OpenTelemetry Jaeger Exporter

Warnings

• gotcha The default `JaegerExporter` sends spans via UDP to a Jaeger Agent, typically on `localhost:6831`. If you need to send spans over HTTP to a Jaeger Collector, you must explicitly use `JaegerHttpExporter` and configure its endpoint.
  Fix: For UDP, ensure a Jaeger Agent is running and accessible at the configured `agent_host_name` and `agent_port`. For HTTP, import `JaegerHttpExporter` and pass `collector_endpoint` to its constructor, e.g., `JaegerHttpExporter(collector_endpoint='http://localhost:14268/api/traces')`.
• gotcha Providing meaningful resource attributes, especially `SERVICE_NAME`, is critical for effectively identifying and filtering your services and applications within the Jaeger UI. Without it, all spans might appear under a generic name.
  Fix: Always initialize `TracerProvider` with a `Resource` that includes `SERVICE_NAME`. Example: `Resource.create({SERVICE_NAME: 'my-application'})`.
• breaking The OpenTelemetry Python SDK, which this exporter relies on, frequently updates its API and internal behavior across major/minor versions. For example, `v1.40.0` included a fix for `NoOpTracer` behavior for `start_span` and `start_as_current_span`, which could subtly affect applications not using a full SDK setup.
  Fix: Always refer to the official OpenTelemetry Python changelog (`opentelemetry-python` repository releases) for specific breaking changes in the SDK or API when upgrading. Ensure your `opentelemetry-sdk` and `opentelemetry-api` packages are compatible with your exporter version.
• gotcha Exporters typically use environment variables (e.g., `OTEL_EXPORTER_JAEGER_AGENT_HOST`, `OTEL_EXPORTER_JAEGER_AGENT_PORT`, `OTEL_EXPORTER_JAEGER_ENDPOINT`) for configuration when using auto-instrumentation or `configure_opentelemetry()`. Programmatic configuration, as shown in the quickstart, will override these.
  Fix: Be mindful of your configuration source. For consistent behavior, choose either environment variables (often for simpler deployments/auto-instrumentation) or programmatic setup, but avoid mixing them in ways that lead to unexpected overrides. Ensure explicit programmatic settings align with your intent.

Install

• pip install opentelemetry-exporter-jaeger opentelemetry-sdk Install with SDK

Imports

• JaegerExporter

  from opentelemetry.exporter.jaeger.thrift import JaegerExporter

  Used for sending spans to a Jaeger Agent via UDP (default). This is the classic Jaeger exporter.
• JaegerHttpExporter

  from opentelemetry.exporter.jaeger.thrift import JaegerHttpExporter

  Used for sending spans to a Jaeger Collector via HTTP (Thrift over HTTP).
• TracerProvider

  from opentelemetry.sdk.trace import TracerProvider

• BatchSpanProcessor

  from opentelemetry.sdk.trace.export import BatchSpanProcessor

• Resource

  from opentelemetry.sdk.resources import Resource

• SERVICE_NAME

  from opentelemetry.sdk.resources import SERVICE_NAME

Quickstart

This quickstart demonstrates how to set up the OpenTelemetry Python SDK with the Jaeger Exporter. It initializes a `TracerProvider` with a `BatchSpanProcessor` and configures the `JaegerExporter` to send traces to a local Jaeger agent via UDP. Resource attributes, particularly `SERVICE_NAME`, are crucial for identifying your service in Jaeger. Environment variables are used for agent host/port to simplify configuration.

import os
from opentelemetry import trace
from opentelemetry.sdk.resources import Resource, SERVICE_NAME
from opentelemetry.sdk.trace import TracerProvider
from opentelemetry.sdk.trace.export import BatchSpanProcessor
from opentelemetry.exporter.jaeger.thrift import JaegerExporter

# Configure resource for your service
resource = Resource.create({
    SERVICE_NAME: os.environ.get('OTEL_SERVICE_NAME', 'my-jaeger-service'),
})

# Configure TracerProvider
provider = TracerProvider(resource=resource)

# Configure Jaeger Exporter (UDP to local agent by default)
# Use environment variables for agent host/port or pass them directly:
# JAEGER_AGENT_HOST, JAEGER_AGENT_PORT
# For HTTP Exporter, use JaegerHttpExporter and JAEGER_COLLECTOR_ENDPOINT
jaeger_exporter = JaegerExporter(
    agent_host_name=os.environ.get('JAEGER_AGENT_HOST', 'localhost'),
    agent_port=int(os.environ.get('JAEGER_AGENT_PORT', 6831)),
)

# Add the exporter to a BatchSpanProcessor
span_processor = BatchSpanProcessor(jaeger_exporter)
provider.add_span_processor(span_processor)

# Set the global tracer provider
trace.set_tracer_provider(provider)

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

with tracer.start_as_current_span('my-operation') as span:
    span.set_attribute('event', 'started')
    print(f"Hello from span: {span.context.trace_id}")
    with tracer.start_as_current_span('inner-operation'):
        print("Inside inner operation")
    span.set_attribute('event', 'finished')

# Ensure all spans are exported before exiting
provider.force_flush()
print("Traces exported to Jaeger.")