Why does maxTimeout have no effect on getStats().average?

[TimDaub]

• According to the docs: https://github.com/diamondio/better-queue#queue-statistics, getStats().average is the time in milliseconds a task spends executing in the queue
• So if maxTimeout is set to Infinity then, we'd expect getStats().average to be anywhere between 0 and Infinity
• However, shouldn't we expect getStats().average to be maximally 5000 if maxTimeout: 5000?

Locally, I've set maxTimeout: 5000 but my average is still climbing. I'm receiving task_timeout in task_failed listener, but I'm wondering why average is higher.
Another contributor opened an issue saying that maxTimeout may not work #75. Is there something to it?

[TimDaub]

I looked into how maxTimeout is implemented and I think it isn't what I had expected.

• What's fine is that a sampling of maxTimeout calls failedBatch peridodically:

  better-queue/lib/queue.js

  Line 604 in 83dca16

  timeout = setTimeout(function () {

• But in failed batch, it seems it just clears out anyways failed tasks?

  better-queue/lib/worker.js

  Line 119 in 83dca16

  Worker.prototype.failedBatch = function (msg) {

• My expectation was that it'd remove all on-going tasks that are older than maxTimeout, but I cannot see how it'd do that. I guess it just clears out all tasks every 5 seconds? I'm confused.

[TimDaub]

e.g. a task has an eta object and so it'd be pretty cool if that was used to filter out all tasks where eta > maxTimeout and then to compensate from niquist-shannon sampling theorem (https://en.wikipedia.org/wiki/Nyquist%E2%80%93Shannon_sampling_theorem) it'd probably be better to call setTimeout every maxTimeout/2 to make sure to catch all false requests frequently.

[leanderlee]

Do you have code that can reproduce this error?

[TimDaub]

I'll create it and get back to you...