​
Installation

To integrate OpenTelemetry (OTEL) tracing with Galileo you need the following dependencies:

pip install opentelemetry-api opentelemetry-sdk opentelemetry-exporter-otlp opentelemetry-instrumentation

​
Authentication

First create an API key to authenticate. Copy your project name and log stream name

import os
headers = {
    "Galileo-API-Key": "7UKL4Qow2w9qTiYEH2HGR9f7MHVlZyTxDmAhxWE5i94",
    "project_name": MY_PROJECT_NAME,
    "log_stream_name": MY_RUN_NAME,  # Matches Galileo's run_name
}
os.environ['OTEL_EXPORTER_OTLP_TRACES_HEADERS'] = ",".join([f"{k}={v}" for k, v in headers.items()])

Once the headers are set, we can initialize the OTEL logging.

​
Python setup

Finally we can setup the OTEL logging to send traces to Galileo. For this we need to set the endpoint and the

from opentelemetry.sdk import trace as trace_sdk
from opentelemetry import trace as trace_api
from opentelemetry.sdk.trace.export import BatchSpanProcessor
from opentelemetry.exporter.otlp.proto.http.trace_exporter import OTLPSpanExporter
from opentelemetry.sdk.trace.export import ConsoleSpanExporter
from openinference.instrumentation.langchain import LangChainInstrumentor

# OTEL tracing setup
endpoint = "http://api.galileo.ai/otel/traces" # Make sure to replace with the correct endpoint
tracer_provider = trace_sdk.TracerProvider()
tracer_provider.add_span_processor(BatchSpanProcessor(OTLPSpanExporter(endpoint)))
tracer_provider.add_span_processor(BatchSpanProcessor(ConsoleSpanExporter()))
trace_api.set_tracer_provider(tracer_provider=tracer_provider)
LangChainInstrumentor().instrument(tracer_provider=tracer_provider)

​
Summary

  1. Set the headers with the API key, project name and log stream name.
  2. Initialize the OTEL logging with the correct endpoint.
  3. Instrument the needed instrumentors.
  4. Start logging with OTEL.

Supported Instrumentors are: OpenAI: openinference-instrumentation-openai LLama Index: openinference-instrumentation-llama-index DSPy: openinference-instrumentation-dspy Bedrock: openinference-instrumentation-bedrock Langchain: openinference-instrumentation-langchain Mistral: openinference-instrumentation-mistralai Guardrails: openinference-instrumentation-guardrails Vertex AI: openinference-instrumentation-vertexai CrewAI: openinference-instrumentation-crewai Haystack: openinference-instrumentation-haystack LiteLLM: openinference-instrumentation-litellm Groq: openinference-instrumentation-groq Instructor:openinference-instrumentation-instructor Anthropic: openinference-instrumentation-anthropic

​
FAQ

​
How do I setup a project?

Log into your console and create a new logging project. Copy the project name and log stream name after setup

​
Which Instrumentors do i need to import?

For these frameworks you need to import the following instrumentors:

OpenAI: from openinference.instrumentation.openai import OpenAIInstrumentor LLama Index: from openinference.instrumentation.llama_index import LlamaIndexInstrumentor DSPy: from openinference.instrumentation.dspy import DSPyInstrumentor Bedrock: from openinference.instrumentation.bedrock import BedrockInstrumentor Langchain: from openinference.instrumentation.langchain import LangChainInstrumentor Mistral: from openinference.instrumentation.mistralai import MistralAIInstrumentor Guardrails: from openinference.instrumentation.guardrails import GuardrailsInstrumentor Vertex AI: from openinference.instrumentation.vertexai import VertexAIInstrumentor CrewAI: from openinference.instrumentation.crewai import CrewAIInstrumentor from openinference.instrumentation.langchain import LangChainInstrumentor Haystack: from openinference.instrumentation.haystack import HaystackInstrumentor LiteLLM: from openinference.instrumentation.litellm import LiteLLMInstrumentor Groq: from openinference.instrumentation.groq import GroqInstrumentor Instructor:

openinference.instrumentation.instructor import InstructorInstrumentor
from openinference.instrumentation.openai import OpenAIInstrumentor
from openai import OpenAI

# Add the needed client so it's using the correct instrumentor
client = instructor.from_openai(OpenAI())

Anthropic: from openinference.instrumentation.anthropic import AnthropicInstrumentor

​
How do I get the correct endpoint?

The endpoint is the console URL but instead of console you replace it with api and the path is /otel/traces