Kubernetes 上的 CPU 和 GPU 资源配额

Question

Team， 我无法理解二手与硬设置的限制。为什么使用部分显示requests.nvidia.com/gpu: "640"我设置 requests.nvidia.com/gpu 的位置："1"？请求相同.cpu 是 1 但显示 80？

apiVersion: v1
kind: ResourceQuota
metadata:
creationTimestamp: "2019-09-01T21:38:31Z"
name: mem-cpu-demo
namespace: maglev-ee630119-711f-487e-975f-c297fd6a7577
resourceVersion: "23545076"
selfLink: /api/v1/namespaces/7577/resourcequotas/mem-cpu-demo
uid: 94a6
spec:
hard:
limits.cpu: "2"
limits.memory: 2Gi
limits.nvidia.com/gpu: "1"
requests.cpu: "1"
requests.memory: 1Gi
requests.nvidia.com/gpu: "1"
status:
hard:
limits.cpu: "2"
limits.memory: 2Gi
limits.nvidia.com/gpu: "1"
requests.cpu: "1"
requests.memory: 1Gi
requests.nvidia.com/gpu: "1"
used:
limits.cpu: "5760"
limits.memory: 34560Gi
requests.cpu: "80"
requests.memory: 80Gi
requests.nvidia.com/gpu: "640"

Answer 1

资源配额是在创建新对象时强制执行的，它对现有 Pod 没有影响，它不会尝试粉碎已经运行的 Pod。 此外，消耗的CPU单位是毫核1千个核心。