
+		},
+	}
+
+	// Add frontend-specific defaults
+	container.LivenessProbe = &corev1.Probe{
+		ProbeHandler: corev1.ProbeHandler{
+			HTTPGet: &corev1.HTTPGetAction{
+				Path: "/health",
+				Port: intstr.FromString(commonconsts.DynamoContainerPortName),
+			},
+		},
+		InitialDelaySeconds: 60,
+		PeriodSeconds:       60,
+		TimeoutSeconds:      30,
+		FailureThreshold:    10,
+	}
+
+	container.ReadinessProbe = &corev1.Probe{
+		ProbeHandler: corev1.ProbeHandler{
+			Exec: &corev1.ExecAction{
+				Command: []string{
+					"/bin/sh",
+					"-c",
+					"curl -s http://localhost:${DYNAMO_PORT}/health | jq -e \".status == \\\"healthy\\\"\"",
+				},
+			},
+		},
+		InitialDelaySeconds: 60,
+		PeriodSeconds:       60,
+		TimeoutSeconds:      30,
+		FailureThreshold:    10,
+	}
+
+	container.Resources = corev1.ResourceRequirements{
+		Requests: corev1.ResourceList{
+			corev1.ResourceCPU:    resource.MustParse("1"),
+			corev1.ResourceMemory: resource.MustParse("2Gi"),
+		},
+		Limits: corev1.ResourceList{
+			corev1.ResourceCPU:    resource.MustParse("1"),
+			corev1.ResourceMemory: resource.MustParse("2Gi"),
+		},
+	}
+
+	// Add standard environment variables
+	container.Env = []corev1.EnvVar{
+		{
+			Name:  commonconsts.EnvDynamoServicePort,
+			Value: fmt.Sprintf("%d", commonconsts.DynamoServicePort),
+		},
+	}
+
+	return container, nil
+}

deploy/cloud/operator/internal/dynamo/component_planner.go

+/*
+ * SPDX-FileCopyrightText: Copyright (c) 2025 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
+ * SPDX-License-Identifier: Apache-2.0
+ */
+
+package dynamo
+
+import (
+	commonconsts "github.com/ai-dynamo/dynamo/deploy/cloud/operator/internal/consts"
+	corev1 "k8s.io/api/core/v1"
+	"k8s.io/apimachinery/pkg/api/resource"
+)
+
+// PlannerDefaults implements ComponentDefaults for Planner components
+type PlannerDefaults struct {
+	*BaseComponentDefaults
+}
+
+func NewPlannerDefaults() *PlannerDefaults {
+	return &PlannerDefaults{&BaseComponentDefaults{}}
+}
+
+func (p *PlannerDefaults) GetBaseContainer(numberOfNodes int32) (corev1.Container, error) {
+	container := p.getCommonContainer()
+
+	// Add planner-specific defaults
+	container.Resources = corev1.ResourceRequirements{
+		Requests: corev1.ResourceList{
+			corev1.ResourceCPU:    resource.MustParse("2"),
+			corev1.ResourceMemory: resource.MustParse("2Gi"),
+		},
+		Limits: corev1.ResourceList{
+			corev1.ResourceCPU:    resource.MustParse("2"),
+			corev1.ResourceMemory: resource.MustParse("2Gi"),
+		},
+	}
+
+	return container, nil
+}
+
+func (p *PlannerDefaults) GetBasePodSpec(numberOfNodes int32) (corev1.PodSpec, error) {
+	podSpec := corev1.PodSpec{
+		ServiceAccountName: commonconsts.PlannerServiceAccountName,
+	}
+	return podSpec, nil
+}

deploy/cloud/operator/internal/dynamo/component_worker.go

+/*
+ * SPDX-FileCopyrightText: Copyright (c) 2025 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
+ * SPDX-License-Identifier: Apache-2.0
+ */
+
+package dynamo
+
+import (
+	"fmt"
+
+	commonconsts "github.com/ai-dynamo/dynamo/deploy/cloud/operator/internal/consts"
+	corev1 "k8s.io/api/core/v1"
+	"k8s.io/apimachinery/pkg/api/resource"
+	"k8s.io/apimachinery/pkg/util/intstr"
+)
+
+// WorkerDefaults implements ComponentDefaults for Worker components
+type WorkerDefaults struct {
+	*BaseComponentDefaults
+}
+
+func NewWorkerDefaults() *WorkerDefaults {
+	return &WorkerDefaults{&BaseComponentDefaults{}}
+}
+
+func (w *WorkerDefaults) GetBaseContainer(numberOfNodes int32) (corev1.Container, error) {
+	container := w.getCommonContainer()
+
+	// Add system port
+	container.Ports = []corev1.ContainerPort{
+		{
+			Protocol:      corev1.ProtocolTCP,
+			Name:          commonconsts.DynamoSystemPortName,
+			ContainerPort: int32(commonconsts.DynamoSystemPort),
+		},
+	}
+
+	// Add worker base defaults
+	container.Resources = corev1.ResourceRequirements{
+		Requests: corev1.ResourceList{
+			corev1.ResourceCPU:    resource.MustParse("10"),
+			corev1.ResourceMemory: resource.MustParse("20Gi"),
+		},
+		Limits: corev1.ResourceList{
+			corev1.ResourceCPU:    resource.MustParse("10"),
