Handle errors in ASP.NET Core Blazor apps

This article describes how Blazor manages unhandled exceptions and how to develop apps that detect and handle errors.

Detailed errors during development

When a Blazor app isn't functioning properly during development, receiving detailed error information from the app assists in troubleshooting and fixing the issue. When an error occurs, Blazor apps display a light yellow bar at the bottom of the screen:

• During development, the bar directs you to the browser console, where you can see the exception.
• In production, the bar notifies the user that an error has occurred and recommends refreshing the browser.

The UI for this error handling experience is part of the Blazor project templates. Not all versions of the Blazor project templates use the data-nosnippet attribute to signal to browsers not to cache the contents of the error UI, but all versions of the Blazor documentation apply the attribute.

In a Blazor Web App, customize the experience in the MainLayout component. Because the Environment Tag Helper (for example, <environment include="Production">...</environment>) isn't supported in Razor components, the following example injects IHostEnvironment to configure error messages for different environments.

At the top of MainLayout.razor:

@inject IHostEnvironment HostEnvironment

Create or modify the Blazor error UI markup:

<div id="blazor-error-ui" data-nosnippet>
    @if (HostEnvironment.IsProduction())
    {
        <span>An error has occurred.</span>
    }
    else
    {
        <span>An unhandled exception occurred.</span>
    }
    <a href="" class="reload">Reload</a>
    <a class="dismiss">🗙</a>
</div>

In a Blazor WebAssembly app, customize the experience in the wwwroot/index.html file:

<div id="blazor-error-ui" data-nosnippet>
    An unhandled error has occurred.
    <a href="" class="reload">Reload</a>
    <a class="dismiss">🗙</a>
</div>

The blazor-error-ui element is normally hidden due to the presence of the display: none style of the blazor-error-ui CSS class in the app's auto-generated stylesheet. When an error occurs, the framework applies display: block to the element.

Detailed circuit errors

This section applies to Blazor Web Apps operating over a circuit.

Client-side errors don't include the call stack and don't provide detail on the cause of the error, but server logs do contain such information. For development purposes, sensitive circuit error information can be made available to the client by enabling detailed errors.

Set CircuitOptions.DetailedErrors to true. For more information and an example, see ASP.NET Core Blazor SignalR guidance.

An alternative to setting CircuitOptions.DetailedErrors is to set the DetailedErrors configuration key to true in the app's Development environment settings file (appsettings.Development.json). Additionally, set SignalR server-side logging (Microsoft.AspNetCore.SignalR) to Debug or Trace for detailed SignalR logging.

appsettings.Development.json:

{
  "DetailedErrors": true,
  "Logging": {
    "LogLevel": {
      "Default": "Information",
      "Microsoft": "Warning",
      "Microsoft.Hosting.Lifetime": "Information",
      "Microsoft.AspNetCore.SignalR": "Debug"
    }
  }
}

The DetailedErrors configuration key can also be set to true using the ASPNETCORE_DETAILEDERRORS environment variable with a value of true on Development/Staging environment servers or on your local system.

Warning

Always avoid exposing error information to clients on the Internet, which is a security risk.

Detailed errors for Razor component server-side rendering

This section applies to Blazor Web Apps.

Use the RazorComponentsServiceOptions.DetailedErrors option to control producing detailed information on errors for Razor component server-side rendering. The default value is false.

The following example enables detailed errors:

builder.Services.AddRazorComponents(options => 
    options.DetailedErrors = builder.Environment.IsDevelopment());

Warning

Only enable detailed errors in the Development environment. Detailed errors may contain sensitive information about the app that malicious users can use in an attack.

The preceding example provides a degree of safety by setting the value of DetailedErrors based on the value returned by IsDevelopment. When the app is in the Development environment, DetailedErrors is set to true. This approach isn't foolproof because it's possible to host a production app on a public server in the Development environment.

Manage unhandled exceptions in developer code

For an app to continue after an error, the app must have error handling logic. Later sections of this article describe potential sources of unhandled exceptions.

In production, don't render framework exception messages or stack traces in the UI. Rendering exception messages or stack traces could:

• Disclose sensitive information to end users.
• Help a malicious user discover weaknesses in an app that can compromise the security of the app, server, or network.

