如何在Azure Web App上运行Go二进制服务？部署故障排查

我有一个运行在8000端口的Go项目，尝试通过Azure DevOps构建后部署至Linux Azure Web App。但访问页面时显示“:( Application Error”，多数尝试中提示“Container didn't respond to HTTP pings on port: 8000”。以下是我的DevOps配置和日志信息：

DevOps配置 #

pool:
  name: Azure Pipelines
  timeoutInMinutes: 30

steps:
- task: GoTool@0
  displayName: 'Use Go 1.22.2'
  inputs:
    version: 1.22.2

- task: Go@0
  displayName: 'go get'
  inputs:
    arguments: '-d'
    workingDirectory: '$(System.DefaultWorkingDirectory)'

- task: Go@0
  displayName: 'go build'
  inputs:
    command: build
    arguments: '-ldflags="-s -w" -o build'
    workingDirectory: '$(System.DefaultWorkingDirectory)'

- task: ArchiveFiles@2
  displayName: 'Archive $(System.DefaultWorkingDirectory)/build'
  inputs:
    rootFolderOrFile: '$(System.DefaultWorkingDirectory)/build'
    includeRootFolder: false

- task: AzureRmWebAppDeployment@4
  displayName: 'Azure App Service Deploy: MyServerProject'
  inputs:
    azureSubscription: 'Visual Studio Enterprise: ...'
    appType: webAppLinux
    WebAppName: MyServerProject
    packageForLinux: '$(Build.ArtifactStagingDirectory)/$(Build.BuildId).zip'
    RuntimeStack: 'DOTNETCORE|7.0'
    StartupCommand: 'my-server'
    AppSettings: '-PORT 8000 -WEBSITES_PORT 8000 -WEBSITES_CONTAINER_START_TIME_LIMIT 1800'

日志信息 #

2024-05-06T19:44:07.724140609Z    _____
2024-05-06T19:44:07.734185163Z   /  _  \ __________ _________   ____
2024-05-06T19:44:07.734190363Z  /  /_\  \\___   /  |  \_  __ \_/ __ \
2024-05-06T19:44:07.734193663Z /    |    \/    /|  |  /|  | \/\  ___/
2024-05-06T19:44:07.734196563Z \____|__  /_____ \____/ |__|    \___  >
2024-05-06T19:44:07.734199763Z         \/      \/                  \/
2024-05-06T19:44:07.734203064Z A P P   S E R V I C E   O N   L I N U X
2024-05-06T19:44:07.734206264Z
2024-05-06T19:44:07.734208964Z Documentation: http://aka.ms/webapp-linux
2024-05-06T19:44:07.734211764Z Dotnet quickstart: https://aka.ms/dotnet-qs
2024-05-06T19:44:07.734214564Z ASP .NETCore Version: 7.0.13
2024-05-06T19:44:07.734217364Z Note: Any data outside '/home' is not persisted
2024-05-06T19:44:09.748366682Z Starting OpenBSD Secure Shell server: sshd.
2024-05-06T19:44:10.140172606Z Starting periodic command scheduler: cron.
2024-05-06T19:44:10.149816058Z Running oryx create-script -appPath /home/site/wwwroot -output /opt/startup/startup.sh -defaultAppFilePath /defaulthome/hostingstart/hostingstart.dll     -bindPort 8000 -bindPort2 '' -userStartupCommand 'my-server'
2024-05-06T19:44:10.341022895Z Could not find build manifest file at '/home/site/wwwroot/oryx-manifest.toml'
2024-05-06T19:44:10.369622450Z Could not find operation ID in manifest. Generating an operation id...
2024-05-06T19:44:10.369672950Z Build Operation ID: 054d28e9-d3b7-43b9-8e77-ea26b35fa7fd
2024-05-06T19:44:11.658474634Z
2024-05-06T19:44:11.658510434Z Agent extension
2024-05-06T19:44:11.658515034Z Before if loop >> DotNet Runtime
2024-05-06T19:44:11.670287898Z DotNet Runtime 7.0Writing output script to '/opt/startup/startup.sh'
2024-05-06T19:44:11.949309510Z Running user provided startup command...
2024-05-06T19:44:11.949361811Z my-server: /lib/x86_64-linux-gnu/libc.so.6: version `GLIBC_2.32' not found (required by my-server)
2024-05-06T19:44:11.949386711Z my-server: /lib/x86_64-linux-gnu/libc.so.6: version `GLIBC_2.34' not found (required by my-server)
2024-05-06T19:44:05.685Z INFO  - 7.0_20240321.5.tuxprod Pulling from appsvc/dotnetcore
2024-05-06T19:44:05.706Z INFO  -  Digest: sha256:***
2024-05-06T19:44:05.706Z INFO  -  Status: Image is up to date for 10.1.0.5:13209/appsvc/dotnetcore:7.0_20240321.5.tuxprod
2024-05-06T19:44:05.724Z INFO  - Pull Image successful, Time taken: 0 Seconds
2024-05-06T19:44:05.925Z INFO  - Starting container for site
2024-05-06T19:44:05.926Z INFO  - docker run -d --expose=8000 --name myserverproject_123312 -e WEBSITE_USE_DIAGNOSTIC_SERVER=true -e PORT=8000 -e WEBSITES_PORT=8000 -e WEBSITE_SITE_NAME=myserverProject -e WEBSITE_AUTH_ENABLED=False -e WEBSITE_ROLE_INSTANCE_ID=0 -e WEBSITE_HOSTNAME=myserverproject.azurewebsites.net -e WEBSITE_INSTANCE_ID=*** -e HTTP_LOGGING_ENABLED=1 appsvc/dotnetcore:7.0_20240321.5.tuxprod my-server
2024-05-06T19:44:09.424Z INFO  - Initiating warmup request to container myserverproject_123312 for site myserverproject
2024-05-06T19:44:13.018Z ERROR - Container myserverproject_123312 for site myserverproject has exited, failing site start
2024-05-06T19:44:13.044Z ERROR - Container myserverproject_123312 didn't respond to HTTP pings on port: 8000, failing site start. See container logs for debugging.
2024-05-06T19:44:13.057Z INFO  - Stopping site myserverproject because it failed during startup.