+			corev1.ResourceMemory: resource.MustParse("20Gi"),
+			"nvidia.com/gpu":      resource.MustParse("1"),
+		},
+	}
+
+	container.LivenessProbe = &corev1.Probe{
+		ProbeHandler: corev1.ProbeHandler{
+			HTTPGet: &corev1.HTTPGetAction{
+				Path: "/live",
+				Port: intstr.FromString(commonconsts.DynamoSystemPortName),
+			},
+		},
+		PeriodSeconds:    5,
+		TimeoutSeconds:   30,
+		FailureThreshold: 1,
+	}
+
+	container.ReadinessProbe = &corev1.Probe{
+		ProbeHandler: corev1.ProbeHandler{
+			HTTPGet: &corev1.HTTPGetAction{
+				Path: "/health",
+				Port: intstr.FromString(commonconsts.DynamoSystemPortName),
+			},
+		},
+		PeriodSeconds:    10,
+		TimeoutSeconds:   30,
+		FailureThreshold: 60,
+	}
+
+	container.StartupProbe = &corev1.Probe{
+		ProbeHandler: corev1.ProbeHandler{
+			HTTPGet: &corev1.HTTPGetAction{
+				Path: "/live",
+				Port: intstr.FromString(commonconsts.DynamoSystemPortName),
+			},
+		},
+		PeriodSeconds:    10,
+		TimeoutSeconds:   5,
+		FailureThreshold: 60,
+	}
+
+	container.Env = []corev1.EnvVar{
+		{
+			Name:  "DYN_SYSTEM_ENABLED",
+			Value: "true",
+		},
+		{
+			Name:  "DYN_SYSTEM_USE_ENDPOINT_HEALTH_STATUS",
+			Value: "[\"generate\"]",
+		},
+		{
+			Name:  "DYN_SYSTEM_PORT",
+			Value: fmt.Sprintf("%d", commonconsts.DynamoSystemPort),
+		},
+	}
+
+	return container, nil
+}

deploy/cloud/operator/internal/dynamo/graph.go

@@ -21,6 +21,7 @@ import (
 	"context"
 	"encoding/json"
 	"fmt"
+	"maps"
 	"regexp"
 	"sort"
 	"strconv"
@@ -191,7 +192,7 @@ func GenerateDynamoComponentsDeployments(ctx context.Context, parentDynamoGraphD
 			// finally set the service account name
 			deployment.Spec.ExtraPodSpec.PodSpec.ServiceAccountName = commonconsts.PlannerServiceAccountName
 		}
-		if deployment.IsMainComponent() && defaultIngressSpec != nil && deployment.Spec.Ingress == nil {
+		if deployment.IsFrontendComponent() && defaultIngressSpec != nil && deployment.Spec.Ingress == nil {
 			deployment.Spec.Ingress = defaultIngressSpec
 		}
 		// merge the envs from the parent deployment with the envs from the service
@@ -219,7 +220,7 @@ func GenerateDynamoComponentsDeployments(ctx context.Context, parentDynamoGraphD
 // updateDynDeploymentConfig updates the runtime config object for the given dynamoDeploymentComponent
 // It updates the port for the given service (if it is the main component)
 func updateDynDeploymentConfig(dynamoDeploymentComponent *v1alpha1.DynamoComponentDeployment, newPort int) error {
-	if dynamoDeploymentComponent.IsMainComponent() {
+	if dynamoDeploymentComponent.IsFrontendComponent() {
 		dynamoDeploymentConfig := dynamoDeploymentComponent.GetDynamoDeploymentConfig()
 		if dynamoDeploymentConfig != nil {
 			var config map[string]any
@@ -668,11 +669,6 @@ func isWorkerComponent(componentType string) bool {
 
 // addStandardEnvVars adds the standard environment variables that are common to both Grove and Controller
 func addStandardEnvVars(container *corev1.Container, controllerConfig controller_common.Config) {
-	container.Env = append(container.Env, corev1.EnvVar{
-		Name:  commonconsts.EnvDynamoServicePort,
-		Value: fmt.Sprintf("%d", commonconsts.DynamoServicePort),
-	})
-
 	if controllerConfig.NatsAddress != "" {
 		container.Env = append(container.Env, corev1.EnvVar{
 			Name:  "NATS_SERVER",
@@ -702,47 +698,60 @@ func GenerateBasePodSpec(
 	multinodeDeploymentType commonconsts.MultinodeDeploymentType,
 	serviceName string,
 ) (corev1.PodSpec, error) {
-	container := corev1.Container{
-		Name:           "main",
-		LivenessProbe:  component.LivenessProbe,
-		ReadinessProbe: component.ReadinessProbe,
-		Env:            component.Envs,
-		Ports: []corev1.ContainerPort{
-			{
-				Protocol:      corev1.ProtocolTCP,
-				Name:          commonconsts.DynamoContainerPortName,
-				ContainerPort: int32(commonconsts.DynamoServicePort),
-			},
-		},
-	}
-	// Add system port for worker components
-	if component.ComponentType == commonconsts.ComponentTypeWorker {
-		container.Ports = append(container.Ports, corev1.ContainerPort{
-			Protocol:      corev1.ProtocolTCP,
-			Name:          commonconsts.DynamoSystemPortName,
