Generally, when we are using any uptime monitoring systems or load balancers, these systems will keep monitoring the health of the application and based on its health condition it will decide to send the request to serve it. For this earlier, we use to create a special endpoint where it will return any error message or code to indicate the health of the API/service.
 

Following is the sample, an endpoint /health where it verifying the database connection and returns the result accordingly.
    [Route("health")] 
    public ActionResult Health() 
    { 
        using (var connection = new SqlConnection(_connectionString)) 
        { 
            try 
            { 
                connection.Open(); 
            } 
            catch (SqlException) 
            { 
                return new HttpStatusCodeResult(503, "Database connection is unhealthy"); 
            } 
        } 
     
        return new EmptyResult(); 
    } 

When we ran the application with endpoint /health, it will display an empty message with 200 status code and 503 status code when there is any connectivity issue while connecting to the database.

Now, based on these resulted status codes, monitoring systems can take appropriate actions like removing this particular services instance from its list of healthy services so that no requests will be redirected to it until it becomes healthy (in our case, when database connectivity issue resolves).
 
We need to keep adding more external resource health checks accordingly.
 
Since .NET Core 2.2, we no need to add a special controller for health check endpoint, instead, the framework itself providing Health Check services as follows.
 
NuGet Package
You have to install following NuGet package
 
Install-Package Microsoft.Extensions.Diagnostics.HealthChecks
 
Once the package is installed, we need to add the following lines at ConfigureServices() and Configure() methods in Startup.cs file.
    public void ConfigureServices(IServiceCollection services) 
    { 
        services.AddHealthChecks(); 
    } 
      
    public void Configure(IApplicationBuilder app, IWebHostEnvironment env) 
    { 
        app.UseHealthChecks("/Health"); 
    } 

As observed, we are providing the endpoint name in Configure() method. These lines of code will enable a dynamic endpoint "/Health" and display either Healthy/UnHealthy results based on its health state.
 
But, where can we write our custom logic to replicate the above? Yes, we have many features to customize our needs from logic to display results.
 
Adding Custom Logic
 
We can perform in two ways the following are those.
 
Option 1
 
In ConfigureServices method,
    public void ConfigureServices(IServiceCollection services) { 
     services.AddHealthChecks() 
      .AddCheck("sql", () => { 
     
       using(var connection = new SqlConnection(_connectionString)) { 
        try { 
         connection.Open(); 
        } catch (SqlException) { 
         return HealthCheckResult.Unhealthy(); 
        } 
       } 
     
       return HealthCheckResult.Healthy(); 
     
      }); 
    } 

Here, we can use an anonymous method to write the custom logic using AddCheck() method. This will expect a HealthCheckResult object as a result. This object will contain 3 options,
    Healthy
    Unhealthy
    Degraded

Based on the result we need to return appropriately so that runtime will return the status code accordingly. For example, in the above code, if database connection passes it will return a 200 status code (Healthy) and 503 status code (Unhealthy) if failed.
 
Option 2 - In a separate class
The class should implement an IHealthCheck interface and implement CheckHealthAsync() method as follows,
    public class DatabaseHealthCheck: IHealthCheck { 
     public Task < HealthCheckResult > CheckHealthAsync(HealthCheckContext context, CancellationToken cancellationToken = 
      default) { 
      using(var connection = new SqlConnection(_connectionString)) { 
       try { 
        connection.Open(); 
       } catch (SqlException) { 
        return HealthCheckResult.Healthy(); 
       } 
      } 
     
      return HealthCheckResult.Healthy(); 
     
     } 
    } 

Once we created the class, we need to mention this class in ConfigureServices() method using AddTask<T> method as follows by giving some valid unique names.
    public void ConfigureServices(IServiceCollection services) 
    { 
         services.AddControllers(); 
     
         services.AddHealthChecks() 
              .AddCheck<DatabaseHealthCheck>("sql"); 
    } 

Now, our code is clean and we can add any number of Health Tasks as above and it will be run in the order how we declared here.
 
Custom Status Code
As discussed above, by default it will send 200 status code if we return Healthy and 503 for Unhealthy. The Healthcheck service even provides scope for us to change this default behavior by providing custom status code using its options object as follows.
    var options = new HealthCheckOptions(); 
    options.ResultStatusCodes[HealthStatus.Unhealthy] = 420; 
    app.UseHealthChecks("/Health", options); 

In this example, I replace the status code for the Unhealthy state with 420.
 