Unhandled exceptions for circuits

This section applies to server-side apps operating over a circuit.

Razor components with server interactivity enabled are stateful on the server. While users interact with the component on the server, they maintain a connection to the server known as a circuit. The circuit holds active component instances, plus many other aspects of state, such as:

• The most recent rendered output of components.
• The current set of event-handling delegates that could be triggered by client-side events.

If a user opens the app in multiple browser tabs, the user creates multiple independent circuits.

Blazor treats most unhandled exceptions as fatal to the circuit where they occur. If a circuit is terminated due to an unhandled exception, the user can only continue to interact with the app by reloading the page to create a new circuit. Circuits outside of the one that's terminated, which are circuits for other users or other browser tabs, aren't affected. This scenario is similar to a desktop app that crashes. The crashed app must be restarted, but other apps aren't affected.

The framework terminates a circuit when an unhandled exception occurs for the following reasons:

• An unhandled exception often leaves the circuit in an undefined state.
• The app's normal operation can't be guaranteed after an unhandled exception.
• Security vulnerabilities may appear in the app if the circuit continues in an undefined state.

Global exception handling

For approaches to handling exceptions globally, see the following sections:

• Error boundaries: Applies to all Blazor apps.
• Alternative global exception handling: Applies to Blazor Server, Blazor WebAssembly, and Blazor Web Apps (8.0 or later) that adopt a global interactive render mode.

Error boundaries

Error boundaries provide a convenient approach for handling exceptions. The ErrorBoundary component:

• Renders its child content when an error hasn't occurred.
• Renders error UI when an unhandled exception is thrown by any component within the error boundary.

To define an error boundary, use the ErrorBoundary component to wrap one or more other components. The error boundary manages unhandled exceptions thrown by the components that it wraps.

<ErrorBoundary>
    ...
</ErrorBoundary>

To implement an error boundary in a global fashion, add the boundary around the body content of the app's main layout.

In MainLayout.razor:

<article class="content px-4">
    <ErrorBoundary>
        @Body
    </ErrorBoundary>
</article>

In Blazor Web Apps with the error boundary only applied to a static MainLayout component, the boundary is only active during static server-side rendering (static SSR). The boundary doesn't activate just because a component further down the component hierarchy is interactive.

An interactive render mode can't be applied to the MainLayout component because the component's Body parameter is a RenderFragment delegate, which is arbitrary code and can't be serialized. To enable interactivity broadly for the MainLayout component and the rest of the components further down the component hierarchy, the app must adopt a global interactive render mode by applying the interactive render mode to the HeadOutlet and Routes component instances in the app's root component, which is typically the App component. The following example adopts the Interactive Server (InteractiveServer) render mode globally.

In Components/App.razor:

<HeadOutlet @rendermode="InteractiveServer" />

...

<Routes @rendermode="InteractiveServer" />

If you prefer not to enable global interactivity, place the error boundary farther down the component hierarchy. The important concepts to keep in mind are that wherever the error boundary is placed:

• If the component where the error boundary is placed isn't interactive, the error boundary is only capable of activating on the server during static SSR. For example, the boundary can activate when an error is thrown in a component lifecycle method but not for an event triggered by user interactivity within the component, such as an error thrown by a button click handler.
• If the component where the error boundary is placed is interactive, the error boundary is capable of activating for interactive components that it wraps.

Note

The preceding considerations aren't relevant for standalone Blazor WebAssembly apps because the client-side rendering (CSR) of a Blazor WebAssembly app is completely interactive.

Consider the following example, where an exception thrown by an embedded counter component is caught by an error boundary in the Home component, which adopts an interactive render mode.

EmbeddedCounter.razor:

<h1>Embedded Counter</h1>

<p role="status">Current count: @currentCount</p>
<button class="btn btn-primary" @onclick="IncrementCount">Click me</button>

@code {
    private int currentCount = 0;

    private void IncrementCount()
    {
        currentCount++;

        if (currentCount > 5)
        {
            throw new InvalidOperationException("Current count is too big!");
        }
    }
}

Home.razor:

@page "/"
@rendermode InteractiveServer

<PageTitle>Home</PageTitle>

<h1>Home</h1>

<ErrorBoundary>
    <EmbeddedCounter />
</ErrorBoundary>

If the unhandled exception is thrown for a currentCount over five:

• The error is logged normally (System.InvalidOperationException: Current count is too big!).
• The exception is handled by the error boundary.
• The default error UI is rendered by the error boundary.

