resource request
The amount of CPU and memory a container says it needs, which the scheduler reserves on a node before placing the pod.
In more detail
The scheduler only places a pod on a node whose allocatable capacity minus existing requests can fit it, even if the node is idle. Requests do not cap usage; limits do. Autoscaling percentages and quotas are also measured against requests.
Example
resources:
requests:
cpu: 100m
memory: 128MiRelated terms
- HorizontalPodAutoscaler — The Kubernetes object that raises or lowers a workload's replica count automatically to keep a metric, such as CPU use, near a target.
- kube-scheduler — The part of Kubernetes that decides which machine each new pod runs on.
- resource limit — A cap on how much memory, processor time or number of processes a container may use.
- millicpu — A thousandth of one CPU, written with an m suffix, so 100m means one tenth of a processor and 500m means half.
- allocatable capacity — The share of a node's CPU and memory available to pods after space is reserved for the operating system and Kubernetes itself.
Taught in
- Generators Write The YAML, You Decide — Certified Kubernetes Administrator (CKA) Exam Prep
- Contexts, Namespaces And One-Line Answers — Certified Kubernetes Administrator (CKA) Exam Prep
- Self-Healing And Autoscaling: Who Restarts, Who Replaces, Who Adds — Certified Kubernetes Administrator (CKA) Exam Prep
- Monitoring With The Built-in CLI — Certified Kubernetes Application Developer (CKAD) Exam Prep
- Observability: Timed Drill And Checkpoint — Certified Kubernetes Application Developer (CKAD) Exam Prep
- Requests, Limits And Quotas — Certified Kubernetes Application Developer (CKAD) Exam Prep
- Configuration And Security: Timed Drill And Checkpoint — Certified Kubernetes Application Developer (CKAD) Exam Prep
- Requests, Limits, Quotas And Horizontal Autoscaling — Kubernetes: Production-Grade Container Orchestration
- Package The Service As A Helm Chart — Kubernetes: Production-Grade Container Orchestration