Custom Response
The beauty of this tool is, we can even customize our output for more clear detailed information about each Health check task. This will be very useful in case we have multiple health check tasks to analyze which task made the complete service heath status to Unhealthy.
 
We can achieve this via HealthCheckOptions ResponseWriter property.
    public void Configure(IApplicationBuilder app, IWebHostEnvironment env) { 
     app.UseHealthChecks("/Health", new Microsoft.AspNetCore.Diagnostics.HealthChecks.HealthCheckOptions() { 
      ResponseWriter = CustomResponseWriter 
     }); 
    } 
     
    private static Task CustomResponseWriter(HttpContext context, HealthReport healthReport) { 
     context.Response.ContentType = "application/json"; 
     
     var result = JsonConvert.SerializeObject(new { 
      status = healthReport.Status.ToString(), 
       errors = healthReport.Entries.Select(e => new { 
        key = e.Key, value = e.Value.Status.ToString() 
       }) 
     }); 
     return context.Response.WriteAsync(result); 
     
    } 

Now, the above code will display the result in JSON format will all the tasks information. Here, Key represents the Task name we have given (in our case "sql") and value is either Healthy/Unhealthy.
 
Health Check UI
We can even see the health check results on-screen visually by installing the following NuGet package
Install-Package AspNetCore.HealthChecks.UI
 
Once installed need to call respective service methods in ConfigureServices() and Configure() methods accordingly.
    public void ConfigureServices(IServiceCollection services) 
    { 
        services.AddHealthChecksUI(); 
    } 
     
    public void Configure(IApplicationBuilder app, IHostingEnvironment env) 
    { 
        app.UseHealthChecksUI(); 
    } 

Once configured, you can run the application and point to /healthchecks-ui endpoint which display a UI as follows,

.NET Core always interesting. In the past few months, we have just launched ASP.NET Core 2.2 on our hosting environment, and now Microsoft has support latest ASP.NET Core 3. This post will be taking the Contacts project used in the ASP.NET Basics series and migrating it from .NET Core 2.2 to .NET Core 3.0.

Installation

If you are a Visual Studio user you can get .NET Core 3.0 by installing at least Visual Studio 16.3. For those not using Visual Studio, you can download and install .NET Core 3.0 SDK from here. As with previous versions, the SDK is available for Windows, Linux, and Mac.

After installation is complete you can runt the following command from a command prompt to see all the versions of the .NET Core SDK you have installed.

dotnet --list-sdks

You should see 3.0.100 listed. If you are like me you might also see a few preview versions of the SDK that can be uninstalled at this point.

Project File Changes

Right-click on the project and select Edit projectName.csproj.

Change the TargetFramework to netcoreapp3.0.

Before:
<TargetFramework>netcoreapp2.2</TargetFramework>

After:
<TargetFramework>netcoreapp3.0</TargetFramework>

The packages section has a lot of changes. Microsoft.AspNetCore.App is now gone and part of .NET Core without needing a specific reference. The other thing to note is that Entity Framework Core is no longer “in the box” so you will see a lot of references add to make Entity Framework Core usable.

Before:
<PackageReference Include="Microsoft.AspNetCore.App" />
<PackageReference Include="Microsoft.VisualStudio.Web.CodeGeneration.Design" Version="2.2.0" PrivateAssets="All" />
<PackageReference Include="Swashbuckle.AspNetCore" Version="4.0.1" />

After:
<PackageReference Include="Microsoft.EntityFrameworkCore" Version="3.0.0" />
<PackageReference Include="Microsoft.AspNetCore.Diagnostics.EntityFrameworkCore" Version="3.0.0" />
<PackageReference Include="Microsoft.AspNetCore.Identity.EntityFrameworkCore" Version="3.0.0" />
<PackageReference Include="Microsoft.AspNetCore.Identity.UI" Version="3.0.0" />
<PackageReference Include="Microsoft.EntityFrameworkCore.SqlServer" Version="3.0.0" />
<PackageReference Include="Microsoft.EntityFrameworkCore.Tools" Version="3.0.0" />
<PackageReference Include="Microsoft.VisualStudio.Web.CodeGeneration.Design" Version="3.0.0" PrivateAssets="All" />
<PackageReference Include="Swashbuckle.AspNetCore" Version="5.0.0-rc4" />

The last thing to note is that Swashbuckle doesn’t have a final version ready for .NET Core 3 so you will have to make sure you are using version 5 rc2 at a minimum.

