deployment-e2e-testing

作者: microsoft

编写Aspire部署端到端测试的指南。当需要创建、修改或调试部署到Azure的部署E2E测试时使用。

npx skills add https://github.com/microsoft/aspire --skill deployment-e2e-testing

Aspire Deployment End-to-End Testing

This skill provides patterns and practices for writing end-to-end tests that deploy Aspire applications to real Azure infrastructure.

Overview

Deployment E2E tests extend the CLI E2E testing patterns with actual Azure deployments. They use the Hex1b terminal automation library to drive the Aspire CLI and verify that deployed applications work correctly.

Location: tests/Aspire.Deployment.EndToEnd.Tests/

Supported Platforms: Linux only (Hex1b requirement).

Prerequisites:

  • Azure subscription with appropriate permissions
  • OIDC authentication (CI) or Azure CLI authentication (local)

Relationship to CLI E2E Tests

Deployment tests build on the CLI E2E testing skill. Before working with deployment tests, familiarize yourself with:

Key differences from CLI E2E tests:

AspectCLI E2E TestsDeployment E2E Tests
Duration5-15 minutes15-45 minutes
ResourcesLocal onlyAzure resources
AuthenticationNoneAzure OIDC/CLI
CleanupTemp directoriesAzure resource groups
TriggersPR, pushNightly, manual, deploy-test/*

Key Components

Core Classes

  • DeploymentE2ETestHelpers (Helpers/DeploymentE2ETestHelpers.cs): Terminal factory and environment helpers
  • DeploymentE2EAutomatorHelpers (Helpers/DeploymentE2EAutomatorHelpers.cs): Async extension methods on Hex1bTerminalAutomator for deployment scenarios
  • Hex1bAutomatorTestHelpers (shared): Common async extension methods on Hex1bTerminalAutomator (WaitForSuccessPromptAsync, AspireNewAsync, etc.)
  • AzureAuthenticationHelpers (Helpers/AzureAuthenticationHelpers.cs): Azure auth and resource naming
  • DeploymentReporter (Helpers/DeploymentReporter.cs): GitHub step summary reporting
  • SequenceCounter (Helpers/SequenceCounter.cs): Prompt tracking (same as CLI E2E)

Test Architecture

Each deployment test:

  1. Validates Azure authentication (skip if not available locally)
  2. Generates a unique resource group name
  3. Creates a project using aspire new
  4. Deploys using aspire deploy
  5. Verifies deployed endpoints work
  6. Reports results to GitHub step summary
  7. Cleans up Azure resources

Test Structure

public sealed class MyDeploymentTests(ITestOutputHelper output)
{
    [Fact]
    public async Task DeployMyScenario()
    {
        // 1. Validate prerequisites
        var subscriptionId = AzureAuthenticationHelpers.TryGetSubscriptionId();
        if (string.IsNullOrEmpty(subscriptionId))
        {
            Assert.Skip("Azure subscription not configured.");
        }

        if (!AzureAuthenticationHelpers.IsAzureAuthAvailable())
        {
            if (DeploymentE2ETestHelpers.IsRunningInCI)
            {
                Assert.Fail("Azure auth not available in CI.");
            }
            Assert.Skip("Azure auth not available. Run 'az login'.");
        }

        // 2. Setup
        var resourceGroupName = AzureAuthenticationHelpers.GenerateResourceGroupName("my-scenario");
        var workspace = TemporaryWorkspace.Create(output);
        var startTime = DateTime.UtcNow;

        try
        {
            // 3. Build terminal and run deployment
            using var terminal = DeploymentE2ETestHelpers.CreateTestTerminal();
            var pendingRun = terminal.RunAsync(TestContext.Current.CancellationToken);

            var counter = new SequenceCounter();
            var auto = new Hex1bTerminalAutomator(terminal, defaultTimeout: TimeSpan.FromSeconds(500));

            await auto.PrepareEnvironmentAsync(workspace, counter);
            if (DeploymentE2ETestHelpers.IsRunningInCI)
            {
                await auto.SourceAspireCliEnvironmentAsync(counter);
            }
            await auto.AspireNewAsync("MyProject", counter, useRedisCache: false);
            // ... deployment steps ...

