diff --git a/recipes/components/dynamo-platform/values.yaml b/recipes/components/dynamo-platform/values.yaml index 86560e3f9..11105c163 100644 --- a/recipes/components/dynamo-platform/values.yaml +++ b/recipes/components/dynamo-platform/values.yaml @@ -12,30 +12,31 @@ # See the License for the specific language governing permissions and # limitations under the License. -# Dynamo Platform Helm values (v0.9.1) +# Dynamo Platform Helm values (v1.0.1) # NVIDIA Dynamo inference serving platform: operator and grove. # Provides OpenAI-compatible endpoints, KV-cache-aware routing, # disaggregated prefill/decode, and SLA-driven autoscaling. -# -# NOTE: etcd and NATS are disabled below. The Dynamo team confirmed these -# are no longer required (Kubernetes-native discovery replaces etcd, ZMQ -# replaces NATS for KV events). However, the current chart (v0.9.0) still -# renders --etcdAddr in operator args, and the runtime SDK still tries to -# connect to etcd for lease creation. This is a known upstream issue. -# Workloads must set DYN_STORE_KV=mem and DYN_EVENT_PLANE=zmq to bypass. + +# --- Global subchart controls (new in 1.0) --- +# Disable etcd — Kubernetes-native discovery replaces it for the operator. +# Runtime pods bypass etcd via DYN_STORE_KV=mem on workloads. +global: + etcd: + install: false + # Disable kai-scheduler sub-chart install — managed as a separate AICR component. + # Keep enabled: true so the Dynamo operator detects and uses the external scheduler. + kai-scheduler: + install: false + enabled: true + # Disable grove sub-chart install — managed as a separate AICR component. + # Keep enabled: true so the Dynamo operator uses the external grove for pod lifecycle. + grove: + install: false + enabled: true dynamo-operator: - controllerManager: - manager: - image: - # Pin to 0.9.0 — chart default renders 0.7.1 which lacks --enable-webhooks - tag: "0.9.0" - kubeRbacProxy: - image: - # gcr.io/kubebuilder/kube-rbac-proxy is no longer available after gcr.io deprecation. - # TODO: remove once upstream dynamo chart updates the default image reference. - repository: registry.k8s.io/kubebuilder/kube-rbac-proxy - tag: v0.15.0 + # Upgrade CRDs via the platform chart (no separate dynamo-crds 1.0 chart) + upgradeCRD: true # Use Kubernetes-native service discovery (operator-level only; # runtime pods still require DYN_STORE_KV=mem to skip etcd leases) @@ -55,16 +56,12 @@ dynamo-operator: kai-scheduler: enabled: false -# Grove operator for pod lifecycle management (required by DynamoGraphDeployment) -grove: - enabled: true +# Disable NATS sub-chart — ZMQ event plane replaces it. +# Runtime pods bypass NATS via DYN_EVENT_PLANE=zmq on workloads. +nats: + enabled: false -# Disable etcd — Kubernetes-native discovery replaces it for the operator. +# Disable etcd sub-chart — Kubernetes-native discovery replaces it for the operator. # Runtime pods bypass etcd via DYN_STORE_KV=mem on workloads. etcd: enabled: false - -# Disable NATS — ZMQ event plane replaces it. -# Runtime pods bypass NATS via DYN_EVENT_PLANE=zmq on workloads. -nats: - enabled: false diff --git a/recipes/overlays/gb200-eks-ubuntu-inference-dynamo.yaml b/recipes/overlays/gb200-eks-ubuntu-inference-dynamo.yaml index ca78d352b..9c858c0e6 100644 --- a/recipes/overlays/gb200-eks-ubuntu-inference-dynamo.yaml +++ b/recipes/overlays/gb200-eks-ubuntu-inference-dynamo.yaml @@ -53,7 +53,7 @@ spec: - name: dynamo-platform type: Helm source: https://helm.ngc.nvidia.com/nvidia/ai-dynamo - version: "0.9.0" + version: "1.0.1" valuesFile: components/dynamo-platform/values.yaml dependencyRefs: - dynamo-crds diff --git a/recipes/overlays/h100-aks-ubuntu-inference-dynamo.yaml b/recipes/overlays/h100-aks-ubuntu-inference-dynamo.yaml index 35b61b33c..90385731b 100644 --- a/recipes/overlays/h100-aks-ubuntu-inference-dynamo.yaml +++ b/recipes/overlays/h100-aks-ubuntu-inference-dynamo.yaml @@ -49,7 +49,7 @@ spec: - name: dynamo-platform type: Helm source: https://helm.ngc.nvidia.com/nvidia/ai-dynamo - version: "0.9.0" + version: "1.0.1" valuesFile: components/dynamo-platform/values.yaml dependencyRefs: - dynamo-crds diff --git a/recipes/overlays/h100-eks-ubuntu-inference-dynamo.yaml b/recipes/overlays/h100-eks-ubuntu-inference-dynamo.yaml index b41414f03..0fb98c237 100644 --- a/recipes/overlays/h100-eks-ubuntu-inference-dynamo.yaml +++ b/recipes/overlays/h100-eks-ubuntu-inference-dynamo.yaml @@ -53,7 +53,7 @@ spec: - name: dynamo-platform type: Helm source: https://helm.ngc.nvidia.com/nvidia/ai-dynamo - version: "0.9.0" + version: "1.0.1" valuesFile: components/dynamo-platform/values.yaml dependencyRefs: - dynamo-crds diff --git a/recipes/overlays/h100-gke-cos-inference-dynamo.yaml b/recipes/overlays/h100-gke-cos-inference-dynamo.yaml index a379b0369..f0d2b8ee5 100644 --- a/recipes/overlays/h100-gke-cos-inference-dynamo.yaml +++ b/recipes/overlays/h100-gke-cos-inference-dynamo.yaml @@ -49,7 +49,7 @@ spec: - name: dynamo-platform type: Helm source: https://helm.ngc.nvidia.com/nvidia/ai-dynamo - version: "0.9.0" + version: "1.0.1" valuesFile: components/dynamo-platform/values.yaml dependencyRefs: - dynamo-crds diff --git a/recipes/overlays/h100-kind-inference-dynamo.yaml b/recipes/overlays/h100-kind-inference-dynamo.yaml index 42a9cf36c..ff15bc238 100644 --- a/recipes/overlays/h100-kind-inference-dynamo.yaml +++ b/recipes/overlays/h100-kind-inference-dynamo.yaml @@ -43,7 +43,7 @@ spec: - name: dynamo-platform type: Helm source: https://helm.ngc.nvidia.com/nvidia/ai-dynamo - version: "0.9.0" + version: "1.0.1" valuesFile: components/dynamo-platform/values.yaml dependencyRefs: - dynamo-crds diff --git a/recipes/registry.yaml b/recipes/registry.yaml index 66d747550..3e9a69588 100644 --- a/recipes/registry.yaml +++ b/recipes/registry.yaml @@ -367,7 +367,7 @@ components: helm: defaultRepository: https://helm.ngc.nvidia.com/nvidia/ai-dynamo defaultChart: dynamo-platform - defaultVersion: "0.9.1" + defaultVersion: "1.0.1" defaultNamespace: dynamo-system nodeScheduling: system: