Azure Resource Detector for OpenTelemetry

Warnings

• gotcha The Azure Resource Detector requires appropriate Azure credentials and permissions (e.g., Contributor role on the resource group or specific Reader permissions) to query Azure Resource Manager APIs. Without correct authentication (e.g., Managed Identity or `AZURE_CLIENT_ID`, `AZURE_TENANT_ID`, `AZURE_CLIENT_SECRET` environment variables), the detector will fail to gather metadata, resulting in missing or incomplete resource attributes.
  Fix: Ensure the application's identity has the necessary permissions to read Azure resource metadata. For local development, set `AZURE_CLIENT_ID`, `AZURE_TENANT_ID`, and `AZURE_CLIENT_SECRET` environment variables. For Azure deployments, prefer Managed Identities.
• gotcha Resource attributes are collected at SDK initialization time and are immutable throughout the application's lifecycle. Any changes to the Azure environment or the running resource's metadata after the application starts will not be reflected in the telemetry emitted by the current application instance.
  Fix: Understand that resource attributes are static for a given application run. If dynamic changes are needed, consider custom attributes on spans/metrics or restarting the application to re-initialize the SDK and re-detect resources.
• gotcha The Azure Resource Detector makes network calls to Azure APIs during application startup to fetch resource metadata. This can introduce latency during the initialization phase of your application. Ensure robust network connectivity and consider the performance implications in latency-sensitive applications.
  Fix: Monitor application startup times. While usually negligible, in environments with strict startup time requirements or slow network egress, this could be a factor. The detector typically caches some results, but the initial fetch is network-dependent.

Install

• pip install opentelemetry-sdk opentelemetry-resource-detector-azure Install core SDK and Azure detector

Imports

• AzureResourceDetector

  from opentelemetry.sdk.extension.resource.azure import AzureResourceDetector

• Resource

  from opentelemetry.sdk.resources import Resource

• TracerProvider

  from opentelemetry.sdk.trace import TracerProvider

• ConsoleSpanExporter

  from opentelemetry.sdk.trace.export import ConsoleSpanExporter

• SimpleSpanProcessor

  from opentelemetry.sdk.trace.export import SimpleSpanProcessor

Quickstart

This quickstart demonstrates how to initialize the `AzureResourceDetector` and integrate it with the OpenTelemetry Python SDK. The detector automatically enriches telemetry with Azure-specific attributes. It requires Azure credentials (via environment variables or Managed Identity) to function correctly. The output on the console will include the automatically detected Azure resource attributes.

import os
from opentelemetry import trace
from opentelemetry.sdk.resources import Resource
from opentelemetry.sdk.trace import TracerProvider
from opentelemetry.sdk.trace.export import ConsoleSpanExporter, SimpleSpanProcessor
from opentelemetry.sdk.extension.resource.azure import AzureResourceDetector

# Configure Azure credentials via environment variables for testing purposes
# In a real Azure environment, Managed Identity is often preferred.
# For local testing, ensure these are set:
# os.environ['AZURE_CLIENT_ID'] = os.environ.get('AZURE_CLIENT_ID', '')
# os.environ['AZURE_TENANT_ID'] = os.environ.get('AZURE_TENANT_ID', '')
# os.environ['AZURE_CLIENT_SECRET'] = os.environ.get('AZURE_CLIENT_SECRET', '')

# Instantiate and register the Azure Resource Detector
# The detector makes network calls to Azure APIs, potentially adding startup latency.
# Resource attributes are collected at SDK initialization and are immutable.
resource = Resource.create([AzureResourceDetector()])

# Set up a TracerProvider with the detected resource
provider = TracerProvider(resource=resource)
processor = SimpleSpanProcessor(ConsoleSpanExporter())
provider.add_span_processor(processor)
trace.set_tracer_provider(provider)

# Get a tracer and create a simple span
tracer = trace.get_tracer(__name__)

with tracer.start_as_current_span("azure-resource-test-span") as span:
    span.set_attribute("my.custom.attribute", "hello")
    print("Span created. Check console for resource attributes from Azure.")

# Ensure all spans are processed and exported on shutdown
provider.shutdown()