Kubernetes resources

Kubernetes resource configuration (requests and limits) is critical for cluster scheduling and runtime behavior, yet many teams set values arbitrarily, leading to issues like Pending pods, OOMKilled restarts, CPU throttling, and evictions. Requests determine pod placement by the scheduler while limits cap actual consumption; misconfiguring either causes the cluster to make decisions based on false assumptions. A common production failure occurs when memory requests/limits are set far below actual application usage, causing node overcommitment and container kills under load.