- use in-cache Pod instead of real-time Pod (by calling API server) to mark it as unschedulable in internal schedulingQ - remove the backoff logic as now we don't call API server - the whole logic is changed to a synchronous call
53 KiB
53 KiB