Temporal OpenTelemetry Samples

This project demonstrates how to integrate Temporal with OpenTelemetry for complete observability of your workflow executions. It includes:

• 📊 Complete observability stack: OpenTelemetry traces, metrics, and logs
• ⚡ Temporal Workflow: HTTP GET workflow with full telemetry instrumentation
• 🔄 OpenTelemetry Collector: Central hub for processing and routing telemetry data
• 🚀 Two deployment options:
  • Open-source tools (Jaeger, Prometheus, and Elasticsearch)
  • Dynatrace

💡 New to OpenTelemetry? Check out our OpenTelemetry Primer to understand the concepts, benefits, and why it matters for Temporal developers.

Usage

Prerequisites

• uv
• Docker

Quick Start

1. Install Python and dependencies:

   # Install Python 3.12 if needed
   uv python install 3.12
   
   # Install all project dependencies
   uv sync --dev

2. Start the observability stack:

   # Starts Temporal server + Jaeger + Prometheus + Elasticsearch + OTEL Collector
   uv run poe up

   Wait for all services to start (about 30-60 seconds). You'll see logs from multiple containers.

3. Run the demo workflow (in another new terminal):

   uv run poe client

   This continuously executes HTTP workflows every 2 seconds. You'll see workflow IDs being printed.

Explore Your Telemetry Data

Now that everything is running, explore the observability data:

🔍 Distributed Traces → Jaeger

• Go to http://localhost:16686/
• Click "Find Traces" to see traces
• Click on any trace to see the complete Temporal Workflow → Activity call chain

📊 Metrics → Prometheus

• Go to http://localhost:9090/
• Try queries like:
  • temporal_workflow_completed_total - completed workflows
  • temporal_worker_task_slots_available - available execution slots in Worker

📝 Logs → Elasticsearch

• Query logs: curl http://localhost:9200/temporal-logs/_search | jq
• Or browse with: curl http://localhost:9200/_cat/indices

⚙️ Temporal UI → Temporal Web

• Go to http://localhost:8233/
• View running workflows, activity history, and task queues

🛠️ Deployment Options

By default, this project uses open-source observability platforms (Jaeger, Prometheus, and Elasticsearch) to storage Temporal traces, metrics, and logs.

To learn more about how this solution works with third-party observability vendors, use the following step-by-step instructions:

1. Dynatrace

To see the OpenTelemetry outputs from Nexus operations, try running the Nexus sample.

🏗️ Want to understand the architecture? See our Architecture Documentation for a comprehensive overview of how Temporal and OpenTelemetry components work together.

🤝 Contributing

We welcome contributions from the community! Whether you want to add new OpenTelemetry integrations, improve documentation, or enhance existing examples, check out our Contributing Guide to get started.