Connection to Azure Application Insights with OpenTelemetry

[Manonich]

Question
How must the OpenTelemetry exporter be configured so that the traces will be sent to Azure Monitor?
Is it possible to add an example for the use case or explain how it can be realized with the available functionality.

Problem
There is no example in the documention or in the code. Other sdk's for azure provide a method UseAzureMonitor() and they work with the environment variable APPLICATIONINSIGHTS_CONNECTION_STRING.
https://learn.microsoft.com/en-us/azure/azure-monitor/app/opentelemetry-enable?tabs=aspnetcore

Situation
Connection string with following structure is known:
"InstrumentationKey=my_key;IngestionEndpoint=my_ingestion_endpoint;LiveEndpoint=my_live_endpoint;ApplicationId=my_app_id"

Following code don't work:

     opentelemetry::sdk::resource::ResourceAttributes attributes = {
          {"service.name", serviceName},
          {"service.version", serviceVersion}
     };
     
     auto resource = opentelemetry::sdk::resource::Resource::Create(attributes);

     opentelemetry::exporter::otlp::OtlpHttpExporterOptions options;
     options.url = "my_live_endpoint";
     options.http_headers.insert(std::make_pair("InstrumentationKey", "my_key"));
     options.http_headers.insert(std::make_pair("ApplicationId", "my_app_id"));

     // Create OTLP HTTP exporter
     auto exporter = opentelemetry::exporter::otlp::OtlpHttpExporterFactory::Create(options);

     // Create test processor
     auto processor = opentelemetry::sdk::trace::SimpleSpanProcessorFactory::Create(std::move(exporter));

    // Create tracer provider
    std::shared_ptr<opentelemetry::trace::TracerProvider> tracerProviderOpenTelemetry = opentelemetry::sdk::trace::TracerProviderFactory::Create(std::move(processor), resource);

     // Create Azure tracer provider
     auto tracerProviderAzure = Azure::Core::Tracing::OpenTelemetry::OpenTelemetryProvider::Create(tracerProviderOpenTelemetry);

    // Create Azure Tracer
    auto tracerAzure = Azure::Core::Tracing::_internal::TracerProviderImplGetter::TracerImplFromTracer(tracerProviderAzure)->CreateTracer(serviceName, serviceVersion);
    auto span = tracerAzure->CreateSpan("AZURE_SPAN", {});
    span->AddAttribute("test.suite", "Hello World!");
    span->End({});

[LarryOsterman]

I implemented similar functionality in the EventHubs stress test:

azure-sdk-for-cpp/sdk/eventhubs/azure-messaging-eventhubs/test/eventhubs-stress-test/src/eventhubs_stress_test.cpp

Lines 89 to 119 in 738d142

	// Create OTLP exporter instance
	auto exporter = otlp::OtlpHttpExporterFactory::Create(opts);
	auto processor = trace_sdk::SimpleSpanProcessorFactory::Create(std::move(exporter));
	auto resource{GetTraceResource(stressScenarioName)};
	#if defined(_MSC_VER)
	#pragma warning(push)
	#pragma warning(disable : 4996)
	#elif defined(__clang__)
	#pragma clang diagnostic push
	#pragma clang diagnostic ignored "-Wdeprecated-declarations"
	#elif defined(__GNUC__)
	#pragma GCC diagnostic push
	#pragma GCC diagnostic ignored "-Wdeprecated-declarations"
	#endif
	// https://github.com/Azure/azure-sdk-for-cpp/issues/5784
	std::shared_ptr<opentelemetry::trace::TracerProvider> provider
	= trace_sdk::TracerProviderFactory::Create(std::move(processor), std::move(resource));
	#ifdef _MSC_VER
	#pragma warning(pop)
	#elif defined(__clang__)
	#pragma clang diagnostic pop
	#elif defined(__GNUC__)
	#pragma GCC diagnostic pop
	#endif
	// Set the global trace provider
	trace::Provider::SetTracerProvider(provider);
	}

We then used the azure monitor exporter to export to Azure Monitor.

Does this help?

[Manonich]

Thanks for the tip. I'll test the collector variant and give feedback.

[LarryOsterman]

Did you copy the connection string from your monitor instance for use in the exporter?

[Manonich]

I work with the connection string.

I've also tested the approach with the opentelemetry collector. Unfortunately, that doesn't work either. The data is displayed in the debug exporter, but nothing appears in Application Insights. I've run the collector both in the container and directly on the computer.

receivers:
  otlp:
    protocols:
      http:
        endpoint: 0.0.0.0:4318
      grpc:
        endpoint: 0.0.0.0:4317

exporters:
  azuremonitor:
    connection_string: "InstrumentationKey=my_key;IngestionEndpoint=my_url"
  debug:
    verbosity: detailed

service:
  pipelines:
    traces:
      receivers: [otlp]
      exporters: [azuremonitor, debug]
    metrics:
      receivers: [otlp]
      exporters: [azuremonitor, debug]
    logs:
      receivers: [otlp]
      exporters: [azuremonitor, debug]