agg.yaml 885 Bytes
Newer Older
1
# SPDX-FileCopyrightText: Copyright (c) 2025-2026 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
2
# SPDX-License-Identifier: Apache-2.0
Alec's avatar
Alec committed
3

4
5
6
apiVersion: nvidia.com/v1alpha1
kind: DynamoGraphDeployment
metadata:
7
  name: vllm-agg
8
9
10
spec:
  services:
    Frontend:
11
      componentType: frontend
12
13
14
      replicas: 1
      extraPodSpec:
        mainContainer:
15
          image: nvcr.io/nvidia/ai-dynamo/vllm-runtime:my-tag
16
17
    VllmDecodeWorker:
      envFromSecret: hf-token-secret
18
      componentType: worker
19
      replicas: 1
20
21
22
      resources:
        limits:
          gpu: "1"
23
24
      extraPodSpec:
        mainContainer:
25
          image: nvcr.io/nvidia/ai-dynamo/vllm-runtime:my-tag
26
          workingDir: /workspace/examples/backends/vllm
27
          command:
28
29
30
            - python3
            - -m
            - dynamo.vllm
31
          args:
32
33
            - --model
            - Qwen/Qwen3-0.6B