-			ContainerPort: int32(commonconsts.DynamoSystemPort),
-		})
+	// Start with base container generated per component type
+	componentDefaults := ComponentDefaultsFactory(component.ComponentType, numberOfNodes)
+	container, err := componentDefaults.GetBaseContainer(numberOfNodes)
+	if err != nil {
+		return corev1.PodSpec{}, fmt.Errorf("failed to get base container: %w", err)
 	}
-	// First merge the mainContainer from extraPodSpec to get the base command and args
+
 	if component.ExtraPodSpec != nil && component.ExtraPodSpec.MainContainer != nil {
 		main := component.ExtraPodSpec.MainContainer.DeepCopy()
 		if main != nil {
 			// merge the extraPodSpec from the parent deployment with the extraPodSpec from the service
-			err := mergo.Merge(&container, *main, mergo.WithOverride)
+			err = mergo.Merge(&container, *main, mergo.WithOverride)
 			if err != nil {
 				return corev1.PodSpec{}, fmt.Errorf("failed to merge extraPodSpec: %w", err)
 			}
+
+			// main container fields that require special handling
 			container.Env = MergeEnvs(component.Envs, container.Env)
+			// Note: startup probe does not have its own top level field so it must be passed in extraPodSpec.MainContainer
+			// We want to overwrite entirely if provided rather than merge
+			if main.StartupProbe != nil {
+				container.StartupProbe = main.StartupProbe
+			}
 		}
 	}
 
-	resourcesConfig, err := controller_common.GetResourcesConfig(component.Resources)
+	// Merge probes entirely if they are passed (no partial merge)
+	if component.LivenessProbe != nil {
+		container.LivenessProbe = component.LivenessProbe.DeepCopy()
+	}
+	if component.ReadinessProbe != nil {
+		container.ReadinessProbe = component.ReadinessProbe.DeepCopy()
+	}
+
+	overrideResources, err := controller_common.GetResourcesConfig(component.Resources)
 	if err != nil {
 		return corev1.PodSpec{}, fmt.Errorf("failed to get resources config: %w", err)
 	}
-	if resourcesConfig != nil {
-		container.Resources = *resourcesConfig
+	// Requests
+	if overrideResources != nil && len(overrideResources.Requests) > 0 {
+		if container.Resources.Requests == nil {
+			container.Resources.Requests = corev1.ResourceList{}
+		}
+		maps.Copy(container.Resources.Requests, overrideResources.Requests)
 	}
+
+	// Limits
+	if overrideResources != nil && len(overrideResources.Limits) > 0 {
+		if container.Resources.Limits == nil {
+			container.Resources.Limits = corev1.ResourceList{}
+		}
+		maps.Copy(container.Resources.Limits, overrideResources.Limits)
+	}
+
 	imagePullSecrets := []corev1.LocalObjectReference{}
 	if secretsRetriever != nil && component.ExtraPodSpec != nil && component.ExtraPodSpec.MainContainer != nil && component.ExtraPodSpec.MainContainer.Image != "" {
 		secretsName, err := secretsRetriever.GetSecrets(namespace, component.ExtraPodSpec.MainContainer.Image)
@@ -780,15 +789,26 @@ func GenerateBasePodSpec(
 	shmVolume, shmVolumeMount := generateSharedMemoryVolumeAndMount(&container.Resources)
 	volumes = append(volumes, shmVolume)
 	container.VolumeMounts = append(container.VolumeMounts, shmVolumeMount)
+
 	// Apply backend-specific container modifications
 	backend := BackendFactory(backendFramework)
 	if backend == nil {
 		return corev1.PodSpec{}, fmt.Errorf("unsupported backend framework: %s", backendFramework)
 	}
 	backend.UpdateContainer(&container, numberOfNodes, role, component, multinodeDeploymentType, serviceName)
-	var podSpec corev1.PodSpec
+
+	// get base podspec from component
+	podSpec, err := componentDefaults.GetBasePodSpec(numberOfNodes)
+	if err != nil {
+		return corev1.PodSpec{}, fmt.Errorf("failed to get base podspec: %w", err)
+	}
+
 	if component.ExtraPodSpec != nil && component.ExtraPodSpec.PodSpec != nil {
-		podSpec = *component.ExtraPodSpec.PodSpec.DeepCopy()
+		// merge extraPodSpec PodSpec with base podspec
+		err := mergo.Merge(&podSpec, component.ExtraPodSpec.PodSpec.DeepCopy(), mergo.WithOverride)
+		if err != nil {
+			return corev1.PodSpec{}, fmt.Errorf("failed to merge extraPodSpec: %w", err)
+		}
 	}
 	podSpec.Containers = append(podSpec.Containers, container)
 	podSpec.Volumes = append(podSpec.Volumes, volumes...)