The following is my full project file for reference.

<Project Sdk="Microsoft.NET.Sdk.Web">
  <PropertyGroup>
    <TargetFramework>netcoreapp3.0</TargetFramework>
    <AspNetCoreHostingModel>InProcess</AspNetCoreHostingModel>
    <UserSecretsId>aspnet-Contacts-cd2c7b27-e79c-43c7-b3ef-1ecb04374b70</UserSecretsId>
  </PropertyGroup>
  <ItemGroup>
    <PackageReference Include="Microsoft.EntityFrameworkCore" Version="3.0.0" />
    <PackageReference Include="Microsoft.AspNetCore.Diagnostics.EntityFrameworkCore" Version="3.0.0" />
    <PackageReference Include="Microsoft.AspNetCore.Identity.EntityFrameworkCore" Version="3.0.0" />
    <PackageReference Include="Microsoft.AspNetCore.Identity.UI" Version="3.0.0" />
    <PackageReference Include="Microsoft.EntityFrameworkCore.SqlServer" Version="3.0.0" />
    <PackageReference Include="Microsoft.EntityFrameworkCore.Tools" Version="3.0.0" />
    <PackageReference Include="Microsoft.VisualStudio.Web.CodeGeneration.Design" Version="3.0.0" PrivateAssets="All" />
    <PackageReference Include="Swashbuckle.AspNetCore" Version="5.0.0-rc4" />
  </ItemGroup>
</Project>

Program Changes

In Program.cs some changes to the way the host is constructed. The over version may or may not have worked, but I created a new app and pulled this out of it just to make sure I’m using the current set up.

using Microsoft.AspNetCore.Hosting;
using Microsoft.Extensions.Hosting;
namespace Contacts
{
    public class Program
    {
        public static void Main(string[] args)
        {
            CreateHostBuilder(args).Build().Run();
        }
        public static IHostBuilder CreateHostBuilder(string[] args) =>
            Host.CreateDefaultBuilder(args)
                .ConfigureWebHostDefaults(webBuilder =>
                                          {
                                              webBuilder.UseStartup<Startup>();
                                          });
    }
}

Startup Changes

In Startup.cs we have quite a few changes to make. As long as you haven’t do any customization in the constructor you can replace it with the following.

public Startup(IConfiguration configuration)
{
        Configuration = configuration;
}

Next, they type on the configuration property changed from IConfigurationRoot to IConfiguration.

Before:
public IConfigurationRoot Configuration { get; }

After:
public IConfiguration Configuration { get; }

Moving on to the ConfigureServices function has a couple of changes to make. The first is a result of updating to the newer version of the Swagger package where the Info class has been replaced with OpenApiInfo.

Before:
services.AddSwaggerGen(c =>
{
        c.SwaggerDoc("v1", new Info { Title = "Contacts API", Version = "v1"});
});

After:
services.AddSwaggerGen(c =>
{
        c.SwaggerDoc("v1", new OpenApiInfo { Title = "Contacts API", Version = "v1" })
});

Next, we are going to move from using UserMvc to the new AddControllersWithViews which is one of the new more targeted ways to add just the bits of the framework you need.

Before:
services.AddMvc().SetCompatibilityVersion(CompatibilityVersion.Version_2_2);

After:
services.AddControllersWithViews();

Now in the Configure function, the function signature needs to be updated and the logging factory bits removed. If you do need to configure logging that should be handled as part of the HostBuilder.

Before:
public void Configure(IApplicationBuilder app, IHostingEnvironment env, ILoggerFactory loggerFactory)
{
loggerFactory.AddConsole(Configuration.GetSection("Logging"));
      loggerFactory.AddDebug();

After:
public void Configure(IApplicationBuilder app, IWebHostEnvironment env)
{

For the next set of changes, I’m just going to show the result and not the before. The UseCors may or may not apply but the addition of UserRouting and the replacement of UseMvc with UserEndpoint will if you want to use the new endpoint routing features.

app.UseStaticFiles();
app.UseRouting();
app.UseCors(builder =>
            {
                builder.AllowAnyHeader();
                builder.AllowAnyMethod();
                builder.AllowAnyOrigin();
            }
           );
app.UseAuthentication();
app.UseAuthorization();
app.UseEndpoints(endpoints =>
                 {
                     endpoints.MapControllerRoute(
                                                  name: "default",
                                                  pattern: "{controller=Home}/{action=Index}/{id?}");
                     endpoints.MapRazorPages();
                 });

Other Miscellaneous Changes

The only other change I had was the removal of @using Microsoft.AspNetCore.Http.Authentication in a few cshtml files related to login.

