Implement stable http semantic conventions

[JamieDanielson]

This is part of Semantic Convention Update: Tracking Issue #4572.

#4690 updated semantic conventions to 1.27, and includes both stable and experimental attributes.

This issue is for updating instrumentations to allow for dual emitting stable and experimental HTTP attributes.

• Add environment variable option OTEL_SEMCONV_STABILITY_OPT_IN that includes http and http/dup options as suggested in the Migration Plan.
• Instrumentations should default to current experimental HTTP attributes, with the option to emit only the stable HTTP attributes (with http as the env var value) or dual emit experimental and stable HTTP attributes (http/dup).

The next major version bump should drop this dual emit for HTTP attributes and default to only the stable HTTP attributes.

• http client spans (Update HTTP client span semconv to 1.27 #4940)
• http server spans (Http server semconv span stable #4978)
• grpc
• fetch
• xhr

[joselopezgr]

Hi @JamieDanielson,

I wanted to check if there's any update on Issue #4897 or if it's still a WIP.

I'm implementing OpenTelemetry via @opentelemetry-api in a TypeScript/NestJS project, and I’m encountering a similar issue where HTTP attributes aren't updating.

Here's the span I'm seeing:

And here’s how it looks in my local Grafana:

Could you confirm if there's a solution for this or if it's still being worked on? 😃

Thank you so much!

[pitoniak32]

Hey all! Is there any update on this? It seems the blocking item has been closed?

[JamieDanielson]

I have updated the issue description to include a checklist for items in flight, as well as remove the "blocked" note.

[JamieDanielson]

As of right now, the http client spans have the updated option using OTEL_SEMCONV_STABILITY_OPT_IN environment variable set to either http or http/dup. The instrumentation is not yet released, and we are working through the http server spans now, with a few others soon to follow. We hope to have updated instrumentation available soon!

[github-actions]

This issue is stale because it has been open 60 days with no activity. Remove stale label or comment or this will be closed in 14 days.

[pichlermarc]

Something that came up on #5217: I think we should have code-only stability-opt-in config, as opposed to pseudo-env var config via window for the xhr/fetch instrumentation.