feat: improve CRD documentation (#3504)

Signed-off-by: Julien Mancuso <jmancuso@nvidia.com>

deploy/cloud/operator/Makefile

@@ -287,10 +287,10 @@ generate-api-docs: crd-ref-docs ## Generate API reference documentation from CRD
 		--renderer=markdown \
 		--output-path=./docs/api_reference.md
 	@echo "✅ Generated API reference at ./docs/api_reference.md"
-	# concatenate header.md and api_reference.md
-	cat docs/header.md ./docs/api_reference.md > ../../../docs/kubernetes/api_reference.md
+	# concatenate header.md, api_reference.md, and footer.md
+	cat docs/header.md ./docs/api_reference.md docs/footer.md > ../../../docs/kubernetes/api_reference.md
 	rm ./docs/api_reference.md
-	@echo "✅ Concatenated header.md and api_reference.md"
+	@echo "✅ Concatenated header.md, api_reference.md, and footer.md"
 
 .PHONY: coverage
 coverage: test

deploy/cloud/operator/api/v1alpha1/dynamocomponentdeployment_types.go

@@ -20,8 +20,6 @@
 package v1alpha1
 
 import (
-	"strings"
-
 	dynamoCommon "github.com/ai-dynamo/dynamo/deploy/cloud/operator/api/dynamo/common"
 	commonconsts "github.com/ai-dynamo/dynamo/deploy/cloud/operator/internal/consts"
 	corev1 "k8s.io/api/core/v1"
@@ -38,12 +36,6 @@ const (
 
 // DynamoComponentDeploymentSpec defines the desired state of DynamoComponentDeployment
 type DynamoComponentDeploymentSpec struct {
-	// DynamoComponent selects the Dynamo component from the archive to deploy.
-	// Typically corresponds to a component defined in the packaged Dynamo artifacts.
-	DynamoComponent string `json:"dynamoComponent,omitempty"`
-	// contains the tag of the DynamoComponent: for example, "my_package:MyService"
-	DynamoTag string `json:"dynamoTag,omitempty"`
-
 	// BackendFramework specifies the backend framework (e.g., "sglang", "vllm", "trtllm")
 	// +kubebuilder:validation:Enum=sglang;vllm;trtllm
 	BackendFramework string `json:"backendFramework,omitempty"`
@@ -53,10 +45,6 @@ type DynamoComponentDeploymentSpec struct {
 	DynamoComponentDeploymentSharedSpec `json:",inline"`
 }
 
-type DynamoComponentDeploymentOverridesSpec struct {
-	DynamoComponentDeploymentSharedSpec `json:",inline"`
-}
-
 type DynamoComponentDeploymentSharedSpec struct {
 	// INSERT ADDITIONAL SPEC FIELDS - desired state of cluster
 	// Important: Run "make" to regenerate code after modifying this file
@@ -229,7 +217,7 @@ func (s *DynamoComponentDeployment) SetSpec(spec any) {
 }
 
 func (s *DynamoComponentDeployment) IsFrontendComponent() bool {
-	return strings.HasSuffix(s.Spec.DynamoTag, s.Spec.ServiceName) || s.Spec.ComponentType == commonconsts.ComponentTypeFrontend
+	return s.Spec.ComponentType == commonconsts.ComponentTypeFrontend
 }
 
 func (s *DynamoComponentDeployment) GetDynamoDeploymentConfig() []byte {

deploy/cloud/operator/api/v1alpha1/dynamocomponentdeployment_types_test.go

@@ -46,9 +46,8 @@ func TestDynamoComponentDeployment_IsFrontendComponent(t *testing.T) {
 			name: "main component",
 			fields: fields{
 				Spec: DynamoComponentDeploymentSpec{
-					DynamoTag: "dynamo-component:main",
 					DynamoComponentDeploymentSharedSpec: DynamoComponentDeploymentSharedSpec{
-						ServiceName: "main",
+						ComponentType: commonconsts.ComponentTypeFrontend,
 					},
 				},
 			},
@@ -58,9 +57,8 @@ func TestDynamoComponentDeployment_IsFrontendComponent(t *testing.T) {
 			name: "not main component",
 			fields: fields{
 				Spec: DynamoComponentDeploymentSpec{
-					DynamoTag: "dynamo-component:main",
 					DynamoComponentDeploymentSharedSpec: DynamoComponentDeploymentSharedSpec{
-						ServiceName: "not-main",
+						ComponentType: commonconsts.ComponentTypeWorker,
 					},
 				},
 			},

deploy/cloud/operator/api/v1alpha1/dynamographdeployment_types.go

@@ -29,20 +29,14 @@ import (
 
 // DynamoGraphDeploymentSpec defines the desired state of DynamoGraphDeployment.
 type DynamoGraphDeploymentSpec struct {
-	// DynamoGraph selects the graph (workflow/topology) to deploy. This must match
-	// a graph name packaged with the Dynamo archive.
-	DynamoGraph string `json:"dynamoGraph,omitempty"`
 	// PVCs defines a list of persistent volume claims that can be referenced by components.
 	// Each PVC must have a unique name that can be referenced in component specifications.
 	// +kubebuilder:validation:Optional
 	PVCs []PVC `json:"pvcs,omitempty"`
-	// Services allows per-service overrides of the component deployment settings.
-	// - key: name of the service defined by the DynamoComponent
-	// - value: overrides for that service
-	// If not set for a service, the default DynamoComponentDeployment values are used.
+	// Services are the services to deploy as part of this deployment.
 	// +kubebuilder:validation:Optional
-	Services map[string]*DynamoComponentDeploymentOverridesSpec `json:"services,omitempty"`
-	// Envs are environment variables applied to all services in the graph unless
+	Services map[string]*DynamoComponentDeploymentSharedSpec `json:"services,omitempty"`
+	// Envs are environment variables applied to all services in the deployment unless
 	// overridden by service-specific configuration.
 	// +kubebuilder:validation:Optional
 	Envs []corev1.EnvVar `json:"envs,omitempty"`

deploy/cloud/operator/api/v1alpha1/zz_generated.deepcopy.go

@@ -173,22 +173,6 @@ func (in *DynamoComponentDeploymentList) DeepCopyObject() runtime.Object {
 	return nil
 }
 
-// DeepCopyInto is an autogenerated deepcopy function, copying the receiver, writing into out. in must be non-nil.
-func (in *DynamoComponentDeploymentOverridesSpec) DeepCopyInto(out *DynamoComponentDeploymentOverridesSpec) {
-	*out = *in
-	in.DynamoComponentDeploymentSharedSpec.DeepCopyInto(&out.DynamoComponentDeploymentSharedSpec)
-}
-
-// DeepCopy is an autogenerated deepcopy function, copying the receiver, creating a new DynamoComponentDeploymentOverridesSpec.
-func (in *DynamoComponentDeploymentOverridesSpec) DeepCopy() *DynamoComponentDeploymentOverridesSpec {
-	if in == nil {
-		return nil
-	}
-	out := new(DynamoComponentDeploymentOverridesSpec)
-	in.DeepCopyInto(out)
-	return out
-}
-
 // DeepCopyInto is an autogenerated deepcopy function, copying the receiver, writing into out. in must be non-nil.
 func (in *DynamoComponentDeploymentSharedSpec) DeepCopyInto(out *DynamoComponentDeploymentSharedSpec) {
 	*out = *in
@@ -406,15 +390,15 @@ func (in *DynamoGraphDeploymentSpec) DeepCopyInto(out *DynamoGraphDeploymentSpec
 	}
 	if in.Services != nil {
 		in, out := &in.Services, &out.Services
-		*out = make(map[string]*DynamoComponentDeploymentOverridesSpec, len(*in))
+		*out = make(map[string]*DynamoComponentDeploymentSharedSpec, len(*in))
 		for key, val := range *in {
-			var outVal *DynamoComponentDeploymentOverridesSpec
+			var outVal *DynamoComponentDeploymentSharedSpec
 			if val == nil {
 				(*out)[key] = nil
 			} else {
 				inVal := (*in)[key]
 				in, out := &inVal, &outVal
-				*out = new(DynamoComponentDeploymentOverridesSpec)
+				*out = new(DynamoComponentDeploymentSharedSpec)
 				(*in).DeepCopyInto(*out)
 			}
 			(*out)[key] = outVal

deploy/cloud/operator/docs/crd-ref-docs-config.yaml

@@ -32,7 +32,7 @@ processor:
     - "ParseError$"
     # Ignore only the override wrapper type to reduce repetition
     # Keep SharedSpec so embedded fields are documented once
-    - "DynamoComponentDeploymentOverridesSpec$"
+    # - "DynamoComponentDeploymentOverridesSpec$"
     - "DynamoComponentDeploymentStatus$"
     - "BaseStatus$"
 