疑问 #

是否可以实现该部署？若可行，我遗漏了哪些配置？

完全可以实现部署，问题出在三个核心配置错误上：

1. 错误选择了.NET Runtime栈 #

你当前指定的RuntimeStack: 'DOTNETCORE|7.0'是.NET的运行环境，但你的项目是Go语言，应该选择Go对应的Runtime栈，或者使用通用的Linux容器栈。

修改AzureRmWebAppDeployment任务的RuntimeStack参数：

RuntimeStack: 'GO|1.22'

如果找不到对应版本，也可以使用NONE，让Oryx自动检测Go环境。

2. Go编译时未做静态编译适配目标环境 #

日志里明确显示GLIBC_2.32和GLIBC_2.34版本缺失，这是因为你在Azure Pipelines代理环境中编译的Go程序依赖了较高版本的GLIBC，但Azure Web App的.NET容器里的GLIBC版本更低。

解决方法是修改go build命令，添加静态编译参数，让程序不依赖系统GLIBC：

arguments: '-ldflags="-s -w" -o build -tags netgo -ldflags "-linkmode external -extldflags -static"'

同时可以禁用CGO，彻底避免依赖系统库：在Go build任务前添加环境变量设置：

- task: Bash@3
  displayName: 'Disable CGO'
  inputs:
    targetType: 'inline'
    script: 'export CGO_ENABLED=0'

3. 启动命令中的程序名称不匹配 #

你在StartupCommand: 'my-server'指定的程序名是my-server，但编译时输出的程序名是build（-o build），部署后解压到容器里的文件名还是build，所以启动命令应该改为：

StartupCommand: './build'

额外优化建议 #

• 可以在Azure Web App的配置页面直接设置WEBSITES_PORT为8000，不需要在AppSettings里重复指定；
• 延长容器启动超时时间的设置WEBSITES_CONTAINER_START_TIME_LIMIT 1800可以保留，避免程序启动慢导致超时。

内容的提问来源于stack exchange，提问作者Pablo