[passionworkeer]

What does this PR do?

LangfuseSpanProcessor forwards flush_at to BatchSpanProcessor as
max_export_batch_size and never sets max_queue_size. OpenTelemetry keeps its own
default queue size and rejects any larger batch, so a flush_at above the queue size
failed client construction with

ValueError: max_export_batch_size must be less than or equal to max_queue_size.

The message never mentions flush_at, which makes the failure hard to attribute to a
Langfuse setting.

This passes max_queue_size=max(flush_at, default_queue_size), where default_queue_size
is OpenTelemetry's own resolved default (BatchSpanProcessor._default_max_queue_size(),
which honours OTEL_BSP_MAX_QUEUE_SIZE). A batch that already fits leaves the queue
exactly as it is today; only a batch that does not fit widens it.

Fixes langfuse/langfuse#19539

Type of change

• Bug fix (non-breaking change which fixes an issue)

Verification

Regression tests in tests/unit/test_flush_at_batch_size.py. On the unpatched main
4 of the 5 fail or error with the ValueError above; all 5 pass with the fix:

case	before	after
flush_at=5000 constructs and flushes	ERROR ×2	PASSED ×2
finished spans still exported with flush_at=5000	FAILED	PASSED
OTEL_BSP_MAX_QUEUE_SIZE=1000 + flush_at=1500	FAILED	PASSED
OTEL_BSP_MAX_QUEUE_SIZE=5000 + flush_at=3000 keeps queue at 5000	PASSED (guard)	PASSED

Queue capacity sweep (probe script, not committed — it reads OpenTelemetry internals):

OTEL_BSP_MAX_QUEUE_SIZE	flush_at	before	after
unset	512	2048	2048
unset	2048	2048	2048
unset	3000	ValueError	3000
1000	1500	ValueError	1500
5000	3000	5000	5000 (regression guard, unchanged)
1000	5000	ValueError	5000
1000	5000	ValueError	5000

Other checks:

• tests/unit/test_span_processor.py + test_resource_manager.py + test_otel.py:
  145 passed, 2 skipped.
• Full tests/unit: 988 passed, 2 skipped. The remaining 2 failures and 18 errors are all
  in tests/unit/test_prompt.py / test_prompt_atexit.py and reproduce identically on the
  unpatched main in this environment (no LANGFUSE_PUBLIC_KEY, no literal python on
  PATH); they are unrelated to this change.
• ruff check . clean, ruff format --check clean on both changed files,
  mypy langfuse clean.

Note for reviewers

_default_max_queue_size() is an OpenTelemetry internal. It is used here because it is
what determines the constraint being fixed, and because the queue default is configurable
through OTEL_BSP_MAX_QUEUE_SIZE — hard-coding 2048 would both miss smaller configured
queues and narrow larger ones. The declared range is opentelemetry-sdk>=1.45.0,<2; the
alternative would be duplicating OpenTelemetry's environment parsing.

An alternative fix would be to validate flush_at and raise a clear error naming the
setting instead of widening the queue. Happy to switch if you prefer that — this PR picks
widening because it keeps every currently working configuration working.

Checklist

• I have self-reviewed the diff
• I have added or updated tests
• I have not hand-edited generated files under langfuse/api/
• No new dependencies, no public API change

Confidence Score: 4/5

The PR appears safe to merge, with a non-blocking test isolation issue.

Summary

Widens the span queue when flush_at exceeds OpenTelemetry's configured queue size.

• Large flush_at settings fit in the span queue.

_{Reviews (1) · Last reviewed commit: "fix(otel): widen the span queue when flu..." · Reviewed by Greptile}

[claude]

Claude Code Review

This pull request is from a fork — automated review is disabled. A repository maintainer can comment @claude review to run a one-time review.

[Copilot]

Copilot was unable to review this pull request because the user who requested the review has insufficient remaining quota. Try requesting a review at a lower effort level.