            await auto.TypeAsync("exit");
            await auto.EnterAsync();
            await pendingRun;

            // 4. Report success
            var duration = DateTime.UtcNow - startTime;
            DeploymentReporter.ReportDeploymentSuccess(
                nameof(DeployMyScenario),
                resourceGroupName,
                deploymentUrls,
                duration);
        }
        catch (Exception ex)
        {
            DeploymentReporter.ReportDeploymentFailure(
                nameof(DeployMyScenario),
                resourceGroupName,
                ex.Message);
            throw;
        }
        finally
        {
            // 5. Cleanup Azure resources
            await CleanupResourceGroupAsync(resourceGroupName);
        }
    }
}

Extension Methods

DeploymentE2EAutomatorHelpers Extensions on Hex1bTerminalAutomator

MethodDescription
PrepareEnvironmentAsync(workspace, counter)Sets up the terminal environment with custom prompt and workspace directory
InstallAspireCliFromPullRequestAsync(prNumber, counter)Downloads and installs the Aspire CLI from a PR build artifact
InstallAspireCliReleaseAsync(counter)Installs the latest released Aspire CLI
SourceAspireCliEnvironmentAsync(counter)Adds ~/.aspire/bin to PATH so the aspire command is available

These extend Hex1bTerminalAutomator and are used alongside the shared Hex1bAutomatorTestHelpers methods (WaitForSuccessPromptAsync, AspireNewAsync, etc.) documented in the CLI E2E Testing Skill.

Azure Authentication

In CI (GitHub Actions)

Tests use OIDC (Workload Identity Federation) for authentication:

- name: Azure Login (OIDC)
  uses: azure/login@v2
  with:
    client-id: ${{ secrets.AZURE_CLIENT_ID }}
    tenant-id: ${{ secrets.AZURE_TENANT_ID }}
    subscription-id: ${{ vars.ASPIRE_DEPLOYMENT_TEST_SUBSCRIPTION }}

The test code automatically detects CI and uses DefaultAzureCredential which picks up the OIDC session.

Local Development

Authenticate with Azure CLI before running tests:

# Login to Azure
az login

# Set your subscription
az account set --subscription "your-subscription-id"

# Set environment variable
export ASPIRE_DEPLOYMENT_TEST_SUBSCRIPTION="your-subscription-id"

# Run tests
dotnet test tests/Aspire.Deployment.EndToEnd.Tests/

Authentication Helpers

// Check if auth is available
if (!AzureAuthenticationHelpers.IsAzureAuthAvailable())
{
    Assert.Skip("Azure auth not available");
}

// Get subscription ID
var subscriptionId = AzureAuthenticationHelpers.GetSubscriptionId();

// Generate unique resource group name
var rgName = AzureAuthenticationHelpers.GenerateResourceGroupName("my-test");
// Result: "aspire-e2e-my-test-20240115-abc12345"

// Check auth type
if (AzureAuthenticationHelpers.IsOidcConfigured())
{
    // Using OIDC (CI)
}
else
{
    // Using Azure CLI (local)
}

Resource Group Naming

Resource groups are named with a consistent pattern for easy identification and cleanup:

{prefix}-{testname}-{date}-{runid}

Example: aspire-e2e-aca-starter-20240115-12345678

Components:

  • prefix: From ASPIRE_DEPLOYMENT_TEST_RG_PREFIX (default: aspire-e2e)
  • testname: Sanitized test name (lowercase, alphanumeric, hyphens)
  • date: UTC date in YYYYMMDD format
  • runid: GitHub run ID or random GUID suffix

Reporting Results

GitHub Step Summary

Tests automatically write to the GitHub step summary:

