Container Service for Kubernetes：部署SGLang PD分離推理服務

部署PD分離架構推理服務。SGLang Prefill Server和Decode Server互動時序圖如下所示。

• 收到使用者推理請求後，Prefill Server將會建立一個Sender對象而Decode Server會建立一個Receiver對象。

• Prefill和Decode通過Handshake建立串連，Decode首先分配一塊顯存地址用於接收KVCache，Prefill Server完成計算後將KVCache傳送給Decode Server，Decode Server收到KVCache後繼續計算後續Token，直到完成使用者的推理請求。

步驟一：準備Qwen3-32B模型檔案

步驟二：部署SGLang PD分離架構的推理服務

本文使用RBG部署2P1D SGLang推理服務，部署架構圖如下所示。

1. 建立sglang_pd.yaml檔案。

   展開查看YAML程式碼範例。

   apiVersion: workloads.x-k8s.io/v1alpha1
   kind: RoleBasedGroup
   metadata:
     name: sglang-pd
   spec:
     roles:
       - name: scheduler
         replicas: 1
         dependencies: [ "decode", "prefill" ]
         template:
           spec:
             volumes:
               - name: model
                 persistentVolumeClaim:
                   claimName: llm-model
             containers:
               - name: scheduler
                 image: anolis-registry.cn-zhangjiakou.cr.aliyuncs.com/anolis-docker-images/docker-temp:0.3.4.post2-sglang0.4.10.post2-pytorch2.7.1.8-cuda12.8.1-py312-alinux3.2104
                 command:
                   - sh
                   - -c
                   - python3 -m sglang.srt.disaggregation.mini_lb --prefill http://sglang-pd-prefill-0.sglang-pd-prefill:8000 http://sglang-pd-prefill-1.sglang-pd-prefill:8000 --prefill-bootstrap-ports 34000 34000 --decode http://sglang-pd-decode-0.sglang-pd-decode:8000 --host 0.0.0.0 --port 8000
                 volumeMounts:
                   - mountPath: /models/Qwen3-32B/
                     name: model
   
       - name: prefill
         replicas: 2
         template:
           metadata:
             labels:
               alibabacloud.com/inference-workload: sglang-pd-prefill
               alibabacloud.com/inference_backend: sglang
           spec:
             volumes:
               - name: model
                 persistentVolumeClaim:
                   claimName: llm-model
               - name: dshm
                 emptyDir:
                   medium: Memory
                   sizeLimit: 15Gi
             containers:
               - name: sglang-prefill
                 image: anolis-registry.cn-zhangjiakou.cr.aliyuncs.com/anolis-docker-images/docker-temp:0.3.4.post2-sglang0.4.10.post2-pytorch2.7.1.8-cuda12.8.1-py312-alinux3.2104
                 imagePullPolicy: Always
                 env:
                   - name: POD_IP
                     valueFrom:
                       fieldRef:
                         fieldPath: status.podIP
                 command:
                   - sh
                   - -c
                   - python3 -m sglang.launch_server --tp 2 --model-path /models/Qwen3-32B/ --disaggregation-mode prefill --port 8000 --disaggregation-bootstrap-port 34000 --host $(POD_IP) --enable-metrics
                 ports:
                   - containerPort: 8000
                     name: http
                   - containerPort: 34000
                     name: bootstrap
                 readinessProbe:
                   initialDelaySeconds: 30
                   periodSeconds: 10
                   tcpSocket:
                     port: 8000
                 resources:
                   limits:
                     nvidia.com/gpu: "2"
                     aliyun/erdma: 1
                     memory: "16Gi"
                     cpu: "4"
                   requests:
                     nvidia.com/gpu: "2"
                     aliyun/erdma: 1
                     memory: "16Gi"
                     cpu: "4"
                 volumeMounts:
                   - mountPath: /models/Qwen3-32B/
                     name: model
                   - mountPath: /dev/shm
                     name: dshm
   
