Native OpenTelemetry Instrumentation for GenAI Semantic Conventions #12279
Unanswered
srinjoy356
asked this question in
Questions
Replies: 0 comments
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Uh oh!
There was an error while loading. Please reload this page.
Hi Haystack team,
We are currently working on adding OpenTelemetry instrumentation for Haystack in the
open-telemetry/opentelemetry-python-genairepository. We are instrumenting core Haystack components (LLMs, Embedders, Pipelines, Agents, and Tools) to emit telemetry that conforms to the official OpenTelemetry GenAI Semantic Conventions.Our maintainers requested that we reach out to see if the Haystack project would be interested in supporting this natively within Haystack itself, rather than us maintaining it externally as auto-instrumentation.
Native instrumentation is often more robust because it avoids monkey-patching and can hook directly into the component lifecycle.
Is this something the Haystack team would be interested in incorporating into the core library or an official integration package? We would be happy to share our approach and findings from the external instrumentation we've built so far.
Thanks!
All reactions