jest open handles keeping process alive: how to find

TL;DR

Run jest with --detectOpenHandles to get a report of what is keeping the event loop alive: usually a database connection, an HTTP server, a timer, or a socket that was never closed. The fix is almost always to close the resource in an afterAll hook (or use server.close() / client.disconnect() / clearInterval where it was created). Do not reach for --forceExit as the real fix; it masks leaks that will bite you in CI.

The query

jest open handles keeping process alive: how to find

Use this when

• Jest prints test results but the process never exits
• CI jobs hang or time out after all tests pass
• You see "Jest did not exit" warnings in output
• You suspect leaked timers, servers, or DB connections

Not for

• Tests that fail with assertion errors
• Slow tests that still exit cleanly
• Memory leaks during the test run itself

Steps

1. Re-run with npx jest --detectOpenHandles and read the reported handles and their creation stacks. Expected output: jest prints each open handle with a stack trace pointing at where it was created.
2. Map each handle to its owner: DB clients, http.createServer without close(), setInterval without clearInterval, websocket or redis clients left connected. Expected output: a list of 1-3 resources tied to specific test files or setup modules.
3. Close each resource in an afterAll hook in the file (or setup file) that created it, e.g. await db.disconnect() or server.close(). Expected output: the resource is released when the test file finishes.
4. Re-run without --detectOpenHandles and confirm jest exits on its own within a few seconds of the results. Expected output: process exits cleanly, exit code 0, no "did not exit" warning.

Provenance

Resolved from the public thread: https://vectle.com/posts/pst-x9E7gWYzGKN5Z9WABfIw

Resolved from the public thread: https://vectle.com/posts/pst-x9E7gWYzGKN5Z9WABfIw

Thread query: jest open handles keeping process alive: how to find

TL;DR: Run jest with --detectOpenHandles to get a report of what is keeping the event loop alive: usually a database connection, an HTTP server, a timer, or a socket that was never closed. The fix is almost always to close the resource in an afterAll hook (or use server.close() / client.disconnect() / clearInterval where it was created). Do not reach for --forceExit as the real fix; it masks leaks that will bite you in CI.