The ErrorBoundary component renders an empty <div> element using the blazor-error-boundary CSS class for its error content. The colors, text, and icon for the default UI are defined in the app's stylesheet in the wwwroot folder, so you're free to customize the error UI.

To change the default error content:

• Wrap the components of the error boundary in the ChildContent property.
• Set the ErrorContent property to the error content.

The following example wraps the EmbeddedCounter component and supplies custom error content:

<ErrorBoundary>
    <ChildContent>
        <EmbeddedCounter />
    </ChildContent>
    <ErrorContent>
        <p class="errorUI">😈 A rotten gremlin got us. Sorry!</p>
    </ErrorContent>
</ErrorBoundary>

For the preceding example, the app's stylesheet presumably includes an errorUI CSS class to style the content. The error content is rendered from the ErrorContent property without a block-level element. A block-level element, such as a division (<div>) or a paragraph (<p>) element, can wrap the error content markup, but it isn't required.

Optionally, use the context (@context) of the ErrorContent to obtain error data:

<ErrorContent>
    @context.HelpLink
</ErrorContent>

The ErrorContent can also name the context. In the following example, the context is named exception:

<ErrorContent Context="exception">
    @exception.HelpLink
</ErrorContent>

Warning

Always avoid exposing error information to clients on the Internet, which is a security risk.

If the error boundary is defined in the app's layout, the error UI is seen regardless of which page the user navigates to after the error occurs. We recommend narrowly scoping error boundaries in most scenarios. If you broadly scope an error boundary, you can reset it to a non-error state on subsequent page navigation events by calling the error boundary's Recover method.

In MainLayout.razor:

• Add a field for the ErrorBoundary to capture a reference to it with the @ref attribute directive.
• In the OnParameterSet lifecycle method, you can trigger a recovery on the error boundary with Recover to clear the error when the user navigates to a different component.

...

<ErrorBoundary @ref="errorBoundary">
    @Body
</ErrorBoundary>

...

@code {
    private ErrorBoundary? errorBoundary;

    protected override void OnParametersSet()
    {
        errorBoundary?.Recover();
    }
}

To avoid the infinite loop where recovering merely rerenders a component that throws the error again, don't call Recover from rendering logic. Only call Recover when:

• The user performs a UI gesture, such as selecting a button to indicate that they want to retry a procedure or when the user navigates to a new component.
• Additional logic that executes also clears the exception. When the component is rerendered, the error doesn't reoccur.

The following example permits the user to recover from the exception with a button:

<ErrorBoundary @ref="errorBoundary">
    <ChildContent>
        <EmbeddedCounter />
    </ChildContent>
    <ErrorContent>
        <div class="alert alert-danger" role="alert">
            <p class="fs-3 fw-bold">😈 A rotten gremlin got us. Sorry!</p>
            <p>@context.HelpLink</p>
            <button class="btn btn-info" @onclick="_ => errorBoundary?.Recover()">
                Clear
            </button>
        </div>
    </ErrorContent>
</ErrorBoundary>

@code {
    private ErrorBoundary? errorBoundary;
}

You can also subclass ErrorBoundary for custom processing by overriding OnErrorAsync. The following example merely logs the error, but you can implement any error handling code you wish. You can remove the line that returns a CompletedTask if your code awaits an asynchronous task.

CustomErrorBoundary.razor:

@inherits ErrorBoundary
@inject ILogger<CustomErrorBoundary> Logger

@if (CurrentException is null)
{
    @ChildContent
}
else if (ErrorContent is not null)
{
    @ErrorContent(CurrentException)
}

@code {
    protected override Task OnErrorAsync(Exception ex)
    {
        Logger.LogError(ex, "😈 A rotten gremlin got us. Sorry!");
        return Task.CompletedTask;
    }
}

The preceding example can also be implemented as a class.

