Can "Pod not yet started" K8s events be mapped to queued state instead of running state?

[ketozhang]

When autoscaling with K8s, I have a few minutes of lag between the node appearing as a K8s resources and when the node is ready. For many minutes, this produces logs reporting "Pod not yet started" which likely comes from the K8s event API.

[2024-06-11, 14:12:16 PDT] {pod_manager.py:378} WARNING - Pod not yet started: $SOME_POD_NAME
...
[2024-06-11, 14:18:12 PDT] {pod_manager.py:378} WARNING - Pod not yet started: $SOME_POD_NAME

In the gantt view, I can see the time interval of the above logs coincides with the running state instead of the queued state.

In the docs, the states are defined as

queued: The task has been assigned to an Executor and is awaiting a worker
running: The task is running on a worker (or on a local/synchronous executor)

...and I'm not sure where "task has been assigned but worker is not ready" falls.

---

Are there any configs that can modify this behavior or am I missing some other mechanic?