​
Overview

Elastic APM is a powerful solution for monitoring the performance of your applications. As Elastic APM is OpenTelemetry (OTel) compatible by default, you can easily integrate it with Langtrace to monitor your LLM applications.

​
Environment Variables

Set up the following environment variables to enable OpenTelemetry tracing and send traces to Elastic APM:

export OTEL_EXPORTER_OTLP_ENDPOINT="abcd1234.apm.us-central1.gcp.cloud.es.io:443"  # Your Elastic OTEL endpoint
export OTEL_EXPORTER_OTLP_HEADERS="Authorization=Bearer <secret>"
export OTEL_RESOURCE_ATTRIBUTES="service.name=rag-app-demo,service.version=0.0.1,deployment.environment=production"

We support all OpenTelemetry environment variables. You can also use the legacy format:

export LANGTRACE_API_HOST="<https://your-elastic-apm-endpoint:443/v1/traces>"
export LANGTRACE_HEADERS="Authorization=Bearer your-elastic-apm-token"
export OTEL_SERVICE_NAME=your-elastic-service

​
RAG Application Example

Here’s an example of monitoring a Retrieval-Augmented Generation (RAG) chatbot application that uses Elastic Search for document search and Azure OpenAI for generating responses.

​
Initialize Langtrace

from langtrace_python_sdk import langtrace
langtrace.init()

# Rest of your application code

​
Configure Azure OpenAI (if using)

Set up the necessary Azure OpenAI environment variables:

export LLM_TYPE=azure
export OPENAI_VERSION=<version>       # e.g., 2023-05-15
export OPENAI_BASE_URL=<base_url>     # Azure OpenAI endpoint URL
export OPENAI_API_KEY=<api_key>       # Azure OpenAI API key
export OPENAI_ENGINE=<deployment_name> # Azure deployment name

​
Monitoring in Elastic APM

After setting up your application, you can monitor your RAG application’s performance in the Elastic APM dashboard. The interface provides deep insights into your application’s behavior:

The main dashboard displays:

  • Service health metrics
  • Transaction throughput
  • Error rates and patterns
  • Infrastructure metrics

Detailed trace analysis reveals the complete request journey:

The trace view enables you to:

  • Track request flows
  • Measure operation durations
  • Identify performance bottlenecks
  • Monitor error propagation

For detailed span inspection:

The span analysis shows:

  • Precise timing measurements
  • Request/response payloads
  • Error context and details
  • Custom metadata and annotations

Use these traces for:

  • Debugging issues
  • Performance monitoring
  • System optimization
  • Understanding user interactions

​
Troubleshooting

• Traces not visible in Elastic APM:

  • Ensure that all required environment variables are set correctly
  • Verify your OTEL_EXPORTER_OTLP_ENDPOINT or LANGTRACE_API_HOST includes /v1/traces at the end
  • Check if your authentication token is valid and properly formatted
  • Verify your service name is correctly set in OTEL_RESOURCE_ATTRIBUTES

• Missing spans or incomplete traces:

  • Ensure Langtrace is initialized before any LLM operations
  • Check if all required dependencies are installed
  • Verify your application has proper permissions to send traces

​
Additional Resources

OverviewGrafana Cloud