- Just like other schedulers such as an OS scheduler, Kubernetes scheduler maintains several different queues: activeQ, backoffQ and unscheduableQ. Why? For better efficiency.
- The activeQ and backoffQ are both priority queues, implemented as a heap. For activeQ, it
- When does a pod go to backoff queue and when does it go to unschedule queue?
- The state transitioning of a pod among these three queues?
- Does it make sense to have more queues?