deploy/cloud/operator/docs/footer.md

+# Operator Default Values Injection
+
+The Dynamo operator automatically applies default values to various fields when they are not explicitly specified in your deployments. These defaults include:
+
+- **Health Probes**: Startup, liveness, and readiness probes are configured differently for frontend, worker, and planner components. For example, worker components receive a startup probe with a 2-hour timeout (720 failures × 10 seconds) to accommodate long model loading times.
+
+- **Shared Memory**: All components receive an 8Gi shared memory volume mounted at `/dev/shm` by default (can be disabled or resized via the `sharedMemory` field).
+
+- **Environment Variables**: Components automatically receive environment variables like `DYN_NAMESPACE`, `DYN_PARENT_DGD_K8S_NAME`, `DYNAMO_PORT`, and backend-specific variables.
+
+- **Pod Configuration**: Default `terminationGracePeriodSeconds` of 60 seconds and `restartPolicy: Always`.
+
+- **Autoscaling**: When enabled without explicit metrics, defaults to CPU-based autoscaling with 80% target utilization.
+
+- **Backend-Specific Behavior**: For multinode deployments, probes are automatically modified or removed for worker nodes depending on the backend framework (VLLM, SGLang, or TensorRT-LLM).
+
+## Pod Specification Defaults
+
+All components receive the following pod-level defaults unless overridden:
+
+- **`terminationGracePeriodSeconds`**: `60` seconds
+- **`restartPolicy`**: `Always`
+
+## Shared Memory Configuration
+
+Shared memory is enabled by default for all components:
+
+- **Enabled**: `true` (unless explicitly disabled via `sharedMemory.disabled`)
+- **Size**: `8Gi`
+- **Mount Path**: `/dev/shm`
+- **Volume Type**: `emptyDir` with `memory` medium
+
+To disable shared memory or customize the size, use the `sharedMemory` field in your component specification.
+
+## Health Probes by Component Type
+
+The operator applies different default health probes based on the component type.
+
+### Frontend Components
+
+Frontend components receive the following probe configurations:
+
+**Liveness Probe:**
+- **Type**: HTTP GET
+- **Path**: `/health`
+- **Port**: `http` (8000)
+- **Initial Delay**: 60 seconds
+- **Period**: 60 seconds
+- **Timeout**: 30 seconds
+- **Failure Threshold**: 10
+
+**Readiness Probe:**
+- **Type**: Exec command
+- **Command**: `curl -s http://localhost:${DYNAMO_PORT}/health | jq -e ".status == \"healthy\""`
+- **Initial Delay**: 60 seconds
+- **Period**: 60 seconds
+- **Timeout**: 30 seconds
+- **Failure Threshold**: 10
+
+### Worker Components
+
+Worker components receive the following probe configurations:
+
+**Liveness Probe:**
+- **Type**: HTTP GET
+- **Path**: `/live`
+- **Port**: `system` (9090)
+- **Period**: 5 seconds
+- **Timeout**: 30 seconds
+- **Failure Threshold**: 1
+
+**Readiness Probe:**
+- **Type**: HTTP GET
+- **Path**: `/health`
+- **Port**: `system` (9090)
+- **Period**: 10 seconds
+- **Timeout**: 30 seconds
+- **Failure Threshold**: 60
+
+**Startup Probe:**
+- **Type**: HTTP GET
+- **Path**: `/live`
+- **Port**: `system` (9090)
+- **Period**: 10 seconds
+- **Timeout**: 5 seconds
+- **Failure Threshold**: 720 (allows up to 2 hours for startup: 10s × 720 = 7200s)
+
+:::{note}
+For larger models (typically >70B parameters) or slower storage systems, you may need to increase the `failureThreshold` to allow more time for model loading. Calculate the required threshold based on your expected startup time: `failureThreshold = (expected_startup_seconds / period)`. Override the startup probe in your component specification if the default 2-hour window is insufficient.
+:::
+
+### Multinode Deployment Probe Modifications
+
+For multinode deployments, the operator modifies probes based on the backend framework and node role:
+
+#### VLLM Backend
+- **Worker nodes**: All probes (liveness, readiness, startup) are removed
+
+#### SGLang Backend
+- **Worker nodes**: All probes (liveness, readiness, startup) are removed
+
+#### TensorRT-LLM Backend
+- **Leader nodes**: All probes remain unchanged
+- **Worker nodes**:
+  - Liveness and startup probes are removed
+  - Readiness probe is replaced with a TCP socket check on SSH port (2222):
+    - **Initial Delay**: 20 seconds
+    - **Period**: 20 seconds
+    - **Timeout**: 5 seconds
+    - **Failure Threshold**: 10
+
+## Environment Variables
+
+The operator automatically injects environment variables based on component type and configuration:
+
+### All Components
+
+- **`DYN_NAMESPACE`**: The Dynamo namespace for the component
+- **`DYN_PARENT_DGD_K8S_NAME`**: The parent DynamoGraphDeployment Kubernetes resource name
+- **`DYN_PARENT_DGD_K8S_NAMESPACE`**: The parent DynamoGraphDeployment Kubernetes namespace
+
+### Frontend Components
+
+- **`DYNAMO_PORT`**: `8000`
+- **`DYN_HTTP_PORT`**: `8000`
+
+### Worker Components
+
+- **`DYN_SYSTEM_ENABLED`**: `true`
+- **`DYN_SYSTEM_USE_ENDPOINT_HEALTH_STATUS`**: `["generate"]`
+- **`DYN_SYSTEM_PORT`**: `9090`
+
+### Planner Components
+
+- **`PLANNER_PROMETHEUS_PORT`**: `9085`
+
+### VLLM Backend (with compilation cache)
+
+When a volume mount is configured with `useAsCompilationCache: true`:
+- **`VLLM_CACHE_ROOT`**: Set to the mount point of the cache volume
+
+## Service Account
+
+Planner components automatically receive the following service account:
+
+- **`serviceAccountName`**: `planner-serviceaccount`
+
+## Image Pull Secrets
+
+The operator automatically discovers and injects image pull secrets for container images. When a component specifies a container image, the operator:
+
+1. Scans all Kubernetes secrets of type `kubernetes.io/dockerconfigjson` in the component's namespace
+2. Extracts the docker registry server URLs from each secret's authentication configuration
+3. Matches the container image's registry host against the discovered registry URLs
+4. Automatically injects matching secrets as `imagePullSecrets` in the pod specification
+
+This eliminates the need to manually specify image pull secrets for each component. The operator maintains an internal index of docker secrets and their associated registries, refreshing this index periodically.
+
+**To disable automatic image pull secret discovery** for a specific component, add the following annotation:
+
+```yaml
+annotations:
+  nvidia.com/disable-image-pull-secret-discovery: "true"
+```
+
+## Autoscaling Defaults
+
+When autoscaling is enabled but no metrics are specified, the operator applies:
+
+- **Default Metric**: CPU utilization
+- **Target Average Utilization**: `80%`
+
+## Port Configurations
+
+Default container ports are configured based on component type:
+
+### Frontend Components
+- **Port**: 8000
+- **Protocol**: TCP
+- **Name**: `http`
+
+### Worker Components
+- **Port**: 9090
+- **Protocol**: TCP
+- **Name**: `system`
+
+### Planner Components
+- **Port**: 9085
+- **Protocol**: TCP
+- **Name**: `metrics`
+
+## Backend-Specific Configurations
+
+### VLLM
+- **Ray Head Port**: 6379 (for multinode deployments)
+
+### SGLang
+- **Distribution Init Port**: 29500 (for multinode deployments)
+
+### TensorRT-LLM
+- **SSH Port**: 2222 (for multinode MPI communication)
+- **OpenMPI Environment**: `OMPI_MCA_orte_keep_fqdn_hostnames=1`
+
+## Implementation Reference
+
+For users who want to understand the implementation details or contribute to the operator, the default values described in this document are set in the following source files:
+
+- **Health Probes & Pod Specifications**: [`internal/dynamo/graph.go`](https://github.com/ai-dynamo/dynamo/blob/main/deploy/cloud/operator/internal/dynamo/graph.go) - Contains the main logic for applying default probes, environment variables, shared memory, and pod configurations
+- **Component-Specific Defaults**:
+  - [`internal/dynamo/component_frontend.go`](https://github.com/ai-dynamo/dynamo/blob/main/deploy/cloud/operator/internal/dynamo/component_frontend.go)
+  - [`internal/dynamo/component_worker.go`](https://github.com/ai-dynamo/dynamo/blob/main/deploy/cloud/operator/internal/dynamo/component_worker.go)
+  - [`internal/dynamo/component_planner.go`](https://github.com/ai-dynamo/dynamo/blob/main/deploy/cloud/operator/internal/dynamo/component_planner.go)
+- **Image Pull Secrets**: [`internal/secrets/docker.go`](https://github.com/ai-dynamo/dynamo/blob/main/deploy/cloud/operator/internal/secrets/docker.go) - Implements the docker secret indexer and automatic discovery
+- **Backend-Specific Behavior**:
+  - [`internal/dynamo/backend_vllm.go`](https://github.com/ai-dynamo/dynamo/blob/main/deploy/cloud/operator/internal/dynamo/backend_vllm.go)
+  - [`internal/dynamo/backend_sglang.go`](https://github.com/ai-dynamo/dynamo/blob/main/deploy/cloud/operator/internal/dynamo/backend_sglang.go)
+  - [`internal/dynamo/backend_trtllm.go`](https://github.com/ai-dynamo/dynamo/blob/main/deploy/cloud/operator/internal/dynamo/backend_trtllm.go)
+- **Constants & Annotations**: [`internal/consts/consts.go`](https://github.com/ai-dynamo/dynamo/blob/main/deploy/cloud/operator/internal/consts/consts.go) - Defines annotation keys and other constants
+
+## Notes
+
+- All these defaults can be overridden by explicitly specifying values in your DynamoComponentDeployment or DynamoGraphDeployment resources
+- User-specified probes (via `livenessProbe`, `readinessProbe`, or `startupProbe` fields) take precedence over operator defaults
+- For multinode deployments, some defaults are modified or removed as described above to accommodate distributed execution patterns
+- The `extraPodSpec.mainContainer` field can be used to override probe configurations set by the operator
\ No newline at end of file

deploy/cloud/operator/internal/controller/dynamocomponentdeployment_controller_test.go

@@ -688,8 +688,6 @@ func TestDynamoComponentDeploymentReconciler_generateLeaderWorkerSet(t *testing.
 							},
 						},
 						Spec: v1alpha1.DynamoComponentDeploymentSpec{
-							DynamoComponent:  "test-lws-component",
-							DynamoTag:        "test-tag",
 							BackendFramework: string(dynamo.BackendFrameworkVLLM),
 							DynamoComponentDeploymentSharedSpec: v1alpha1.DynamoComponentDeploymentSharedSpec{
 								Envs: []corev1.EnvVar{
@@ -990,7 +988,6 @@ func TestDynamoComponentDeploymentReconciler_generateLeaderWorkerSet(t *testing.
 					dynamoComponentDeployment: &v1alpha1.DynamoComponentDeployment{
 						ObjectMeta: metav1.ObjectMeta{Name: "test-lws-nil-id", Namespace: "default"},
 						Spec: v1alpha1.DynamoComponentDeploymentSpec{
-							DynamoComponent: "test-comp", DynamoTag: "test",
 							DynamoComponentDeploymentSharedSpec: v1alpha1.DynamoComponentDeploymentSharedSpec{
 								Multinode: &v1alpha1.MultinodeSpec{
 									NodeCount: 2,
@@ -1034,7 +1031,6 @@ func TestDynamoComponentDeploymentReconciler_generateLeaderWorkerSet(t *testing.
 					dynamoComponentDeployment: &v1alpha1.DynamoComponentDeployment{
 						ObjectMeta: metav1.ObjectMeta{Name: "test-lws-leader-err", Namespace: "default"},
 						Spec: v1alpha1.DynamoComponentDeploymentSpec{
-							DynamoComponent: "test-comp", DynamoTag: "test",
 							DynamoComponentDeploymentSharedSpec: v1alpha1.DynamoComponentDeploymentSharedSpec{
 								Multinode: &v1alpha1.MultinodeSpec{
 									NodeCount: 2,

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

@@ -27,7 +27,7 @@ func isPythonCommand(cmd string) bool {
 	return matched
 }
 
-func (b *SGLangBackend) UpdateContainer(container *corev1.Container, numberOfNodes int32, role Role, component *v1alpha1.DynamoComponentDeploymentOverridesSpec, serviceName string, multinodeDeployer MultinodeDeployer) {
+func (b *SGLangBackend) UpdateContainer(container *corev1.Container, numberOfNodes int32, role Role, component *v1alpha1.DynamoComponentDeploymentSharedSpec, serviceName string, multinodeDeployer MultinodeDeployer) {
 	// Check for volumeMounts with useAsCompilationCache=true
 	for _, volumeMount := range component.VolumeMounts {
 		if volumeMount.UseAsCompilationCache {
@@ -112,7 +112,7 @@ func (b *SGLangBackend) UpdateContainer(container *corev1.Container, numberOfNod
 	}
 }
 
-func (b *SGLangBackend) UpdatePodSpec(podSpec *corev1.PodSpec, numberOfNodes int32, role Role, component *v1alpha1.DynamoComponentDeploymentOverridesSpec, serviceName string) {
+func (b *SGLangBackend) UpdatePodSpec(podSpec *corev1.PodSpec, numberOfNodes int32, role Role, component *v1alpha1.DynamoComponentDeploymentSharedSpec, serviceName string) {
 	// do nothing
 }
 

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

@@ -192,7 +192,7 @@ func TestSGLangBackend_PythonCommandInjection(t *testing.T) {
 				Args:    append([]string{}, tt.initialArgs...),
 			}
 
-			backend.UpdateContainer(container, tt.numberOfNodes, tt.role, &v1alpha1.DynamoComponentDeploymentOverridesSpec{}, "test-service", tt.multinodeDeployer)
+			backend.UpdateContainer(container, tt.numberOfNodes, tt.role, &v1alpha1.DynamoComponentDeploymentSharedSpec{}, "test-service", tt.multinodeDeployer)
 
 			if !reflect.DeepEqual(container.Command, tt.expectedCommand) {
 				t.Errorf("UpdateContainer() command = %v, want %v", container.Command, tt.expectedCommand)
@@ -317,7 +317,7 @@ func TestSGLangBackend_ShellCommandInjection(t *testing.T) {
 				Args:    append([]string{}, tt.initialArgs...),
 			}
 
-			backend.UpdateContainer(container, tt.numberOfNodes, tt.role, &v1alpha1.DynamoComponentDeploymentOverridesSpec{}, "test-service", tt.multinodeDeployer)
+			backend.UpdateContainer(container, tt.numberOfNodes, tt.role, &v1alpha1.DynamoComponentDeploymentSharedSpec{}, "test-service", tt.multinodeDeployer)
 
 			if !reflect.DeepEqual(container.Args, tt.expectedArgs) {
 				t.Errorf("UpdateContainer() args = %v, want %v", container.Args, tt.expectedArgs)
@@ -491,7 +491,7 @@ func TestSGLangBackend_ProbeRemoval(t *testing.T) {
 				StartupProbe:   startupProbe,
 			}
 