       - name: decode
         replicas: 1
         template:
           metadata:
             labels:
               alibabacloud.com/inference-workload: sglang-pd-decode
               alibabacloud.com/inference_backend: sglang
           spec:
             volumes:
               - name: model
                 persistentVolumeClaim:
                   claimName: llm-model
               - name: dshm
                 emptyDir:
                   medium: Memory
                   sizeLimit: 15Gi
             containers:
               - name: sglang-decode
                 image: anolis-registry.cn-zhangjiakou.cr.aliyuncs.com/anolis-docker-images/docker-temp:0.3.4.post2-sglang0.4.10.post2-pytorch2.7.1.8-cuda12.8.1-py312-alinux3.2104
                 imagePullPolicy: Always
                 env:
                   - name: POD_IP
                     valueFrom:
                       fieldRef:
                         fieldPath: status.podIP
                 command:
                   - sh
                   - -c
                   - python3 -m sglang.launch_server --tp 2 --model-path /models/Qwen3-32B/ --disaggregation-mode decode --port 8000 --host $(POD_IP) --enable-metrics
                 ports:
                   - containerPort: 8000
                     name: http
                 readinessProbe:
                   initialDelaySeconds: 30
                   periodSeconds: 10
                   tcpSocket:
                     port: 8000
                 resources:
                   limits:
                     nvidia.com/gpu: "2"
                     aliyun/erdma: 1
                     memory: "16Gi"
                     cpu: "4"
                   requests:
                     nvidia.com/gpu: "2"
                     aliyun/erdma: 1
                     memory: "16Gi"
                     cpu: "4"
                 volumeMounts:
                   - mountPath: /models/Qwen3-32B/
                     name: model
                   - mountPath: /dev/shm
                     name: dshm
   ---
   apiVersion: v1
   kind: Service
   metadata:
     labels:
       app: sglang-pd
     name: sglang-pd
     namespace: default
   spec:
     ports:
       - name: http
         port: 8000
         protocol: TCP
         targetPort: 8000
     selector:
       rolebasedgroup.workloads.x-k8s.io/name: sglang-pd
       rolebasedgroup.workloads.x-k8s.io/role: scheduler
     type: ClusterIP
   

2. 部署SGLang PD分離推理服務。

   kubectl create -f sglang_pd.yaml

步驟三：驗證推理服務

1. 執行以下命令，在推理服務與本地環境之間建立連接埠轉寄。

   重要

   kubectl port-forward建立的連接埠轉寄不具備生產層級的可靠性、安全性和擴充性，因此僅適用於開發和調試目的，不適合在生產環境使用。更多關於Kubernetes叢集內生產可用的網路方案的資訊，請參見Ingress管理。

   kubectl port-forward svc/sglang-pd 8000:8000

   預期輸出：

   Forwarding from 127.0.0.1:8000 -> 8000
   Forwarding from [::1]:8000 -> 8000

2. 執行以下命令，向模型推理服務發送了一條樣本的模型推理請求。

   curl http://localhost:8000/v1/chat/completions -H "Content-Type: application/json"  -d '{"model": "/models/Qwen3-32B", "messages": [{"role": "user", "content": "測試一下"}], "max_tokens": 30, "temperature": 0.7, "top_p": 0.9, "seed": 10}'

   預期輸出：

   {"id":"29f3fdac693540bfa7808fc1a8701758","object":"chat.completion","created":1753695366,"model":"/models/Qwen3-32B","choices":[{"index":0,"message":{"role":"assistant","content":"<think>\n好的，使用者讓我測試一下，我需要先確認他們的具體需求。可能他們想測試我的功能，比如回答問題、產生內容","reasoning_content":null,"tool_calls":null},"logprobs":null,"finish_reason":"length","matched_stop":null}],"usage":{"prompt_tokens":10,"total_tokens":40,"completion_tokens":30,"prompt_tokens_details":null}}

   輸出結果表明模型可以根據給定的輸入（在這個例子中是一條測試訊息）產生相應的回複。