name: dotnet-worker-services
version: "1.0.0"
category: "Distributed"
description: "Build long-running .NET background services with BackgroundService, Generic Host, graceful shutdown, configuration, logging, and deployment patterns suited to workers and daemons."
compatibility: "Requires a worker, hosted service, or background-processing scenario."

.NET Worker Services

Trigger On

• building long-running background services or scheduled workers
• adding hosted services to an app or extracting them into a worker process
• reviewing graceful shutdown, cancellation, queue processing, or health behavior

Documentation

• Worker Services in .NET
• Background tasks with hosted services in ASP.NET Core
• Create Windows Service using BackgroundService
• App health checks in .NET
• Health checks in ASP.NET Core

References

• patterns.md - BackgroundService patterns, graceful shutdown, and health check implementations
• anti-patterns.md - Common worker service mistakes and how to avoid them

Workflow

1. Use BackgroundService as your base class:
2. Provides standard StartAsync/StopAsync handling
3. Focus on implementing ExecuteAsync only

4. Proper cancellation token management built-in

5. Handle scoped dependencies correctly:

6. Create service scopes for scoped services

7. No scope is created by default in hosted services

8. Implement graceful shutdown:

9. Propagate cancellation tokens throughout
10. Complete work promptly when token fires

11. Avoid ungraceful shutdown at timeout

12. Keep execution loop thin:

13. Move business logic to testable services
14. Handle exceptions to prevent service crashes

15. Use PeriodicTimer for scheduled work

16. Add observability:

17. Use health checks for readiness/liveness
18. Expose metrics and structured logging
19. Consider distributed locks for multi-instance

Basic BackgroundService Pattern

Simple Worker

public class Worker : BackgroundService
{
    private readonly ILogger<Worker> _logger;

    public Worker(ILogger<Worker> logger)
    {
        _logger = logger;
    }

    protected override async Task ExecuteAsync(CancellationToken stoppingToken)
    {
        _logger.LogInformation("Worker starting");

        while (!stoppingToken.IsCancellationRequested)
        {
            try
            {
                _logger.LogInformation("Worker running at: {Time}", DateTimeOffset.Now);
                await DoWorkAsync(stoppingToken);
                await Task.Delay(TimeSpan.FromSeconds(5), stoppingToken);
            }
            catch (OperationCanceledException) when (stoppingToken.IsCancellationRequested)
            {
                // Graceful shutdown, not an error
                break;
            }
            catch (Exception ex)
            {
                _logger.LogError(ex, "Error in worker iteration");
                // Continue or break based on error severity
            }
        }

        _logger.LogInformation("Worker stopping");
    }

    private async Task DoWorkAsync(CancellationToken cancellationToken)
    {
        // Business logic here
    }
}

Using PeriodicTimer (Recommended)

public class TimedWorker : BackgroundService
{
    private readonly ILogger<TimedWorker> _logger;
    private readonly IServiceScopeFactory _scopeFactory;
    private readonly TimeSpan _period = TimeSpan.FromMinutes(1);

    public TimedWorker(ILogger<TimedWorker> logger, IServiceScopeFactory scopeFactory)
    {
        _logger = logger;
        _scopeFactory = scopeFactory;
    }

    protected override async Task ExecuteAsync(CancellationToken stoppingToken)
    {
        using var timer = new PeriodicTimer(_period);

        while (await timer.WaitForNextTickAsync(stoppingToken))
        {
            try
            {
                await using var scope = _scopeFactory.CreateAsyncScope();
                var processor = scope.ServiceProvider.GetRequiredService<IDataProcessor>();
                await processor.ProcessAsync(stoppingToken);
            }
            catch (Exception ex)
            {
                _logger.LogError(ex, "Error processing scheduled task");
            }
        }
    }
}

Handling Scoped Dependencies

Correct Pattern with Scope Factory

public class ScopedWorker : BackgroundService
{
    private readonly IServiceScopeFactory _scopeFactory;
    private readonly ILogger<ScopedWorker> _logger;

    public ScopedWorker(IServiceScopeFactory scopeFactory, ILogger<ScopedWorker> logger)
    {
        _scopeFactory = scopeFactory;
        _logger = logger;
    }

    protected override async Task ExecuteAsync(CancellationToken stoppingToken)
    {
        while (!stoppingToken.IsCancellationRequested)
        {
            // Create scope for each unit of work
            await using var scope = _scopeFactory.CreateAsyncScope();

            var dbContext = scope.ServiceProvider.GetRequiredService<AppDbContext>();
            var service = scope.ServiceProvider.GetRequiredService<IScopedService>();

            await service.ProcessAsync(dbContext, stoppingToken);
            await Task.Delay(TimeSpan.FromSeconds(10), stoppingToken);
        }
    }
}

Queue Processing Pattern

Message Queue Worker

public class QueueWorker : BackgroundService
{
    private readonly ILogger<QueueWorker> _logger;
    private readonly IServiceScopeFactory _scopeFactory;
    private readonly IBackgroundTaskQueue _taskQueue;

    public QueueWorker(
        ILogger<QueueWorker> logger,
        IServiceScopeFactory scopeFactory,
        IBackgroundTaskQueue taskQueue)
    {
        _logger = logger;
        _scopeFactory = scopeFactory;
        _taskQueue = taskQueue;
    }

    protected override async Task ExecuteAsync(CancellationToken stoppingToken)
    {
        _logger.LogInformation("Queue Worker started");

        while (!stoppingToken.IsCancellationRequested)
        {
            var workItem = await _taskQueue.DequeueAsync(stoppingToken);

            try
            {
                await using var scope = _scopeFactory.CreateAsyncScope();
                await workItem(scope.ServiceProvider, stoppingToken);
            }
            catch (Exception ex)
            {
                _logger.LogError(ex, "Error processing queued work item");
                // Handle poison message - retry, dead-letter, etc.
            }
        }
    }
}