deploy/cloud/operator/internal/dynamo/graph_test.go

@@ -63,7 +63,7 @@ func TestGenerateDynamoComponentsDeployments(t *testing.T) {
 							"service1": {
 								DynamoComponentDeploymentSharedSpec: v1alpha1.DynamoComponentDeploymentSharedSpec{
 									DynamoNamespace: &[]string{"default"}[0],
-									ComponentType:   "main",
+									ComponentType:   "frontend",
 									Replicas:        &[]int32{3}[0],
 									Resources: &common.Resources{
 										Requests: &common.ResourceItem{
@@ -107,7 +107,7 @@ func TestGenerateDynamoComponentsDeployments(t *testing.T) {
 						DynamoComponentDeploymentSharedSpec: v1alpha1.DynamoComponentDeploymentSharedSpec{
 							ServiceName:     "service1",
 							DynamoNamespace: &[]string{"default"}[0],
-							ComponentType:   "main",
+							ComponentType:   "frontend",
 							Replicas:        &[]int32{3}[0],
 							Resources: &common.Resources{
 								Requests: &common.ResourceItem{
@@ -171,7 +171,7 @@ func TestGenerateDynamoComponentsDeployments(t *testing.T) {
 							"service1": {
 								DynamoComponentDeploymentSharedSpec: v1alpha1.DynamoComponentDeploymentSharedSpec{
 									DynamoNamespace: nil,
-									ComponentType:   "main",
+									ComponentType:   "frontend",
 									Replicas:        &[]int32{3}[0],
 									Resources: &common.Resources{
 										Requests: &common.ResourceItem{
@@ -215,7 +215,7 @@ func TestGenerateDynamoComponentsDeployments(t *testing.T) {
 						DynamoComponentDeploymentSharedSpec: v1alpha1.DynamoComponentDeploymentSharedSpec{
 							ServiceName:     "service1",
 							DynamoNamespace: &[]string{"dynamo-test-dynamographdeployment"}[0],
-							ComponentType:   "main",
+							ComponentType:   "frontend",
 							Replicas:        &[]int32{3}[0],
 							Resources: &common.Resources{
 								Requests: &common.ResourceItem{
@@ -279,7 +279,7 @@ func TestGenerateDynamoComponentsDeployments(t *testing.T) {
 							"service1": {
 								DynamoComponentDeploymentSharedSpec: v1alpha1.DynamoComponentDeploymentSharedSpec{
 									DynamoNamespace: &[]string{"default"}[0],
-									ComponentType:   "main",
+									ComponentType:   "frontend",
 									Replicas:        &[]int32{3}[0],
 									Resources: &common.Resources{
 										Requests: &common.ResourceItem{
@@ -325,7 +325,7 @@ func TestGenerateDynamoComponentsDeployments(t *testing.T) {
 							"service1": {
 								DynamoComponentDeploymentSharedSpec: v1alpha1.DynamoComponentDeploymentSharedSpec{
 									DynamoNamespace: nil,
-									ComponentType:   "main",
+									ComponentType:   "frontend",
 									Replicas:        &[]int32{3}[0],
 									Resources: &common.Resources{
 										Requests: &common.ResourceItem{
@@ -373,7 +373,7 @@ func TestGenerateDynamoComponentsDeployments(t *testing.T) {
 						DynamoComponentDeploymentSharedSpec: v1alpha1.DynamoComponentDeploymentSharedSpec{
 							ServiceName:     "service1",
 							DynamoNamespace: &[]string{"dynamo-test-dynamographdeployment"}[0],
-							ComponentType:   "main",
+							ComponentType:   "frontend",
 							Replicas:        &[]int32{3}[0],
 							Resources: &common.Resources{
 								Requests: &common.ResourceItem{
@@ -447,7 +447,7 @@ func TestGenerateDynamoComponentsDeployments(t *testing.T) {
 							"service1": {
 								DynamoComponentDeploymentSharedSpec: v1alpha1.DynamoComponentDeploymentSharedSpec{
 									DynamoNamespace: nil,
-									ComponentType:   "main",
+									ComponentType:   "frontend",
 									Replicas:        &[]int32{3}[0],
 									Resources: &common.Resources{
 										Requests: &common.ResourceItem{
@@ -491,7 +491,7 @@ func TestGenerateDynamoComponentsDeployments(t *testing.T) {
 						DynamoComponentDeploymentSharedSpec: v1alpha1.DynamoComponentDeploymentSharedSpec{
 							ServiceName:     "service1",
 							DynamoNamespace: &[]string{"dynamo-test-dynamographdeployment"}[0],
-							ComponentType:   "main",
+							ComponentType:   "frontend",
 							Replicas:        &[]int32{3}[0],
 							Resources: &common.Resources{
 								Requests: &common.ResourceItem{
@@ -574,7 +574,7 @@ func TestGenerateDynamoComponentsDeployments(t *testing.T) {
 							"service1": {
 								DynamoComponentDeploymentSharedSpec: v1alpha1.DynamoComponentDeploymentSharedSpec{
 									DynamoNamespace: &[]string{"default"}[0],