-			backend.UpdateContainer(container, tt.numberOfNodes, tt.role, &v1alpha1.DynamoComponentDeploymentOverridesSpec{}, "test-service", tt.multinodeDeployer)
+			backend.UpdateContainer(container, tt.numberOfNodes, tt.role, &v1alpha1.DynamoComponentDeploymentSharedSpec{}, "test-service", tt.multinodeDeployer)
 
 			if tt.expectProbesRemoved {
 				if container.LivenessProbe != nil {
@@ -523,15 +523,14 @@ func TestSGLangBackend_UpdateContainer_UseAsCompilationCache(t *testing.T) {
 
 	tests := []struct {
 		name                       string
-		component                  *v1alpha1.DynamoComponentDeploymentOverridesSpec
+		component                  *v1alpha1.DynamoComponentDeploymentSharedSpec
 		volumeMounts               []corev1.VolumeMount
 		expectNoEnvVarChanges      bool
 		expectLoggedPartialSupport bool
 	}{
 		{
 			name: "SGLang backend with useAsCompilationCache volume mount",
-			component: &v1alpha1.DynamoComponentDeploymentOverridesSpec{
-				DynamoComponentDeploymentSharedSpec: v1alpha1.DynamoComponentDeploymentSharedSpec{
+			component: &v1alpha1.DynamoComponentDeploymentSharedSpec{
 				VolumeMounts: []v1alpha1.VolumeMount{
 					{
 						Name:                  "sglang-cache",
@@ -540,15 +539,13 @@ func TestSGLangBackend_UpdateContainer_UseAsCompilationCache(t *testing.T) {
 					},
 				},
 			},
-			},
 			volumeMounts:               []corev1.VolumeMount{},
 			expectNoEnvVarChanges:      true, // SGLang doesn't set env vars yet
 			expectLoggedPartialSupport: true,
 		},
 		{
 			name: "SGLang backend with useAsCompilationCache at custom volume mount",
-			component: &v1alpha1.DynamoComponentDeploymentOverridesSpec{
-				DynamoComponentDeploymentSharedSpec: v1alpha1.DynamoComponentDeploymentSharedSpec{
+			component: &v1alpha1.DynamoComponentDeploymentSharedSpec{
 				VolumeMounts: []v1alpha1.VolumeMount{
 					{
 						Name:                  "custom-cache",
@@ -557,15 +554,13 @@ func TestSGLangBackend_UpdateContainer_UseAsCompilationCache(t *testing.T) {
 					},
 				},
 			},
-			},
 			volumeMounts:               []corev1.VolumeMount{},
 			expectNoEnvVarChanges:      true, // SGLang doesn't set env vars yet
 			expectLoggedPartialSupport: true,
 		},
 		{
 			name: "SGLang backend without useAsCompilationCache volume mount",
-			component: &v1alpha1.DynamoComponentDeploymentOverridesSpec{
-				DynamoComponentDeploymentSharedSpec: v1alpha1.DynamoComponentDeploymentSharedSpec{
+			component: &v1alpha1.DynamoComponentDeploymentSharedSpec{
 				VolumeMounts: []v1alpha1.VolumeMount{
 					{
 						Name:       "regular-volume",
@@ -573,18 +568,15 @@ func TestSGLangBackend_UpdateContainer_UseAsCompilationCache(t *testing.T) {
 					},
 				},
 			},
-			},
 			volumeMounts:               []corev1.VolumeMount{},
 			expectNoEnvVarChanges:      true,
 			expectLoggedPartialSupport: false,
 		},
 		{
 			name: "SGLang backend with no volume mounts",
-			component: &v1alpha1.DynamoComponentDeploymentOverridesSpec{
-				DynamoComponentDeploymentSharedSpec: v1alpha1.DynamoComponentDeploymentSharedSpec{
+			component: &v1alpha1.DynamoComponentDeploymentSharedSpec{
 				VolumeMounts: nil,
 			},
-			},
 			volumeMounts:               []corev1.VolumeMount{},
 			expectNoEnvVarChanges:      true,
 			expectLoggedPartialSupport: false,

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

@@ -18,7 +18,7 @@ type TRTLLMBackend struct {
 	MpiRunSecretName string
 }
 
-func (b *TRTLLMBackend) UpdateContainer(container *corev1.Container, numberOfNodes int32, role Role, component *v1alpha1.DynamoComponentDeploymentOverridesSpec, serviceName string, multinodeDeployer MultinodeDeployer) {
+func (b *TRTLLMBackend) UpdateContainer(container *corev1.Container, numberOfNodes int32, role Role, component *v1alpha1.DynamoComponentDeploymentSharedSpec, serviceName string, multinodeDeployer MultinodeDeployer) {
 	// Check for volumeMounts with useAsCompilationCache=true
 	for _, volumeMount := range component.VolumeMounts {
 		if volumeMount.UseAsCompilationCache {
@@ -75,7 +75,7 @@ func (b *TRTLLMBackend) UpdateContainer(container *corev1.Container, numberOfNod
 	}
 }
 
-func (b *TRTLLMBackend) UpdatePodSpec(podSpec *corev1.PodSpec, numberOfNodes int32, role Role, component *v1alpha1.DynamoComponentDeploymentOverridesSpec, serviceName string) {
+func (b *TRTLLMBackend) UpdatePodSpec(podSpec *corev1.PodSpec, numberOfNodes int32, role Role, component *v1alpha1.DynamoComponentDeploymentSharedSpec, serviceName string) {
 	// Add SSH keypair volume for TRTLLM multinode deployments
 	if numberOfNodes > 1 {
 		sshVolume := corev1.Volume{
@@ -102,7 +102,7 @@ func (b *TRTLLMBackend) addSSHVolumeMount(container *corev1.Container) {
 }
 
 // setupLeaderContainer configures the leader node with SSH setup and mpirun command
-func (b *TRTLLMBackend) setupLeaderContainer(container *corev1.Container, numberOfNodes int32, serviceName string, component *v1alpha1.DynamoComponentDeploymentOverridesSpec, multinodeDeployer MultinodeDeployer) {
+func (b *TRTLLMBackend) setupLeaderContainer(container *corev1.Container, numberOfNodes int32, serviceName string, component *v1alpha1.DynamoComponentDeploymentSharedSpec, multinodeDeployer MultinodeDeployer) {
 	// Generate the list of worker hostnames
 	workerHosts := b.generateWorkerHostnames(numberOfNodes, serviceName, multinodeDeployer)
 

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

@@ -21,7 +21,7 @@ func TestTRTLLMBackend_UpdateContainer(t *testing.T) {
 		numberOfNodes          int32
 		role                   Role
 		multinodeDeployer      MultinodeDeployer
-		component              *v1alpha1.DynamoComponentDeploymentOverridesSpec
+		component              *v1alpha1.DynamoComponentDeploymentSharedSpec
 		expectedVolumeMounts   []corev1.VolumeMount
 		expectedCommand        []string
 		expectedArgs           []string
@@ -36,7 +36,7 @@ func TestTRTLLMBackend_UpdateContainer(t *testing.T) {
 			numberOfNodes:          1,
 			role:                   RoleMain,
 			multinodeDeployer:      &GroveMultinodeDeployer{},
-			component:              &v1alpha1.DynamoComponentDeploymentOverridesSpec{},
+			component:              &v1alpha1.DynamoComponentDeploymentSharedSpec{},
 			expectedVolumeMounts:   []corev1.VolumeMount{},
 			expectedCommand:        []string{},
 			expectedArgs:           []string{"python3", "--model", "test"},
@@ -51,15 +51,13 @@ func TestTRTLLMBackend_UpdateContainer(t *testing.T) {
 			numberOfNodes:     3,
 			role:              RoleLeader,
 			multinodeDeployer: &GroveMultinodeDeployer{},
-			component: &v1alpha1.DynamoComponentDeploymentOverridesSpec{
-				DynamoComponentDeploymentSharedSpec: v1alpha1.DynamoComponentDeploymentSharedSpec{
+			component: &v1alpha1.DynamoComponentDeploymentSharedSpec{
 				Resources: &common.Resources{
 					Requests: &common.ResourceItem{
 						GPU: "2",
 					},
 				},
 			},
-			},
 			expectedVolumeMounts: []corev1.VolumeMount{
 				{Name: mpiRunSecretName, MountPath: "/ssh-pk", ReadOnly: true},
 			},
@@ -78,7 +76,7 @@ func TestTRTLLMBackend_UpdateContainer(t *testing.T) {
 			numberOfNodes:     3,
 			role:              RoleWorker,
 			multinodeDeployer: &GroveMultinodeDeployer{},
-			component:         &v1alpha1.DynamoComponentDeploymentOverridesSpec{},
+			component:         &v1alpha1.DynamoComponentDeploymentSharedSpec{},
 			expectedVolumeMounts: []corev1.VolumeMount{
 				{Name: mpiRunSecretName, MountPath: "/ssh-pk", ReadOnly: true},
 			},
@@ -107,15 +105,13 @@ func TestTRTLLMBackend_UpdateContainer(t *testing.T) {
 			numberOfNodes:     2,
 			role:              RoleLeader,
 			multinodeDeployer: &LWSMultinodeDeployer{},
-			component: &v1alpha1.DynamoComponentDeploymentOverridesSpec{
-				DynamoComponentDeploymentSharedSpec: v1alpha1.DynamoComponentDeploymentSharedSpec{
+			component: &v1alpha1.DynamoComponentDeploymentSharedSpec{
 				Resources: &common.Resources{
 					Limits: &common.ResourceItem{
 						GPU: "1",
 					},
 				},
 			},
-			},
 			expectedVolumeMounts: []corev1.VolumeMount{
 				{Name: mpiRunSecretName, MountPath: "/ssh-pk", ReadOnly: true},
 			},
@@ -350,7 +346,7 @@ func TestTRTLLMBackend_UpdatePodSpec(t *testing.T) {
 					},
 				},
 			}
-			component := &v1alpha1.DynamoComponentDeploymentOverridesSpec{}
+			component := &v1alpha1.DynamoComponentDeploymentSharedSpec{}
 
