openclaw-test-heap-leaks

Analyze test memory growth by collecting repeated .heapsnapshot files from Vitest workers and comparing before/after states using heapsnapshot-delta.mjs to identify retained modules or runtime object leaks.

Vitest worker OOMs often stem from retained transformed modules or real data leaks across iterations. Comparing heap snapshots from the same worker PID isolates whether module retention or runtime object accumulation drives RSS growth.

Compare heap snapshots by PID by organizing snapshots into per-PID groups or lane directories, then run heapsnapshot-delta.mjs to quantify memory deltas between iterations for each specific worker.

Yes, apply heap snapshot tooling to CI-style Linux test runs to generate per-file RSS summaries and compare snapshots by worker PID, allowing you to isolate hotspot test files and patch cleanup logic.

You need at least two heap snapshots to identify test memory growth. Capturing repeated snapshots across iterations allows heapsnapshot-delta.mjs to compare before and after states and distinguish transformed-module retention from real data leaks.

The best way to distinguish transformed-module retention from actual memory leaks is to compare heap snapshots across iterations using heapsnapshot-delta.mjs, which quantifies memory deltas and separates module growth from runtime object leaks.