This post tells the story of how contributors from across the cloud native community worked together to enhance Dapr’s OpenTelemetry integration, particularly around asynchronous workflows. It also highlights the ongoing effort to align Dapr with the OpenTelemetry semantic conventions using OpenTelemetry Weaver, and explores how this collaboration can serve as a useful example for other CNCF projects. None of this work started as a formal initiative. It emerged through discussion, experimentation, and a shared goal of making telemetry easier to understand and more consistent across the ecosystem.

OpenTelemetry's platform is an observability framework, offering insights into distributed tracing, metrics collection, and log management for cloud-native applications. Through documentation, tutorials, and community contributions, OpenTelemetry provides insights into instrumenting applications to monitor performance and troubleshoot issues. Developers and DevOps teams can learn about distributed tracing standards, telemetry data formats, and integration with observability tools to gain visibility into complex distributed systems.

OpenTelemetry

Dapr's workflow engine struggled with trace context propagation across long-lived gRPC streams, causing fragmented traces where workflow, activity, and user spans appeared disconnected. Contributors solved this by embedding W3C trace context directly into workflow activity messages in durabletask-go, updating SDKs to restore context, and leveraging OpenTelemetry's automatic instrumentation. The result is continuous, hierarchical traces in Jaeger that show the complete workflow lifecycle. Dapr is now adopting OpenTelemetry Weaver to standardize semantic conventions across components and exploring better span kind modeling for asynchronous operations. This collaborative effort demonstrates patterns applicable to other CNCF projects facing similar telemetry challenges.

Improving Async Workflow Observability in Dapr

The challenges of propagating traces throughout complex orchestrations

The complexities of a typical workflow orchestration

Restoring context across the workflow boundary

From trace continuity to semantic alignment

Why this work matters across the cloud native ecosystem