In this blog, we are going to discuss vulnerable versions of .Net Core. Microsoft releases the information about security breaches in ASP.Net Core. It informs developers which version they need to update to remove this vulnerability. Microsoft is aware of DOS Attack in the OData library. If you are using OData library in your application in the sense attacker can exploit.

We have two types of dependencies in .net core,

• Direct dependencies
• transitive dependencies

Direct dependencies are dependencies where you specifically add a package to your project, transitive dependencies occur when you add a package to your project that in turn relies on another package.

Mitigation policy
Open your application through visual studio and go to package manager console and run the below command.
command :-  dotnet --info 

By running the above command you will come to know which package we need to update as per Microsoft security guidelines.
 
Direct dependencies
By editing your Cs.proj file we can fix the issue or we can update Nuget Package manager.
 
Transitive dependencies
Transitive dependencies occur when any vulnerable package is referring or relies on another package. By examining the project.asset.json file you can fix the issue.

In this blog, we have discussed vulnerable versions of .Net Core. As per Microsoft security advice it is better to update packages which are in your application.

How Do I Enable Response Compression with GRPC?

There are two main approaches that I’ve found so far to enable the compression of gRPC responses. You can either configure this at the server level so that all gRPC services apply compression to responses, or at a per-service level.

Server Level Options

services.AddGrpc(o =>
{
    o.ResponseCompressionLevel = CompressionLevel.Optimal;
    o.ResponseCompressionAlgorithm = "gzip";
});

When registering the gRPC services into the dependency injection container with the AddGrpc method inside ConfigureServices, it’s possible to set properties on the GrpcServiceOptions. At this level, the options affect all gRPC services which the server implements.

Using the overload of the AddGrpc extension method, we can supply an Action<GrpcServiceOptions>. In the above snippet we’ve set the compression algorithm to “gzip”. We can also optionally control the CompressionLevel which trades the time needed to compress the data against the final size that is achieved by compression. If not specified the current implementation defaults to using CompressionLevel.Fastest. In the preceding snippet, we’ve chosen to allow more time for the compression to reduce the bytes to the smallest possible size.

Service Level Options

services.AddGrpc()
    .AddServiceOptions<WeatherService>(o =>
        {
            o.ResponseCompressionLevel = CompressionLevel.Optimal;
            o.ResponseCompressionAlgorithm = "gzip";
        });

After calling AddGrpc, an IGrpcServerBuilder is returned. We can call an extension method on that builder called AddServiceOptions to provide per service options. This method is generic and accepts the type of the gRPC service that the options should apply.

In the preceding example, we have decided to provide options specifically for calls that are handled by the WeatherService implementation. The same options are available at this level as we discussed for the server level configuration. In this scenario, if we mapped other gRPC services within this server, they would not receive the compression options.

Making Requests from A GRPC Client

Now that response compression is enabled, we need to ensure our requests state that our client accepts compressed content. In fact, this is enabled by default when using a GrpcChannel created using the ForAddress method so we have nothing to do in our client code.

var channel = GrpcChannel.ForAddress("https://localhost:5005");

Channels created in this way already send a “grpc-accept-encoding” header which includes the gzip compression type. The server reads this header and determines that the client allows gzipped responses to be returned.

One way to visualise the effect of compression is to enable trace level logging for our application while in development. We can achieve this by modifying the appsettings.Development.json file as follows:

{
  "Logging": {
    "LogLevel": {
        "Default": "Debug",
        "System": "Information",
        "Grpc": "Trace",
        "Microsoft": "Trace"
    }
  }
}

When running our server, we now get much more verbose console logging.

