开启掘金成长之旅！这是我参与「掘金日新计划 · 12 月更文挑战」的第20天，点击查看活动详情

内存资源限制实践

1.设置容器的内存请求和限制

1.创建一个拥有一个容器的pod。容器将会请求100MiB内存，并且内存会被限制在200MiB以内。

[root@k8s-master /k8s/deploy]# vim memory-requests.yaml
apiVersion: v1
kind: Pod
metadata:
  name: memory-requests
spec:
  containers:
  - name: memory-requests-container
    image: polinux/stress
    command: ["stress"]
    args: ["--vm","1","--vm-bytes","150M","--vm-hang","1"]
    resources:
      requests:
        memory: "100Mi"
      limits:
        memory: "200Mi"
[root@k8s-master /k8s/deploy]# kubectl apply -f memory-requests.yaml
pod/memory-requests created

2.检查Pod的信息

[root@k8s-master /k8s/deploy]# kubectl get pod memory-requests
NAME              READY   STATUS    RESTARTS   AGE
memory-requests   1/1     Running   0          50s
[root@k8s-master /k8s/deploy]# kubectl get pod memory-requests -o yaml
......
spec:
  containers:
  - args:
    - --vm
    - "1"
    - --vm-bytes
    - 150M
    - --vm-hang
    - "1"
    command:
    - stress
    image: polinux/stress
    imagePullPolicy: Always
    name: memory-requests-container
    resources:
      limits:
        memory: 200Mi
      requests:
        memory: 100Mi
......
显示该Pod中的容器的内存请求为100MiB,内存限制为200MiB

3.获取该Pod的指标数据

[root@k8s-master /k8s/deploy]# kubectl top pod
NAME             CPU(cores)         MEMORY(bytes)
memory-requests  46m                150Mi
输出结果显示Pod正在使用的内存约为150MiB.大于Pod请求的100MiB,但又在Pod限制的200MiB之内

4.删除Pod

[root@k8s-master /k8s/deploy]# kubectl delete pod memory-requests
pod "memory-requests" deleted

2.运行超过容器内存限制的应用

当节点拥有足够的可用内存时，容器可以使用其请求的内存。但是，容器不允许使用超过其限制的内存。如果容器分配的内存超过其限制，该容器会成为被终止的候选容器。如果容器继续消耗超出其限制的内存，则终止容器。如果终止的容器可以被重启，则kubelet会重新启动它。

1.创建一个Pod,其拥有一个容器，该容器的内存请求为100MiB,内存限制为200MiB,尝试分配超出其限制的内存
[root@k8s-master /k8s/deploy]# vim memory-requests.yaml

apiVersion: v1
kind: Pod
metadata:
  name: memory-requests
spec:
  containers:
  - name: memory-requests-container
    image: polinux/stress
    command: ["stress"]
    args: ["--vm","1","--vm-bytes","250M","--vm-hang","1"]
    resources:
      requests:
        memory: "100Mi"
      limits:
        memory: "200Mi"
[root@k8s-master /k8s/deploy]# kubectl apply -f  memory-requests.yaml
pod/memory-requests created

2.查看Pod，此时容器可能正在运行或被杀死。重复前面的命令，直到容器被杀掉
[root@k8s-master /k8s/deploy]# kubectl get pod memory-requests
NAME              READY   STATUS             RESTARTS      AGE
memory-requests   0/1     CrashLoopBackOff   2 (18s ago)   84s

3.查看容器的详细信息
[root@k8s-master /k8s/deploy]# kubectl describe pod memory-requests
.....
        State:          Waiting
      Reason:       CrashLoopBackOff
    Last State:     Terminated
      Reason:       OOMKilled
.....

4.删除pod
[root@k8s-master ~]# kubectl delete pod memory-requests
pod "memory-requests" deleted

3.超过节点的内存分配

Pod的调度基于请求。只有当节点拥有足够满足Pod内存请求的内存时，才会被Pod调度至节点上运行。

1.创建一个Pod，其拥有一个请求1000GiB内存的容器，这应该超过了集群中任何一台节点所拥有的内存。
[root@k8s-master /k8s/deploy]# vim memory-requests.yaml
apiVersion: v1
kind: Pod
metadata:
  name: memory-requests
spec:
  containers:
  - name: memory-requests-container
    image: polinux/stress
    command: ["stress"]
    args: ["--vm","1","--vm-bytes","250M","--vm-hang","1"]
    resources:
      requests:
        memory: "100Gi"
      limits:
        memory: "200Gi"


2.查看Pod状态，发现处于Penging 状态。这就意外着，该pod没有被调度至任何节点上运行
[root@k8s-master /k8s/deploy]# kubectl get pod memory-requests
NAME              READY   STATUS    RESTARTS   AGE
memory-requests   0/1     Pending   0          4m56s

3.查看pod详情，输出结果显示，由于节点内存不足，该容器无法被调度
[root@k8s-master /k8s/deploy]# kubectl describe pod memory-requests
......
Events:
  Type     Reason            Age    From               Message
  ----     ------            ----   ----               -------
  Warning  FailedScheduling  6m13s  default-scheduler  0/4 nodes are available: 1 node(s) had taint {node-role.kubernetes.io/master: }, that the pod didn't tolerate, 3Insufficient memory.
  Warning  FailedScheduling  4m51s  default-scheduler  0/4 nodes are available: 1 node(s) had taint {node-role.kubernetes.io/master: }, that the pod didn't tolerate, 3Insufficient memory.
  
4.删除pod
[root@k8s-master /k8s/deploy]# kubectl delete -f memory-requests.yaml
pod "memory-requests" deleted

4.如果没有指定内存限制

如果没有为容器指定内存限制，容器可无限制地使用其所在节点的所有可用内存，进而可能导致该节点调用OOM Killer。此外，如果发生OOM Kill,没有配置资源限制的容器将被杀掉的可行性更大。

在kubernetes中，可用通过LimitRange自动为其容器设定，所使用的内存资源最大最小值

kubernetes-内存资源限制实践