[DavidTraina]

What does this PR do?

Since langfuse/langfuse#19043, the observations v2 API keeps scanning until it fills the page or reaches its 45 s budget, so slow pages are expected. This is especially true for selective filters on large projects. Batch evaluation fetched pages with the client-wide timeout, which defaults to 5 s.

Fern retries only 408/409/429/5xx responses, not httpx timeouts. So the run stopped on the first slow page with completed=False. Resuming sent the identical request and timed out the same way, so the run never made progress. The error log also said Failed to fetch batch after 3 retries after a single attempt.

Fix

• Send timeout_in_seconds=60 per request on observation page fetches.
• Leave the global timeout alone. Raising LANGFUSE_TIMEOUT would also lengthen the OTLP export deadline, and with it flush() and shutdown() when the host is unreachable.
• Log Failed to fetch batch: <error> without the misleading retry count.

The per-request option also applies when the user passes a custom httpx_client.

Related PRs

Independent follow-ups to langfuse/langfuse#19043, mergeable in any order: #2097 (batch evaluation stops on empty pages), #2098 (docs), #2099 (test helpers), #2101 (pagination helper).

Type of change

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

Verification

List the main commands you ran:

uv run --frozen ruff check .                                  # All checks passed!
uv run --frozen ruff format --check .                         # 593 files already formatted
uv run --frozen mypy langfuse --no-error-summary              # no issues
uv run --frozen pytest -n auto --dist worksteal tests/unit    # 1003 passed, 2 skipped

I also ran run_batched_evaluation against a local stub server that answers after 7 s. Before the change it failed after 5.0 s with one request (completed=False). With timeout_in_seconds=60 the same call succeeded after 7.0 s.

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.
• 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: 5/5

This PR appears safe to merge.

Summary

Batch evaluation now gives each observation page request a 60-second timeout without changing the client-wide timeout.

• Batch evaluation waits up to 60 seconds for each observation page.
• Batch evaluation logs fetch failures without a retry count.

_{Reviews (1) · Last reviewed commit: "fix(batch-evaluation): use a 60 s timeou..." · 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.