Skip to content

fix(tracing): prevent RecursionError in truncation of truncated previews #2857

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Closed
Bahtya wants to merge 3 commits into from
+11 −0
Closed

fix(tracing): prevent RecursionError in truncation of truncated previews #2857

Changes from 1 commit
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
a2b1218
fix(tracing): prevent RecursionError in truncation of truncated previews
Bahtya Apr 8, 2026
95620ca
fix: use sentinel key to identify truncated preview dicts
Bahtya Apr 8, 2026
f6163df
fix: validate sentinel value and preview fields for truncated detection
Bahtya Apr 8, 2026
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
5 changes: 5 additions & 0 deletions src/agents/tracing/processors.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -291,6 +291,11 @@ def _truncate_json_value_for_limit(self, value: Any, max_bytes: int) -> Any:
return self._truncate_string_for_json_limit(value, max_bytes)

if isinstance(value, dict):
# Avoid re-truncating our own truncated preview dicts, which
# would cause infinite recursion (the preview dict contains bool
# values that get wrapped into new preview dicts, and so on).
if value.get("truncated") is True and "original_type" in value and "preview" in value:
return value
Copy link
Copy Markdown

@chatgpt-codex-connector chatgpt-codex-connector bot Apr 8, 2026

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

P2 Badge Narrow truncated-preview detection to SDK-generated shape

The early return in _truncate_json_value_for_limit now treats any dict with truncated=True, original_type, and preview as an internal preview object and skips truncation. If user-provided span data legitimately uses those keys and also contains large additional fields, root-level truncation returns the oversized dict unchanged, which can exceed _OPENAI_TRACING_MAX_FIELD_BYTES and cause export rejection or dropped tracing data. This check should only bypass truncation for SDK-created preview dicts (for example by requiring the exact key set or using a dedicated sentinel).

Useful? React with 👍 / 👎.

Comment on lines +302 to +303
Copy link
Copy Markdown

@chatgpt-codex-connector chatgpt-codex-connector bot Apr 8, 2026

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

P2 Badge Restrict preview bypass to SDK-owned objects

Fresh evidence: although this guard now also checks original_type and preview, it still short-circuits on any user-supplied dict that includes "__truncated_preview__": true and those keys. If that dict also contains large extra fields and exceeds _OPENAI_TRACING_MAX_FIELD_BYTES, _truncate_json_value_for_limit returns it unchanged instead of delegating to _truncate_mapping_for_json_limit, so sanitization can still emit oversized span fields that get rejected or dropped by trace ingestion.

Useful? React with 👍 / 👎.

return self._truncate_mapping_for_json_limit(value, max_bytes)

if isinstance(value, list):
Expand Down