[hassiebp]

What does this PR do?

When trace attributes are passed through LangChain invocation metadata (config={"metadata": {"langfuse_session_id": ...}}) and the runnable is called with ainvoke, only the root observation got session_id (and user_id, tags, trace_name, propagated metadata). Every child generation, tool and chain had none, so those observations dropped out of session-scoped views such as the Sessions page tool filters. Sync invoke and an explicit propagate_attributes(...) wrapper both worked.

Root cause: CallbackHandler is a sync handler with run_inline=False. In async runs LangChain calls it via run_in_executor(None, copy_context().run, ...), so each callback runs in a fresh copy of the caller's context. The root on_chain_start enters propagate_attributes, but that context change stays in the root callback's copy, so the child callbacks never see it and their spans start without the propagated keys.

Fix:

• The root run stores the OTel context it built after entering propagate_attributes on _RootRunState.propagation_context.
• Child start_observation calls (chain, tool, retriever, LLM) run inside _use_root_propagated_attributes. It fills in only the langfuse.propagated.* keys that are missing from the current context. Scalar values already set in the current context win, and metadata/tags are merged. If nothing differs, nothing is attached, so the sync path does not change.

Fixes LFE-17917

Type of change

• Bug fix
• New feature
• Breaking change
• Refactor
• Documentation update
• Tooling, CI, or repo maintenance

Verification

test_root_chain_metadata_propagates_session_id_to_children_on_ainvoke failed before the fix (KeyError: 'session.id' on ChatPromptTemplate) and passes after it. test_child_run_keeps_attributes_propagated_in_its_own_context checks that a child's own propagate_attributes still wins over the root value.

uv run --frozen ruff check .                       # All checks passed!
uv run --frozen mypy langfuse --no-error-summary   # exit 0
uv run --frozen pytest -n auto tests/unit/test_langchain.py tests/unit/test_langchain_utils.py   # 31 passed
uv run --frozen pytest -n auto --dist worksteal tests/unit   # 974 passed, 2 skipped

I did not run e2e or live-provider tests. The behavior is covered by unit tests with the in-memory exporter.

Checklist

• I self-reviewed the diff using code_review.md.
• I added or updated tests for behavior changes.
• I updated docs, examples, or .env.template if needed. (Not needed: this brings async in line with the documented behavior.)
• I did not hand-edit generated files; if generated files changed, I used the upstream regeneration path.
• I did not commit secrets or credentials.

  

Confidence Score: 4/5

Fix the remaining async session override case before merging.

Summary

Saves the root chain's propagated attributes and makes them available while child observations start.

• Async LangChain child observations inherit root trace attributes.

Diagram

%%{init: {'theme': 'neutral'}}%%
flowchart TD
  A["Caller context: session outer"] --> B["Root callback: invocation sets session inner"]
  B --> C["Save root context"]
  A --> D["Async child callback inherits session outer"]
  C --> E["Merge saved root attributes"]
  D --> E
  E --> F["Existing value wins: child keeps session outer"]

Loading

_{Reviews (1) · Last reviewed commit: "fix(langchain): propagate root trace att..." · Reviewed by Greptile}

[github-actions]

@claude review

[chatgpt-codex-connector]

Codex Review Summary

This comment shows the latest Codex review activity on this pull request.

Review	Status	Commit	Review trigger
📝 Code Review	✅ Completed 2026-10-28T09:57:23.118356Z	9c582ab	PR opened

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review" or "@codex security review".

Codex reacts with 👀 while any review is running, comments if it has suggestions, and reacts with 👍 once all reviews finish with no findings.

[claude]

Claude Code Review

No review was started: this request came from a bot account. Manual reviews can only be requested by someone with write access to this repository. Ask a maintainer to comment @claude review, or have your automation post the comment from a user account with write access.

_{Tip: disable this comment in your organization's Code Review settings.}

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 9c582abcd8

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

[claude]

Nothing blocking. The comments below are optional suggestions. There is no need to push a fix for them before merging.