 			// Call UpdatePodSpec
 			backend.UpdatePodSpec(podSpec, tt.numberOfNodes, tt.role, component, "test-service")
@@ -550,7 +546,7 @@ func TestTRTLLMBackend_setupLeaderContainer(t *testing.T) {
 		numberOfNodes     int32
 		multinodeDeployer MultinodeDeployer
 		serviceName       string
-		component         *v1alpha1.DynamoComponentDeploymentOverridesSpec
+		component         *v1alpha1.DynamoComponentDeploymentSharedSpec
 		initialArgs       []string
 		initialCommand    []string
 		expected          string
@@ -560,15 +556,13 @@ func TestTRTLLMBackend_setupLeaderContainer(t *testing.T) {
 			numberOfNodes:     3,
 			multinodeDeployer: &GroveMultinodeDeployer{},
 			serviceName:       "test-service",
-			component: &v1alpha1.DynamoComponentDeploymentOverridesSpec{
-				DynamoComponentDeploymentSharedSpec: v1alpha1.DynamoComponentDeploymentSharedSpec{
+			component: &v1alpha1.DynamoComponentDeploymentSharedSpec{
 				Resources: &common.Resources{
 					Requests: &common.ResourceItem{
 						GPU: "2",
 					},
 				},
 			},
-			},
 			initialArgs:    []string{"python3", "--model", "test"},
 			initialCommand: []string{},
 			expected:       "mkdir -p ~/.ssh && ls -la /ssh-pk/ && cp /ssh-pk/private.key ~/.ssh/id_rsa && cp /ssh-pk/private.key.pub ~/.ssh/id_rsa.pub && cp /ssh-pk/private.key.pub ~/.ssh/authorized_keys && chmod 600 ~/.ssh/id_rsa ~/.ssh/authorized_keys && chmod 644 ~/.ssh/id_rsa.pub ~/.ssh/authorized_keys && printf 'Host *\\nIdentityFile ~/.ssh/id_rsa\\nStrictHostKeyChecking no\\nPort 2222\\n' > ~/.ssh/config && mpirun --oversubscribe -n 6 -H $(GROVE_PCSG_NAME)-$(GROVE_PCSG_INDEX)-test-service-ldr-0.$(GROVE_HEADLESS_SERVICE),$(GROVE_PCSG_NAME)-$(GROVE_PCSG_INDEX)-test-service-wkr-0.$(GROVE_HEADLESS_SERVICE),$(GROVE_PCSG_NAME)-$(GROVE_PCSG_INDEX)-test-service-wkr-1.$(GROVE_HEADLESS_SERVICE) --mca pml ob1 --mca plm_rsh_args \"-p 2222 -o StrictHostKeyChecking=no -i ~/.ssh/id_rsa\" -x CUDA_VISIBLE_DEVICES -x HF_DATASETS_CACHE -x HF_ENDPOINT -x HF_HOME -x HF_TOKEN -x HOME -x HUGGING_FACE_HUB_TOKEN -x LD_LIBRARY_PATH -x MODEL_PATH -x NCCL_DEBUG -x NCCL_IB_DISABLE -x NCCL_P2P_DISABLE -x PATH -x PYTHONPATH -x TENSORRT_LLM_CACHE_DIR -x TOKENIZERS_PARALLELISM -x TRANSFORMERS_CACHE -x USER bash -c 'source /opt/dynamo/venv/bin/activate && trtllm-llmapi-launch python3 --model test'",
@@ -578,7 +572,7 @@ func TestTRTLLMBackend_setupLeaderContainer(t *testing.T) {
 			numberOfNodes:     2,
 			multinodeDeployer: &LWSMultinodeDeployer{},
 			serviceName:       "worker",
-			component:         &v1alpha1.DynamoComponentDeploymentOverridesSpec{},
+			component:         &v1alpha1.DynamoComponentDeploymentSharedSpec{},
 			initialArgs:       []string{},
 			initialCommand:    []string{"python", "-m", "worker"},
 			expected:          "mkdir -p ~/.ssh && ls -la /ssh-pk/ && cp /ssh-pk/private.key ~/.ssh/id_rsa && cp /ssh-pk/private.key.pub ~/.ssh/id_rsa.pub && cp /ssh-pk/private.key.pub ~/.ssh/authorized_keys && chmod 600 ~/.ssh/id_rsa ~/.ssh/authorized_keys && chmod 644 ~/.ssh/id_rsa.pub ~/.ssh/authorized_keys && printf 'Host *\\nIdentityFile ~/.ssh/id_rsa\\nStrictHostKeyChecking no\\nPort 2222\\n' > ~/.ssh/config && mpirun --oversubscribe -n 0 -H $(LWS_LEADER_ADDRESS),$(LWS_WORKER_1_ADDRESS) --mca pml ob1 --mca plm_rsh_args \"-p 2222 -o StrictHostKeyChecking=no -i ~/.ssh/id_rsa\" -x CUDA_VISIBLE_DEVICES -x HF_DATASETS_CACHE -x HF_ENDPOINT -x HF_HOME -x HF_TOKEN -x HOME -x HUGGING_FACE_HUB_TOKEN -x LD_LIBRARY_PATH -x MODEL_PATH -x NCCL_DEBUG -x NCCL_IB_DISABLE -x NCCL_P2P_DISABLE -x PATH -x PYTHONPATH -x TENSORRT_LLM_CACHE_DIR -x TOKENIZERS_PARALLELISM -x TRANSFORMERS_CACHE -x USER bash -c 'source /opt/dynamo/venv/bin/activate && trtllm-llmapi-launch python -m worker'",
@@ -588,15 +582,13 @@ func TestTRTLLMBackend_setupLeaderContainer(t *testing.T) {
 			numberOfNodes:     2,
 			multinodeDeployer: &GroveMultinodeDeployer{},
 			serviceName:       "test",
-			component: &v1alpha1.DynamoComponentDeploymentOverridesSpec{
-				DynamoComponentDeploymentSharedSpec: v1alpha1.DynamoComponentDeploymentSharedSpec{
+			component: &v1alpha1.DynamoComponentDeploymentSharedSpec{
 				Resources: &common.Resources{
 					Limits: &common.ResourceItem{
 						GPU: "1",
 					},
 				},
 			},
-			},
 			initialArgs:    []string{"launch", "--config", "test.yaml"},
 			initialCommand: []string{"sh", "-c"},
 			expected:       "mkdir -p ~/.ssh && ls -la /ssh-pk/ && cp /ssh-pk/private.key ~/.ssh/id_rsa && cp /ssh-pk/private.key.pub ~/.ssh/id_rsa.pub && cp /ssh-pk/private.key.pub ~/.ssh/authorized_keys && chmod 600 ~/.ssh/id_rsa ~/.ssh/authorized_keys && chmod 644 ~/.ssh/id_rsa.pub ~/.ssh/authorized_keys && printf 'Host *\\nIdentityFile ~/.ssh/id_rsa\\nStrictHostKeyChecking no\\nPort 2222\\n' > ~/.ssh/config && mpirun --oversubscribe -n 2 -H $(GROVE_PCSG_NAME)-$(GROVE_PCSG_INDEX)-test-ldr-0.$(GROVE_HEADLESS_SERVICE),$(GROVE_PCSG_NAME)-$(GROVE_PCSG_INDEX)-test-wkr-0.$(GROVE_HEADLESS_SERVICE) --mca pml ob1 --mca plm_rsh_args \"-p 2222 -o StrictHostKeyChecking=no -i ~/.ssh/id_rsa\" -x CUDA_VISIBLE_DEVICES -x HF_DATASETS_CACHE -x HF_ENDPOINT -x HF_HOME -x HF_TOKEN -x HOME -x HUGGING_FACE_HUB_TOKEN -x LD_LIBRARY_PATH -x MODEL_PATH -x NCCL_DEBUG -x NCCL_IB_DISABLE -x NCCL_P2P_DISABLE -x PATH -x PYTHONPATH -x TENSORRT_LLM_CACHE_DIR -x TOKENIZERS_PARALLELISM -x TRANSFORMERS_CACHE -x USER bash -c 'source /opt/dynamo/venv/bin/activate && trtllm-llmapi-launch launch --config test.yaml'",
@@ -606,15 +598,13 @@ func TestTRTLLMBackend_setupLeaderContainer(t *testing.T) {
 			numberOfNodes:     2,
 			multinodeDeployer: &GroveMultinodeDeployer{},
 			serviceName:       "test",
-			component: &v1alpha1.DynamoComponentDeploymentOverridesSpec{
-				DynamoComponentDeploymentSharedSpec: v1alpha1.DynamoComponentDeploymentSharedSpec{
+			component: &v1alpha1.DynamoComponentDeploymentSharedSpec{
 				Resources: &common.Resources{
 					Limits: &common.ResourceItem{
 						GPU: "1",
 					},
 				},
 			},
-			},
 			initialArgs:    []string{"-m", "dynamo.trtllm", "--model-path", "test"},
 			initialCommand: []string{"python3"},
 			expected:       "mkdir -p ~/.ssh && ls -la /ssh-pk/ && cp /ssh-pk/private.key ~/.ssh/id_rsa && cp /ssh-pk/private.key.pub ~/.ssh/id_rsa.pub && cp /ssh-pk/private.key.pub ~/.ssh/authorized_keys && chmod 600 ~/.ssh/id_rsa ~/.ssh/authorized_keys && chmod 644 ~/.ssh/id_rsa.pub ~/.ssh/authorized_keys && printf 'Host *\\nIdentityFile ~/.ssh/id_rsa\\nStrictHostKeyChecking no\\nPort 2222\\n' > ~/.ssh/config && mpirun --oversubscribe -n 2 -H $(GROVE_PCSG_NAME)-$(GROVE_PCSG_INDEX)-test-ldr-0.$(GROVE_HEADLESS_SERVICE),$(GROVE_PCSG_NAME)-$(GROVE_PCSG_INDEX)-test-wkr-0.$(GROVE_HEADLESS_SERVICE) --mca pml ob1 --mca plm_rsh_args \"-p 2222 -o StrictHostKeyChecking=no -i ~/.ssh/id_rsa\" -x CUDA_VISIBLE_DEVICES -x HF_DATASETS_CACHE -x HF_ENDPOINT -x HF_HOME -x HF_TOKEN -x HOME -x HUGGING_FACE_HUB_TOKEN -x LD_LIBRARY_PATH -x MODEL_PATH -x NCCL_DEBUG -x NCCL_IB_DISABLE -x NCCL_P2P_DISABLE -x PATH -x PYTHONPATH -x TENSORRT_LLM_CACHE_DIR -x TOKENIZERS_PARALLELISM -x TRANSFORMERS_CACHE -x USER bash -c 'source /opt/dynamo/venv/bin/activate && trtllm-llmapi-launch python3 -m dynamo.trtllm --model-path test'",
@@ -624,15 +614,13 @@ func TestTRTLLMBackend_setupLeaderContainer(t *testing.T) {
 			numberOfNodes:     2,
 			multinodeDeployer: &GroveMultinodeDeployer{},
 			serviceName:       "test",
-			component: &v1alpha1.DynamoComponentDeploymentOverridesSpec{
-				DynamoComponentDeploymentSharedSpec: v1alpha1.DynamoComponentDeploymentSharedSpec{
+			component: &v1alpha1.DynamoComponentDeploymentSharedSpec{
 				Resources: &common.Resources{
 					Limits: &common.ResourceItem{
 						GPU: "1",
 					},
 				},
 			},
-			},
 			initialArgs:    []string{"--model-path", "Qwen/Qwen3-0.6B", "--served-model-name", "Qwen/Qwen3-0.6B", "--disaggregation-mode", "prefill"},
 			initialCommand: []string{"python3", "-m", "dynamo.trtllm"},
 			expected:       "mkdir -p ~/.ssh && ls -la /ssh-pk/ && cp /ssh-pk/private.key ~/.ssh/id_rsa && cp /ssh-pk/private.key.pub ~/.ssh/id_rsa.pub && cp /ssh-pk/private.key.pub ~/.ssh/authorized_keys && chmod 600 ~/.ssh/id_rsa ~/.ssh/authorized_keys && chmod 644 ~/.ssh/id_rsa.pub ~/.ssh/authorized_keys && printf 'Host *\\nIdentityFile ~/.ssh/id_rsa\\nStrictHostKeyChecking no\\nPort 2222\\n' > ~/.ssh/config && mpirun --oversubscribe -n 2 -H $(GROVE_PCSG_NAME)-$(GROVE_PCSG_INDEX)-test-ldr-0.$(GROVE_HEADLESS_SERVICE),$(GROVE_PCSG_NAME)-$(GROVE_PCSG_INDEX)-test-wkr-0.$(GROVE_HEADLESS_SERVICE) --mca pml ob1 --mca plm_rsh_args \"-p 2222 -o StrictHostKeyChecking=no -i ~/.ssh/id_rsa\" -x CUDA_VISIBLE_DEVICES -x HF_DATASETS_CACHE -x HF_ENDPOINT -x HF_HOME -x HF_TOKEN -x HOME -x HUGGING_FACE_HUB_TOKEN -x LD_LIBRARY_PATH -x MODEL_PATH -x NCCL_DEBUG -x NCCL_IB_DISABLE -x NCCL_P2P_DISABLE -x PATH -x PYTHONPATH -x TENSORRT_LLM_CACHE_DIR -x TOKENIZERS_PARALLELISM -x TRANSFORMERS_CACHE -x USER bash -c 'source /opt/dynamo/venv/bin/activate && trtllm-llmapi-launch python3 -m dynamo.trtllm --model-path Qwen/Qwen3-0.6B --served-model-name Qwen/Qwen3-0.6B --disaggregation-mode prefill'",
@@ -642,15 +630,13 @@ func TestTRTLLMBackend_setupLeaderContainer(t *testing.T) {
 			numberOfNodes:     2,
 			multinodeDeployer: &GroveMultinodeDeployer{},
 			serviceName:       "test",
-			component: &v1alpha1.DynamoComponentDeploymentOverridesSpec{
-				DynamoComponentDeploymentSharedSpec: v1alpha1.DynamoComponentDeploymentSharedSpec{
+			component: &v1alpha1.DynamoComponentDeploymentSharedSpec{
 				Resources: &common.Resources{
 					Limits: &common.ResourceItem{
 						GPU: "1",
 					},
 				},
 			},
-			},
 			initialArgs:    []string{"-m", "dynamo.trtllm", "--model-path", "test"},
 			initialCommand: []string{"/usr/bin/python3.8"},
 			expected:       "mkdir -p ~/.ssh && ls -la /ssh-pk/ && cp /ssh-pk/private.key ~/.ssh/id_rsa && cp /ssh-pk/private.key.pub ~/.ssh/id_rsa.pub && cp /ssh-pk/private.key.pub ~/.ssh/authorized_keys && chmod 600 ~/.ssh/id_rsa ~/.ssh/authorized_keys && chmod 644 ~/.ssh/id_rsa.pub ~/.ssh/authorized_keys && printf 'Host *\\nIdentityFile ~/.ssh/id_rsa\\nStrictHostKeyChecking no\\nPort 2222\\n' > ~/.ssh/config && mpirun --oversubscribe -n 2 -H $(GROVE_PCSG_NAME)-$(GROVE_PCSG_INDEX)-test-ldr-0.$(GROVE_HEADLESS_SERVICE),$(GROVE_PCSG_NAME)-$(GROVE_PCSG_INDEX)-test-wkr-0.$(GROVE_HEADLESS_SERVICE) --mca pml ob1 --mca plm_rsh_args \"-p 2222 -o StrictHostKeyChecking=no -i ~/.ssh/id_rsa\" -x CUDA_VISIBLE_DEVICES -x HF_DATASETS_CACHE -x HF_ENDPOINT -x HF_HOME -x HF_TOKEN -x HOME -x HUGGING_FACE_HUB_TOKEN -x LD_LIBRARY_PATH -x MODEL_PATH -x NCCL_DEBUG -x NCCL_IB_DISABLE -x NCCL_P2P_DISABLE -x PATH -x PYTHONPATH -x TENSORRT_LLM_CACHE_DIR -x TOKENIZERS_PARALLELISM -x TRANSFORMERS_CACHE -x USER bash -c 'source /opt/dynamo/venv/bin/activate && trtllm-llmapi-launch /usr/bin/python3.8 -m dynamo.trtllm --model-path test'",
@@ -660,15 +646,13 @@ func TestTRTLLMBackend_setupLeaderContainer(t *testing.T) {
 			numberOfNodes:     2,
 			multinodeDeployer: &GroveMultinodeDeployer{},
 			serviceName:       "test",
-			component: &v1alpha1.DynamoComponentDeploymentOverridesSpec{
-				DynamoComponentDeploymentSharedSpec: v1alpha1.DynamoComponentDeploymentSharedSpec{
+			component: &v1alpha1.DynamoComponentDeploymentSharedSpec{
 				Resources: &common.Resources{
 					Requests: &common.ResourceItem{
 						GPU: "1",
 					},
 				},
 			},
-			},
 			initialArgs:    []string{"serve", "--model", "test"},
 			initialCommand: []string{},
 			expected:       "mkdir -p ~/.ssh && ls -la /ssh-pk/ && cp /ssh-pk/private.key ~/.ssh/id_rsa && cp /ssh-pk/private.key.pub ~/.ssh/id_rsa.pub && cp /ssh-pk/private.key.pub ~/.ssh/authorized_keys && chmod 600 ~/.ssh/id_rsa ~/.ssh/authorized_keys && chmod 644 ~/.ssh/id_rsa.pub ~/.ssh/authorized_keys && printf 'Host *\\nIdentityFile ~/.ssh/id_rsa\\nStrictHostKeyChecking no\\nPort 2222\\n' > ~/.ssh/config && mpirun --oversubscribe -n 2 -H $(GROVE_PCSG_NAME)-$(GROVE_PCSG_INDEX)-test-ldr-0.$(GROVE_HEADLESS_SERVICE),$(GROVE_PCSG_NAME)-$(GROVE_PCSG_INDEX)-test-wkr-0.$(GROVE_HEADLESS_SERVICE) --mca pml ob1 --mca plm_rsh_args \"-p 2222 -o StrictHostKeyChecking=no -i ~/.ssh/id_rsa\" -x CUDA_VISIBLE_DEVICES -x HF_DATASETS_CACHE -x HF_ENDPOINT -x HF_HOME -x HF_TOKEN -x HOME -x HUGGING_FACE_HUB_TOKEN -x LD_LIBRARY_PATH -x MODEL_PATH -x NCCL_DEBUG -x NCCL_IB_DISABLE -x NCCL_P2P_DISABLE -x PATH -x PYTHONPATH -x TENSORRT_LLM_CACHE_DIR -x TOKENIZERS_PARALLELISM -x TRANSFORMERS_CACHE -x USER bash -c 'source /opt/dynamo/venv/bin/activate && trtllm-llmapi-launch serve --model test'",
@@ -678,15 +662,13 @@ func TestTRTLLMBackend_setupLeaderContainer(t *testing.T) {
 			numberOfNodes:     2,
 			multinodeDeployer: &GroveMultinodeDeployer{},
 			serviceName:       "test",
-			component: &v1alpha1.DynamoComponentDeploymentOverridesSpec{
-				DynamoComponentDeploymentSharedSpec: v1alpha1.DynamoComponentDeploymentSharedSpec{
+			component: &v1alpha1.DynamoComponentDeploymentSharedSpec{
 				Resources: &common.Resources{
 					Requests: &common.ResourceItem{
 						GPU: "1",
 					},
 				},
 			},
-			},
 			initialArgs:    []string{"serve", "--model", "test"},
 			initialCommand: []string{},
 			expected:       "mkdir -p ~/.ssh && ls -la /ssh-pk/ && cp /ssh-pk/private.key ~/.ssh/id_rsa && cp /ssh-pk/private.key.pub ~/.ssh/id_rsa.pub && cp /ssh-pk/private.key.pub ~/.ssh/authorized_keys && chmod 600 ~/.ssh/id_rsa ~/.ssh/authorized_keys && chmod 644 ~/.ssh/id_rsa.pub ~/.ssh/authorized_keys && printf 'Host *\\nIdentityFile ~/.ssh/id_rsa\\nStrictHostKeyChecking no\\nPort 2222\\n' > ~/.ssh/config && mpirun --oversubscribe -n 2 -H $(GROVE_PCSG_NAME)-$(GROVE_PCSG_INDEX)-test-ldr-0.$(GROVE_HEADLESS_SERVICE),$(GROVE_PCSG_NAME)-$(GROVE_PCSG_INDEX)-test-wkr-0.$(GROVE_HEADLESS_SERVICE) --mca pml ob1 --mca plm_rsh_args \"-p 2222 -o StrictHostKeyChecking=no -i ~/.ssh/id_rsa\" -x CUDA_VISIBLE_DEVICES -x CUSTOM_VAR -x HF_DATASETS_CACHE -x HF_ENDPOINT -x HF_HOME -x HF_TOKEN -x HOME -x HUGGING_FACE_HUB_TOKEN -x LD_LIBRARY_PATH -x MODEL_PATH -x NCCL_DEBUG -x NCCL_IB_DISABLE -x NCCL_P2P_DISABLE -x PATH -x PYTHONPATH -x TENSORRT_LLM_CACHE_DIR -x TOKENIZERS_PARALLELISM -x TRANSFORMERS_CACHE -x USER bash -c 'source /opt/dynamo/venv/bin/activate && trtllm-llmapi-launch serve --model test'",
@@ -874,15 +856,14 @@ func TestTRTLLMBackend_UpdateContainer_UseAsCompilationCache(t *testing.T) {
 
