# SPDX-FileCopyrightText: Copyright (c) 2025-2026 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
# SPDX-License-Identifier: Apache-2.0
#
# vLLM-based GlobalPlanner test: 2 prefill pools + 1 decode pool.
# Each pool SLA Planner reads router histogram metrics from cluster Prometheus
# and delegates scaling decisions to the GlobalPlanner.
#
# Architecture:
#   DGD gp-ctrl:      Frontend + GlobalRouter + GlobalPlanner
#   DGD gp-prefill-0: LocalRouter + VllmPrefillWorker (TP1) + Planner
#   DGD gp-prefill-1: LocalRouter + VllmPrefillWorker (TP2) + Planner
#   DGD gp-decode-0:  LocalRouter + VllmDecodeWorker  (TP1) + Planner
#
# Prerequisites:
#   - Cluster Prometheus deployed and scraping LocalRouter pods via PodMonitor
#   - HuggingFace token secret: kubectl create secret generic hf-token-secret \
#       --from-literal=HF_TOKEN=<your-token> -n ${K8S_NAMESPACE}
#
# Usage:
#   export K8S_NAMESPACE=... DYNAMO_IMAGE=... DYNAMO_VLLM_IMAGE=... MODEL_NAME=... STORAGE_CLASS_NAME=...
#   envsubst < global-planner-vllm-test.yaml | kubectl apply   -n ${K8S_NAMESPACE} -f -
#   envsubst < global-planner-vllm-test.yaml | kubectl delete  -n ${K8S_NAMESPACE} -f -
apiVersion: rbac.authorization.k8s.io/v1
kind: ClusterRoleBinding
metadata:
  name: ${K8S_NAMESPACE}-planner
roleRef:
  apiGroup: rbac.authorization.k8s.io
  kind: ClusterRole
  name: dynamo-platform-dynamo-operator-planner
subjects:
  - kind: ServiceAccount
    name: default
    namespace: ${K8S_NAMESPACE}
---
apiVersion: v1
kind: ConfigMap
metadata:
  name: gp-global-router-config
data:
  global_router_config.json: |
    {
      "num_prefill_pools": 2,
      "num_decode_pools": 1,
      "prefill_pool_dynamo_namespaces": [
        "${K8S_NAMESPACE}-gp-prefill-0",
        "${K8S_NAMESPACE}-gp-prefill-1"
      ],
      "decode_pool_dynamo_namespaces": [
        "${K8S_NAMESPACE}-gp-decode-0"
      ],
      "prefill_pool_selection_strategy": {
        "ttft_min": 10, "ttft_max": 3000, "ttft_resolution": 2,
        "isl_min": 0,   "isl_max": 32000, "isl_resolution": 2,
        "prefill_pool_mapping": [[0,1],[0,1]]
      },
      "decode_pool_selection_strategy": {
        "itl_min": 10,  "itl_max": 500,   "itl_resolution": 2,
        "context_length_min": 0, "context_length_max": 32000, "context_length_resolution": 2,
        "decode_pool_mapping": [[0,0],[0,0]]
      }
    }
---
# Shared model cache — ReadWriteMany PVC mounted into all vLLM worker pods.
# The model is downloaded once and reused across pods and restarts.
# Set storageClassName to a RWX-capable storage class available in your cluster
# (e.g. azurefile-csi-premium on AKS, nfs-csi on Nebius, efs-sc on EKS).
apiVersion: v1
kind: PersistentVolumeClaim
metadata:
  name: hf-model-cache
spec:
  accessModes:
    - ReadWriteMany
  storageClassName: ${STORAGE_CLASS_NAME}
  resources:
    requests:
      storage: 50Gi
---
apiVersion: nvidia.com/v1alpha1
kind: DynamoGraphDeployment
metadata:
  name: gp-ctrl
spec:
  services:
    Frontend:
      componentType: frontend
      replicas: 1
      extraPodSpec:
        imagePullSecrets:
          - name: docker-imagepullsecret
        mainContainer:
          image: ${DYNAMO_IMAGE}
          workingDir: /workspace
          command:
            - python3
            - -m
            - dynamo.frontend
          args:
            - --router-mode
            - round-robin
            - --namespace
            - ${K8S_NAMESPACE}-gp-ctrl
            - --model-name
            - ${MODEL_NAME}

    GlobalRouter:
      componentType: default
      replicas: 1
      extraPodSpec:
        imagePullSecrets:
          - name: docker-imagepullsecret
        tolerations:
          - key: "karpenter.sh/disrupted"
            operator: "Exists"
            effect: "NoSchedule"
        affinity:
          nodeAffinity:
            preferredDuringSchedulingIgnoredDuringExecution:
              - weight: 100
                preference:
                  matchExpressions:
                    - key: karpenter.sh/nodepool
                      operator: In
                      values:
                        - general-medium-storage
        volumes:
          - name: global-router-config
            configMap:
              name: gp-global-router-config
        mainContainer:
          image: ${DYNAMO_IMAGE}
          workingDir: /workspace
          command:
            - python3
            - -m
            - dynamo.global_router
          args:
            - --config
            - /config/global_router_config.json
            - --model-name
            - ${MODEL_NAME}
            - --namespace
            - ${K8S_NAMESPACE}-gp-ctrl
          volumeMounts:
            - name: global-router-config
              mountPath: /config
              readOnly: true

    GlobalPlanner:
      componentType: default
      replicas: 1
      extraPodSpec:
        imagePullSecrets:
          - name: docker-imagepullsecret
        mainContainer:
          image: ${DYNAMO_IMAGE}
          command:
            - python3
            - -m
            - dynamo.global_planner
---
apiVersion: nvidia.com/v1alpha1
kind: DynamoGraphDeployment
metadata:
  name: gp-prefill-0
spec:
  services:
    LocalRouter:
      componentType: default
      replicas: 1
      extraPodSpec:
        imagePullSecrets:
          - name: docker-imagepullsecret
        mainContainer:
          image: ${DYNAMO_IMAGE}
          workingDir: /workspace
          env:
            - name: DYN_SYSTEM_PORT
              value: "9090"
          command:
            - python3
            - -m
            - dynamo.router
          args:
            - --endpoint
            - ${K8S_NAMESPACE}-gp-prefill-0.prefill.generate
            - --router-block-size
            - "16"
            - --no-router-track-active-blocks

    VllmPrefillWorker:
      envFromSecret: hf-token-secret
      componentType: worker
      subComponentType: prefill
      replicas: 1
      resources:
        limits:
          gpu: "1"
      extraPodSpec:
        volumes:
          - name: hf-model-cache
            persistentVolumeClaim:
              claimName: hf-model-cache
        mainContainer:
          image: ${DYNAMO_VLLM_IMAGE}
          workingDir: /workspace/examples/backends/vllm
          command:
            - python3
            - -m
            - dynamo.vllm
          args:
            - --model
            - ${MODEL_NAME}
            - --tensor-parallel-size
            - "1"
            - --is-prefill-worker
          volumeMounts:
            - name: hf-model-cache
              mountPath: /home/dynamo/.cache/huggingface/hub

    Planner:
      componentType: planner
      replicas: 1
      extraPodSpec:
        imagePullSecrets:
          - name: docker-imagepullsecret
        mainContainer:
          image: ${DYNAMO_IMAGE}
          command:
            - python3
            - -m
            - dynamo.planner
          args:
            - --config
            - '{"environment":"global-planner","global_planner_namespace":"${K8S_NAMESPACE}-gp-ctrl","backend":"vllm","mode":"prefill","enable_load_scaling":false,"enable_throughput_scaling":true,"throughput_metrics_source":"router","ttft":2000,"max_gpu_budget":-1,"prefill_engine_num_gpu":1,"model_name":"${MODEL_NAME}","profile_results_dir":"/workspace/tests/planner/profiling_results/H200_TP1P_TP1D"}'
---
apiVersion: nvidia.com/v1alpha1
kind: DynamoGraphDeployment
metadata:
  name: gp-prefill-1
spec:
  services:
    LocalRouter:
      componentType: default
      replicas: 1
      extraPodSpec:
        imagePullSecrets:
          - name: docker-imagepullsecret
        mainContainer:
          image: ${DYNAMO_IMAGE}
          workingDir: /workspace
          env:
            - name: DYN_SYSTEM_PORT
              value: "9090"
          command:
            - python3
            - -m
            - dynamo.router
          args:
            - --endpoint
            - ${K8S_NAMESPACE}-gp-prefill-1.prefill.generate
            - --router-block-size
            - "16"
            - --no-router-track-active-blocks

    VllmPrefillWorker:
      envFromSecret: hf-token-secret
      componentType: worker
      subComponentType: prefill
      replicas: 1
      resources:
        limits:
          gpu: "2"
      extraPodSpec:
        volumes:
          - name: hf-model-cache
            persistentVolumeClaim:
              claimName: hf-model-cache
        mainContainer:
          image: ${DYNAMO_VLLM_IMAGE}
          workingDir: /workspace/examples/backends/vllm
          command:
            - python3
            - -m
            - dynamo.vllm
          args:
            - --model
            - ${MODEL_NAME}
            - --tensor-parallel-size
            - "2"
            - --is-prefill-worker
          volumeMounts:
            - name: hf-model-cache
              mountPath: /home/dynamo/.cache/huggingface/hub

    Planner:
      componentType: planner
      replicas: 1
      extraPodSpec:
        imagePullSecrets:
          - name: docker-imagepullsecret
        mainContainer:
          image: ${DYNAMO_IMAGE}
          command:
            - python3
            - -m
            - dynamo.planner
          args:
            - --config
            - '{"environment":"global-planner","global_planner_namespace":"${K8S_NAMESPACE}-gp-ctrl","backend":"vllm","mode":"prefill","enable_load_scaling":false,"enable_throughput_scaling":true,"throughput_metrics_source":"router","ttft":2000,"max_gpu_budget":-1,"prefill_engine_num_gpu":2,"model_name":"${MODEL_NAME}","profile_results_dir":"/workspace/tests/planner/profiling_results/H200_TP1P_TP1D"}'
---
apiVersion: nvidia.com/v1alpha1
kind: DynamoGraphDeployment
metadata:
  name: gp-decode-0
spec:
  services:
    LocalRouter:
      componentType: default
      replicas: 1
      extraPodSpec:
        imagePullSecrets:
          - name: docker-imagepullsecret
        mainContainer:
          image: ${DYNAMO_IMAGE}
          workingDir: /workspace
          env:
            - name: DYN_SYSTEM_PORT
              value: "9090"
          command:
            - python3
            - -m
            - dynamo.router
          args:
            - --endpoint
            - ${K8S_NAMESPACE}-gp-decode-0.backend.generate
            - --router-block-size
            - "16"
            - --router-kv-overlap-score-weight
            - "0"

    VllmDecodeWorker:
      envFromSecret: hf-token-secret
      componentType: worker
      subComponentType: decode
      replicas: 1
      resources:
        limits:
          gpu: "1"
      extraPodSpec:
        volumes:
          - name: hf-model-cache
            persistentVolumeClaim:
              claimName: hf-model-cache
        mainContainer:
          image: ${DYNAMO_VLLM_IMAGE}
          workingDir: /workspace/examples/backends/vllm
          command:
            - python3
            - -m
            - dynamo.vllm
          args:
            - --model
            - ${MODEL_NAME}
            - --tensor-parallel-size
            - "1"
          volumeMounts:
            - name: hf-model-cache
              mountPath: /home/dynamo/.cache/huggingface/hub

    Planner:
      componentType: planner
      replicas: 1
      extraPodSpec:
        imagePullSecrets:
          - name: docker-imagepullsecret
        mainContainer:
          image: ${DYNAMO_IMAGE}
          command:
            - python3
            - -m
            - dynamo.planner
          args:
            - --config
            - '{"environment":"global-planner","global_planner_namespace":"${K8S_NAMESPACE}-gp-ctrl","backend":"vllm","mode":"decode","enable_load_scaling":false,"enable_throughput_scaling":true,"throughput_metrics_source":"router","itl":200,"max_gpu_budget":-1,"decode_engine_num_gpu":1,"model_name":"${MODEL_NAME}","profile_results_dir":"/workspace/tests/planner/profiling_results/H200_TP1P_TP1D"}'