When a timeout is happens the node process is hanging and not finalized.

I have a simple node script that needs to retrieve data for etcd. I have added a policy to deal with timeout (beside a circuit breaker and retry policy).

• node v14.17.6
• npm 6.14.15
• etcd3 1.1.0

The code i am trying to execute is the following:

const timeout = 5000;

const getAllKeys = async () => {

  const timeoutHost = Policy.timeout(timeout, TimeoutStrategy.Aggressive);
  const timeoutHostListener = timeoutHost.onTimeout(() => console.log("timeout while connecting to the host"));
  const circuitBreakerHost = Policy.handleAll().circuitBreaker(5_000, new ConsecutiveBreaker(3));

  const timeoutGlobal = Policy.timeout(timeout, TimeoutStrategy.Cooperative);
  const timeoutGlobalListener = timeoutGlobal.onTimeout(
    () => console.log("timeout on connection to etcd"),
  );
  const retryGlobal = Policy.handleWhen(isRecoverableError).retry().attempts(3);

  const etcd = new Etcd3({
    hosts,
    faultHandling: {
      host: () => Policy.wrap(
        timeoutHost,
        circuitBreakerHost,
      ),
      global: Policy.wrap(
        timeoutGlobal,
        retryGlobal,
      ),
    },
  });

  try {
    const result = await etcd.getAll();
    return result;
  } catch (e) {
    throw e;
  } finally {
    timeoutHostListener.dispose();
    timeoutGlobalListener.dispose();
    etcd.close();
  }
};

getAllKeys().then((watcher) => {
  console.log(`Data: ${JSON.stringify(watcher)}`);
}).catch((err) => {
  console.log("error: " + err);
});

When I have a timeout error, the result is the following:

timeout on connection to etcd
error: Error: Operation timed out after 5000ms

The node process is not able to exit and it is hanging while when there is no timeout the node process finalizes correctly. Not sure if I am missing something or it is a bug.

Reproduce the timeout using the shown getAllKeys script and inspect the lifecycle around the timeout policies and etcd.close() in the finally block. Trace which resources remain open after the timeout, then verify that the process exits after cleanup while preserving the reported timeout behavior.