Host ASP.NET Core in a Windows Service

An ASP.NET Core app can be hosted on Windows as a Windows Service without using IIS. When hosted as a Windows Service, the app automatically starts after server reboots.

Prerequisites

• ASP.NET Core SDK 7.0 or later
• PowerShell 6.2 or later

Worker Service template

The ASP.NET Core Worker Service template provides a starting point for writing long running service apps. To use the template as a basis for a Windows Service app:

1. Create a Worker Service app from the .NET Core template.
2. Install the NuGet package Microsoft.Extensions.Hosting.WindowsServices.
3. Follow the guidance in the App configuration section to update the Worker Service app so that it can run as a Windows Service.

Visual Studio

1. Create a new project.
2. Select Worker Service. Select Next.
3. Provide a project name in the Project name field or accept the default project name. Select Create.
4. In the Create a new Worker service dialog, select Create.

Visual Studio for Mac

1. Create a new project.
2. Select App under .NET Core in the sidebar.
3. Select Worker under ASP.NET Core. Select Next.
4. Select .NET Core 3.0 or later for the Target Framework. Select Next.
5. Provide a name in the Project Name field. Select Create.

.NET CLI

Use the Worker Service (worker) template with the dotnet new command from a command shell. In the following example, a Worker Service app is created named ContosoWorker. A folder for the ContosoWorker app is created automatically when the command is executed.

dotnet new worker -o ContosoWorker

App configuration

Update Program.cs to call AddWindowsService. When the app is running as a Windows Service, AddWindowsService:

• Sets the host lifetime to WindowsServiceLifetime.
• Sets the content root to AppContext.BaseDirectory. For more information, see the Current directory and content root section.
• Enables logging to the event log:
  • The application name is used as the default source name.
  • The default log level is Warning or higher for an app based on an ASP.NET Core template that calls CreateDefaultBuilder to build the host.
  • Override the default log level with the Logging:EventLog:LogLevel:Default key in appsettings.json/appsettings.{Environment}.json or other configuration provider.
  • Only administrators can create new event sources. When an event source can't be created using the application name, a warning is logged to the Application source and event logs are disabled.

Consider the following ServiceA class:

namespace SampleApp.Services;

public class ServiceA : BackgroundService
{
    public ServiceA(ILoggerFactory loggerFactory)
    {
        Logger = loggerFactory.CreateLogger<ServiceA>();
    }

    public ILogger Logger { get; }

    protected override async Task ExecuteAsync(CancellationToken stoppingToken)
    {
        Logger.LogInformation("ServiceA is starting.");

        stoppingToken.Register(() => Logger.LogInformation("ServiceA is stopping."));

        while (!stoppingToken.IsCancellationRequested)
        {
            Logger.LogInformation("ServiceA is doing background work.");

            await Task.Delay(TimeSpan.FromSeconds(5), stoppingToken);
        }

        Logger.LogInformation("ServiceA has stopped.");
    }
}

The following Program.cs calls AddHostedService to register ServiceA:

using SampleApp.Services;

var builder = WebApplication.CreateBuilder(args);

builder.Services.AddRazorPages();

builder.Services.AddWindowsService();
builder.Services.AddHostedService<ServiceA>();

var app = builder.Build();

app.MapRazorPages();

app.Run();

The following sample apps accompany this topic:

• Background Worker Service Sample: A non-web app sample based on the Worker Service template that uses hosted services for background tasks.
• Web App Service Sample: A Razor Pages web app sample that runs as a Windows Service with hosted services for background tasks.

For MVC guidance, see the articles under Overview of ASP.NET Core MVC and Migrate from ASP.NET Core 2.2 to 3.0.

Deployment type

For information and advice on deployment scenarios, see .NET Core application deployment.

SDK

For a web app-based service that uses the Razor Pages or MVC frameworks, specify the Web SDK in the project file:

<Project Sdk="Microsoft.NET.Sdk.Web">

If the service only executes background tasks (for example, hosted services), specify the Worker SDK in the project file:

<Project Sdk="Microsoft.NET.Sdk.Worker">

Framework-dependent deployment (FDD)

Framework-dependent deployment (FDD) relies on the presence of a shared system-wide version of .NET Core on the target system. When the FDD scenario is adopted following the guidance in this article, the SDK produces an executable (.exe), called a framework-dependent executable.

If using the Web SDK, a web.config file, which is normally produced when publishing an ASP.NET Core app, is unnecessary for a Windows Services app. To disable the creation of the web.config file, add the <IsTransformWebConfigDisabled> property set to true.

<PropertyGroup>
  <TargetFramework>net7.0</TargetFramework>
  <IsTransformWebConfigDisabled>true</IsTransformWebConfigDisabled>
</PropertyGroup>

Self-contained deployment (SCD)

Self-contained deployment (SCD) doesn't rely on the presence of a shared framework on the host system. The runtime and the app's dependencies are deployed with the app.