 	tests := []struct {
 		name                       string
-		component                  *v1alpha1.DynamoComponentDeploymentOverridesSpec
+		component                  *v1alpha1.DynamoComponentDeploymentSharedSpec
 		volumeMounts               []corev1.VolumeMount
 		expectNoEnvVarChanges      bool
 		expectLoggedPartialSupport bool
 	}{
 		{
 			name: "TensorRT-LLM backend with useAsCompilationCache volume mount",
-			component: &v1alpha1.DynamoComponentDeploymentOverridesSpec{
-				DynamoComponentDeploymentSharedSpec: v1alpha1.DynamoComponentDeploymentSharedSpec{
+			component: &v1alpha1.DynamoComponentDeploymentSharedSpec{
 				VolumeMounts: []v1alpha1.VolumeMount{
 					{
 						Name:                  "trtllm-cache",
@@ -891,15 +872,13 @@ func TestTRTLLMBackend_UpdateContainer_UseAsCompilationCache(t *testing.T) {
 					},
 				},
 			},
-			},
 			volumeMounts:               []corev1.VolumeMount{},
 			expectNoEnvVarChanges:      true, // TensorRT-LLM doesn't set env vars yet
 			expectLoggedPartialSupport: true,
 		},
 		{
 			name: "TensorRT-LLM backend with useAsCompilationCache at custom mount point",
-			component: &v1alpha1.DynamoComponentDeploymentOverridesSpec{
-				DynamoComponentDeploymentSharedSpec: v1alpha1.DynamoComponentDeploymentSharedSpec{
+			component: &v1alpha1.DynamoComponentDeploymentSharedSpec{
 				VolumeMounts: []v1alpha1.VolumeMount{
 					{
 						Name:                  "custom-cache",
@@ -908,15 +887,13 @@ func TestTRTLLMBackend_UpdateContainer_UseAsCompilationCache(t *testing.T) {
 					},
 				},
 			},
-			},
 			volumeMounts:               []corev1.VolumeMount{},
 			expectNoEnvVarChanges:      true, // TensorRT-LLM doesn't set env vars yet
 			expectLoggedPartialSupport: true,
 		},
 		{
 			name: "TensorRT-LLM backend without useAsCompilationCache",
-			component: &v1alpha1.DynamoComponentDeploymentOverridesSpec{
-				DynamoComponentDeploymentSharedSpec: v1alpha1.DynamoComponentDeploymentSharedSpec{
+			component: &v1alpha1.DynamoComponentDeploymentSharedSpec{
 				VolumeMounts: []v1alpha1.VolumeMount{
 					{
 						Name:       "regular-volume",
@@ -924,18 +901,15 @@ func TestTRTLLMBackend_UpdateContainer_UseAsCompilationCache(t *testing.T) {
 					},
 				},
 			},
-			},
 			volumeMounts:               []corev1.VolumeMount{},
 			expectNoEnvVarChanges:      true,
 			expectLoggedPartialSupport: false,
 		},
 		{
 			name: "TensorRT-LLM backend with no volume mounts",
-			component: &v1alpha1.DynamoComponentDeploymentOverridesSpec{
-				DynamoComponentDeploymentSharedSpec: v1alpha1.DynamoComponentDeploymentSharedSpec{
+			component: &v1alpha1.DynamoComponentDeploymentSharedSpec{
 				VolumeMounts: nil,
 			},
-			},
 			volumeMounts:               []corev1.VolumeMount{},
 			expectNoEnvVarChanges:      true,
 			expectLoggedPartialSupport: false,

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

@@ -15,7 +15,7 @@ const (
 
 type VLLMBackend struct{}
 
-func (b *VLLMBackend) UpdateContainer(container *corev1.Container, numberOfNodes int32, role Role, component *v1alpha1.DynamoComponentDeploymentOverridesSpec, serviceName string, multinodeDeployer MultinodeDeployer) {
+func (b *VLLMBackend) UpdateContainer(container *corev1.Container, numberOfNodes int32, role Role, component *v1alpha1.DynamoComponentDeploymentSharedSpec, serviceName string, multinodeDeployer MultinodeDeployer) {
 	isMultinode := numberOfNodes > 1
 
 	if isMultinode {
@@ -60,7 +60,7 @@ func (b *VLLMBackend) UpdateContainer(container *corev1.Container, numberOfNodes
 	}
 }
 