-									ComponentType:   "main",
+									ComponentType:   "frontend",
 									Replicas:        &[]int32{3}[0],
 									Resources: &common.Resources{
 										Requests: &common.ResourceItem{
@@ -625,7 +625,7 @@ func TestGenerateDynamoComponentsDeployments(t *testing.T) {
 						DynamoComponentDeploymentSharedSpec: v1alpha1.DynamoComponentDeploymentSharedSpec{
 							ServiceName:     "service1",
 							DynamoNamespace: &[]string{"default"}[0],
-							ComponentType:   "main",
+							ComponentType:   "frontend",
 							Replicas:        &[]int32{3}[0],
 							Resources: &common.Resources{
 								Requests: &common.ResourceItem{
@@ -1121,6 +1121,15 @@ func Test_mergeEnvs(t *testing.T) {
 	}
 }
 
+func sortEnvVars(envs []corev1.EnvVar) []corev1.EnvVar {
+	sorted := make([]corev1.EnvVar, len(envs))
+	copy(sorted, envs)
+	sort.Slice(sorted, func(i, j int) bool {
+		return sorted[i].Name < sorted[j].Name
+	})
+	return sorted
+}
+
 func TestGenerateGrovePodGangSet(t *testing.T) {
 	type args struct {
 		ctx              context.Context
@@ -1159,7 +1168,7 @@ func TestGenerateGrovePodGangSet(t *testing.T) {
 						Services: map[string]*v1alpha1.DynamoComponentDeploymentOverridesSpec{
 							"Frontend": {
 								DynamoComponentDeploymentSharedSpec: v1alpha1.DynamoComponentDeploymentSharedSpec{
-									ComponentType: "main", // Frontend component
+									ComponentType: "frontend", // Frontend component
 									ExtraPodMetadata: &common.ExtraPodMetadata{
 										Annotations: map[string]string{
 											"nvidia.com/annotation1": "annotation1",
@@ -1308,7 +1317,7 @@ func TestGenerateGrovePodGangSet(t *testing.T) {
 								Labels: map[string]string{
 									commonconsts.KubeLabelDynamoSelector:      "test-dynamo-graph-deployment-frontend",
 									commonconsts.KubeLabelMetricsEnabled:      commonconsts.KubeLabelValueTrue,
-									commonconsts.KubeLabelDynamoComponentType: commonconsts.ComponentTypeMain,
+									commonconsts.KubeLabelDynamoComponentType: commonconsts.ComponentTypeFrontend,
 									"nvidia.com/label1":                       "label1",
 									"nvidia.com/label2":                       "label2",
 								},
@@ -1503,10 +1512,6 @@ func TestGenerateGrovePodGangSet(t *testing.T) {
 														Name:  "PLANNER_ENV_1",
 														Value: "2",
 													},
-													{
-														Name:  "DYNAMO_PORT",
-														Value: fmt.Sprintf("%d", commonconsts.DynamoServicePort),
-													},
 													{
 														Name:  "NATS_SERVER",
 														Value: "nats-address",
@@ -1537,13 +1542,6 @@ func TestGenerateGrovePodGangSet(t *testing.T) {
 														MountPath: "/dev/shm",
 													},
 												},
-												Ports: []corev1.ContainerPort{
-													{
-														Protocol:      corev1.ProtocolTCP,
-														Name:          commonconsts.DynamoContainerPortName,
-														ContainerPort: int32(commonconsts.DynamoServicePort),
-													},
-												},
 											},
 										},
 									},
@@ -1594,6 +1592,7 @@ func TestGenerateGrovePodGangSet(t *testing.T) {
 											GPU:    "1",
 										},
 									},
+									ComponentType: commonconsts.ComponentTypeFrontend,
 									Envs: []corev1.EnvVar{
 										{
 											Name:  "FRONTEND_ENV_1",
@@ -1813,11 +1812,6 @@ func TestGenerateGrovePodGangSet(t *testing.T) {
 													"python3 -m dynamo.sglang.worker --dist-init-addr ${GROVE_PCSG_NAME}-${GROVE_PCSG_INDEX}-worker-ldr-0.${GROVE_HEADLESS_SERVICE}:29500 --nnodes 3 --node-rank 0 --custom-flag custom-value",
 												},
 												Ports: []corev1.ContainerPort{
-													{
-														Protocol:      corev1.ProtocolTCP,
-														Name:          commonconsts.DynamoContainerPortName,
-														ContainerPort: int32(commonconsts.DynamoServicePort),
-													},
 													{
 														Protocol:      corev1.ProtocolTCP,
 														Name:          commonconsts.DynamoSystemPortName,
@@ -1830,12 +1824,20 @@ func TestGenerateGrovePodGangSet(t *testing.T) {
 														Value: "1",
 													},
 													{
-														Name:  "WORKER_ENV_1",
-														Value: "1",
+														Name:  "DYN_SYSTEM_ENABLED",
+														Value: "true",
 													},
 													{
-														Name:  "DYNAMO_PORT",
