jaeger-spotbugs-benchmark

name	jaeger-spotbugs-benchmark
description	Launch Jaeger, run SpotBugs benchmark traces, capture Jaeger UI screenshots via the shared `scripts/capture-jaeger-trace-screenshot.mjs` helper, and report bottlenecks by rule, jar, and class for inspequte. Use when asked to profile `scripts/bench-spotbugs.sh`, inspect Jaeger traces, save screenshots under `target/bench`, and explain the slowest components.

jaeger spotbugs benchmark

Inputs

• Repository root with scripts/bench-spotbugs.sh.
• Docker available locally.
• Jaeger UI reachable at http://localhost:16686.
• OTLP HTTP endpoint http://localhost:4318/.
• Node.js + npm available for screenshot capture.
• Execute all commands from repository root.

Outputs

• Jaeger container running.
• Benchmark log in target/bench/spotbugs.log.
• Jaeger trace screenshot in target/bench/.
• Trace JSON export in target/bench/.
• Bottleneck report naming the slowest rule, jar, and class.

Workflow

1. Start Jaeger:
   • Run .codex/skills/jaeger-spotbugs-benchmark/scripts/start-jaeger.sh.
2. Run benchmark with tracing enabled:
   • Run .codex/skills/jaeger-spotbugs-benchmark/scripts/run-bench-spotbugs-with-otel.sh.
   • Override repeat count with positional arg when needed.
   • Override endpoint with OTEL_ENDPOINT only when explicitly requested.
3. Export latest trace JSON and resolve trace ID:
   • trace_json="$(.codex/skills/jaeger-spotbugs-benchmark/scripts/export-jaeger-trace.sh)"
   • trace_id="$(basename "${trace_json}" .json)"
   • trace_id="${trace_id#jaeger-trace-}"
4. Prepare Playwright runtime once per environment:
   • npm install --no-save --no-package-lock playwright@1.53.0
   • npx playwright install --with-deps chromium
5. Capture screenshot via shared script:
   • screenshot_png="$(JAEGER_OUT_DIR=target/bench node scripts/capture-jaeger-trace-screenshot.mjs "${trace_id}")"
6. Analyze trace:
   • .codex/skills/jaeger-spotbugs-benchmark/scripts/analyze-trace-json.sh "${trace_json}"
7. Report bottleneck:
   • Include trace ID and screenshot path.
   • Identify one slowest rule, one slowest jar, and one slowest class.
   • Include concrete timing evidence from trace analysis.
   • Add a short investigation note tying screenshot timeline shape to the extracted slow spans.

Reporting Template

Use this exact format:

## Bottleneck Report
- Trace ID: <trace-id>
- Screenshot: target/bench/jaeger-trace-<trace-id>.png
- Slowest rule: <rule-id> (<total-ms> ms total across <span-count> spans)
- Slowest jar: <jar-path-or-name> (<total-ms> ms total across <span-count> spans)
- Slowest class: <class-name-or-entry> (<total-ms> ms total across <span-count> spans)
- Investigation:
  - Screenshot evidence: <what the expanded timeline shows>
  - Trace evidence: <why these three are the bottleneck>
  - Next action: <one concrete optimization direction>

Guardrails

• Use the exact Jaeger container image: jaegertracing/all-in-one:latest.
• Keep screenshot files only under target/bench.
• Do not report a bottleneck without numeric timing evidence.
• Prefer Jaeger API export plus screenshot observation over guesswork.