containersresourcesrequestscpu

In Kubernetes, containers declare resource requirements for CPU as part of their resource specifications. The cpu value under requests represents the amount of compute resource that the container is guaranteed by the scheduler. This helps the system place pods on nodes where sufficient resources are available.

CPU units are defined in cores. Values can be expressed as integers or decimal fractions, commonly using milli-cores like 100m for 0.1 CPU. For example, a request of 250m means the container is guaranteed 0.25 CPU. The corresponding limits.cpu value, when set, defines the maximum CPU the container is allowed to use. If a limit is not set, the container can potentially use more than its request if excess CPU is available on the node, but there is no enforced upper bound unless a limit is specified.

Requests influence scheduling, as the Kubernetes scheduler uses them to determine on which node a Pod can be placed. Limits control runtime consumption and can prevent a single container from overusing CPU and starving others. The combination of requests and limits determines the Pod’s QoS class: Guaranteed (requests equal to limits for all resources), Burstable (requests less than limits for at least one resource), or BestEffort (no requests or limits).

Typical usage patterns involve setting reasonable CPU requests to reflect baseline needs, and configuring limits to cap maximum usage under contention. Inadequate requests can lead to unschedulable pods, while overly high requests can underutilize cluster capacity.

Example (plain text):

containers:

- name: app

image: example/app

resources:

requests:

cpu: "120m"

limits:

cpu: "250m"

Observability tools and commands (such as kubectl describe pod and metrics servers) help verify that CPU requests and usage align with expectations.

---