Cannot use top-level queues directly

[ron-arbo-hpe]

Hi, I am playing around with queues and have noticed my workloads that reference top level queues are not able to be scheduled.

I see the following events in the pod:

Events:
  Type     Reason             Age                    From           Message
  ----     ------             ----                   ----           -------
  Warning  PodGrouperWarning  12m                    pod-grouper    Pod "qa-build-preemptible" not found
  Warning  Unschedulable      2m11s (x599 over 12m)  kai-scheduler  Unable to schedule pod

I have checked all the logs. The only concerning thing I see is an error in the pod-grouper pod saying "failed to find top owner" and "pod ... not found", like in the events above. The pod definitely exists (these are the events from the pod itself, after all), so I am not sure why it is not found.

The quickstart guide tells us to reference the child queue of the default-parent-queue in our workloads, so is this a known limitation?

If so, why is that the case? I don't see any documentation about it, maybe I am missing something.
Also, a more informative pod event would be very helpful here.

Thanks!

[enoodle]

Hello @ron-arbo-hpe 👋

It is true that kai-scheduler is currently implemented this way, but this was not a design decision but an implementation detail that stuck. Improving it could be a great way to contribute to the project.

[srujanreddya]

@enoodle Let me peak in and work on it?