-func (b *VLLMBackend) UpdatePodSpec(podSpec *corev1.PodSpec, numberOfNodes int32, role Role, component *v1alpha1.DynamoComponentDeploymentOverridesSpec, serviceName string) {
+func (b *VLLMBackend) UpdatePodSpec(podSpec *corev1.PodSpec, numberOfNodes int32, role Role, component *v1alpha1.DynamoComponentDeploymentSharedSpec, serviceName string) {
 	// do nothing
 }
 

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

@@ -16,7 +16,7 @@ func TestVLLMBackend_UpdateContainer(t *testing.T) {
 		name                  string
 		numberOfNodes         int32
 		role                  Role
-		component             *v1alpha1.DynamoComponentDeploymentOverridesSpec
+		component             *v1alpha1.DynamoComponentDeploymentSharedSpec
 		multinodeDeployer     MultinodeDeployer
 		initialArgs           []string
 		initialLivenessProbe  *corev1.Probe
@@ -31,7 +31,7 @@ func TestVLLMBackend_UpdateContainer(t *testing.T) {
 			name:              "single node does not modify args",
 			numberOfNodes:     1,
 			role:              RoleMain,
-			component:         &v1alpha1.DynamoComponentDeploymentOverridesSpec{},
+			component:         &v1alpha1.DynamoComponentDeploymentSharedSpec{},
 			multinodeDeployer: &GroveMultinodeDeployer{},
 			initialArgs:       []string{"python3", "-m", "dynamo.vllm"},
 			expectNotModified: true,
@@ -40,7 +40,7 @@ func TestVLLMBackend_UpdateContainer(t *testing.T) {
 			name:                "multinode leader prepends ray start --head",
 			numberOfNodes:       3,
 			role:                RoleLeader,
-			component:           &v1alpha1.DynamoComponentDeploymentOverridesSpec{},
+			component:           &v1alpha1.DynamoComponentDeploymentSharedSpec{},
 			multinodeDeployer:   &GroveMultinodeDeployer{},
 			initialArgs:         []string{"python3", "-m", "dynamo.vllm", "--model", "test"},
 			expectContains:      []string{"ray start --head --port=6379 &&", "python3", "-m", "dynamo.vllm", "--model", "test"},
@@ -50,7 +50,7 @@ func TestVLLMBackend_UpdateContainer(t *testing.T) {
 			name:                "multinode worker replaces args with ray start --block",
 			numberOfNodes:       3,
 			role:                RoleWorker,
-			component:           &v1alpha1.DynamoComponentDeploymentOverridesSpec{},
+			component:           &v1alpha1.DynamoComponentDeploymentSharedSpec{},
 			multinodeDeployer:   &GroveMultinodeDeployer{},
 			initialArgs:         []string{"python3", "-m", "dynamo.vllm", "--model", "test"},
 			expectedArgs:        []string{"ray start --address=$(GROVE_PCSG_NAME)-$(GROVE_PCSG_INDEX)-test-service-ldr-0.$(GROVE_HEADLESS_SERVICE):6379 --block"},
@@ -60,7 +60,7 @@ func TestVLLMBackend_UpdateContainer(t *testing.T) {
 			name:                "multinode worker with LWS deployment type",
 			numberOfNodes:       2,
 			role:                RoleWorker,
-			component:           &v1alpha1.DynamoComponentDeploymentOverridesSpec{},
+			component:           &v1alpha1.DynamoComponentDeploymentSharedSpec{},
 			multinodeDeployer:   &LWSMultinodeDeployer{},
 			initialArgs:         []string{"python3", "-m", "dynamo.vllm"},
 			expectedArgs:        []string{"ray start --address=$(LWS_LEADER_ADDRESS):6379 --block"},
@@ -70,7 +70,7 @@ func TestVLLMBackend_UpdateContainer(t *testing.T) {
 			name:              "multinode leader with no initial args",
 			numberOfNodes:     2,
 			role:              RoleLeader,
-			component:         &v1alpha1.DynamoComponentDeploymentOverridesSpec{},
+			component:         &v1alpha1.DynamoComponentDeploymentSharedSpec{},
 			multinodeDeployer: &GroveMultinodeDeployer{},
 			initialArgs:       []string{},
 			expectNotModified: true, // Should not modify empty args
@@ -79,7 +79,7 @@ func TestVLLMBackend_UpdateContainer(t *testing.T) {
 			name:              "multinode main role (non-leader/worker) does not modify args",
 			numberOfNodes:     3,
 			role:              RoleMain,
-			component:         &v1alpha1.DynamoComponentDeploymentOverridesSpec{},
+			component:         &v1alpha1.DynamoComponentDeploymentSharedSpec{},
 			multinodeDeployer: &GroveMultinodeDeployer{},
 			initialArgs:       []string{"python3", "-m", "dynamo.frontend"},
 			expectNotModified: true,
@@ -131,7 +131,7 @@ func TestVLLMBackend_UpdateContainer_UseAsCompilationCache(t *testing.T) {
 
 	tests := []struct {
 		name                  string
-		component             *v1alpha1.DynamoComponentDeploymentOverridesSpec
+		component             *v1alpha1.DynamoComponentDeploymentSharedSpec
 		volumeMounts          []corev1.VolumeMount
 		expectCacheEnvVar     bool
 		expectCacheEnvVarName string
@@ -139,8 +139,7 @@ func TestVLLMBackend_UpdateContainer_UseAsCompilationCache(t *testing.T) {
 	}{
 		{
 			name: "VLLM backend with useAsCompilationCache volume mount",
-			component: &v1alpha1.DynamoComponentDeploymentOverridesSpec{
-				DynamoComponentDeploymentSharedSpec: v1alpha1.DynamoComponentDeploymentSharedSpec{
+			component: &v1alpha1.DynamoComponentDeploymentSharedSpec{
 				VolumeMounts: []v1alpha1.VolumeMount{
 					{
 						Name:                  "vllm-cache",
@@ -149,7 +148,6 @@ func TestVLLMBackend_UpdateContainer_UseAsCompilationCache(t *testing.T) {
 					},
 				},
 			},
-			},
 			volumeMounts:          []corev1.VolumeMount{},
 			expectCacheEnvVar:     true,
 			expectCacheEnvVarName: "VLLM_CACHE_ROOT",
@@ -157,8 +155,7 @@ func TestVLLMBackend_UpdateContainer_UseAsCompilationCache(t *testing.T) {
 		},
 		{
 			name: "VLLM backend with useAsCompilationCache at custom mount point",
-			component: &v1alpha1.DynamoComponentDeploymentOverridesSpec{
-				DynamoComponentDeploymentSharedSpec: v1alpha1.DynamoComponentDeploymentSharedSpec{
+			component: &v1alpha1.DynamoComponentDeploymentSharedSpec{
 				VolumeMounts: []v1alpha1.VolumeMount{
 					{
 						Name:                  "custom-cache",
@@ -167,7 +164,6 @@ func TestVLLMBackend_UpdateContainer_UseAsCompilationCache(t *testing.T) {
 					},
 				},
 			},
-			},
 			volumeMounts:          []corev1.VolumeMount{},
 			expectCacheEnvVar:     true,
 			expectCacheEnvVarName: "VLLM_CACHE_ROOT",
@@ -175,8 +171,7 @@ func TestVLLMBackend_UpdateContainer_UseAsCompilationCache(t *testing.T) {
 		},
 		{
 			name: "VLLM backend without useAsCompilationCache",
-			component: &v1alpha1.DynamoComponentDeploymentOverridesSpec{
-				DynamoComponentDeploymentSharedSpec: v1alpha1.DynamoComponentDeploymentSharedSpec{
+			component: &v1alpha1.DynamoComponentDeploymentSharedSpec{
 				VolumeMounts: []v1alpha1.VolumeMount{
 					{
 						Name:       "regular-volume",
@@ -184,17 +179,14 @@ func TestVLLMBackend_UpdateContainer_UseAsCompilationCache(t *testing.T) {
 					},
 				},
 			},
-			},
 			volumeMounts:      []corev1.VolumeMount{},
 			expectCacheEnvVar: false,
 		},
 		{
 			name: "VLLM backend with no volume mounts",
-			component: &v1alpha1.DynamoComponentDeploymentOverridesSpec{
-				DynamoComponentDeploymentSharedSpec: v1alpha1.DynamoComponentDeploymentSharedSpec{
+			component: &v1alpha1.DynamoComponentDeploymentSharedSpec{
 				VolumeMounts: nil,
 			},
-			},
 			volumeMounts:      []corev1.VolumeMount{},
 			expectCacheEnvVar: false,
 		},

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

@@ -125,7 +125,7 @@ func GenerateDynamoComponentsDeployments(ctx context.Context, parentDynamoGraphD
 	}
 	for componentName, component := range parentDynamoGraphDeployment.Spec.Services {
 		deployment := &v1alpha1.DynamoComponentDeployment{}
-		deployment.Spec.DynamoComponentDeploymentSharedSpec = component.DynamoComponentDeploymentSharedSpec
+		deployment.Spec.DynamoComponentDeploymentSharedSpec = *component
 		deployment.Name = GetDynamoComponentName(parentDynamoGraphDeployment, componentName)
 		deployment.Spec.BackendFramework = parentDynamoGraphDeployment.Spec.BackendFramework
 		deployment.Namespace = parentDynamoGraphDeployment.Namespace
@@ -597,18 +597,18 @@ const (
 // Backend interface for modular backend logic
 // Each backend (SGLang, VLLM, etc.) implements this interface
 type Backend interface {
-	UpdateContainer(container *corev1.Container, numberOfNodes int32, role Role, component *v1alpha1.DynamoComponentDeploymentOverridesSpec, serviceName string, multinodeDeployer MultinodeDeployer)
-	UpdatePodSpec(podSpec *corev1.PodSpec, numberOfNodes int32, role Role, component *v1alpha1.DynamoComponentDeploymentOverridesSpec, serviceName string)
+	UpdateContainer(container *corev1.Container, numberOfNodes int32, role Role, component *v1alpha1.DynamoComponentDeploymentSharedSpec, serviceName string, multinodeDeployer MultinodeDeployer)
+	UpdatePodSpec(podSpec *corev1.PodSpec, numberOfNodes int32, role Role, component *v1alpha1.DynamoComponentDeploymentSharedSpec, serviceName string)
 }
 
 // NoopBackend does no processing - used for non-worker components like frontend, planner, router
 type NoopBackend struct{}
 
-func (b *NoopBackend) UpdateContainer(container *corev1.Container, numberOfNodes int32, role Role, component *v1alpha1.DynamoComponentDeploymentOverridesSpec, serviceName string, multinodeDeployer MultinodeDeployer) {
+func (b *NoopBackend) UpdateContainer(container *corev1.Container, numberOfNodes int32, role Role, component *v1alpha1.DynamoComponentDeploymentSharedSpec, serviceName string, multinodeDeployer MultinodeDeployer) {
 	// No-op: frontend, planner, router, etc. don't need backend-specific processing
 }
 
-func (b *NoopBackend) UpdatePodSpec(podSpec *corev1.PodSpec, numberOfNodes int32, role Role, component *v1alpha1.DynamoComponentDeploymentOverridesSpec, serviceName string) {
+func (b *NoopBackend) UpdatePodSpec(podSpec *corev1.PodSpec, numberOfNodes int32, role Role, component *v1alpha1.DynamoComponentDeploymentSharedSpec, serviceName string) {
 	// No-op: frontend, planner, router, etc. don't need backend-specific processing
 }
 
@@ -691,7 +691,7 @@ func addStandardEnvVars(container *corev1.Container, controllerConfig controller
 //
 //nolint:gocyclo
 func GenerateBasePodSpec(
-	component *v1alpha1.DynamoComponentDeploymentOverridesSpec,
+	component *v1alpha1.DynamoComponentDeploymentSharedSpec,
 	backendFramework BackendFramework,
 	secretsRetriever SecretsRetriever,
 	parentGraphDeploymentName string,
@@ -861,7 +861,7 @@ func setMetricsLabels(labels map[string]string, dynamoGraphDeployment *v1alpha1.
 	labels[commonconsts.KubeLabelMetricsEnabled] = commonconsts.KubeLabelValueTrue
 }
 
