ipex-llm/docker/llm/serving/cpu/kubernetes/README.md

Deployment ipex-llm serving service in K8S environment

Image

To deploy IPEX-LLM-serving cpu in Kubernetes environment, please use this image: intelanalytics/ipex-llm-serving-cpu:2.5.0-SNAPSHOT

Before deployment

Models

In this document, we will use vicuna-7b-v1.5 as the deployment model.

After downloading the model, please change name from vicuna-7b-v1.5 to vicuna-7b-v1.5-ipex-llm to use ipex-llm as the backend. The ipex-llm backend will be used if model path contains ipex-llm. Otherwise, the original transformer-backend will be used.

You can download the model from here.

Kubernetes config

We recommend to setup your kubernetes cluster before deployment. Mostly importantly, please set cpu-management-policy to static by using this tutorial. Also, it would be great to also set the topology management policy to single-numa-node.

Machine config

Set hyper-threading to off, ensure that only physical cores are used during deployment.

Deployment

Reminder on OMP_NUM_THREADS

The entrypoint of the image will try to set OMP_NUM_THREADS to the correct number by reading configs from the runtime. However, this only happens correctly if the core-binding feature is enabled. If not, please set environment variable OMP_NUM_THREADS manually in the yaml file.

Vllm usage

If you want to use the vllm AsyncLLMEngine for serving, you should set the args -w vllm_worker in worker part of deployment.yaml.

PersistentVolume

We use the following yaml file for PersistentVolume deployment:

apiVersion: v1
kind: PersistentVolume
metadata:
  name: models-pv
  labels:
    app: models
spec:
  capacity:
    storage: 10Gi #Modify according to model size
  accessModes:
    - ReadWriteMany
  storageClassName: models
  nfs:
    path: YOUR_NFS_PATH
    server: YOUR_NFS_SERVER

Then you should upload model to YOUR_NFS_PATH

Controller

We use the following yaml file for controller deployment:

apiVersion: v1
kind: Pod
metadata:
  name: ipex-llm-fschat-a1234bd-controller
  labels:
    fastchat-appid: a1234bd
    fastchat-app-type: controller
spec:
  dnsPolicy: "ClusterFirst"
  containers:
  - name: fastchat-controller # fixed
    image: intelanalytics/ipex-llm-serving-cpu:2.5.0-SNAPSHOT
    imagePullPolicy: IfNotPresent
    env:
    - name: CONTROLLER_HOST # fixed
      value: "0.0.0.0"
    - name: CONTROLLER_PORT # fixed
      value: "21005"
    - name: API_HOST # fixed
      value: "0.0.0.0"
    - name: API_PORT # fixed
      value: "8000"
    - name: "GRADIO_PORT" # You can change this port
      value: "8002"
    ports:
      - containerPort: 21005
        name: con-port
      - containerPort: 8000
        name: api-port
    resources:
      requests:
        memory: 16Gi
        cpu: 4
      limits:
        memory: 16Gi
        cpu: 4
    args: ["-m", "controller"]
  restartPolicy: "Never"
---
# Service for the controller
apiVersion: v1
kind: Service
metadata:
  name: ipex-llm-a1234bd-fschat-controller-service
spec:
  # You may also want to change this to use the cluster's feature
  type: NodePort
  selector:
    fastchat-appid: a1234bd
    fastchat-app-type: controller
  ports:
    - name: cont-port
      protocol: TCP
      port: 21005
      targetPort: 21005
    - name: api-port
      protocol: TCP
      port: 8000
      targetPort: 8000

Worker

We use the following deployment for worker deployment:

apiVersion: apps/v1
kind: Deployment
metadata:
  name: ipex-llm-fschat-a1234bd-worker-deployment
spec:
  # Change this to the number you want
  replicas: 1
  selector:
    matchLabels:
      fastchat: worker
  template:
    metadata:
      labels:
        fastchat: worker
    spec:
      dnsPolicy: "ClusterFirst"
      containers:
      - name: fastchat-worker # fixed
        image: intelanalytics/ipex-llm-serving-cpu:2.5.0-SNAPSHOT
        imagePullPolicy: IfNotPresent
        env:
        - name: CONTROLLER_HOST # fixed
          value: ipex-llm-a1234bd-fschat-controller-service
        - name: CONTROLLER_PORT # fixed
          value: "21005"
        - name: WORKER_HOST # fixed
          valueFrom:
            fieldRef:
              fieldPath: status.podIP
        - name: WORKER_PORT # fixed
          value: "21841"
        - name: MODEL_PATH 
          value: "/llm/models/vicuna-7b-v1.5-ipex-llm/" # change this to your model
        - name: OMP_NUM_THREADS
          value: "16"
        resources:
          requests:
            memory: 32Gi
            cpu: 16
          limits:
            memory: 32Gi
            cpu: 16
        args: ["-m", "worker"] # add , "-w", "vllm_worker" if vllm_worker is expected
        volumeMounts:
          - name: llm-models
            mountPath: /llm/models/
      restartPolicy: "Always"
      volumes:
      - name: llm-models
        persistentVolumeClaim:
          claimName: models-pvc

You may want to change the MODEL_PATH variable in the yaml. Also, please remember to change the volume path accordingly.

Use gradio_web_ui

We have set port using GRADIO_PORT envrionment variable in deployment.yaml, you can use this command

k port-forward ipex-llm-fschat-a1234bd-controller --address 0.0.0.0 8002:8002

Then visit http://YOUR_HOST_IP:8002 to access ui.

Testing

Check pod ip and port mappings

If you need to access the serving on host , you can use kubectl get nodes -o wide to get internal ip and kubectl get service to get port mappings.

Using openai-python

First, install openai-python:

pip install --upgrade openai

Then, interact with model vicuna-7b-v1.5-ipex-llm:

import openai
openai.api_key = "EMPTY"
openai.api_base = "http://localhost:8000/v1"

model = "vicuna-7b-v1.5-ipex-llm"
prompt = "Once upon a time"

# create a completion
completion = openai.Completion.create(model=model, prompt=prompt, max_tokens=64)
# print the completion
print(prompt + completion.choices[0].text)

# create a chat completion
completion = openai.ChatCompletion.create(
  model=model,
  messages=[{"role": "user", "content": "Hello! What is your name?"}]
)
# print the completion
print(completion.choices[0].message.content)

cURL

cURL is another good tool for observing the output of the api.

Before using cURL, you should set your http_proxy and https_proxy to empty

export http_proxy=
export https_proxy=

For the following examples, you may also change the service deployment address.

List Models:

curl http://localhost:8000/v1/models

If you have jq installed, you can use it to format the output like this:

curl http://localhost:8000/v1/models | jq

Chat Completions:

curl http://localhost:8000/v1/chat/completions \
  -H "Content-Type: application/json" \
  -d '{
    "model": "YOUR_MODEL",
    "messages": [{"role": "user", "content": "Hello! What is your name?"}]
  }'

Text Completions:

curl http://localhost:8000/v1/completions \
  -H "Content-Type: application/json" \
  -d '{
    "model": "YOUR_MODEL",
    "prompt": "Once upon a time",
    "max_tokens": 41,
    "temperature": 0.5
  }'

Embeddings:

curl http://localhost:8000/v1/embeddings \
  -H "Content-Type: application/json" \
  -d '{
    "model": "YOUR_MODEL",
    "input": "Hello world!"
  }'