ExceptionHandler middleware mutes http_server_request_duration_seconds_count metrics on exceptions

[davidroth]

Is there an existing issue for this?

• I have searched the existing issues

Describe the bug

Consider the following .NET 8 minimal api application with OTEL metrics enabled.

builder.Services.AddProblemDetails();

builder.Services
        .AddOpenTelemetry()
        .WithMetrics(metrics =>
        {
            metrics.AddAspNetCoreInstrumentation();
            metrics.AddPrometheusExporter();
        })
....

app.MapGet("/crash", () =>
{
    throw new InvalidOperationException("An error occured");
})

When I run the application in Developer mode, I see that the http_server_request_duration_seconds_count metrics is emmited for the failed request and also enriched because of this line of code: DiagnosticsTelemetry.ReportUnhandledException(_logger, context, ex);.

This is the enriched metric:

http_server_request_duration_seconds_count{error_type="System.InvalidOperationException",http_request_method="GET",http_response_status_code="500",http_route="/crash",network_protocol_version="2",url_scheme="https"} 2 1702390457828

So http_server_request_duration_seconds_count is increased for every invocation on the "/crash" route and also enriched with the error_type.

Now, as soon as I start the app with the "Production" environment, where the following line kicks in:

if (!app.Environment.IsDevelopment())
{
    app.UseExceptionHandler(new ExceptionHandlerOptions() { });
 }

the http_server_request_duration_seconds_count metric is now longer emitted for the failed requests.

Expected Behavior

I dont know why ExceptionHandlerMiddlewareImpl "mutes" the http_server_request_duration_seconds_count metric for endpoints with exceptions. Maybe I'm missing something and it makes sense, but for me the behaviour doesn't make sense.

Is this possibly a bug or as designed?

Steps To Reproduce

Run the following app:

using OpenTelemetry.Metrics;

var builder = WebApplication.CreateBuilder(args);

builder.Services.AddEndpointsApiExplorer();
builder.Services.AddSwaggerGen();

builder.Services
        .AddOpenTelemetry()
        .WithMetrics(metrics =>
        {
            metrics.AddAspNetCoreInstrumentation();
            metrics.AddPrometheusExporter();
        });

builder.Services.AddProblemDetails();

var app = builder.Build();

app.UseSwagger();
app.UseSwaggerUI();
app.UseExceptionHandler(new ExceptionHandlerOptions() { });

app.UseHttpsRedirection();
app.MapPrometheusScrapingEndpoint();

app.MapGet("/crash", () =>
{
    if (true)
    {
        throw new InvalidOperationException("XY");
    }
    return "OK";
})
.WithName("Crash")
.WithOpenApi();

app.Run();

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

  <PropertyGroup>
    <TargetFramework>net8.0</TargetFramework>
    <Nullable>enable</Nullable>
    <ImplicitUsings>enable</ImplicitUsings>
    <InvariantGlobalization>true</InvariantGlobalization>
  </PropertyGroup>

  <ItemGroup>
    <PackageReference Include="Microsoft.AspNetCore.OpenApi" Version="8.0.0" />
    <PackageReference Include="OpenTelemetry.Exporter.Prometheus.AspNetCore" Version="1.7.0-rc.1" />
    <PackageReference Include="OpenTelemetry.Extensions.Hosting" Version="1.7.0" />
    <PackageReference Include="OpenTelemetry.Instrumentation.AspNetCore" Version="1.6.0-rc.1" />
    <PackageReference Include="Swashbuckle.AspNetCore" Version="6.4.0" />
  </ItemGroup>

</Project>

1. Invoke the "/crash" endpoint multiple times
2. Open /metrics endpoint and search for http_server_request_duration_seconds_count metric

.NET Version

.NET 8

[BrennanConroy]

Looks like a dupe of #52648

[davidroth]

Duplicate of #52648