-func generateComponentContext(component *v1alpha1.DynamoComponentDeploymentOverridesSpec, parentGraphDeploymentName string, namespace string, numberOfNodes int32) ComponentContext {
+func generateComponentContext(component *v1alpha1.DynamoComponentDeploymentSharedSpec, parentGraphDeploymentName string, namespace string, numberOfNodes int32) ComponentContext {
 	componentContext := ComponentContext{
 		numberOfNodes:                  numberOfNodes,
 		ParentGraphDeploymentName:      parentGraphDeploymentName,
@@ -875,7 +875,7 @@ func generateComponentContext(component *v1alpha1.DynamoComponentDeploymentOverr
 
 // GeneratePodSpecForComponent creates a PodSpec for Grove deployments (simplified wrapper)
 func GeneratePodSpecForComponent(
-	component *v1alpha1.DynamoComponentDeploymentOverridesSpec,
+	component *v1alpha1.DynamoComponentDeploymentSharedSpec,
 	backendFramework BackendFramework,
 	secretsRetriever SecretsRetriever,
 	dynamoDeployment *v1alpha1.DynamoGraphDeployment,
@@ -1004,7 +1004,7 @@ func GenerateGrovePodCliqueSet(
 	return controller_common.CanonicalizePodCliqueSet(gangSet), nil
 }
 
-func generateLabels(component *v1alpha1.DynamoComponentDeploymentOverridesSpec, dynamoDeployment *v1alpha1.DynamoGraphDeployment, componentName string) (map[string]string, error) {
+func generateLabels(component *v1alpha1.DynamoComponentDeploymentSharedSpec, dynamoDeployment *v1alpha1.DynamoGraphDeployment, componentName string) (map[string]string, error) {
 	labels := make(map[string]string)
 	labels[commonconsts.KubeLabelDynamoSelector] = GetDynamoComponentName(dynamoDeployment, componentName)
 	labels[commonconsts.KubeLabelDynamoGraphDeploymentName] = dynamoDeployment.Name
@@ -1033,7 +1033,7 @@ func generateLabels(component *v1alpha1.DynamoComponentDeploymentOverridesSpec,
 	return labels, nil
 }
 
-func generateAnnotations(component *v1alpha1.DynamoComponentDeploymentOverridesSpec) (map[string]string, error) {
+func generateAnnotations(component *v1alpha1.DynamoComponentDeploymentSharedSpec) (map[string]string, error) {
 	annotations := make(map[string]string)
 	if component.Annotations != nil {
 		err := mergo.Merge(&annotations, component.Annotations, mergo.WithOverride)
@@ -1147,7 +1147,7 @@ func determineBackendFramework(
 // 3. Return error if worker has neither detection nor explicit config
 // Also validates consistency between detected and explicit if both exist
 func getBackendFrameworkFromComponent(
-	component *v1alpha1.DynamoComponentDeploymentOverridesSpec,
+	component *v1alpha1.DynamoComponentDeploymentSharedSpec,
 	dynamoDeployment *v1alpha1.DynamoGraphDeployment,
 ) (BackendFramework, error) {
 	// Extract command/args from component
@@ -1170,10 +1170,8 @@ func getBackendFrameworkFromComponent(
 
 // ConvertDynamoComponentDeploymentToSpec converts a DynamoComponentDeployment to our component spec interface
 // This is a helper for the controller to use our backend logic
-func ConvertDynamoComponentDeploymentToSpec(dynComponent *v1alpha1.DynamoComponentDeployment) *v1alpha1.DynamoComponentDeploymentOverridesSpec {
-	return &v1alpha1.DynamoComponentDeploymentOverridesSpec{
-		DynamoComponentDeploymentSharedSpec: *dynComponent.Spec.DynamoComponentDeploymentSharedSpec.DeepCopy(),
-	}
+func ConvertDynamoComponentDeploymentToSpec(dynComponent *v1alpha1.DynamoComponentDeployment) *v1alpha1.DynamoComponentDeploymentSharedSpec {
+	return dynComponent.Spec.DynamoComponentDeploymentSharedSpec.DeepCopy()
 }
 
 // GetBackendFrameworkFromDynamoComponent determines backend framework for a DynamoComponentDeployment

docs/kubernetes/api_reference.md

@@ -85,6 +85,7 @@ DynamoComponentDeployment is the Schema for the dynamocomponentdeployments API
 
 _Appears in:_
 - [DynamoComponentDeploymentSpec](#dynamocomponentdeploymentspec)
+- [DynamoGraphDeploymentSpec](#dynamographdeploymentspec)
 
 | Field | Description | Default | Validation |
 | --- | --- | --- | --- |
@@ -92,12 +93,13 @@ _Appears in:_
 | `labels` _object (keys:string, values:string)_ | Labels to add to generated Kubernetes resources for this component. |  |  |
 | `serviceName` _string_ | The name of the component |  |  |
 | `componentType` _string_ | ComponentType indicates the role of this component (for example, "main"). |  |  |
+| `subComponentType` _string_ | SubComponentType indicates the sub-role of this component (for example, "prefill"). |  |  |
 | `dynamoNamespace` _string_ | Dynamo namespace of the service (allows to override the Dynamo namespace of the service defined in annotations inside the Dynamo archive) |  |  |
 | `resources` _[Resources](#resources)_ | Resources requested and limits for this component, including CPU, memory,<br />GPUs/devices, and any runtime-specific resources. |  |  |
 | `autoscaling` _[Autoscaling](#autoscaling)_ | Autoscaling config for this component (replica range, target utilization, etc.). |  |  |
 | `envs` _[EnvVar](https://kubernetes.io/docs/reference/generated/kubernetes-api/v1.28/#envvar-v1-core) array_ | Envs defines additional environment variables to inject into the component containers. |  |  |
 | `envFromSecret` _string_ | EnvFromSecret references a Secret whose key/value pairs will be exposed as<br />environment variables in the component containers. |  |  |
-| `pvc` _[PVC](#pvc)_ | PVC config describing volumes to be mounted by the component. |  |  |
+| `volumeMounts` _[VolumeMount](#volumemount) array_ | VolumeMounts references PVCs defined at the top level for volumes to be mounted by the component. |  |  |
 | `ingress` _[IngressSpec](#ingressspec)_ | Ingress config to expose the component outside the cluster (or through a service mesh). |  |  |
 | `sharedMemory` _[SharedMemorySpec](#sharedmemoryspec)_ | SharedMemory controls the tmpfs mounted at /dev/shm (enable/disable and size). |  |  |
 | `extraPodMetadata` _[ExtraPodMetadata](#extrapodmetadata)_ | ExtraPodMetadata adds labels/annotations to the created Pods. |  |  |
@@ -121,19 +123,18 @@ _Appears in:_
 
 | Field | Description | Default | Validation |
 | --- | --- | --- | --- |
-| `dynamoComponent` _string_ | DynamoComponent selects the Dynamo component from the archive to deploy.<br />Typically corresponds to a component defined in the packaged Dynamo artifacts. |  |  |
-| `dynamoTag` _string_ | contains the tag of the DynamoComponent: for example, "my_package:MyService" |  |  |
 | `backendFramework` _string_ | BackendFramework specifies the backend framework (e.g., "sglang", "vllm", "trtllm") |  | Enum: [sglang vllm trtllm] <br /> |
 | `annotations` _object (keys:string, values:string)_ | Annotations to add to generated Kubernetes resources for this component<br />(such as Pod, Service, and Ingress when applicable). |  |  |
 | `labels` _object (keys:string, values:string)_ | Labels to add to generated Kubernetes resources for this component. |  |  |
 | `serviceName` _string_ | The name of the component |  |  |
 | `componentType` _string_ | ComponentType indicates the role of this component (for example, "main"). |  |  |
+| `subComponentType` _string_ | SubComponentType indicates the sub-role of this component (for example, "prefill"). |  |  |
 | `dynamoNamespace` _string_ | Dynamo namespace of the service (allows to override the Dynamo namespace of the service defined in annotations inside the Dynamo archive) |  |  |
 | `resources` _[Resources](#resources)_ | Resources requested and limits for this component, including CPU, memory,<br />GPUs/devices, and any runtime-specific resources. |  |  |
 | `autoscaling` _[Autoscaling](#autoscaling)_ | Autoscaling config for this component (replica range, target utilization, etc.). |  |  |
 | `envs` _[EnvVar](https://kubernetes.io/docs/reference/generated/kubernetes-api/v1.28/#envvar-v1-core) array_ | Envs defines additional environment variables to inject into the component containers. |  |  |
 | `envFromSecret` _string_ | EnvFromSecret references a Secret whose key/value pairs will be exposed as<br />environment variables in the component containers. |  |  |
-| `pvc` _[PVC](#pvc)_ | PVC config describing volumes to be mounted by the component. |  |  |
+| `volumeMounts` _[VolumeMount](#volumemount) array_ | VolumeMounts references PVCs defined at the top level for volumes to be mounted by the component. |  |  |
 | `ingress` _[IngressSpec](#ingressspec)_ | Ingress config to expose the component outside the cluster (or through a service mesh). |  |  |
 | `sharedMemory` _[SharedMemorySpec](#sharedmemoryspec)_ | SharedMemory controls the tmpfs mounted at /dev/shm (enable/disable and size). |  |  |
 | `extraPodMetadata` _[ExtraPodMetadata](#extrapodmetadata)_ | ExtraPodMetadata adds labels/annotations to the created Pods. |  |  |
@@ -176,8 +177,9 @@ _Appears in:_
 
 | Field | Description | Default | Validation |
 | --- | --- | --- | --- |
-| `dynamoGraph` _string_ | DynamoGraph selects the graph (workflow/topology) to deploy. This must match<br />a graph name packaged with the Dynamo archive. |  |  |
-| `envs` _[EnvVar](https://kubernetes.io/docs/reference/generated/kubernetes-api/v1.28/#envvar-v1-core) array_ | Envs are environment variables applied to all services in the graph unless<br />overridden by service-specific configuration. |  | Optional: {} <br /> |
+| `pvcs` _[PVC](#pvc) array_ | PVCs defines a list of persistent volume claims that can be referenced by components.<br />Each PVC must have a unique name that can be referenced in component specifications. |  | Optional: {} <br /> |
+| `services` _object (keys:string, values:[DynamoComponentDeploymentSharedSpec](#dynamocomponentdeploymentsharedspec))_ | Services are the services to deploy as part of this deployment. |  | Optional: {} <br /> |
+| `envs` _[EnvVar](https://kubernetes.io/docs/reference/generated/kubernetes-api/v1.28/#envvar-v1-core) array_ | Envs are environment variables applied to all services in the deployment unless<br />overridden by service-specific configuration. |  | Optional: {} <br /> |
 | `backendFramework` _string_ | BackendFramework specifies the backend framework (e.g., "sglang", "vllm", "trtllm"). |  | Enum: [sglang vllm trtllm] <br /> |
 
 
@@ -266,17 +268,15 @@ _Appears in:_
 
 
 _Appears in:_
-- [DynamoComponentDeploymentSharedSpec](#dynamocomponentdeploymentsharedspec)
-- [DynamoComponentDeploymentSpec](#dynamocomponentdeploymentspec)
+- [DynamoGraphDeploymentSpec](#dynamographdeploymentspec)
 
 | Field | Description | Default | Validation |
 | --- | --- | --- | --- |
 | `create` _boolean_ | Create indicates to create a new PVC |  |  |
-| `name` _string_ | Name is the name of the PVC |  |  |
-| `storageClass` _string_ | StorageClass to be used for PVC creation. Leave it as empty if the PVC is already created. |  |  |
-| `size` _[Quantity](#quantity)_ | Size of the NIM cache in Gi, used during PVC creation |  |  |
-| `volumeAccessMode` _[PersistentVolumeAccessMode](https://kubernetes.io/docs/reference/generated/kubernetes-api/v1.28/#persistentvolumeaccessmode-v1-core)_ | VolumeAccessMode is the volume access mode of the PVC |  |  |
-| `mountPoint` _string_ |  |  |  |
+| `name` _string_ | Name is the name of the PVC |  | Required: {} <br /> |
+| `storageClass` _string_ | StorageClass to be used for PVC creation. Required when create is true. |  |  |
+| `size` _[Quantity](#quantity)_ | Size of the volume in Gi, used during PVC creation. Required when create is true. |  |  |
+| `volumeAccessMode` _[PersistentVolumeAccessMode](https://kubernetes.io/docs/reference/generated/kubernetes-api/v1.28/#persistentvolumeaccessmode-v1-core)_ | VolumeAccessMode is the volume access mode of the PVC. Required when create is true. |  |  |
 