// Task queue interface
public interface IBackgroundTaskQueue
{
    ValueTask QueueBackgroundWorkItemAsync(
        Func<IServiceProvider, CancellationToken, ValueTask> workItem);

    ValueTask<Func<IServiceProvider, CancellationToken, ValueTask>> DequeueAsync(
        CancellationToken cancellationToken);
}

Health Checks for Workers

Adding Health Check Endpoint

// Program.cs
var builder = Host.CreateApplicationBuilder(args);

builder.Services.AddHostedService<Worker>();

// Add health checks
builder.Services.AddHealthChecks()
    .AddCheck<WorkerHealthCheck>("worker_health")
    .AddResourceUtilizationHealthCheck();

// Add HTTP endpoint for health checks
builder.Services.AddHealthChecksUI();

// Or use simple TCP listener for Kubernetes
builder.Services.AddSingleton<TcpHealthProbeService>();
builder.Services.AddHostedService(sp => sp.GetRequiredService<TcpHealthProbeService>());

var host = builder.Build();
host.Run();

Custom Health Check

public class WorkerHealthCheck : IHealthCheck
{
    private readonly WorkerState _workerState;

    public WorkerHealthCheck(WorkerState workerState)
    {
        _workerState = workerState;
    }

    public Task<HealthCheckResult> CheckHealthAsync(
        HealthCheckContext context,
        CancellationToken cancellationToken = default)
    {
        if (_workerState.LastSuccessfulRun > DateTime.UtcNow.AddMinutes(-5))
        {
            return Task.FromResult(HealthCheckResult.Healthy(
                $"Last successful run: {_workerState.LastSuccessfulRun}"));
        }

        return Task.FromResult(HealthCheckResult.Unhealthy(
            $"No successful run since: {_workerState.LastSuccessfulRun}"));
    }
}

// Shared state
public class WorkerState
{
    public DateTime LastSuccessfulRun { get; set; } = DateTime.UtcNow;
    public bool IsProcessing { get; set; }
}

Graceful Shutdown Pattern

Proper Shutdown Handling

public class GracefulWorker : BackgroundService
{
    private readonly ILogger<GracefulWorker> _logger;
    private int _currentWorkItemId;

    protected override async Task ExecuteAsync(CancellationToken stoppingToken)
    {
        _logger.LogInformation("Worker starting");

        while (!stoppingToken.IsCancellationRequested)
        {
            _currentWorkItemId = GetNextWorkItemId();

            try
            {
                // Pass cancellation token to all async operations
                await ProcessWorkItemAsync(_currentWorkItemId, stoppingToken);
            }
            catch (OperationCanceledException) when (stoppingToken.IsCancellationRequested)
            {
                _logger.LogInformation(
                    "Shutdown requested, stopping after work item {Id}", _currentWorkItemId);
                break;
            }
        }
    }

    public override async Task StopAsync(CancellationToken cancellationToken)
    {
        _logger.LogInformation("Worker stopping gracefully");
        await base.StopAsync(cancellationToken);
        _logger.LogInformation("Worker stopped");
    }
}

Windows Service Deployment

Configuring as Windows Service

// Program.cs
var builder = Host.CreateApplicationBuilder(args);

builder.Services.AddWindowsService(options =>
{
    options.ServiceName = "My Worker Service";
});

builder.Services.AddHostedService<Worker>();

var host = builder.Build();
host.Run();

Project File Settings

<Project Sdk="Microsoft.NET.Sdk.Worker">
  <PropertyGroup>
    <TargetFramework>net8.0</TargetFramework>
    <RuntimeIdentifier>win-x64</RuntimeIdentifier>
    <PublishSingleFile>true</PublishSingleFile>
    <SelfContained>true</SelfContained>
  </PropertyGroup>
</Project>

Best Practices

1. Use BackgroundService as base class - Handles StartAsync/StopAsync boilerplate and cancellation management
2. Create scopes for scoped dependencies - Use IServiceScopeFactory to resolve scoped services like DbContext
3. Propagate cancellation tokens everywhere - Pass to all async methods for responsive shutdown
4. Wrap work in try-catch - Unhandled exceptions stop the service completely
5. Use PeriodicTimer for timed tasks - Cleaner than Task.Delay with proper cancellation support
6. Add health checks - Essential for Kubernetes liveness/readiness probes
7. Avoid blocking StartAsync - Long initialization delays other hosted services
8. Call base methods when overriding - Always call await base.StartAsync() and await base.StopAsync()
9. Publish as single file for Windows Service - Reduces deployment complexity and errors
10. Consider scaling requirements - Separate worker projects if independent scaling is needed

Anti-Patterns to Avoid

Deliver

• well-behaved worker processes and hosted services
• predictable startup and shutdown behavior
• proper scoped dependency handling
• health checks for production observability
• retry and poison-message handling for queue work

Validate

• cancellation token propagated and shutdown honored
• scoped services resolved within proper scopes
• exception handling prevents service crashes
• health checks report accurate worker status
• runtime behavior visible through logs or telemetry
• no blocking calls in async context