info: Microsoft.AspNetCore.Routing.EndpointMiddleware[0]
      Executing endpoint 'gRPC - /WeatherForecast.WeatherForecasts/GetWeather'
dbug: Grpc.AspNetCore.Server.ServerCallHandler[1]
      Reading message.
dbug: Microsoft.AspNetCore.Server.Kestrel[25]
      Connection id "0HLQB6EMBPUIA", Request id "0HLQB6EMBPUIA:00000001": started reading request body.
dbug: Microsoft.AspNetCore.Server.Kestrel[26]
      Connection id "0HLQB6EMBPUIA", Request id "0HLQB6EMBPUIA:00000001": done reading request body.
trce: Grpc.AspNetCore.Server.ServerCallHandler[3]
      Deserializing 0 byte message to 'Google.Protobuf.WellKnownTypes.Empty'.
trce: Grpc.AspNetCore.Server.ServerCallHandler[4]
      Received message.
dbug: Grpc.AspNetCore.Server.ServerCallHandler[6]
      Sending message.
trce: Grpc.AspNetCore.Server.ServerCallHandler[9]
      Serialized 'WeatherForecast.WeatherReply' to 2851 byte message.
trce: Microsoft.AspNetCore.Server.Kestrel[37]
      Connection id "0HLQB6EMBPUIA" sending HEADERS frame for stream ID 1 with length 104 and flags END_HEADERS
trce: Grpc.AspNetCore.Server.ServerCallHandler[10]
      Compressing message with 'gzip' encoding.
trce: Grpc.AspNetCore.Server.ServerCallHandler[7]
      Message sent.
info: Microsoft.AspNetCore.Routing.EndpointMiddleware[1]
      Executed endpoint 'gRPC - /WeatherForecast.WeatherForecasts/GetWeather'
trce: Microsoft.AspNetCore.Server.Kestrel[37]
      Connection id "0HLQB6EMBPUIA" sending DATA frame for stream ID 1 with length 978 and flags NONE
trce: Microsoft.AspNetCore.Server.Kestrel[37]
      Connection id "0HLQB6EMBPUIA" sending HEADERS frame for stream ID 1 with length 15 and flags END_STREAM, END_HEADERS
info: Microsoft.AspNetCore.Hosting.Diagnostics[2]
      Request finished in 2158.9035ms 200 application/grpc

On line 16 of this log, we can see that the WeatherReply, essentially an array of 100 WeatherData items in this sample, has been serialised to protocol buffers and has a size of 2851 bytes.

Later, in line 20, we can see that the message has been compressed with gzip encoding and on line 26, we can see the size of the data frame for this call which is 978 bytes. The data, in this case, has compressed quite well (66% reduction) because the repeated WeatherData items contain text and many of the values repeat within the message.

In this example, gzip compression has a good effect on the over the wire size of the data.

Disable Response Compression within A Service Method Implementation

It’s possible to control the response compression on a per-method basis. At this time, I’ve only found a way to do this on an opt-out approach. When compression is enabled for a service or server, we can opt-out of compression within the service method implementation.

Let’s look at the server log output when calling a service method which streams WeatherData messages from the server.

info: WeatherForecast.Grpc.Server.Services.WeatherService[0]
      Sending WeatherData response
dbug: Grpc.AspNetCore.Server.ServerCallHandler[6]
      Sending message.
trce: Grpc.AspNetCore.Server.ServerCallHandler[9]
      Serialized 'WeatherForecast.WeatherData' to 30 byte message.
trce: Grpc.AspNetCore.Server.ServerCallHandler[10]
      Compressing message with 'gzip' encoding.
trce: Microsoft.AspNetCore.Server.Kestrel[37]
      Connection id "0HLQBMRRH10JQ" sending DATA frame for stream ID 1 with length 50 and flags NONE
trce: Grpc.AspNetCore.Server.ServerCallHandler[7]
      Message sent.

On line 6, we can see that an individual WeatherData message is 30 bytes in size. On line 8, this gets compressed, and on line 10, we can see that the data length is now 50 bytes, larger than the original message. In this case, there is no gain from gzip compression, and we see an increase in the overall message size sent over the wire.

We can avoid compression for a particular message by setting the WriteOptions for the call within the service method.

public override async Task GetWeatherStream(Empty _, IServerStreamWriter<WeatherData> responseStream, ServerCallContext context)
{
    context.WriteOptions = new WriteOptions(WriteFlags.NoCompress);

    // implementation of the method which writes to the stream
}

At the top of our service method, we can set the WriteOptions on the ServerCallContext. We pass in a new WriteOptions instance which the WriteFlags value set to NoCompress. These write options are used for the next write.