 
 #### SharedMemorySpec
@@ -297,3 +297,247 @@ _Appears in:_
 | `size` _[Quantity](#quantity)_ |  |  |  |
 
 
+#### VolumeMount
+
+
+
+VolumeMount references a PVC defined at the top level for volumes to be mounted by the component
+
+
+
+_Appears in:_
+- [DynamoComponentDeploymentSharedSpec](#dynamocomponentdeploymentsharedspec)
+- [DynamoComponentDeploymentSpec](#dynamocomponentdeploymentspec)
+
+| Field | Description | Default | Validation |
+| --- | --- | --- | --- |
+| `name` _string_ | Name references a PVC name defined in the top-level PVCs map |  | Required: {} <br /> |
+| `mountPoint` _string_ | MountPoint specifies where to mount the volume.<br />If useAsCompilationCache is true and mountPoint is not specified,<br />a backend-specific default will be used. |  |  |
+| `useAsCompilationCache` _boolean_ | UseAsCompilationCache indicates this volume should be used as a compilation cache.<br />When true, backend-specific environment variables will be set and default mount points may be used. | false |  |
+
+
+# Operator Default Values Injection
+
+The Dynamo operator automatically applies default values to various fields when they are not explicitly specified in your deployments. These defaults include:
+
+- **Health Probes**: Startup, liveness, and readiness probes are configured differently for frontend, worker, and planner components. For example, worker components receive a startup probe with a 2-hour timeout (720 failures × 10 seconds) to accommodate long model loading times.
+
+- **Shared Memory**: All components receive an 8Gi shared memory volume mounted at `/dev/shm` by default (can be disabled or resized via the `sharedMemory` field).
+
+- **Environment Variables**: Components automatically receive environment variables like `DYN_NAMESPACE`, `DYN_PARENT_DGD_K8S_NAME`, `DYNAMO_PORT`, and backend-specific variables.
+
+- **Pod Configuration**: Default `terminationGracePeriodSeconds` of 60 seconds and `restartPolicy: Always`.
+
+- **Autoscaling**: When enabled without explicit metrics, defaults to CPU-based autoscaling with 80% target utilization.
+
+- **Backend-Specific Behavior**: For multinode deployments, probes are automatically modified or removed for worker nodes depending on the backend framework (VLLM, SGLang, or TensorRT-LLM).
+
+## Pod Specification Defaults
+
+All components receive the following pod-level defaults unless overridden:
+
+- **`terminationGracePeriodSeconds`**: `60` seconds
+- **`restartPolicy`**: `Always`
+
+## Shared Memory Configuration
+
+Shared memory is enabled by default for all components:
+
+- **Enabled**: `true` (unless explicitly disabled via `sharedMemory.disabled`)
+- **Size**: `8Gi`
+- **Mount Path**: `/dev/shm`
+- **Volume Type**: `emptyDir` with `memory` medium
+
+To disable shared memory or customize the size, use the `sharedMemory` field in your component specification.
+
+## Health Probes by Component Type
+
+The operator applies different default health probes based on the component type.
+
+### Frontend Components
+
+Frontend components receive the following probe configurations:
+
+**Liveness Probe:**
+- **Type**: HTTP GET
+- **Path**: `/health`
+- **Port**: `http` (8000)
+- **Initial Delay**: 60 seconds
+- **Period**: 60 seconds
+- **Timeout**: 30 seconds
+- **Failure Threshold**: 10
+
+**Readiness Probe:**
+- **Type**: Exec command
+- **Command**: `curl -s http://localhost:${DYNAMO_PORT}/health | jq -e ".status == \"healthy\""`
+- **Initial Delay**: 60 seconds
+- **Period**: 60 seconds
+- **Timeout**: 30 seconds
+- **Failure Threshold**: 10
+
+### Worker Components
+
+Worker components receive the following probe configurations:
+
+**Liveness Probe:**
+- **Type**: HTTP GET
+- **Path**: `/live`
+- **Port**: `system` (9090)
+- **Period**: 5 seconds
+- **Timeout**: 30 seconds
+- **Failure Threshold**: 1
+
+**Readiness Probe:**
+- **Type**: HTTP GET
+- **Path**: `/health`
+- **Port**: `system` (9090)
+- **Period**: 10 seconds
+- **Timeout**: 30 seconds
+- **Failure Threshold**: 60
+
+**Startup Probe:**
+- **Type**: HTTP GET
+- **Path**: `/live`
+- **Port**: `system` (9090)
+- **Period**: 10 seconds
+- **Timeout**: 5 seconds
+- **Failure Threshold**: 720 (allows up to 2 hours for startup: 10s × 720 = 7200s)
+
+:::{note}
+For larger models (typically >70B parameters) or slower storage systems, you may need to increase the `failureThreshold` to allow more time for model loading. Calculate the required threshold based on your expected startup time: `failureThreshold = (expected_startup_seconds / period)`. Override the startup probe in your component specification if the default 2-hour window is insufficient.
+:::
+
+### Multinode Deployment Probe Modifications
+
+For multinode deployments, the operator modifies probes based on the backend framework and node role:
+
+#### VLLM Backend
+- **Worker nodes**: All probes (liveness, readiness, startup) are removed
+
+#### SGLang Backend
+- **Worker nodes**: All probes (liveness, readiness, startup) are removed
+
+#### TensorRT-LLM Backend
+- **Leader nodes**: All probes remain unchanged
+- **Worker nodes**:
+  - Liveness and startup probes are removed
+  - Readiness probe is replaced with a TCP socket check on SSH port (2222):
+    - **Initial Delay**: 20 seconds
+    - **Period**: 20 seconds
+    - **Timeout**: 5 seconds
+    - **Failure Threshold**: 10
+
+## Environment Variables
+
+The operator automatically injects environment variables based on component type and configuration:
+
+### All Components
+
+- **`DYN_NAMESPACE`**: The Dynamo namespace for the component
+- **`DYN_PARENT_DGD_K8S_NAME`**: The parent DynamoGraphDeployment Kubernetes resource name
+- **`DYN_PARENT_DGD_K8S_NAMESPACE`**: The parent DynamoGraphDeployment Kubernetes namespace
+
+### Frontend Components
+
+- **`DYNAMO_PORT`**: `8000`
+- **`DYN_HTTP_PORT`**: `8000`
+
+### Worker Components
+
+- **`DYN_SYSTEM_ENABLED`**: `true`
+- **`DYN_SYSTEM_USE_ENDPOINT_HEALTH_STATUS`**: `["generate"]`
+- **`DYN_SYSTEM_PORT`**: `9090`
+
+### Planner Components
+
+- **`PLANNER_PROMETHEUS_PORT`**: `9085`
+
+### VLLM Backend (with compilation cache)
+
+When a volume mount is configured with `useAsCompilationCache: true`:
+- **`VLLM_CACHE_ROOT`**: Set to the mount point of the cache volume
+
+## Service Account
+
+Planner components automatically receive the following service account:
+
+- **`serviceAccountName`**: `planner-serviceaccount`
+
+## Image Pull Secrets
+
+The operator automatically discovers and injects image pull secrets for container images. When a component specifies a container image, the operator:
+
+1. Scans all Kubernetes secrets of type `kubernetes.io/dockerconfigjson` in the component's namespace
+2. Extracts the docker registry server URLs from each secret's authentication configuration
+3. Matches the container image's registry host against the discovered registry URLs
+4. Automatically injects matching secrets as `imagePullSecrets` in the pod specification
+
+This eliminates the need to manually specify image pull secrets for each component. The operator maintains an internal index of docker secrets and their associated registries, refreshing this index periodically.
+
+**To disable automatic image pull secret discovery** for a specific component, add the following annotation:
+
+```yaml
+annotations:
+  nvidia.com/disable-image-pull-secret-discovery: "true"
+```
+
+## Autoscaling Defaults
+
+When autoscaling is enabled but no metrics are specified, the operator applies:
+
+- **Default Metric**: CPU utilization
+- **Target Average Utilization**: `80%`
+
+## Port Configurations
+
+Default container ports are configured based on component type:
+
+### Frontend Components
+- **Port**: 8000
+- **Protocol**: TCP
+- **Name**: `http`
+
+### Worker Components
+- **Port**: 9090
+- **Protocol**: TCP
+- **Name**: `system`
+
+### Planner Components
+- **Port**: 9085
+- **Protocol**: TCP
+- **Name**: `metrics`
+
+## Backend-Specific Configurations
+
+### VLLM
+- **Ray Head Port**: 6379 (for multinode deployments)
+
+### SGLang
+- **Distribution Init Port**: 29500 (for multinode deployments)
+
+### TensorRT-LLM
+- **SSH Port**: 2222 (for multinode MPI communication)
+- **OpenMPI Environment**: `OMPI_MCA_orte_keep_fqdn_hostnames=1`
+
+## Implementation Reference
+
+For users who want to understand the implementation details or contribute to the operator, the default values described in this document are set in the following source files:
+
+- **Health Probes & Pod Specifications**: [`internal/dynamo/graph.go`](https://github.com/ai-dynamo/dynamo/blob/main/deploy/cloud/operator/internal/dynamo/graph.go) - Contains the main logic for applying default probes, environment variables, shared memory, and pod configurations
+- **Component-Specific Defaults**:
+  - [`internal/dynamo/component_frontend.go`](https://github.com/ai-dynamo/dynamo/blob/main/deploy/cloud/operator/internal/dynamo/component_frontend.go)
+  - [`internal/dynamo/component_worker.go`](https://github.com/ai-dynamo/dynamo/blob/main/deploy/cloud/operator/internal/dynamo/component_worker.go)
+  - [`internal/dynamo/component_planner.go`](https://github.com/ai-dynamo/dynamo/blob/main/deploy/cloud/operator/internal/dynamo/component_planner.go)
+- **Image Pull Secrets**: [`internal/secrets/docker.go`](https://github.com/ai-dynamo/dynamo/blob/main/deploy/cloud/operator/internal/secrets/docker.go) - Implements the docker secret indexer and automatic discovery
+- **Backend-Specific Behavior**:
+  - [`internal/dynamo/backend_vllm.go`](https://github.com/ai-dynamo/dynamo/blob/main/deploy/cloud/operator/internal/dynamo/backend_vllm.go)
+  - [`internal/dynamo/backend_sglang.go`](https://github.com/ai-dynamo/dynamo/blob/main/deploy/cloud/operator/internal/dynamo/backend_sglang.go)
+  - [`internal/dynamo/backend_trtllm.go`](https://github.com/ai-dynamo/dynamo/blob/main/deploy/cloud/operator/internal/dynamo/backend_trtllm.go)
+- **Constants & Annotations**: [`internal/consts/consts.go`](https://github.com/ai-dynamo/dynamo/blob/main/deploy/cloud/operator/internal/consts/consts.go) - Defines annotation keys and other constants
+
+## Notes
+
+- All these defaults can be overridden by explicitly specifying values in your DynamoComponentDeployment or DynamoGraphDeployment resources
+- User-specified probes (via `livenessProbe`, `readinessProbe`, or `startupProbe` fields) take precedence over operator defaults
+- For multinode deployments, some defaults are modified or removed as described above to accommodate distributed execution patterns
+- The `extraPodSpec.mainContainer` field can be used to override probe configurations set by the operator
\ No newline at end of file