kubectl-ai/modelserving/k8s/llm-server-rpc.yaml at main · mikebz/kubectl-ai · 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
kind: Deployment
apiVersion: apps/v1
metadata:
  name: llm-server-rpc
spec:
  replicas: 1
  selector:
    matchLabels:
      app: llm-server-rpc
  template:
    metadata:
      labels:
        app: llm-server-rpc
    spec:
      serviceAccountName: llm-server-rpc
      containers:
      - name: llm-server-rpc
        image: llamacpp-gemma3-12b-it-cpu:latest # placeholder value, replaced by deployment scripts
        env:
        - name: LLAMA_ARG_N_GPU_LAYERS
          value: "99"
        - name: LLAMA_ARG_FLASH_ATTN
          value: "yes"
        # - name: SERVERS
        #   value: rpc-server-0,rpc-server-1,rpc-server-2,rpc-server-3
        args:
        - --jinja # Needed for tool use, no env var
        - --rpc
        - rpc-server-0.rpc-server:50052,rpc-server-1.rpc-server:50052,rpc-server-2.rpc-server:50052,rpc-server-3.rpc-server:50052

---

kind: ServiceAccount
apiVersion: v1
metadata:
  name: llm-server-rpc

---

kind: Service
apiVersion: v1
metadata:
  name: llm-server-rpc
  labels:
    app: llm-server-rpc
spec:
  selector:
    app: llm-server-rpc
  ports:
  - name: http
    port: 80
    targetPort: 8080
    protocol: TCP