A Windows Runtime Identifier (RID) is included in the <PropertyGroup> that contains the target framework:

<RuntimeIdentifier>win-x64</RuntimeIdentifier>

To publish for multiple RIDs:

• Provide the RIDs in a semicolon-delimited list.
• Use the property name <RuntimeIdentifiers> (plural).

For more information, see .NET Core RID Catalog.

Service user account

To create a user account for a service, use the New-LocalUser cmdlet from an administrative PowerShell 6 command shell.

On Windows 10 October 2018 Update (version 1809/build 10.0.17763) or later:

New-LocalUser -Name {SERVICE NAME}

On Windows OS earlier than the Windows 10 October 2018 Update (version 1809/build 10.0.17763):

powershell -Command "New-LocalUser -Name {SERVICE NAME}"

Provide a strong password when prompted.

Unless the -AccountExpires parameter is supplied to the New-LocalUser cmdlet with an expiration DateTime, the account doesn't expire.

For more information, see Microsoft.PowerShell.LocalAccounts and Service User Accounts.

An alternative approach to managing users when using Active Directory is to use Managed Service Accounts. For more information, see Group Managed Service Accounts Overview.

Log on as a service rights

To establish Log on as a service rights for a service user account:

1. Open the Local Security Policy editor by running secpol.msc.
2. Expand the Local Policies node and select User Rights Assignment.
3. Open the Log on as a service policy.
4. Select Add User or Group.
5. Provide the object name (user account) using either of the following approaches:
   1. Type the user account ({DOMAIN OR COMPUTER NAME\USER}) in the object name field and select OK to add the user to the policy.
   2. Select Advanced. Select Find Now. Select the user account from the list. Select OK. Select OK again to add the user to the policy.
6. Select OK or Apply to accept the changes.

Create and manage the Windows Service

Create a service

Use PowerShell commands to register a service. From an administrative PowerShell 6 command shell, execute the following commands:

$acl = Get-Acl "{EXE PATH}"
$aclRuleArgs = "{DOMAIN OR COMPUTER NAME\USER}", "Read,Write,ReadAndExecute", "ContainerInherit,ObjectInherit", "None", "Allow"
$accessRule = New-Object System.Security.AccessControl.FileSystemAccessRule($aclRuleArgs)
$acl.SetAccessRule($accessRule)
$acl | Set-Acl "{EXE PATH}"

New-Service -Name {SERVICE NAME} -BinaryPathName "{EXE FILE PATH} --contentRoot {EXE FOLDER PATH}" -Credential "{DOMAIN OR COMPUTER NAME\USER}" -Description "{DESCRIPTION}" -DisplayName "{DISPLAY NAME}" -StartupType Automatic

• {EXE PATH}: Path of the app's executable on the host (for example, d:\myservice). Don't include the app's executable file name in the path. A trailing slash isn't required.
• {DOMAIN OR COMPUTER NAME\USER}: Service user account (for example, Contoso\ServiceUser).
• {SERVICE NAME}: Service name (for example, MyService).
• {EXE FILE PATH}: The app's full executable path (for example, d:\myservice\myservice.exe). Include the executable's file name with extension.
• {EXE FOLDER PATH}: The app's full executable folder path (for example d:\myservice).
• {DESCRIPTION}: Service description (for example, My sample service).
• {DISPLAY NAME}: Service display name (for example, My Service).

Start a service

Start a service with the following PowerShell 6 command:

Start-Service -Name {SERVICE NAME}

The command takes a few seconds to start the service.

Determine a service's status

To check the status of a service, use the following PowerShell 6 command:

Get-Service -Name {SERVICE NAME}

The status is reported as one of the following values:

• Starting
• Running
• Stopping
• Stopped

Stop a service

Stop a service with the following PowerShell 6 command:

Stop-Service -Name {SERVICE NAME}

Remove a service

After a short delay to stop a service, remove a service with the following PowerShell 6 command:

Remove-Service -Name {SERVICE NAME}

Proxy server and load balancer scenarios

Services that interact with requests from the Internet or a corporate network and are behind a proxy or load balancer might require additional configuration. For more information, see Configure ASP.NET Core to work with proxy servers and load balancers.

Configure endpoints

By default, ASP.NET Core binds to http://localhost:5000. Configure the URL and port by setting the ASPNETCORE_URLS environment variable.

For additional URL and port configuration approaches, see the relevant server article:

• Configure endpoints for the ASP.NET Core Kestrel web server
• HTTP.sys web server implementation in ASP.NET Core

The preceding guidance covers support for HTTPS endpoints. For example, configure the app for HTTPS when authentication is used with a Windows Service.

Note

Use of the ASP.NET Core HTTPS development certificate to secure a service endpoint isn't supported.

Current directory and content root

The current working directory returned by calling GetCurrentDirectory for a Windows Service is the C:\WINDOWS\system32 folder. The system32 folder isn't a suitable location to store a service's files (for example, settings files). Use one of the following approaches to maintain and access a service's assets and settings files.