-														Value: fmt.Sprintf("%d", commonconsts.DynamoServicePort),
+														Name:  "DYN_SYSTEM_PORT",
+														Value: "9090",
+													},
+													{
+														Name:  "DYN_SYSTEM_USE_ENDPOINT_HEALTH_STATUS",
+														Value: `["generate"]`,
+													},
+													{
+														Name:  "WORKER_ENV_1",
+														Value: "1",
 													},
 													{
 														Name:  "NATS_SERVER",
@@ -1909,11 +1911,6 @@ func TestGenerateGrovePodGangSet(t *testing.T) {
 													"python3 -m dynamo.sglang.worker --dist-init-addr ${GROVE_PCSG_NAME}-${GROVE_PCSG_INDEX}-worker-ldr-0.${GROVE_HEADLESS_SERVICE}:29500 --nnodes 3 --node-rank $((GROVE_PCLQ_POD_INDEX + 1)) --custom-flag custom-value",
 												},
 												Ports: []corev1.ContainerPort{
-													{
-														Protocol:      corev1.ProtocolTCP,
-														Name:          commonconsts.DynamoContainerPortName,
-														ContainerPort: int32(commonconsts.DynamoServicePort),
-													},
 													{
 														Protocol:      corev1.ProtocolTCP,
 														Name:          commonconsts.DynamoSystemPortName,
@@ -1926,12 +1923,20 @@ func TestGenerateGrovePodGangSet(t *testing.T) {
 														Value: "1",
 													},
 													{
-														Name:  "WORKER_ENV_1",
-														Value: "1",
+														Name:  "DYN_SYSTEM_ENABLED",
+														Value: "true",
 													},
 													{
-														Name:  "DYNAMO_PORT",
-														Value: fmt.Sprintf("%d", commonconsts.DynamoServicePort),
+														Name:  "DYN_SYSTEM_PORT",
+														Value: "9090",
+													},
+													{
+														Name:  "DYN_SYSTEM_USE_ENDPOINT_HEALTH_STATUS",
+														Value: `["generate"]`,
+													},
+													{
+														Name:  "WORKER_ENV_1",
+														Value: "1",
 													},
 													{
 														Name:  "NATS_SERVER",
@@ -1967,8 +1972,9 @@ func TestGenerateGrovePodGangSet(t *testing.T) {
 							{
 								Name: "frontend",
 								Labels: map[string]string{
-									commonconsts.KubeLabelMetricsEnabled: commonconsts.KubeLabelValueTrue,
-									commonconsts.KubeLabelDynamoSelector: "test-dynamo-graph-deployment-frontend",
+									commonconsts.KubeLabelMetricsEnabled:      commonconsts.KubeLabelValueTrue,
+									commonconsts.KubeLabelDynamoSelector:      "test-dynamo-graph-deployment-frontend",
+									commonconsts.KubeLabelDynamoComponentType: commonconsts.ComponentTypeFrontend,
 								},
 								Annotations: map[string]string{},
 								Spec: grovev1alpha1.PodCliqueSpec{
@@ -2158,10 +2164,6 @@ func TestGenerateGrovePodGangSet(t *testing.T) {
 														Name:  "PLANNER_ENV_1",
 														Value: "2",
 													},
-													{
-														Name:  "DYNAMO_PORT",
-														Value: fmt.Sprintf("%d", commonconsts.DynamoServicePort),
-													},
 													{
 														Name:  "NATS_SERVER",
 														Value: "nats-address",
@@ -2192,13 +2194,6 @@ func TestGenerateGrovePodGangSet(t *testing.T) {
 														MountPath: "/dev/shm",
 													},
 												},
-												Ports: []corev1.ContainerPort{
-													{
-														Protocol:      corev1.ProtocolTCP,
-														Name:          commonconsts.DynamoContainerPortName,
-														ContainerPort: int32(commonconsts.DynamoServicePort),
-													},
-												},
 											},
 										},
 									},
@@ -2237,7 +2232,8 @@ func TestGenerateGrovePodGangSet(t *testing.T) {
 						Services: map[string]*v1alpha1.DynamoComponentDeploymentOverridesSpec{
 							"Frontend": {
 								DynamoComponentDeploymentSharedSpec: v1alpha1.DynamoComponentDeploymentSharedSpec{
-									Replicas: &[]int32{1}[0],
+									Replicas:      &[]int32{1}[0],
+									ComponentType: commonconsts.ComponentTypeFrontend,
 									Resources: &common.Resources{
 										Requests: &common.ResourceItem{
 											CPU:    "1",
@@ -2492,11 +2488,6 @@ func TestGenerateGrovePodGangSet(t *testing.T) {
 													"ray start --head --port=6379 && python3 -m dynamo.vllm --custom-flag custom-value",
 												},
 												Ports: []corev1.ContainerPort{
-													{
-														Protocol:      corev1.ProtocolTCP,
-														Name:          commonconsts.DynamoContainerPortName,
