如何仅在ADF部署任务失败时运行触发器重新启用阶段？

问题背景 #

我们有三个阶段，第二阶段（ADFDeploy）会先禁用ADF触发器，再执行AzureResourceGroupDeployment@2部署任务。如果该部署任务失败，流水线会直接退出，导致ADF触发器一直处于禁用状态。我们需要实现：

• 仅当第二阶段中**AzureResourceGroupDeployment@2任务失败**时，才运行触发器重新启用阶段；
• 如果第二阶段中除该部署任务外的其他任务失败，流水线直接退出，且不能使用continueOnError参数。

修改步骤 #

1. 调整部署阶段（ADFDeploy） #

给部署任务添加命名和输出变量，用于标记该任务是否失败：

stages:
- stage: ADFDeploy
  displayName: ADF部署阶段
  variables:
    # 你的变量配置
  pool:
    # 你的代理池配置
  jobs:
  - job: ADFDeployJob
    displayName: ADF部署任务
    steps:
      - task: AzurePowerShell@5
        displayName: 'Get All ADFTriggers'
        inputs:
          azureSubscription: '${{ parameters.AzureServiceCon }}'
          ScriptType: 'InlineScript'
          preferredAzurePowerShellVersion: '2.2.8'
          Inline: |
            some code
          azurePowerShellVersion: 'LatestVersion'

      - task: PythonScript@0
        inputs:
          scriptSource: 'filePath'
          scriptPath: '$(System.DefaultWorkingDirectory)/AzurePipelines/YAML/Pipelines/02_ApplicationDeployment/YAML_CD/ADF_CD/reenable-triggers.py'
          pythonInterpreter: 'C:\Python310\python.exe'
        displayName: "Get ADF Triggers"

      - task: PublishPipelineArtifact@1
        inputs:
          targetPath: '$(Build.SourcesDirectory)/triggerName.txt'
          artifact: 'ADFTrgList'
          publishLocation: 'pipeline'      

      - task: AzurePowerShell@5
        displayName: 'Disable adf triggers'
        inputs:
          azureSubscription: '${{ parameters.AzureServiceCon }}'
          ScriptType: 'FilePath'
          ScriptPath: '$(System.DefaultWorkingDirectory)/AzurePipelines/YAML/Pipelines/02_ApplicationDeployment/YAML_CD/ADF_CD/PrePostDeployment.ps1'
          ScriptArguments: '-armTemplate "$(Pipeline.Workspace)/ArmTemplates/ARMTemplateForFactory.json" -ResourceGroupName $(paramResourceGroupName) -DataFactoryName $(paramfactoryName) -predeployment $true -deleteDeployment $false'
          preferredAzurePowerShellVersion: '2.2.8'
          azurePowerShellVersion: 'LatestVersion'

      # 修改后的部署任务：添加命名和输出变量
      - task: AzureResourceGroupDeployment@2
        displayName: 'Azure Deployment:Create ADF'
        name: ADFFactoryDeployment # 给任务命名，用于后续引用输出变量
        outputs:
          deploymentFailed: ${{ failed() }} # 输出变量，任务失败时为true，成功时为false
        inputs:
          azureSubscription: '${{ parameters.AzureServiceCon }}'
          resourceGroupName: "$(paramResourceGroupName)"
          location: "$(paramResourceGroupLocation)"
          csmFile: '$(Pipeline.Workspace)/**/ARMTemplateForFactory.json'
          csmParametersFile: '$(Pipeline.Workspace)/**/ARMTemplateParametersForFactory.json'
          deploymentMode: ${{ parameters.paramDeploymentMode }}
          enabled: ${{ parameters.paramEnabled }}

2. 调整触发器重新启用阶段 #

修改阶段依赖和触发条件，确保仅在部署任务失败时运行；同时添加下载artifact的步骤，确保能读取到触发器列表文件：

stages:
- stage: ADFTriggers
  displayName: Re-EnableADFTriggers-${{ parameters.environmentShortName }}
  dependsOn: ADFDeploy # 改为依赖ADF部署阶段
  # 触发条件：仅当部署任务执行且失败，同时环境符合要求时运行
  condition: |
    and(
      exists(dependencies.ADFDeploy.outputs['ADFDeployJob.ADFFactoryDeployment.deploymentFailed']),
      eq(dependencies.ADFDeploy.outputs['ADFDeployJob.ADFFactoryDeployment.deploymentFailed'], 'true'),
      in(variables['environmentName'], 'Prod', 'QA', 'STest')
    )

  variables:
    - template: ../../../YAML/Variables/${{ parameters.environmentShortName }}/variables-master.yml
  pool:
    name: ${{ parameters.agentPoolName }}
    vmImage: 'windows-latest'
    demands:
      - msbuild
      - visualstudio
      - vstest

  jobs:
  - deployment: ADFTrg_${{ parameters.environmentShortName }}
    displayName: ADFTriggers ${{ parameters.environmentShortName }}
    pool:
      name: ${{ parameters.agentPoolName }}
      vmImage: 'windows-latest'
      demands:
      - msbuild
      - visualstudio
      - vstest
    workspace:
      clean: all
    environment: ADF-TRG-${{ parameters.environmentShortName }}
    strategy: 
      runOnce:
        deploy:
          steps: 
          - checkout: 'YAML'
          # 添加下载artifact的步骤，获取触发器列表文件
          - task: DownloadPipelineArtifact@2
            displayName: '下载ADF触发器列表'
            inputs:
              artifact: 'ADFTrgList'
              path: '$(System.DefaultWorkingDirectory)'
          - task: AzurePowerShell@5
            displayName: '重新启用ADF触发器'
            inputs:
              azureSubscription: '${{ parameters.AzureServiceCon }}'
              ScriptType: 'InlineScript'
              preferredAzurePowerShellVersion: '2.2.8'
              Inline: |
                # 检查触发器名称文件是否存在
                $triggerFile = 'triggerName.txt'
                if (Test-Path $triggerFile) {
                    # 读取文件中的触发器名称
                    $triggerNames = Get-Content -Path $triggerFile

                    # 遍历触发器名称并重新启用
                    foreach ($name in $triggerNames) {
                        Start-AzDataFactoryV2Trigger -DataFactoryName $(paramfactoryName) -ResourceGroupName $(paramResourceGroupName) -TriggerName $name -Force
                        Write-Output "已重新启用触发器: $name"
                    }
                } else {
                    Write-Output "未找到触发器名称文件。"
                }
              azurePowerShellVersion: 'LatestVersion'

逻辑说明 #

1. 部署阶段：

   • 前序任务失败时，流水线会直接终止，不会执行部署任务，也不会触发重新启用阶段；
   • 前序任务成功后，执行部署任务，通过outputs输出任务失败状态。

2. 重新启用阶段：

   • exists()函数确保部署任务已执行（即前序任务都成功）；
   • eq(..., 'true')确保部署任务确实失败；
   • 只有同时满足这两个条件，才会运行重新启用触发器的逻辑；
   • 添加DownloadPipelineArtifact步骤，确保能获取到之前发布的触发器列表文件。

内容的提问来源于stack exchange，提问作者Abhishek Singh