FAQ

Why not clean up in after* hooks?

afterAll runs per worker, and afterEach runs per test. Either can delete a shared resource while other tests or workers still need it. Use the dedicated cleanup callback, which runs after the complete execution.

Why did Global Cache report a signature mismatch?

The same key was requested with a different computation, TTL, or call site during one run. Give each logical computation a unique key and keep one canonical call site per key.

Can I cache a Playwright page or database connection?

No. Values cross an HTTP/JSON boundary. Cache serializable data such as a storage state, identifier, token, or plain response, then use it to configure a local object.

How do I force a value to be recalculated?

Call await globalCache.delete(key). It removes the in-memory and persistent value, so the next get() computes it again.

Should I persist values in CI?

Usually not on ephemeral runners. Set ignoreTTL: true in CI so values remain shared within a run but are not written for reuse by later runs. If your CI restores .global-cache deliberately, keep TTLs enabled.

Do I need to stop the local server?

No. The Playwright integration manages the small local HTTP server in command-line, UI, and VS Code runs. Run-scoped values are refreshed between executions even when Playwright keeps its process open.

Can multiple shards share values?

Yes, if they can reach the same external Core server and use the same GLOBAL_CACHE_RUN_ID. The system launching the shards should use a new run ID for each unrelated execution and clean shared resources only after every shard finishes.