-														ContainerPort: int32(commonconsts.DynamoServicePort),
-													},
 													{
 														Protocol:      corev1.ProtocolTCP,
 														Name:          commonconsts.DynamoSystemPortName,
@@ -2509,12 +2500,20 @@ func TestGenerateGrovePodGangSet(t *testing.T) {
 														Value: "1",
 													},
 													{
-														Name:  "WORKER_ENV_1",
-														Value: "1",
+														Name:  "DYN_SYSTEM_ENABLED",
+														Value: "true",
 													},
 													{
-														Name:  "DYNAMO_PORT",
-														Value: fmt.Sprintf("%d", commonconsts.DynamoServicePort),
+														Name:  "DYN_SYSTEM_PORT",
+														Value: "9090",
+													},
+													{
+														Name:  "DYN_SYSTEM_USE_ENDPOINT_HEALTH_STATUS",
+														Value: `["generate"]`,
+													},
+													{
+														Name:  "WORKER_ENV_1",
+														Value: "1",
 													},
 													{
 														Name:  "NATS_SERVER",
@@ -2591,11 +2590,6 @@ func TestGenerateGrovePodGangSet(t *testing.T) {
 													"ray start --address=${GROVE_PCSG_NAME}-${GROVE_PCSG_INDEX}-worker-ldr-0.${GROVE_HEADLESS_SERVICE}:6379 --block",
 												},
 												Ports: []corev1.ContainerPort{
-													{
-														Protocol:      corev1.ProtocolTCP,
-														Name:          commonconsts.DynamoContainerPortName,
-														ContainerPort: int32(commonconsts.DynamoServicePort),
-													},
 													{
 														Protocol:      corev1.ProtocolTCP,
 														Name:          commonconsts.DynamoSystemPortName,
@@ -2608,12 +2602,20 @@ func TestGenerateGrovePodGangSet(t *testing.T) {
 														Value: "1",
 													},
 													{
-														Name:  "WORKER_ENV_1",
-														Value: "1",
+														Name:  "DYN_SYSTEM_ENABLED",
+														Value: "true",
 													},
 													{
-														Name:  "DYNAMO_PORT",
-														Value: fmt.Sprintf("%d", commonconsts.DynamoServicePort),
+														Name:  "DYN_SYSTEM_PORT",
+														Value: "9090",
+													},
+													{
+														Name:  "DYN_SYSTEM_USE_ENDPOINT_HEALTH_STATUS",
+														Value: `["generate"]`,
+													},
+													{
+														Name:  "WORKER_ENV_1",
+														Value: "1",
 													},
 													{
 														Name:  "NATS_SERVER",
@@ -2649,8 +2651,9 @@ func TestGenerateGrovePodGangSet(t *testing.T) {
 							{
 								Name: "frontend",
 								Labels: map[string]string{
-									commonconsts.KubeLabelMetricsEnabled: commonconsts.KubeLabelValueTrue,
-									commonconsts.KubeLabelDynamoSelector: "test-dynamo-graph-deployment-frontend",
+									commonconsts.KubeLabelDynamoComponentType: commonconsts.ComponentTypeFrontend,
+									commonconsts.KubeLabelMetricsEnabled:      commonconsts.KubeLabelValueTrue,
+									commonconsts.KubeLabelDynamoSelector:      "test-dynamo-graph-deployment-frontend",
 								},
 								Annotations: map[string]string{},
 								Spec: grovev1alpha1.PodCliqueSpec{
@@ -2840,10 +2843,6 @@ func TestGenerateGrovePodGangSet(t *testing.T) {
 														Name:  "PLANNER_ENV_1",
 														Value: "2",
 													},
-													{
-														Name:  "DYNAMO_PORT",
-														Value: fmt.Sprintf("%d", commonconsts.DynamoServicePort),
-													},
 													{
 														Name:  "NATS_SERVER",
 														Value: "nats-address",
@@ -2874,13 +2873,6 @@ func TestGenerateGrovePodGangSet(t *testing.T) {
 														MountPath: "/dev/shm",
 													},
 												},
-												Ports: []corev1.ContainerPort{
-													{
-														Protocol:      corev1.ProtocolTCP,
-														Name:          commonconsts.DynamoContainerPortName,
-														ContainerPort: int32(commonconsts.DynamoServicePort),
-													},
-												},
 											},
 										},
 									},
@@ -2906,6 +2898,19 @@ func TestGenerateGrovePodGangSet(t *testing.T) {
 			sort.Slice(tt.want.Spec.Template.Cliques, func(i, j int) bool {
 				return tt.want.Spec.Template.Cliques[i].Name < tt.want.Spec.Template.Cliques[j].Name
 			})
+
+			// Sort environment variables for all containers in all cliques
+			for _, clique := range got.Spec.Template.Cliques {
+				for i := range clique.Spec.PodSpec.Containers {