With streaming responses, it’s also possible to set this value on the IServerStreamWriter.

public override async Task GetWeatherStream(Empty _, IServerStreamWriter<WeatherData> responseStream, ServerCallContext context)
{   
    responseStream.WriteOptions = new WriteOptions(WriteFlags.NoCompress);

    // implementation of the method which writes to the stream
}

When this option is applied, the logs now show that compression for calls to this service method is not applied.

info: WeatherForecast.Grpc.Server.Services.WeatherService[0]
      Sending WeatherData response
dbug: Grpc.AspNetCore.Server.ServerCallHandler[6]
      Sending message.
trce: Grpc.AspNetCore.Server.ServerCallHandler[9]
      Serialized 'WeatherForecast.WeatherData' to 30 byte message.
trce: Microsoft.AspNetCore.Server.Kestrel[37]
      Connection id "0HLQBMTL1HLM8" sending DATA frame for stream ID 1 with length 35 and flags NONE
trce: Grpc.AspNetCore.Server.ServerCallHandler[7]
      Message sent.

Now the 30 byte message has a length of 35 bytes in the DATA frame. There is a small overhead which accounts for the extra 5 bytes which we don’t need to concern ourselves with here.

Disable Response Compression from A GRPC Client

By default, the gRPC channel includes options that control which encodings it accepts. It is possible to configure these when creating the channel if you wish to disable compression of responses from your client. Generally, I would avoid this and let the server decide what to do, since it knows best what can and cannot be compressed. That said, you may sometimes need to control this from the client.

The only way I’ve found to do this in my exploration of the API to date is to configure the channel by passing in a GrpcChannelOptions instance. One of the properties on this options class is for the CompressionProviders, an IList<ICompressionProvider>. By default, when this is null, the client implementation adds the Gzip compression provider for you automatically. This means that the server can choose to gzip the response message(s) as we have already seen.

private static async Task Main()
{
    using var channel = GrpcChannel.ForAddress("https://localhost:5005", new GrpcChannelOptions { CompressionProviders = new List<ICompressionProvider>() });

    var client = new WeatherForecastsClient(channel);

    var reply = await client.GetWeatherAsync(new Empty());

    foreach (var forecast in reply.WeatherData)
    {
        Console.WriteLine($"{forecast.DateTimeStamp.ToDateTime():s} | {forecast.Summary} | {forecast.TemperatureC} C");
    }

    Console.WriteLine("Press a key to exit");
    Console.ReadKey();
}

In this sample client code, we establish the GrpcChannel and pass in a new instance of GrpcChannelOptions. We set the CompressionProviders property to an empty list. Since we now specify no providers in our channel, when the calls are created and sent via this channel, they won’t include any compression encodings in the “grpc-accept-encoding” header. The server acknowledges this and not apply gzip compression to the response.

Summary

In this post, we’ve explored the possibility of compressing response messages from a gRPC server. We’ve identified that in some cases, but crucially not all, this may result in smaller payloads. We’ve seen that by default clients calls include the gzip “grpc-accept-encoding” value in the headers. If the server is configured to apply compression, it only does so if a supported encoding type is matched from the request header.

We can configure the GrpcChannelOptions when creating a channel for the client, to disable the inclusion of the gzip compression encoding. On the server, we can configure the whole server, or a specific service to enable compression for responses. We can override and disable that on a per service-method level as well.

With the introduction of ASP.NET Core 3.0 the default JSON serializer has been changed from Newtonsoft.Json to System.Text.Json. For projects and libraries switching to the new JSON serializer this change means more performance and the opportunity to rewrite our JsonConverters.

Serialization of concrete classes

Let's start with a simple one that can (de)serialize a concrete class Category. In our example we (de)serialize the property Name only.

public class Category
{
   public string Name { get; }

   public Category(string name)
   {
      Name = name;
   }
}

To implement a custom JSON converter we have to derive from the generic class JsonConverter<T> and to implement 2 methods: Read and Write.

public class CategoryJsonConverter : JsonConverter<Category>
{
   public override Category Read(ref Utf8JsonReader reader,
                                 Type typeToConvert,
                                 JsonSerializerOptions options)
   {
      var name = reader.GetString();

      return new Category(name);
   }

   public override void Write(Utf8JsonWriter writer,
                              Category value,
                              JsonSerializerOptions options)
   {
      writer.WriteStringValue(value.Name);
   }
}

The method Read is using the Utf8JsonReader to fetch a string, i.e. the name, and the method Write is writing a string using an instance of Utf8JsonWriter.

