gpu-lab/workloads/llama-server/deployment.yaml at main · CelestialCreator/gpu-lab · GitHub

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
apiVersion: apps/v1
kind: Deployment
metadata:
  name: llama-server
  namespace: default
  labels:
    app: llama-server
spec:
  replicas: 1
  strategy:
    type: Recreate
  selector:
    matchLabels:
      app: llama-server
  template:
    metadata:
      labels:
        app: llama-server
    spec:
      containers:
      - name: llama-server
        image: localhost:5000/llama-server:latest
        args:
        - --model
        - /workspace/models/Qwen3.5-35B-A3B-GGUF/Qwen3.5-35B-A3B-Q8_0.gguf
        - --port
        - "8080"
        - --host
        - "0.0.0.0"
        - --temp
        - "0.6"
        - --top-p
        - "0.95"
        - --top-k
        - "20"
        - --cache-type-k
        - q8_0
        - --cache-type-v
        - q8_0
        - --ctx-size
        - "262144"
        - -ngl
        - "99"
        - --tensor-split
        - "32,8"
        env:
        - name: NVIDIA_VISIBLE_DEVICES
          value: "${GPU_UUID}"
        - name: CUDA_VISIBLE_DEVICES
          value: "0,1"
        resources:
          limits:
            nvidia.com/gpu: "2"
            memory: 12Gi
          requests:
            nvidia.com/gpu: "2"
            memory: 8Gi
        ports:
        - containerPort: 8080
          name: http
        volumeMounts:
        - name: workspace
          mountPath: /workspace
        - name: dshm
          mountPath: /dev/shm
      volumes:
      - name: workspace
        hostPath:
          path: /home/akshay/llama-workspace
          type: DirectoryOrCreate
      - name: dshm
        emptyDir:
          medium: Memory
          sizeLimit: 1Gi
---
apiVersion: v1
kind: Service
metadata:
  name: llama-server
  namespace: default
spec:
  type: NodePort
  selector:
    app: llama-server
  ports:
  - port: 8080
    targetPort: 8080
    nodePort: 30080