CustomErrorBoundary.cs:

using Microsoft.AspNetCore.Components;
using Microsoft.AspNetCore.Components.Web;

namespace BlazorSample;

public class CustomErrorBoundary : ErrorBoundary
{
    [Inject]
    ILogger<CustomErrorBoundary> Logger {  get; set; } = default!;

    protected override Task OnErrorAsync(Exception ex)
    {
        Logger.LogError(ex, "😈 A rotten gremlin got us. Sorry!");
        return Task.CompletedTask;
    }
}

Either of the preceding implementations used in a component:

<CustomErrorBoundary>
    ...
</CustomErrorBoundary>

Alternative global exception handling

The approach described in this section applies to Blazor Server, Blazor WebAssembly, and Blazor Web Apps that adopt a global interactive render mode (InteractiveServer, InteractiveWebAssembly, or InteractiveAuto). The approach doesn't work with Blazor Web Apps that adopt per-page/component render modes or static server-side rendering (static SSR) because the approach relies on a CascadingValue/CascadingParameter, which don't work across render mode boundaries or with components that adopt static SSR.

An alternative to using Error boundaries (ErrorBoundary) is to pass a custom error component as a CascadingValue to child components. An advantage of using a component over using an injected service or a custom logger implementation is that a cascaded component can render content and apply CSS styles when an error occurs.

The following ProcessError component example merely logs errors, but methods of the component can process errors in any way required by the app, including through the use of multiple error processing methods.

ProcessError.razor:

@inject ILogger<ProcessError> Logger

<CascadingValue Value="this">
    @ChildContent
</CascadingValue>

@code {
    [Parameter]
    public RenderFragment? ChildContent { get; set; }

    public void LogError(Exception ex)
    {
        Logger.LogError("ProcessError.LogError: {Type} Message: {Message}", 
            ex.GetType(), ex.Message);

        // Call StateHasChanged if LogError directly participates in 
        // rendering. If LogError only logs or records the error,
        // there's no need to call StateHasChanged.
        //StateHasChanged();
    }
}

Note

For more information on RenderFragment, see ASP.NET Core Razor components.

When using this approach in a Blazor Web App, open the Routes component and wrap the Router component (<Router>...</Router>) with the ProcessError component. This permits the ProcessError component to cascade down to any component of the app where the ProcessError component is received as a CascadingParameter.

In Routes.razor:

<ProcessError>
    <Router ...>
        ...
    </Router>
</ProcessError>

When using this approach in a Blazor Server or Blazor WebAssembly app, open the App component, wrap the Router component (<Router>...</Router>) with the ProcessError component. This permits the ProcessError component to cascade down to any component of the app where the ProcessError component is received as a CascadingParameter.

In App.razor:

<ProcessError>
    <Router ...>
        ...
    </Router>
</ProcessError>

To process errors in a component:

• Designate the ProcessError component as a CascadingParameter in the @code block. In an example Counter component in an app based on a Blazor project template, add the following ProcessError property:

  [CascadingParameter]
  public ProcessError? ProcessError { get; set; }
  

• Call an error processing method in any catch block with an appropriate exception type. The example ProcessError component only offers a single LogError method, but the error processing component can provide any number of error processing methods to address alternative error processing requirements throughout the app. The following Counter component @code block example includes the ProcessError cascading parameter and traps an exception for logging when the count is greater than five:

  @code {
      private int currentCount = 0;
  
      [CascadingParameter]
      public ProcessError? ProcessError { get; set; }
  
      private void IncrementCount()
      {
          try
          {
              currentCount++;
  
              if (currentCount > 5)
              {
                  throw new InvalidOperationException("Current count is over five!");
              }
          }
          catch (Exception ex)
          {
              ProcessError?.LogError(ex);
          }
      }
  }
  

The logged error:

fail: {COMPONENT NAMESPACE}.ProcessError[0]
ProcessError.LogError: System.InvalidOperationException Message: Current count is over five!

If the LogError method directly participates in rendering, such as showing a custom error message bar or changing the CSS styles of the rendered elements, call StateHasChanged at the end of the LogError method to rerender the UI.

