TOPIC CONTAINERS
Resource Requests & Limits sharing a node politely
IN 10 SECONDS
Requests are the minimum CPU/Memory guaranteed to a container. Limits are the absolute maximum resources a container is permitted to consume before getting throttled or killed.
GOTCHA If you don't configure requests, the scheduler assumes zero resources are needed, packing too many containers onto a single machine.
HOW RESOURCES CONFLICTS OCCUR
01 Request spec pod specifies request: 1 CPU, 2GB RAM. Scheduler guarantees this capacity.
02 Spike activity app consumes more CPU, crossing requests but staying under limits.
03 Memory Leak container memory usage hits the hard limits boundary.
04 OOM Kill orchestrator kills the container instantly with OOMKilled to protect node stability.
POKE IT YOURSELF
kubectl top pod — view real-time CPU/Memory usage of active pods
kubectl describe node NODE_NAME — see resource allocations and allocations percentage
Drill this topic →
~65 sec read