optimizing infrastructure scaling config

we've been hitting weird bottlenecks during peak traffic, especially rapid scale-up events. our kube-scheduler pods often get stuck in a Pending state, impacting overall infrastructure scaling. it looks like this:

Events:\n  Type     Reason            Age   From               Message\n  ----     ------            ----  ----               -------\n  Warning  FailedScheduling  3s    default-scheduler  0/10 nodes are available: 10 Insufficient cpu.\n

we've tried tuning priority_class and resource_limits in our manifests, but it's not quite enough. any advanced kube-scheduler tuning tips for high-throughput scaling? anyone faced this before?