OpenTelemetry
Applications
.NET

Set up .NET applications

Export ASP.NET Core application logs, traces and metrics to Hosted OpenTelemetry. For the endpoint, credentials and viewing requirements, see Application connection details.

Install packages

For an app that does not yet export OTLP:

dotnet add package OpenTelemetry.Extensions.Hosting
dotnet add package OpenTelemetry.Exporter.OpenTelemetryProtocol
dotnet add package OpenTelemetry.Instrumentation.AspNetCore
dotnet add package OpenTelemetry.Instrumentation.Http

Configure the application

Merge these keys into appsettings.json, replacing the stack ID and service name:

{
  "OTEL_SERVICE_NAME": "my-app",
  "OTEL_EXPORTER_OTLP_ENDPOINT": "https://{stackId}-otel.logit.io",
  "OTEL_EXPORTER_OTLP_PROTOCOL": "http/protobuf"
}

Build authentication and set the standard SDK header in the app's process environment. Alternatively, put the dashboard's whole Header value into OTEL_EXPORTER_OTLP_HEADERS using local user secrets. Keep credentials out of committed settings files.

ASP.NET Core reads appsettings.json and then appsettings.{Environment}.json. Environment variables override those settings. A console app without configuration binding must use process environment variables for the service name, endpoint, protocol and header.

Register OpenTelemetry

Add the namespaces and registration before builder.Build() in your ASP.NET Core startup code:

using OpenTelemetry;
using OpenTelemetry.Logs;
using OpenTelemetry.Metrics;
using OpenTelemetry.Trace;
 
// After creating the application's builder:
builder.Services.AddOpenTelemetry()
    .WithTracing(tracing => tracing
        .AddAspNetCoreInstrumentation()
        .AddHttpClientInstrumentation())
    .WithMetrics(metrics => metrics
        .AddAspNetCoreInstrumentation()
        .AddHttpClientInstrumentation())
    .UseOtlpExporter();
 
builder.Logging.AddOpenTelemetry(logging =>
{
    logging.IncludeFormattedMessage = true;
    logging.IncludeScopes = true;
});

UseOtlpExporter() enables export for logs, metrics and traces and reads the shared configuration keys above. With the base endpoint in OTEL_EXPORTER_OTLP_ENDPOINT, it appends each signal's /v1/ path. Metrics export roughly every 60 seconds.

If OpenTelemetry is already registered

Keep the existing registration and service name. Calling UseOtlpExporter() twice, or combining it with AddOtlpExporter(), throws an error.

  • With an existing UseOtlpExporter(), update its configuration. If it uses the protocol and base URL overload, use HttpProtobuf and the Hosted base URL.
  • With existing AddOtlpExporter() callbacks, remove a shared assignment to Endpoint and use the shared configuration keys above. If a callback must set an endpoint explicitly, use the full URL ending in /v1/logs, /v1/traces or /v1/metrics for that signal and set Protocol to HttpProtobuf.
  • If AddOtlpExporter() is already used for traces and metrics, add a logging exporter only if one is missing. Do not add it when UseOtlpExporter() is registered.
  • For Serilog or NLog, keep your logger and configure its OTLP sink for the full /v1/logs URL. A sink's header map uses Authorization with the value Basic followed by the Base64 token, with a real space.
  • If the app already sends to a Collector, update that Collector's HTTP exporter using the Hosted OpenTelemetry Collector example.

Setting separate username or API key settings does not configure the exporter automatically. The exporter needs the complete Basic auth header.

Zero-code alternative

The .NET auto-instrumentation profiler is an alternative to the SDK registration above. On Linux or macOS, install it with otel-dotnet-auto-install.sh and source instrument.sh. On Windows, use PowerShell 5.1 with OpenTelemetry.DotNet.Auto.psm1, run Install-OpenTelemetryCore, then register it for the current session, Windows service or IIS as appropriate.

The profiler reads process environment variables, including the endpoint, http/protobuf protocol and header; it does not read appsettings.json. Use the installation instructions offered by dashboard onboarding for this alternative.

Restart the app, generate an application log and a request, then check the result.

ApplicationsPython