Use ContentRootPath or ContentRootFileProvider

Use IHostEnvironment.ContentRootPath or ContentRootFileProvider to locate an app's resources.

When the app runs as a service, UseWindowsService sets the ContentRootPath to AppContext.BaseDirectory.

The app's default settings files, appsettings.json and appsettings.{Environment}.json, are loaded from the app's content root by calling CreateDefaultBuilder during host construction.

For other settings files loaded by developer code in ConfigureAppConfiguration, there's no need to call SetBasePath. In the following example, the custom_settings.json file exists in the app's content root and is loaded without explicitly setting a base path:

public class Program
{
    public static void Main(string[] args)
    {
        CreateHostBuilder(args).Build().Run();
    }

    public static IHostBuilder CreateHostBuilder(string[] args) =>
        Host.CreateDefaultBuilder(args)
            .UseWindowsService()
            .ConfigureAppConfiguration((hostingContext, config) =>
            {
                config.AddJsonFile("custom_settings.json");
            })
            .ConfigureServices((hostContext, services) =>
            {
                services.AddHostedService<Worker>();
            });
}

Don't attempt to use GetCurrentDirectory to obtain a resource path because a Windows Service app returns the C:\WINDOWS\system32 folder as its current directory.

Store a service's files in a suitable location on disk

Specify an absolute path with SetBasePath when using an IConfigurationBuilder to the folder containing the files.

Troubleshoot

To troubleshoot a Windows Service app, see Troubleshoot and debug ASP.NET Core projects.

Common errors

• An old or pre-release version of PowerShell is in use.
• The registered service doesn't use the app's published output from the dotnet publish command. Output of the dotnet build command isn't supported for app deployment. Published assets are found in either of the following folders depending on the deployment type:
  • bin/Release/{TARGET FRAMEWORK}/publish (FDD)
  • bin/Release/{TARGET FRAMEWORK}/{RUNTIME IDENTIFIER}/publish (SCD)
• The service isn't in the RUNNING state.
• The paths to resources that the app uses (for example, certificates) are incorrect. The base path of a Windows Service is c:\Windows\System32.
• The user doesn't have Log on as a service rights.
• The user's password is expired or incorrectly passed when executing the New-Service PowerShell command.
• The app requires ASP.NET Core authentication but isn't configured for secure connections (HTTPS).
• The request URL port is incorrect or not configured correctly in the app.

System and Application Event Logs

Access the System and Application Event Logs:

1. Open the Start menu, search for Event Viewer, and select the Event Viewer app.
2. In Event Viewer, open the Windows Logs node.
3. Select System to open the System Event Log. Select Application to open the Application Event Log.
4. Search for errors associated with the failing app.

Run the app at a command prompt

Many startup errors don't produce useful information in the event logs. You can find the cause of some errors by running the app at a command prompt on the hosting system. To log additional detail from the app, lower the log level or run the app in the Development environment.

Clear package caches

A functioning app may fail immediately after upgrading either the .NET Core SDK on the development machine or changing package versions within the app. In some cases, incoherent packages may break an app when performing major upgrades. Most of these issues can be fixed by following these instructions:

1. Delete the bin and obj folders.

2. Clear the package caches by executing dotnet nuget locals all --clear from a command shell.

   Clearing package caches can also be accomplished with the nuget.exe tool and executing the command nuget locals all -clear. nuget.exe isn't a bundled install with the Windows desktop operating system and must be obtained separately from the NuGet website.

3. Restore and rebuild the project.

4. Delete all of the files in the deployment folder on the server prior to redeploying the app.

Slow or unresponsive app

A crash dump is a snapshot of the system's memory and can help determine the cause of an app crash, startup failure, or slow app.

App crashes or encounters an exception

Obtain and analyze a dump from Windows Error Reporting (WER):

1. Create a folder to hold crash dump files at c:\dumps.

2. Run the EnableDumps PowerShell script with the application executable name:

   .\EnableDumps {APPLICATION EXE} c:\dumps
   

3. Run the app under the conditions that cause the crash to occur.

4. After the crash has occurred, run the DisableDumps PowerShell script:

   .\DisableDumps {APPLICATION EXE}
   

After an app crashes and dump collection is complete, the app is allowed to terminate normally. The PowerShell script configures WER to collect up to five dumps per app.

Warning

Crash dumps might take up a large amount of disk space (up to several gigabytes each).

App is unresponsive, fails during startup, or runs normally

When an app stops responding but doesn't crash, fails during startup, or runs normally, see User-Mode Dump Files: Choosing the Best Tool to select an appropriate tool to produce the dump.

Analyze the dump

A dump can be analyzed using several approaches. For more information, see Analyzing a User-Mode Dump File.

Additional resources

• View or download sample code (how to download)
• Kestrel endpoint configuration (includes HTTPS configuration and SNI support)
• .NET Generic Host in ASP.NET Core
• Troubleshoot and debug ASP.NET Core projects