openclaw-test-heap-leaks

name	openclaw-test-heap-leaks
description	Investigate OpenClaw pnpm test memory growth, Vitest OOMs, RSS spikes, and heap snapshot deltas.

OpenClaw Test Heap Leaks

Use this skill for test-memory investigations. Do not guess from RSS alone when heap snapshots are available. Treat snapshot-name deltas as triage evidence, not proof, until retainers or dominators support the call.

Read ../openclaw-test-performance/SKILL.md first for the current test-performance commands and proof routing.

For runtime fixes (e.g., closure leaks in long-running services like the gateway), see Validating runtime fixes below — that uses a dedicated harness rather than the unit-test profiling workflow.

Workflow

1. Reproduce the failing shape first.

   • Match the real entrypoint and worker budget. For a broad unit-fast baseline with per-config max RSS and top-file timing, start with:

     pnpm test:perf:groups \
       --config test/vitest/vitest.unit-fast.config.ts \
       --allow-failures \
       --output .artifacts/test-perf/unit-fast-memory.json
     

   • For a suspected file, rerun that file with one worker and collect wall/RSS evidence: /usr/bin/time -l pnpm test <file> --maxWorkers=1 --reporter=verbose.

   • Current pnpm test execution is planned by scripts/test-projects.mts. Record the printed Vitest config or shard and preserve that shape when the report is configuration- or worker-budget-specific.

2. Collect the strongest available heap evidence.

   • Run pnpm test:perf:profile:runner -- --output-dir .artifacts/test-perf/vitest-runner-profile -- <file> for a CPU profile plus a sampling heap profile of the unit runner. Open the heap profile in DevTools and inspect the largest allocation families.