Because the approaches in this section handle errors with a try-catch statement, an app's SignalR connection between the client and server isn't broken when an error occurs and the circuit remains alive. Other unhandled exceptions remain fatal to a circuit. For more information, see the section on how a circuit reacts to unhandled exceptions.

Log errors with a persistent provider

If an unhandled exception occurs, the exception is logged to ILogger instances configured in the service container. Blazor apps log console output with the Console Logging Provider. Consider logging to a location on the server (or backend web API for client-side apps) with a provider that manages log size and log rotation. Alternatively, the app can use an Application Performance Management (APM) service, such as Azure Application Insights (Azure Monitor).

Note

Native Application Insights features to support client-side apps and native Blazor framework support for Google Analytics might become available in future releases of these technologies. For more information, see Support App Insights in Blazor WASM Client Side (microsoft/ApplicationInsights-dotnet #2143) and Web analytics and diagnostics (includes links to community implementations) (dotnet/aspnetcore #5461). In the meantime, a client-side app can use the Application Insights JavaScript SDK with JS interop to log errors directly to Application Insights from a client-side app.

During development in a Blazor app operating over a circuit, the app usually sends the full details of exceptions to the browser's console to aid in debugging. In production, detailed errors aren't sent to clients, but an exception's full details are logged on the server.

You must decide which incidents to log and the level of severity of logged incidents. Hostile users might be able to trigger errors deliberately. For example, don't log an incident from an error where an unknown ProductId is supplied in the URL of a component that displays product details. Not all errors should be treated as incidents for logging.

For more information, see the following articles:

• ASP.NET Core Blazor logging
• Handle errors in ASP.NET Core‡
• Create web APIs with ASP.NET Core

‡Applies to server-side Blazor apps and other server-side ASP.NET Core apps that are web API backend apps for Blazor. Client-side apps can trap and send error information on the client to a web API, which logs the error information to a persistent logging provider.

Places where errors may occur

Framework and app code may trigger unhandled exceptions in any of the following locations, which are described further in the following sections of this article:

• Component instantiation
• Lifecycle methods
• Rendering logic
• Event handlers
• Component disposal
• JavaScript interop
• Prerendering

Component instantiation

When Blazor creates an instance of a component:

• The component's constructor is invoked.
• The constructors of DI services supplied to the component's constructor via the @inject directive or the [Inject] attribute are invoked.

An error in an executed constructor or a setter for any [Inject] property results in an unhandled exception and stops the framework from instantiating the component. If the app is operating over a circuit, the circuit fails. If constructor logic may throw exceptions, the app should trap the exceptions using a try-catch statement with error handling and logging.

Lifecycle methods

During the lifetime of a component, Blazor invokes lifecycle methods. If any lifecycle method throws an exception, synchronously or asynchronously, the exception is fatal to a circuit. For components to deal with errors in lifecycle methods, add error handling logic.

In the following example where OnParametersSetAsync calls a method to obtain a product:

• An exception thrown in the ProductRepository.GetProductByIdAsync method is handled by a try-catch statement.
• When the catch block is executed:
  • loadFailed is set to true, which is used to display an error message to the user.
  • The error is logged.

@page "/product-details/{ProductId:int?}"
@inject ILogger<ProductDetails> Logger
@inject IProductRepository Product

<PageTitle>Product Details</PageTitle>

<h1>Product Details Example</h1>

@if (details != null)
{
    <h2>@details.ProductName</h2>
    <p>
        @details.Description
        <a href="@details.Url">Company Link</a>
    </p>
    
}
else if (loadFailed)
{
    <h1>Sorry, we could not load this product due to an error.</h1>
}
else
{
    <h1>Loading...</h1>
}

@code {
    private ProductDetail? details;
    private bool loadFailed;

    [Parameter]
    public int ProductId { get; set; }

    protected override async Task OnParametersSetAsync()
    {
        try
        {
            loadFailed = false;

            // Reset details to null to display the loading indicator
            details = null;

            details = await Product.GetProductByIdAsync(ProductId);
        }
        catch (Exception ex)
        {
            loadFailed = true;
            Logger.LogWarning(ex, "Failed to load product {ProductId}", ProductId);
        }
    }

    public class ProductDetail
    {
        public string? ProductName { get; set; }
        public string? Description { get; set; }
        public string? Url { get; set; }
    }

