fix(bench): move collector config out of workflows/ + unthrottle telemetrygen (PR-N1.2)

[jensholdgaard]

What

Fixes the two failures PR-N1 (#70) hit on its first real dispatch.

1. Collector config was in the workflow-scan directory

.github/workflows/bench-collector.yaml (the OTel collector config from PR-N1) made GitHub try to parse it as a workflow definition on every push to main — everything under .github/workflows/ is scanned as a workflow — producing a failing run on each push (example: run 26706450810).

Moved to .github/bench-collector.yaml (one level up, still version-controlled beside bench.yml, no longer workflow-scanned — .github/FUNDING.yml already lives there as precedent). Updated the --config path in bench.yml and both files' self-describing comments.

2. telemetrygen crawled at 1 log/sec

The second dispatch (run 26706451070) got past install but the capture step ran for ~4h before I cancelled it. The worker log showed exported batched logs count: 100 every ~100 seconds = 1 log/sec.

Root cause, traced to source at the pinned v0.153.0:

• cmd/telemetrygen/pkg/logs/config.go:54 → c.Rate = 1 (hardcoded default of 1 log/sec).
• cmd/telemetrygen/pkg/logs/logs.go:53-55 → only --rate 0 selects the unbounded rate.Inf path.

1,000,000 records at 1/sec ≈ 11.5 days. Added --rate 0. With rate.Inf, 1M records is seconds.

3. Backstop against future hangs

Wrapped telemetrygen in timeout 600. Any future re-introduction of throttling (or a genuine collector stall) now fails the step in 10 min — with /tmp/collector.log dumped — instead of burning a runner to GitHub's 6h job ceiling. The real exit code is captured via || tg_rc=$? rather than if ! …; then (which reports $? as 0 inside the branch; 124 = the timeout backstop).

CLAUDE.md §6.6

• cargo fmt --all --check, cargo test --all-features (322/0/44), cargo clippy, mdbook build — all green (no Rust changes; baseline preserved).
• Both YAML files validated with yaml.safe_load.
• The workflow itself can't be exercised without dispatching — first clean dispatch happens after merge.

Out of scope

The dispatch + the §9 row in docs/benchmarks.md follow this merge (PR-N2), once we have a non-hung run producing real A1/C1/C2 numbers on ci-runner + telemetrygen-synthetic-v1.

🤖 Generated with Claude Code

Summary by CodeRabbit

• Chores
  • Enhanced benchmark testing workflow with stronger error detection, timeout safeguards, and improved diagnostic logging to help identify and resolve issues more quickly during test execution.
  • Updated and clarified benchmark collector configuration documentation to provide better context on infrastructure setup decisions and explain the configuration organization rationale.

[coderabbitai]

No actionable comments were generated in the recent review. 🎉

ℹ️ Recent review info

⚙️ Run configuration

Configuration used: defaults

Review profile: CHILL

Plan: Pro Plus

Run ID: 15143be3-f361-4d8b-b066-93362ff8cf02

📥 Commits

Reviewing files that changed from the base of the PR and between 4ba907c and 58e9e50.

📒 Files selected for processing (2)

• .github/bench-collector.yaml
• .github/workflows/bench.yml

---

📝 Walkthrough

Walkthrough

Benchmark workflow infrastructure is reorganized: the OpenTelemetry Collector config moves from .github/workflows/bench-collector.yaml to .github/bench-collector.yaml with explanatory comments. The telemetrygen logs capture step is enhanced with unbounded rate, timeout guarding, proper exit code handling, and structured error reporting.

Changes

Benchmark Infrastructure Improvements

Layer / File(s)	Summary
Config file reorganization and documentation .github/bench-collector.yaml, .github/workflows/bench.yml	Collector config is placed under .github/ with a header comment documenting the rationale (avoiding GitHub Actions workflow parsing failures), and the workflow comment and config reference are updated to point to .github/bench-collector.yaml.
Telemetrygen capture resilience .github/workflows/bench.yml	Telemetrygen logs generation is expanded with --rate 0 for unbounded capture, wrapped in timeout 600, modified to preserve the command's actual exit code despite set -e, and enhanced with ::error:: annotation and collector log dump on failure.

Estimated code review effort

🎯 2 (Simple) | ⏱️ ~12 minutes

Possibly related PRs

• jensholdgaard/ourios#70: Main PR refines the telemetrygen OTLP capture wiring introduced by this PR, updating the path reference and capture step with exit handling and timeout protection.
• jensholdgaard/ourios#57: Both PRs modify the benchmark workflow; the retrieved PR introduces on-demand bench running, while the main PR updates the telemetrygen capture step to use the reorganized collector config.

Poem

🐇 A config takes flight from workflows to rest,
In .github/ home, it settles the best!
With timeout and rate, telemetrygen sings,
Error logs flutter on swift bunny wings. ✨

🚥 Pre-merge checks | ✅ 5

✅ Passed checks (5 passed)

Check name	Status	Explanation
Title check	✅ Passed	The title accurately and specifically summarizes the main changes: moving collector config out of workflows and unthrottling telemetrygen with explicit rate parameter.
Description check	✅ Passed	The description comprehensively covers what the PR does, why, and related changes, but lacks completion of the repository's required checklist sections.
Docstring Coverage	✅ Passed	No functions found in the changed files to evaluate docstring coverage. Skipping docstring coverage check.
Linked Issues check	✅ Passed	Check skipped because no linked issues were found for this pull request.
Out of Scope Changes check	✅ Passed	Check skipped because no linked issues were found for this pull request.

_{✏️ Tip: You can configure your own custom pre-merge checks in the settings.}

✨ Finishing Touches

🧪 Generate unit tests (beta)

• Create PR with unit tests
• Commit unit tests in branch fix/bench-collector-path-and-rate

---

_{Comment @coderabbitai help to get the list of available commands and usage tips.}

[Copilot]

Pull request overview

This PR fixes the benchmark workflow’s telemetry capture path so the OTel collector config is no longer parsed as a workflow and telemetry generation completes within a bounded time.

Changes:

• Moves the collector config reference from .github/workflows/ to .github/.
• Adds --rate 0 to unthrottle telemetrygen logs.
• Wraps telemetry generation with a 10-minute timeout and collector-log dump on failure.

Reviewed changes

Copilot reviewed 2 out of 2 changed files in this pull request and generated no comments.

File	Description
.github/workflows/bench.yml	Updates collector config path and hardens telemetrygen execution with unbounded rate plus timeout handling.
.github/bench-collector.yaml	Updates comments to document why the collector config lives outside .github/workflows/.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.