OpenTelemetryDemo

Evaluation Data to OpenTelemetry

This sample shows how to send feature flag evaluation data to an OpenTelemetry-compatible backend using Microsoft.FeatureManagement.Telemetry.OpenTelemetry, exported to Azure Monitor via the Azure Monitor OpenTelemetry Distro (Azure.Monitor.OpenTelemetry.AspNetCore). Evaluation data is emitted each time a feature or variant is evaluated with telemetry enabled.

This package is published alongside Microsoft.FeatureManagement.Telemetry.ApplicationInsights (see the VariantAndTelemetryDemo sample) so that applications not yet ready to move off the Application Insights SDK aren't forced to. New applications are encouraged to use Microsoft.FeatureManagement.Telemetry.OpenTelemetry, the industry-standard, vendor-neutral observability framework.

Quickstart

To run this sample, follow these steps:

1. Set up a new Application Insights resource in Azure, and from Overview copy the Connection String.
2. Place the connection string in appsettings.json at APPLICATIONINSIGHTS_CONNECTION_STRING (or set it as an environment variable of the same name instead).
3. Set the project as the startup project (or cd into this directory).
4. Run the project (dotnet run) and browse to the app.
5. Head to the Application Insights resource in the Azure Portal to see the emitted telemetry (logs, traces, and metrics).

A connection string is required for telemetry to be exported. Without one, the app still runs and requests succeed, but UseAzureMonitor() is skipped and nothing is exported.

About the App

This app uses Microsoft.FeatureManagement.Telemetry.OpenTelemetry alongside the OpenTelemetry .NET SDK and the Azure Monitor OpenTelemetry Distro to export logs, traces, and metrics. See Program.cs for how tracing/logging/metrics and Azure Monitor are wired up via UseAzureMonitor().

Registration

builder.Services.AddFeatureManagement()
    .WithTargeting();

var telemetry = builder.Services.AddOpenTelemetry();

telemetry.AddFeatureManagementProcessors();

if (!string.IsNullOrEmpty(connectionString))
{
    telemetry.UseAzureMonitor(o => o.ConnectionString = connectionString);
}

Register targeting processors before exporters: call AddFeatureManagementProcessors() before UseAzureMonitor(). This ensures TargetingId is added to spans and logs before they are exported.

Targeting Id

In order to connect evaluation events with other telemetry from the user, a targeting id needs to be emitted. This sample uses the provided TargetingHttpContextMiddleware, which reads the targeting context and adds TargetingId to both the HttpContext and the current Activity's baggage as a request comes in. TargetingActivityProcessor and TargetingLogProcessor (automatically registered by AddFeatureManagementProcessors()) then read that baggage to enrich spans and logs respectively.

Sample App Usage

Sample steps to try out the app:

1. Run the app. When the app is first started a user id will be generated and stored in an authentication cookie (see RandomizeUser).
2. When the page is loaded, the ImageRating feature is evaluated, which defines three variants, emitting a FeatureEvaluation custom event and trace.
3. Select a rating for the loaded image and click vote. A Vote custom event and an ImageRating metric will be emitted.
4. Go to Checkout and click "Check Out", which emits a checkout custom event and a checkoutAmount metric.
5. Head to the Application Insights resource in the Azure Portal.
   1. Try going to Logs > New Query and run the query customEvents. This should show the custom events emitted, each carrying a TargetingId property.
   2. Try going to Metrics. Under Metric find Custom > ImageRating and checkoutAmount.
   3. From the Metrics window, out-of-the-box metrics like Page Views and Server Requests can be viewed thanks to the ASP.NET Core request instrumentation UseAzureMonitor() adds automatically.