    /*
    * Register the service in Program.cs:
    * using static BlazorSample.Components.Pages.ProductDetails;
    * builder.Services.AddScoped<IProductRepository, ProductRepository>();
    */

    public interface IProductRepository
    {
        public Task<ProductDetail> GetProductByIdAsync(int id);
    }

    public class ProductRepository : IProductRepository
    {
        public Task<ProductDetail> GetProductByIdAsync(int id)
        {
            return Task.FromResult(
                new ProductDetail()
                {
                    ProductName = "Flowbee ",
                    Description = "The Revolutionary Haircutting System You've Come to Love!",
                    Url = "https://flowbee.com/"
                });
        }
    }
}

Rendering logic

The declarative markup in a Razor component file (.razor) is compiled into a C# method called BuildRenderTree. When a component renders, BuildRenderTree executes and builds up a data structure describing the elements, text, and child components of the rendered component.

Rendering logic can throw an exception. An example of this scenario occurs when @someObject.PropertyName is evaluated but @someObject is null. For Blazor apps operating over a circuit, an unhandled exception thrown by rendering logic is fatal to the app's circuit.

To prevent a NullReferenceException in rendering logic, check for a null object before accessing its members. In the following example, person.Address properties aren't accessed if person.Address is null:

@if (person.Address != null)
{
    <div>@person.Address.Line1</div>
    <div>@person.Address.Line2</div>
    <div>@person.Address.City</div>
    <div>@person.Address.Country</div>
}

The preceding code assumes that person isn't null. Often, the structure of the code guarantees that an object exists at the time the component is rendered. In those cases, it isn't necessary to check for null in rendering logic. In the prior example, person might be guaranteed to exist because person is created when the component is instantiated, as the following example shows:

@code {
    private Person person = new();

    ...
}

Event handlers

Client-side code triggers invocations of C# code when event handlers are created using:

• @onclick
• @onchange
• Other @on... attributes
• @bind

Event handler code might throw an unhandled exception in these scenarios.

If the app calls code that could fail for external reasons, trap exceptions using a try-catch statement with error handling and logging.

If an event handler throws an unhandled exception (for example, a database query fails) that isn't trapped and handled by developer code:

• The framework logs the exception.
• In a Blazor app operating over a circuit, the exception is fatal to the app's circuit.

Component disposal

A component may be removed from the UI, for example, because the user has navigated to another page. When a component that implements System.IDisposable is removed from the UI, the framework calls the component's Dispose method.

If the component's Dispose method throws an unhandled exception in a Blazor app operating over a circuit, the exception is fatal to the app's circuit.

If disposal logic may throw exceptions, the app should trap the exceptions using a try-catch statement with error handling and logging.

For more information on component disposal, see ASP.NET Core Razor component lifecycle.

JavaScript interop

IJSRuntime is registered by the Blazor framework. IJSRuntime.InvokeAsync allows .NET code to make asynchronous calls to the JavaScript (JS) runtime in the user's browser.

The following conditions apply to error handling with InvokeAsync:

• If a call to InvokeAsync fails synchronously, a .NET exception occurs. A call to InvokeAsync may fail, for example, because the supplied arguments can't be serialized. Developer code must catch the exception. If app code in an event handler or component lifecycle method doesn't handle an exception in a Blazor app operating over a circuit, the resulting exception is fatal to the app's circuit.
• If a call to InvokeAsync fails asynchronously, the .NET Task fails. A call to InvokeAsync may fail, for example, because the JS-side code throws an exception or returns a Promise that completed as rejected. Developer code must catch the exception. If using the await operator, consider wrapping the method call in a try-catch statement with error handling and logging. Otherwise in a Blazor app operating over a circuit, the failing code results in an unhandled exception that's fatal to the app's circuit.
• Calls to InvokeAsync must complete within a certain period or else the call times out. The default timeout period is one minute. The timeout protects the code against a loss in network connectivity or JS code that never sends back a completion message. If the call times out, the resulting System.Threading.Tasks fails with an OperationCanceledException. Trap and process the exception with logging.