// Report success with URLs
DeploymentReporter.ReportDeploymentSuccess(
    testName: "DeployStarterToACA",
    resourceGroupName: "aspire-e2e-...",
    deploymentUrls: new Dictionary<string, string>
    {
        ["Dashboard"] = "https://dashboard.azurecontainerapps.io",
        ["Web Frontend"] = "https://webfrontend.azurecontainerapps.io"
    },
    duration: TimeSpan.FromMinutes(15));

// Report failure
DeploymentReporter.ReportDeploymentFailure(
    testName: "DeployStarterToACA",
    resourceGroupName: "aspire-e2e-...",
    errorMessage: "Deployment timed out",
    logs: "Full deployment logs...");

Asciinema Recordings

Tests generate recordings for debugging:

var recordingPath = DeploymentE2ETestHelpers.GetTestResultsRecordingPath(nameof(MyTest));
// CI: $GITHUB_WORKSPACE/testresults/recordings/MyTest.cast
// Local: /tmp/aspire-deployment-e2e/recordings/MyTest.cast

Cleanup

Always cleanup Azure resources in a finally block:

private static async Task CleanupResourceGroupAsync(string resourceGroupName)
{
    var process = new Process
    {
        StartInfo = new ProcessStartInfo
        {
            FileName = "az",
            Arguments = $"group delete --name {resourceGroupName} --yes --no-wait",
            RedirectStandardOutput = true,
            RedirectStandardError = true,
            UseShellExecute = false
        }
    };

    process.Start();
    await process.WaitForExitAsync();

    if (process.ExitCode != 0)
    {
        var error = await process.StandardError.ReadToEndAsync();
        throw new InvalidOperationException($"Cleanup failed: {error}");
    }
}

Workflow Triggers

Deployment tests are triggered by:

  1. Nightly schedule (03:00 UTC) - Runs on main
  2. Manual dispatch - Via GitHub Actions UI
  3. Push to deploy-test/* - For rapid iteration

Iterating on Tests

To iterate quickly during development:

# Create a protected branch
git checkout -b deploy-test/my-feature

# Make changes
# ...

# Push to trigger workflow
git push origin deploy-test/my-feature

Environment Variables

VariableRequiredDescription
ASPIRE_DEPLOYMENT_TEST_SUBSCRIPTIONYesAzure subscription ID
ASPIRE_DEPLOYMENT_TEST_RG_PREFIXNoResource group prefix (default: aspire-e2e)
AZURE_DEPLOYMENT_TEST_TENANT_IDCIAzure AD tenant ID (GitHub secret)
AZURE_DEPLOYMENT_TEST_CLIENT_IDCIOIDC app client ID (GitHub secret)
AZURE_DEPLOYMENT_TEST_SUBSCRIPTION_IDCIAzure subscription ID (GitHub secret)

DO: Always Validate Prerequisites

var subscriptionId = AzureAuthenticationHelpers.TryGetSubscriptionId();
if (string.IsNullOrEmpty(subscriptionId))
{
    Assert.Skip("Subscription not configured");
}

if (!AzureAuthenticationHelpers.IsAzureAuthAvailable())
{
    Assert.Skip("Azure auth not available");
}

DO: Generate Unique Resource Groups

var rgName = AzureAuthenticationHelpers.GenerateResourceGroupName("my-test");

DO: Report to GitHub Summary

DeploymentReporter.ReportDeploymentSuccess(...);
// or
DeploymentReporter.ReportDeploymentFailure(...);

DO: Always Cleanup Resources

try
{
    // ... deployment test ...
}
finally
{
    await CleanupResourceGroupAsync(resourceGroupName);
}

DON'T: Hardcode Subscription IDs

// DON'T
var subscriptionId = "12345-abcde-...";

// DO
var subscriptionId = AzureAuthenticationHelpers.GetSubscriptionId();

DON'T: Skip Cleanup on Failure

// DON'T - cleanup might not run
await DeployAsync();
await CleanupAsync();  // Skipped if deploy throws!

