採用接 cloud function 再打 job

• 建立一個 python job image 用於觸發

  • 快速建立一個 python main.py 可以打印 log 的 file

    import logging
    import time
    
    # 配置日誌
    logging.basicConfig(level=logging.INFO, format='%(asctime)s - %(message)s')
    
    def main():
        logging.info("開始執行腳本")
        for i in range(5):
            logging.info(f"這是第 {i+1} 條日誌訊息")
            time.sleep(1)  # 暫停1秒，模擬處理時間
        logging.info("腳本執行完畢")
    
    if __name__ == "__main__":
        main()
    

  • dockerfile

    FROM python:3.11-slim
    
    WORKDIR /app
    
    COPY main.py .
    
    CMD ["python", "main.py"]
    

  • docker build

    docker build -t python-logger:v1 .
    
    # Test docker
    docker run python-logger:v1  
    2024-10-05 04:48:17,651 - 開始執行腳本
    2024-10-05 04:48:17,653 - 這是第 1 條日誌訊息
    2024-10-05 04:48:18,657 - 這是第 2 條日誌訊息
    2024-10-05 04:48:19,662 - 這是第 3 條日誌訊息
    2024-10-05 04:48:20,664 - 這是第 4 條日誌訊息
    2024-10-05 04:48:21,670 - 這是第 5 條日誌訊息
    2024-10-05 04:48:22,672 - 腳本執行完畢
    

  • create artifact hub with terraform

    # install gcloud from the site: https://cloud.google.com/sdk/docs/install
    
    gcloud init
    
    gcloud auth application-default login
    

    # 添加 Artifact Registry 倉庫
    resource "google_artifact_registry_repository" "my_repo" {
    
      location      = var.region
      repository_id = "test-k8s-job"
      description   = "My artifact repository"
      format        = "DOCKER"
    
    }
    

  • docker push to hub

    #!/bin/bash
    
    # 設置變量
    PROJECT_ID="your-project-id"  # 替換為您的 GCP 項目 ID
    REGION="asia-east1"           # 替換為您的目標區域
    REPO_NAME="test-k8s-job"
    IMAGE_NAME="python-logger"
    IMAGE_TAG="v1"
    
    # 配置 Docker 以使用 gcloud 作為憑證助手
    # gcloud auth configure-docker ${REGION}-docker.pkg.dev
    
    # 標記映像
    docker tag ${IMAGE_NAME}:${IMAGE_TAG} ${REGION}-docker.pkg.dev/${PROJECT_ID}/${REPO_NAME}/${IMAGE_NAME}:${IMAGE_TAG}
    
    # 推送映像
    docker push ${REGION}-docker.pkg.dev/${PROJECT_ID}/${REPO_NAME}/${IMAGE_NAME}:${IMAGE_TAG}
    
    echo "Image pushed successfully to Artifact Registry"
    

• 建立一個 cloud function 用於觸發 job in k8s

  • 在 k8s 上運行 image 看看

      gcloud components install gke-gcloud-auth-plugin
    
    gcloud container clusters get-credentials portal-uat-asia-east1-gke --region asia-east1 --project xxx
    

    由於 cluster 是 private，所以要用跳板機。中間安裝流程直接跳過。

    gcloud compute ssh portal-uat-asia-east1-bastion-vm --tunnel-through-iap --project=xxx
    
    # 確保最後可以拿到此資料，證明連線成功
    kubectl get ns
    

  • 新增 job

    apiVersion: batch/v1
    kind: Job
    metadata:
      name: python-logger-job
    spec:
      template:
        spec:
          containers:
          - name: python-logger
            image: asia-east1-docker.pkg.dev/shopeetwbi/test-k8s-job/python-logger:v1
            # 如果您的 Python 腳本需要參數，可以在這裡添加
            # args: ["arg1", "arg2"]
          restartPolicy: Never
      backoffLimit: 4
    

    kubectl apply -f python-logger-job.yaml
    

    由於我的 gke 是 autopilot，剛起 job 需要分配 node 會 pending 一段時間

    Events:
      Type     Reason            Age                From                                   Message
      ----     ------            ----               ----                                   -------
      Normal   TriggeredScaleUp  41s                cluster-autoscaler                     pod triggered scale-up: [{https://www.googleapis.com/compute/v1/projects/xxxx/zones/asia-east1-c/instanceGroups/gk3-portal-uat-asia-east-nap-rurfc12x-61262313-grp 0->1 (max: 1000)}]
      Warning  FailedScheduling  4s (x11 over 80s)  gke.io/optimize-utilization-scheduler  no nodes available to schedule pods
    

    完成後得到 logs

    kubectl logs python-logger-job-ll8bg
    exec /usr/local/bin/python: exec format error
    

    這可能是因為 arm 打包 image 的問題，我們改成用 gcloud building

    gcloud builds submit --pack image=asia-east1-docker.pkg.dev/shopeetwbi/test-k8s-job/python-logger:v1