Similarly, JS code may initiate calls to .NET methods indicated by the [JSInvokable] attribute. If these .NET methods throw an unhandled exception:

• In a Blazor app operating over a circuit, the exception isn't treated as fatal to the app's circuit.
• The JS-side Promise is rejected.

You have the option of using error handling code on either the .NET side or the JS side of the method call.

For more information, see the following articles:

• Call JavaScript functions from .NET methods in ASP.NET Core Blazor
• Call .NET methods from JavaScript functions in ASP.NET Core Blazor

Prerendering

Razor components are prerendered by default so that their rendered HTML markup is returned as part of the user's initial HTTP request.

In a Blazor app operating over a circuit, prerendering works by:

• Creating a new circuit for all of the prerendered components that are part of the same page.
• Generating the initial HTML.
• Treating the circuit as disconnected until the user's browser establishes a SignalR connection back to the same server. When the connection is established, interactivity on the circuit is resumed and the components' HTML markup is updated.

For prerendered client-side components, prerendering works by:

• Generating initial HTML on the server for all of the prerendered components that are part of the same page.
• Making the component interactive on the client after the browser has loaded the app's compiled code and the .NET runtime (if not already loaded) in the background.

If a component throws an unhandled exception during prerendering, for example, during a lifecycle method or in rendering logic:

• In a Blazor app operating over a circuit, the exception is fatal to the circuit. For prerendered client-side components, the exception prevents rendering the component.
• The exception is thrown up the call stack from the ComponentTagHelper.

Under normal circumstances when prerendering fails, continuing to build and render the component doesn't make sense because a working component can't be rendered.

To tolerate errors that may occur during prerendering, error handling logic must be placed inside a component that may throw exceptions. Use try-catch statements with error handling and logging. Instead of wrapping the ComponentTagHelper in a try-catch statement, place error handling logic in the component rendered by the ComponentTagHelper.

Advanced scenarios

Recursive rendering

Components can be nested recursively. This is useful for representing recursive data structures. For example, a TreeNode component can render more TreeNode components for each of the node's children.

When rendering recursively, avoid coding patterns that result in infinite recursion:

• Don't recursively render a data structure that contains a cycle. For example, don't render a tree node whose children includes itself.
• Don't create a chain of layouts that contain a cycle. For example, don't create a layout whose layout is itself.
• Don't allow an end user to violate recursion invariants (rules) through malicious data entry or JavaScript interop calls.

Infinite loops during rendering:

• Causes the rendering process to continue forever.
• Is equivalent to creating an unterminated loop.

In these scenarios, the Blazor fails and usually attempts to:

• Consume as much CPU time as permitted by the operating system, indefinitely.
• Consume an unlimited amount of memory. Consuming unlimited memory is equivalent to the scenario where an unterminated loop adds entries to a collection on every iteration.

To avoid infinite recursion patterns, ensure that recursive rendering code contains suitable stopping conditions.

Custom render tree logic

Most Razor components are implemented as Razor component files (.razor) and are compiled by the framework to produce logic that operates on a RenderTreeBuilder to render their output. However, a developer may manually implement RenderTreeBuilder logic using procedural C# code. For more information, see ASP.NET Core Blazor advanced scenarios (render tree construction).

Warning

Use of manual render tree builder logic is considered an advanced and unsafe scenario, not recommended for general component development.

If RenderTreeBuilder code is written, the developer must guarantee the correctness of the code. For example, the developer must ensure that:

• Calls to OpenElement and CloseElement are correctly balanced.
• Attributes are only added in the correct places.

Incorrect manual render tree builder logic can cause arbitrary undefined behavior, including crashes, the app or server to stop responding, and security vulnerabilities.

Consider manual render tree builder logic on the same level of complexity and with the same level of danger as writing assembly code or Microsoft Intermediate Language (MSIL) instructions by hand.

Additional resources

• Handle caught exceptions outside of a Razor component's lifecycle
• ASP.NET Core Blazor logging
• Handle errors in ASP.NET Core†
• Create web APIs with ASP.NET Core
• Blazor samples GitHub repository (dotnet/blazor-samples) (how to download)

†Applies to backend ASP.NET Core web API apps that client-side Blazor apps use for logging.