In both cases (i.e. during serialization and deserialization) the converter is not being called if the value is null so I skipped the null checks. The .NET team doesn't do null checks either, see JsonKeyValuePairConverter<TKey, TValue>.

Let's test the new JSON converter. For that we create an instance of JsonSerializerOptions and add our CategoryJsonConverter to the Converters collection. Next, we use the static class JsonSerializer to serialize and to deserialize an instance of Category.

Category category = new Category("my category");

var serializerOptions = new JsonSerializerOptions
{
    Converters = { new CategoryJsonConverter() }
};

// json = "my category"
var json = JsonSerializer.Serialize(category, serializerOptions);

// deserializedCategory.Name = "my category"
var deserializedCategory = JsonSerializer.Deserialize<Category>(json, serializerOptions);

Serialization of generic classes

The next example is slightly more complex. The property we are serializing is a generic type argument, i.e. we can't use methods like reader.GetString() or writer.WriteStringValue(name) because we don't know the type at compile time.

In this example I've changed the class Category to a generic type and renamed the property Name to Key:

public class Category<T>
{
   public T Key { get; }

   public Category(T key)
   {
      Key = key;
   }
}

For serialization of the generic property Key we need to fetch a JsonSerializer<T> using the instance of JsonSerializerOptions.

public class CategoryJsonConverter<T> : JsonConverter<Category<T>>
{
   public override Category<T> Read(ref Utf8JsonReader reader,
                                    Type typeToConvert,
                                    JsonSerializerOptions options)
   {
      var converter = GetKeyConverter(options);
      var key = converter.Read(ref reader, typeToConvert, options);

      return new Category<T>(key);
   }

   public override void Write(Utf8JsonWriter writer,
                              Category<T> value,
                              JsonSerializerOptions options)
   {
      var converter = GetKeyConverter(options);
      converter.Write(writer, value.Key, options);
   }

   private static JsonConverter<T> GetKeyConverter(JsonSerializerOptions options)
   {
      var converter = options.GetConverter(typeof(T)) as JsonConverter<T>;

      if (converter is null)
         throw new JsonException("...");

      return converter;
   }
}

The behavior of the generic JSON converter is the same as before especially if the Key is of type string.

Deciding the concrete JSON converter at runtime

Having several categories with different key types, say, string and int, we need to register them all with the JsonSerializerOptions.

var serializerOptions = new JsonSerializerOptions
                        {
                           Converters =
                           {
                              new CategoryJsonConverter<string>(),
                              new CategoryJsonConverter<int>()
                           }
                        };

If the number of required CategoryJsonConverters grows to big or the concrete types of the Key are not known at compile time then this approach is not an option. To make this decision at runtime we need to implement a JsonConverterFactory. The factory has 2 method: CanConvert(type) that returns true if the factory is responsible for the serialization of the provided type; and CreateConverter(type, options) that should return an instance of type JsonConverter.

public class CategoryJsonConverterFactory : JsonConverterFactory
{
   public override bool CanConvert(Type typeToConvert)
   {
      if (!typeToConvert.IsGenericType)
         return false;

      var type = typeToConvert;

      if (!type.IsGenericTypeDefinition)
         type = type.GetGenericTypeDefinition();

      return type == typeof(Category<>);
   }

   public override JsonConverter CreateConverter(Type typeToConvert,
                                                 JsonSerializerOptions options)
   {
      var keyType = typeToConvert.GenericTypeArguments[0];
      var converterType = typeof(CategoryJsonConverter<>).MakeGenericType(keyType);

      return (JsonConverter)Activator.CreateInstance(converterType);
   }
}

Now, we can remove all registrations of the CategoryJsonConverter<T> from the options and add the newly implemented factory.

Category<int> category = new Category<int>(42);

var serializerOptions = new JsonSerializerOptions
{
    Converters = { new CategoryJsonConverterFactory() }
};

// json = 42
var json = JsonSerializer.Serialize(category, serializerOptions);

// deserialized.Key = 42
var deserialized = JsonSerializer.Deserialize<Category<int>>(json, serializerOptions);

In the end the implementation of a custom converter for System.Text.Json is very similar to the one for Newtonsoft.Json. The biggest difference here is the non-existence of a non-generic JsonConverter but for that we've got the JsonConverterFactory.

Actually, there is a non-generic JsonConverter which is the base class of the JsonConverter<T> and the JsonConverterFactory but we cannot (and should not) use this class directly because its constructor is internal.