+					clique.Spec.PodSpec.Containers[i].Env = sortEnvVars(clique.Spec.PodSpec.Containers[i].Env)
+				}
+			}
+			for _, clique := range tt.want.Spec.Template.Cliques {
+				for i := range clique.Spec.PodSpec.Containers {
+					clique.Spec.PodSpec.Containers[i].Env = sortEnvVars(clique.Spec.PodSpec.Containers[i].Env)
+				}
+			}
+
 			if diff := cmp.Diff(got, tt.want); diff != "" {
 				t.Errorf("GenerateGrovePodGangSet() mismatch (-want +got):\n%s", diff)
 			}
@@ -3018,31 +3023,6 @@ func TestGeneratePodSpecForComponent_SGLang(t *testing.T) {
 			expectError:      false,
 			expectContains:   []string{},
 		},
-		{
-			name: "SGLang with resources",
-			component: &v1alpha1.DynamoComponentDeploymentOverridesSpec{
-				DynamoComponentDeploymentSharedSpec: v1alpha1.DynamoComponentDeploymentSharedSpec{
-					ComponentType: commonconsts.ComponentTypeWorker,
-
-					ExtraPodSpec: &common.ExtraPodSpec{
-						MainContainer: &corev1.Container{
-							Args: []string{"python3", "-m", "dynamo.sglang.worker"},
-							Resources: corev1.ResourceRequirements{
-								Requests: corev1.ResourceList{
-									corev1.ResourceCPU:    resource.MustParse("1"),
-									corev1.ResourceMemory: resource.MustParse("2Gi"),
-								},
-							},
-						},
-					},
-				},
-			},
-			backendFramework: BackendFrameworkSGLang,
-			role:             RoleMain,
-			numberOfNodes:    1,
-			expectError:      false,
-			expectContains:   []string{"python3 -m dynamo.sglang.worker"},
-		},
 	}
 
 	for _, tt := range tests {
@@ -3574,7 +3554,7 @@ func TestDetermineBackendFramework(t *testing.T) {
 	}{
 		{
 			name:          "non-worker component returns noop",
-			componentType: "main",
+			componentType: "frontend",
 			command:       []string{"/bin/sh", "-c"},
 			args:          []string{"echo hello world"},
 			expected:      BackendFrameworkNoop,
@@ -3735,7 +3715,7 @@ func TestGetBackendFrameworkFromComponent(t *testing.T) {
 			name: "non-worker component returns noop",
 			component: &v1alpha1.DynamoComponentDeploymentOverridesSpec{
 				DynamoComponentDeploymentSharedSpec: v1alpha1.DynamoComponentDeploymentSharedSpec{
-					ComponentType: "main", // Frontend component
+					ComponentType: "frontend", // Frontend component
 				},
 			},
 			deployment: &v1alpha1.DynamoGraphDeployment{},
@@ -4145,3 +4125,59 @@ func TestGenerateGrovePodGangSet_StartsAfterDependencies(t *testing.T) {
 		})
 	}
 }
+
+func TestGenerateBasePodSpec_PlannerServiceAccount(t *testing.T) {
+	secretsRetriever := &mockSecretsRetriever{}
+	controllerConfig := controller_common.Config{}
+
+	tests := []struct {
+		name               string
+		component          *v1alpha1.DynamoComponentDeploymentOverridesSpec
+		expectedServiceAcc string
+	}{
+		{
+			name: "Planner component should have planner service account",
+			component: &v1alpha1.DynamoComponentDeploymentOverridesSpec{
+				DynamoComponentDeploymentSharedSpec: v1alpha1.DynamoComponentDeploymentSharedSpec{
+					ComponentType: commonconsts.ComponentTypePlanner,
+				},
+			},
+			expectedServiceAcc: commonconsts.PlannerServiceAccountName,
+		},
+		{
+			name: "Planner service account should not be set for non-planner components",
+			component: &v1alpha1.DynamoComponentDeploymentOverridesSpec{
+				DynamoComponentDeploymentSharedSpec: v1alpha1.DynamoComponentDeploymentSharedSpec{
+					ComponentType: commonconsts.ComponentTypeWorker,
+				},
+			},
+			expectedServiceAcc: "",
+		},
+	}
+
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			podSpec, err := GenerateBasePodSpec(
+				tt.component,
+				BackendFrameworkSGLang,
+				secretsRetriever,
+				"default",
+				RoleMain,
+				1,
+				controllerConfig,
+				commonconsts.MultinodeDeploymentTypeGrove,
+				"test-service",
+			)
+
+			if err != nil {
+				t.Errorf("GenerateBasePodSpec() error = %v", err)
+				return
+			}
+
+			if podSpec.ServiceAccountName != tt.expectedServiceAcc {
+				t.Errorf("GenerateBasePodSpec() serviceAccountName = %v, want %v",
+					podSpec.ServiceAccountName, tt.expectedServiceAcc)
+			}
+		})
+	}
+}

examples/runtime/hello_world/deploy/hello_world.yaml

@@ -27,7 +27,7 @@ spec:
         timeoutSeconds: 2
         failureThreshold: 3
       dynamoNamespace: hello-world
-      componentType: main
+      componentType: frontend
       replicas: 1
       resources:
         requests: