disagg-multinode.yaml

# SPDX-FileCopyrightText: Copyright (c) 2025-2026 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
# SPDX-License-Identifier: Apache-2.0

apiVersion: nvidia.com/v1alpha1
kind: DynamoGraphDeployment
metadata:
  name: sglang-disagg-multinode
spec:
  envs:
  - name: HF_TOKEN
    valueFrom:
      secretKeyRef:
        name: hf-token-secret
        key: HF_TOKEN
  - name: GLOO_SOCKET_IFNAME
    value: "eth0"
  backendFramework: sglang
  services:
    Frontend:
      componentType: frontend
      replicas: 1
      extraPodSpec:
        mainContainer:
          image: my-registry/sglang-runtime:my-tag
    decode:
      multinode:
        nodeCount: 2
      envFromSecret: hf-token-secret
      componentType: worker
      replicas: 1
      resources:
        limits:
          gpu: "4"
      extraPodSpec:
        mainContainer:
          image: my-registry/sglang-runtime:my-tag
          workingDir: /workspace/examples/backends/sglang
          command:
          - python3
          - -m
          - dynamo.sglang
          args:
            - --model-path
            - Qwen/Qwen3-0.6B
            - --served-model-name
            - Qwen/Qwen3-0.6B
            - --tp-size
            - "8"
            - --trust-remote-code
            - --skip-tokenizer-init
            - --disaggregation-mode
            - decode
            - --disaggregation-transfer-backend
            - nixl
            - --disaggregation-bootstrap-port
            - "30001"
            - --host
            - "0.0.0.0"
            - --mem-fraction-static
            - "0.82"
    prefill:
      multinode:
        nodeCount: 2
      envFromSecret: hf-token-secret
      componentType: worker
      replicas: 1
      resources:
        limits:
          gpu: "4"
      extraPodSpec:
        mainContainer:
          image: my-registry/sglang-runtime:my-tag
          workingDir: /workspace/examples/backends/sglang
          command:
          - python3
          - -m
          - dynamo.sglang
          args:
            - --model-path
            - Qwen/Qwen3-0.6B
            - --served-model-name
            - Qwen/Qwen3-0.6B
            - --tp-size
            - "8"
            - --trust-remote-code
            - --skip-tokenizer-init
            - --disaggregation-mode
            - prefill
            - --disaggregation-transfer-backend
            - nixl
            - --disaggregation-bootstrap-port
            - "30001"
            - --mem-fraction-static
            - "0.82"
            - --host
            - "0.0.0.0"