// DO - always cleanup
try
{
    await DeployAsync();
}
finally
{
    await CleanupAsync();  // Always runs
}

Troubleshooting

Authentication Failures

Local: Ensure Azure CLI is authenticated:

az login
az account show

CI: Check OIDC configuration:

  • AZURE_CLIENT_ID secret is set
  • AZURE_TENANT_ID secret is set
  • Workload Identity Federation is configured in Azure AD

Deployment Timeouts

Deployments can take 15-30+ minutes. If tests timeout:

  • Check the asciinema recording for where it stopped
  • Increase timeout in WaitUntil calls
  • Check Azure portal for deployment status

Orphaned Resources

Find and cleanup orphaned test resources:

# List all test resource groups
az group list --query "[?starts_with(name, 'aspire-e2e')]" -o table

# Delete specific resource group
az group delete --name aspire-e2e-xxx --yes

Tenant Rotation

The test tenant rotates every ~90 days. When rotation occurs:

  1. Create new App Registration in new tenant
  2. Configure Workload Identity Federation for the deployment-testing environment
  3. Grant Owner role on subscription (constrained - cannot create other Owner identities)
  4. Update GitHub secrets: AZURE_DEPLOYMENT_TEST_CLIENT_ID, AZURE_DEPLOYMENT_TEST_TENANT_ID, AZURE_DEPLOYMENT_TEST_SUBSCRIPTION_ID

来自 microsoft 的更多技能

oss-growth
microsoft
OSS增长黑客角色
agent-framework-azure-ai-py
microsoft
使用Microsoft Agent Framework Python SDK(agent-framework-azure-ai)构建Azure AI Foundry代理。在创建使用AzureAIAgentsProvider的持久化代理、使用托管工具(代码解释器、文件搜索、网络搜索)、集成MCP服务器、管理对话线程或实现流式响应时使用。涵盖函数工具、结构化输出和多工具代理。
development
airunway-aks-setup
microsoft
在AKS上设置AI Runway——从裸集群到运行模型。涵盖集群验证、控制器安装、GPU评估、提供商设置和首次部署。适用场景:“设置AI Runway”、“接入AKS集群”、“安装AI Runway”、“airunway设置”、“将模型部署到AKS”、“在AKS上进行GPU推理”、“在AKS上配置KAITO”、“在AKS上运行LLM”、“在AKS上使用vLLM”、“在AKS上设置模型服务”、“AI Runway控制器”。
devops
appinsights-instrumentation
microsoft
使用Azure Application Insights对Web应用进行插桩的指南。提供遥测模式、SDK设置和配置参考。适用场景:如何对应用进行插桩、App Insights SDK、遥测模式、什么是App Insights、Application Insights指南、插桩示例、APM最佳实践。
devops
applicationinsights-web-ts
microsoft
使用Application Insights JavaScript SDK(@microsoft/applicationinsights-web)为浏览器/Web应用添加检测。用于真实用户监控(RUM)——页面视图、点击、AJAX/fetch依赖项、异常、自定义事件,以及与后端OpenTelemetry追踪关联的浏览器端GenAI代理追踪。涵盖SDK加载器脚本和npm设置、框架扩展(React、React Native、Angular)、点击分析、遥测初始化器,以及从浏览器发出的代理/工具/模型跨度所遵循的OTel GenAI语义约定。
devops
azure-ai-anomalydetector-java
microsoft
使用适用于 Java 的 Azure AI 异常检测器 SDK 构建异常检测应用程序。在实现单变量/多变量异常检测、时间序列分析或 AI 驱动的监控时使用。
development
azure-ai-language-conversations-py
microsoft
使用azure-ai-language-conversations Python SDK实现对话语言理解(CLU)。当使用ConversationAnalysisClient分析对话意图和实体、构建NLP功能或将语言理解集成到应用程序中时使用。
development
azure-ai-ml-py
microsoft
Azure Machine Learning SDK v2 for Python。用于机器学习工作区、作业、模型、数据集、计算资源和管道。 触发词:“azure-ai-ml”、“MLClient”、“工作区”、“模型注册表”、“训练作业”、“数据集”。
development