Hardcoded 1000ms timeout causes failures in unstable network conditions #42

Description

The current implementation has a hardcoded 1000ms timeout that makes the system fragile in unstable network conditions.

Problem Code

const timeout = setTimeout(async () => {
  await cleanup();
  const val = await ctx.publisher.get(`${ctx.keyPrefix}:sentinel:${streamId}`);
  if (val === DONE_VALUE) {
    resolve(null);
  }
  if (Date.now() - start > 1000) {
    controller.error(new Error("Timeout waiting for ack"));
  }
}, 1000);

Issues

1. Hardcoded timeout is inflexible: The 1000ms timeout cannot adapt to different network environments
2. Too aggressive: 1 second may be insufficient for high-latency or unstable networks
3. Confusing logic: The setTimeout delay and the timeout check both use 1000ms, meaning the timeout error will almost always trigger when the timer fires

Impact

• Systems fail unnecessarily in slower network conditions
• No way for users to configure timeout based on their environment
• Production deployments in high-latency regions become unreliable

Suggested Solutions

Option 1: Configurable Timeout

const timeout = ctx.config.ackTimeout || 5000; // Default to 5s
const timer = setTimeout(async () => {
  await cleanup();
  const val = await ctx.publisher.get(`${ctx.keyPrefix}:sentinel:${streamId}`);
  if (val === DONE_VALUE) {
    resolve(null);
  } else if (Date.now() - start > timeout) {
    controller.error(new Error("Timeout waiting for ack"));
  }
}, timeout);

Option 2: Retry with Backoff

Implement exponential backoff retry logic instead of a single hard timeout.

Environment

• Network latency: Variable (100ms - 2000ms in our production)
• This affects production stability

Would you be open to a PR implementing configurable timeouts?